The qacG gene on plasmid pST94 confers resistance to quaternary ammonium compounds in staphylococci isolated from the food industry


E. Heir, National Institute of Public Health, PO Box 4404 Torshov, N-0403 Oslo, Norway (e-mail:


The 2·3 kb resistance plasmid pST94 revealed a new gene (qacG) encoding resistance to benzalkonium chloride (BC), a commonly used quaternary ammonium disinfectant, and the intercalating dye ethidium bromide (Eb) in staphylococci isolated from the food industry. The 107 amino acid QacG protein showing 69·2% identity to the staphylococcal multi-drug resistance protein Smr is a new member of the small multi-drug resistance (SMR) protein family. QacG conferred resistance via proton dependent efflux. An additional ORF on pST94 encoded a protein with extensive similarity to replication proteins of other Gram-positive bacteria. Gene constructs containing the qacG and smr gene region combined with the smr or qacG promoter, respectively, indicated that QacG is more efficient than Smr and that qacG has a weaker promoter. Resistant qacG-ontaining cells could be adapted to withstand higher concentrations of BC. Adapted qacG-containing cells showed increased resistance mainly to BC. In contrast, adaptation of sensitive cells showed cross-resistance development to a range of compounds. Induction of proton-dependent efflux was observed for BC-adapted staphylococci cells not containing qacG. The ability of sublethal concentrations of BC to develop cross-resistance and induce efflux mechanisms could be of practical significance; it should be considered before use of any new disinfectant and in the design of better disinfection procedures.


Resistance to antiseptics and disinfectants based on quaternary ammonium compounds (QACs) is widely distributed among clinical isolates of staphylococci. Recent reports have also shown that QAC resistance is widespread among staphylo- cocci isolated from certain areas of the food industry ( Sundheim et al. 1992 ; Heir et al. 1995 ). So far, three determinants responsible for staphylococcal resistance to QAC have been identified and characterized ( Lyon & Skurray 1987; Tennent et al. 1989 ; Rouch et al. 1990 ; Littlejohn et al. 1992 ). The products encoded by these determinants (qacA, qacB and smr) belong to two families of membrane transport proteins, the small multi-drug resistance family (SMR) and the major facilitator superfamily (MFS).

Members of the SMR family can be divided into two distinct clusters of small integral transmembrane proteins ( Paulsen et al. 1996a ). One cluster consists of the four resistance proteins Smr ( Grinius et al. 1992 ), QacE, QacE ( Paulsen et al. 1993 ) and EmrE ( Yerushalmi et al. 1995 ; also designated EBR ( Purewal 1991) and MvrC ( Morimyo et al. 1992 )). The best characterized determinant in this family is the Staphylococcus multi-drug resistance gene smr ( Grinius et al. 1992 ) also known as qacC, qacD ( Littlejohn et al. 1991 ) and ebr ( Sasatsu et al. 1989 ). A highly similar gene, qacC, originating from a Staphylococcus sp. strain isolated from the food industry, has been sequenced and shown to differ by only one nucleotide ( Heir et al. 1995 ). For simplicity, these resistance genes are termed smr in this report. Recent experiments have confirmed that at least two proteins of the SMR family, Smr ( Littlejohn et al. 1992 ) and EmrE ( Yerushalmi et al. 1995 ), support drug transport mediated by the proton motive force. The smr gene is located on both conjugative and non-conjugative plasmids in both Staph. aureus and coagulase-negative staphylococci ( Littlejohn et al. 1991 ; Leelaporn et al. 1994 ). The smr gene encodes a protein of 107 amino acids.

The MFS includes the closely related multi-drug resistance determinants qacA and qacB. The qacA and qacB genes differ by only seven nucleotides ( Paulsen et al. 1996a ). Phenotypically, both determinants confer resistance to monovalent organic cations, while qacB has lower or no resistance to divalent cations (e.g. chlorhexidine and pentamidine) compared with qacA.

In the food industry, the production of heat-resistant enterotoxin by staphylococci is of serious concern. Enterotoxin production has been associated with Staph. aureus. More recently, enterotoxin production has also been demonstrated with some CNS ( Hirooka et al. 1988 ; Hoover et al. 1983 ; Bautista et al. 1988 ; Valle et al. 1990 ; Nanu & Narayan 1992; Vernozy et al. 1996 ). Coagulase-negative staphylococci have also been recognized as increasingly important causative organisms of infections associated with implanted foreign bodies and in patients with immuno-compromised host defences. In addition, CNS are regarded as possible genetic reservoirs of resistance determinants ( Jaffe et al. 1982 ; Pfaller & Herwaldt 1988).

Disinfectants based on QACs are widely used in the hospital environment and the food industry. It has been proposed that the occurrence of resistance to antiseptics and disinfectants is caused by the extensive use of these compounds in certain environments ( Stickler & King 1992; Langsrud & Sundheim 1997).

In a previous report ( Heir et al. 1995 ), it was documented that 25 of 191 staphylococci isolates were found to have increased MIC values to BC. Of these, five isolates did not hybridize to gene-specific probes of the previously known QAC resistance determinants qacA/B and smr. Here, we report the cloning, sequence and properties of the Staphylococcus plasmid pST94 harbouring a new resistance gene, qacG, conferring resistance to QAC and ethidium bromide (Eb). In addition, we have studied the development of resistance and cross-resistance when staphylococci are exposed to sublethal concentrations of BC.

Materials and methods

Bacterial strains, plasmids and growth conditions

The bacterial strains and plasmids used in this work are listed in Table 1. Unless otherwise stated, staphylococci strains were cultured in Mueller-Hinton broth (MHB; Difco) at 37 °C with shaking. Wild type strains were determined to be staphylococci as described previously ( Heir et al. 1995 ). Escherichia coli DH5α was grown in Brain Heart Infusion broth (BHI; Difco). When appropriate, growth media were supplemented with selective antibiotic concentrations, 5 mg l−1 chloramphenicol (Staph.) and 100 mg l−1 ampicillin (E. coli).

Table 1.  Bacterial strains and plasmids
Strain or plasmidRelevant characteristics *Reference or source
  • *

    QAC, quarternary ammonium compounds; BC, benzalkonium chloride; Ap, ampicillin; Cm, chloramphenicol.

Staphylococcus spp.
ST94Wild type strain, pST94; QACrThis study
ST94cIsogenic to ST94 but cured for plasmid pST94This study
ST94cAST94c adapted to growth in BCThis study
ST93Wild type strain; QACrThis study
ST51Wild type strain; QACrThis study
Staphylococcus aureus
RN4220Restriction deficient, plasmid-free host strainKreiswirth et al. (1983)
RN4220ARN4220 adapted to growth in BCThis study
RN4220 (pQG1)ARN4220 containing pQG1 and adapted to growth in BC; QACrThis study
Escherichia coli DH5αHost strain for pGEM7Zf (+) Gibco BRL
pST942·267 kb; contains qacGThis study
pSK2653·0 kb cloning vector; pC194 derivative with multiple cloning sites from pUC19; CmrJones & Khan (1986)
pGEM7Zf (+) 3·0 kb cloning vector; AprlacZPromega
pQC30·65 kb smr promoter and gene in pSK265This study
pQG10·61 kb qacG promoter and gene cloned into pSK265This study
pQCG10·65 kb smr promoter with qacG gene in pSK265This study
pQGC0·61 kb qacG promoter with smr gene in pSK265This study

Plasmid curing

Staphylococcus ST94 was grown in MHB containing 2 mg l−1 novobiocin and sub-cultured every 2 d in increasing concentrations of novobiocin. A sample from the last culture was diluted and spread onto MH agar containing 0·5 mg l−1 Eb. Colonies were inspected for fluorescence under u.v. light after 24 h. Red fluorescing colonies accumulating Eb and white non-accumulating colonies were analysed for plasmid content and resistance to BC.

Plasmid DNA isolation and molecular techniques

Plasmids were generally isolated and purified as described previously ( Heir et al. 1995 ). The resistance plasmid pST94 was isolated from agarose gels using low-melting-point agarose (seaPlaque agarose; FMC Bioproducts, Pockland, ME, USA). The agarose containing the plasmid was melted at 65 °C and purified using the Wizard Clean-up system (Promega, Madison, WI, USA). Standard procedures for molecular cloning were used according to the directions of Sambrook et al. (1989) . Overlapping fragments of HindIII and NsiI were cloned into the plasmid cloning vector pGEM7Zf(+) and electroporated into E. coli DH5α by the procedure of Hanahan et al. (1991) . Escherichia coli recombinant plasmids were isolated and purified with the Wizard Minipreps DNA purification system. DNA sequencing was performed by a primer walking strategy using synthetic oligonucleotide primers and the Sequenase version 2·0 sequencing kit (United States Biochemical Corporation, Cleveland, OH, USA). Staphylococcus aureus RN4220 was used as the host of pSK265 and transformed as described by Schenk & Laddaga (1992). The computer analyses were performed using the DNASIS sequence analyses program package (Hitachi, Japan) and Genetics Computer Group package (version 8·0; University of Wisconsin, Madison, WI, USA). In Southern blotting experiments, plasmid DNA was separated on agarose gels and transferred to Hybond N+ membranes (Amersham International, Amersham, UK) by vacuum blotting before hybridization to a qacG specific oligonucleotide (pST94 nucleotide-number 1804–1820) probe as previously described ( Heir et al. 1995 ).

RNA isolation and primer extension

RNA was isolated by the method of Igo & Losick (1986), and A260 and A280 readings were taken to determine the quality and concentration of RNA. Primer extension was performed by the method of Alam et al. (1986) with the following modifications: 5–6 μg of total RNA were annealed with about 1 pmol of pre-labelled 32P primer 5′-TTTGTGAAACCTTCTGA-3′ in a total volume of 11 μl (by cooling from 65 to 30 °C within 15–20 min), then 5 × avian myeloblastosis virus reverse transcriptase buffer (Promega), deoxynucleoside triphosphates (each at a final concentration of 50 μmol l−1) and reverse transcriptase (5 U; Promega) were added to the mixture (to a final volume of 20 μl). The mixture was incubated at 42 °C for 60 min for polymerization. The reaction was stopped with 10 μl of sequenase stop buffer and analysed on a sequencing gel.

Promoter-gene studies

The pQC3 and pQG1 clones containing the smr and qacG gene and promoters, respectively, were constructed by standard PCR techniques using flanking primers with BamHI and EcoRI restriction sites (underlined; localization of primers on the pST94 or pST827 ( Heir et al. 1995 ) plasmids in parenthesis). Primers for the construction of pQC3 were qacCG-1F: 5′-ATCGGATCCCTTAGTTAACATTTCT AAAAC-3′ (pST827 nucleotides 231–251, sense) and qacCG-1D:



(pST827 nucleotides 856–873, antisense) while the primers for construction of pQG1 were qacCG-1 A:


(pST94 nucleotides 1257–1276, sense) and qacCG-1E:



(pST94 nucleotides 1837–1855, antisense). The hybrid clones containing the smr gene preceded by the qacG promoter and vice versa were designated pQCG1 and pQGC, respectively. A ‘gene SOEing’ (Splicing by Overlap Extension) PCR-technique ( Horton & Pease 1991) was used to synthesize the constructs. For construction of the pQCG1 clone, the flanking primers qacCG-1F and qacCG-1E, and the overlapping primers qacCG-1G:



and qacCG1H:



were used. The pQGC1 clone was constructed using the flanking primers qacCG1 A and qacCG1D together with the overlapping primers qacCG1B:



and qacCG1C:



All PCR constructs were digested with EcoRI and BamHI restriction enzymes and purified using the Wizard Clean-up kit (Promega) before ligation into the pSK265 vector and transformation into Staph. aureus RN4220. Both strands of all smr/qacG inserts were sequenced and found to be identical to the originating host DNA.

Adaptation studies

Strains RN4220 and RN4220(pQG1) were serially subcultured to progressively higher, but always subminimal, inhibitory concentrations of BC to give the adapted strains RN4220 A and RN4220QG1 A, respectively. The strains were grown in 5 ml MHB at 32 °C with moderate shaking. The increase in concentration of BC used during serial subculturing was strain-dependent and determined empirically.

Minimal inhibitory concentrations

Minimal inhibitory concentrations (MICs) of wild-type staphylococci and Staph. aureus RN4220 transformants were determined in a microtitre assay as previously described ( Sundheim et al. 1992 ) to a range of antimicrobial agents: BC (QAC), Eb, proflavine, rhodamine 6G (dyes), methyl viologen (paraquat), tetraphenyl phosphonium chloride (TPP) and tetracycline hydrochloride. The lowest concentration of antimicrobial agent totally preventing growth after 24 h was taken to be the MIC. All MIC tests were done in at least triplicate.

Suspension test

The bactericidal effect of disinfectant was measured using the Council of Europe (1987) suspension test for antimicrobial activity of disinfectants. The test was performed as described by Langsrud & Sundheim (1997) with the following modifications. The non-adapted strains RN4220 and RN4220 (pQG1) were grown with shaking at 32 °C. For the corresponding adapted strain RN44220 (pQG1)A, 10 μg ml−1bc was included in the growth medium before incubation. After overnight growth (16–20 h), 50 μl of the cell suspension were inoculated in 5 ml MHB without BC and incubation was continued (4·5–6 h) until a cell density of approximately 5 × 1011–1 × 1012 cfu l−1 was reached. The number of cfu was determined using colony counts on Mueller-Hinton agar after incubation for 24 h at 37 °C. The significance of the bactericidal effect of BC on sensitive and resistant (qacG-harbouring) Staph. aureus RN4220, and unadapted and adapted resistant (qacG-harbouring) Staph. aureus RN4220, respectively, was calculated by the Two sample t-test procedure in MINITAB (MINITAB for Windows release 9, MINITAB Inc., PA, USA).

Efflux studies

The fluorescence technique used by Jones & Midgley (1985) and Midgley (1986) was adapted to measure the efflux of Eb. In short, an overnight (16 h) culture in MHB was used as inoculum to 200 ml MHB and grown for 90–120 min to an O.D.600 of 0·3. Aliquots of 10 ml were harvested by centrifugation (3000 g) and washed twice in 20 mmol l−1 Hepes buffer (pH 7·0). Loading of the cells was done by resuspending and incubating the cells in 10 ml 20 mmol l−1 Hepes (pH 7·0) containing Eb (5 μmol l−1) in the presence or absence of (20 μmol l−1) carbonyl cyanide m-chlorophenylhydrazone (CCCP) at 37 °C with agitation for 30 min. The cell suspension was chilled on ice and the cells washed once in 20 mmol l−1 Hepes (pH 7·0) containing Eb (5 μmol l−1). Washed cells were resuspended in 1 ml ice cold 20 mmol l−1 Hepes (pH 7·0) containing Eb (5 μmol l−1) and stored on ice until efflux measurement. Efflux of Eb was initiated by addition of glucose (10 mmol l−1). All fluorescence measurements were performed at 37 °C with a Perkin-Elmer luminescence spectrophotometer LS 50B (The Perkin-Elmer Corporation, CT, USA). Fluorescence was integrated every 1 s with excitation and emission wavelengths of 520 and 590 nm, respectively.

Nucleotide accession number

The sequence reported has been assigned the EMBL accession number Y16944.


Identification and nucleotide sequence of pST94

Curing experiments with the QAC-resistant Staph. ST94 resulted in a sub-population of Eb-accumulating (Eba) colonies lacking a 2–3 kb plasmid designated pST94. The Eba isolates also showed a typical BC-sensitive phenotype. This putative QAC resistance plasmid was isolated, cloned and sequenced ( Fig. 1). Computer analyses of the pST94 nucleotide sequence revealed the presence of two open reading frames (ORFs) transcribed in opposite directions. The smaller ORF, designated qacG, encoded a putative protein of 107 amino acid residues (QacG). QacG showed 69·2% identity to the Smr protein ( Sasatsu et al. 1989 ; Littlejohn et al. 1991 ; Grinius et al. 1992 ).

Figure 1.

Nucleotide sequence of pST94 and deduced amino acid sequence. Putative −35 and −10 promoter regions and ribosomal binding sites are underlined. The transcriptional start site (tss) of the putative qacG gene is indicated. Imperfect inverted repeat sequences are represented by horizontal arrows below the nucleotide sequence. A stretch of a conserved part of a putative plus ori sequence is shown in bold

The larger ORF, designated rep94, encoded a putative replication protein. Both ORFs (rep94 and qacG) were associated with typical initiation codons (ATG) preceded by candidate ribosomal binding sites (RBS′) and RNA polymerase binding sites. The qacG transcription start point, located 28 bp upstream of the ATG translation start codon, was determined by primer extension analyses. The results supported the putative −35 and −10 regions, although these were difficult to identify because of low homology to consensus promoter sequences ( Graves & Rabinowitz 1986). Regions of dyad symmetry (inverted repeats) with the potential to serve as transcriptional terminators were found downstream of both the qacG and rep94 stop codon.

Comparison of QacG and Smr. Minimal inhibitory concentrations and ethidium bromide efflux experiments

In addition to Staph. ST94, two other QAC-resistant strains (Staph. ST93 and Staph. ST51) isolated from the food industry ( Heir et al. 1995 ) hybridized to a qacG-specific probe. To compare the qacG-harbouring strains with the strains containing smr, the MIC values to BC and Eb were determined. The qacG- and smr-harbouring isolates showed small differences in MIC values to BC (8–10 mg l−1) and Eb (20–40 mg l−1). The variation observed could be due to differences in the promoter and coding sequences. To examine this, Staph. aureus RN4220 (pQC3) and RN4220 (pQG1) clones containing the smr and qacG gene, respectively, and clones harbouring hybrid smr/qacG constructs containing the smr promoter region preceding the qacG gene (pQCG1) and vice versa (pQGC), were constructed ( Table 2). MIC values of Staph. aureus RN4220 cells containing pQCG1 and pQGC varied considerably compared with the clones containing pQG1 and pQC3.

Table 2.  Comparison of minimal inhibitory concentration (MIC) to benzalkonium chloride (BC) and ethidium bromide (Eb) for Staphylococcus aureus RN4220 clones containing different qacG/smr constructs
MIC (mg l−1)
  • *

    Negative control.


All constructed clones were compared for their ability to support efflux of Eb ( Fig. 2). During loading of cells, accumulation of Eb was significantly reduced in Staph. aureus RN4220 cells harbouring pQC3 or pQG1 compared with control cells (Staph. aureus RN4220 (pSK265)). Addition of glucose (10 mmol l−1) activated similar efflux in pQC3- and pQG1-containing cells, while no activation of efflux was observed for the control cells. RN4220 (pQCG1) also showed decreased accumulation of Eb during loading compared with pQC3- and pQG1-harbouring cells. Like RN4220 (pQC3) and RN4220 (pQG1) cells, pQCG1-harbouring cells initiated efflux by the addition of glucose. For the RN4220 (pQGC) cells, only a slight efflux was activated by the addition of glucose. In the presence of CCCP, all cells accumulated Eb to the same level as control cells (not shown).

Figure 2.

Fluorimetric comparison of the efflux activity in Staphylococcus aureus RN4220 harbouring different qacG/smr constructs. Curve A: pSK265, negative control; B: pQGC; C and F: pQC3; D and G: pQG1; E and H: pQCG1. All cells were loaded with ethidium in the absence of CCCP. At time 0 s the following additions were made: A, B, F, G, H, glucose (10 mmol l−1); C, D, E, glucose (10 mmol l−1), CCCP (20 μmol l−1)

Adaptation and cross-resistance

To determine whether the presence of sublethal concentrations of BC could select for resistance to BC and a range of other representative compounds, the QAC-resistant Staph. ST94, Staph. aureus RN4220 (pQG1) and the isogenic, QAC-sensitive control strains Staph. ST94c (plasmid cured) and Staph. aureus RN4220 (pSK265), were adapted to growth at higher concentrations of BC. The results showed that the response to adaptation was strain-dependent ( Table 3). Adapted strains harbouring the qacG resistance gene mainly showed increased resistance to the adaptation drug BC only, while adaptation of the two sensitive staphylococci strains (ST94c and RN4220 (pSK265)) led to elevated MIC values to the majority of compounds tested. In addition, the MICs (representative MIC values in parentheses) to the drugs methyl viologen (500–3000 mg l−1), rhodamine 6G (0·5–1 mg l−1) and tetracycline hydrochloride (2–4 mg l−1), were determined. The effect of adaptation, observed by MIC values of these drugs, was strain-dependent. For rhodamine 6G and tetracycline hydrochloride, only a slight increase in MIC, or no increase at all, was observed for the adapted strains while for methylviologen, no differences, or a slight decrease in MIC, were observed for the adapted strains relative to the non-adapted strains (data not shown). The adapted strain Staph. ST94cA was further studied to determine whether the observed elevated MIC values could be due to the initiation of an energy-linked efflux system. The fluorimetric assay ( Fig. 3) showed that ST94cA cells conferred efflux of Eb. Efflux was, however, inhibited in the presence of the protonophore CCCP. This indicates that the efflux is energized by the proton motive force. Efflux of Eb, activated by glucose and inhibited by CCCP, was also observed for Staph. aureus RN4220 (pSK265) cells adapted to growth in BC (results not shown).

Table 3.  Minimum inhibitory concentrations (MIC) of tetraphenylphosphonium chloride (TPP), proflavine, benzalkonium chloride (BC) and ethidium bromide (Eb) to adapted and non-adapted staphylococci
MIC (mg l−1)
StrainResistance-markerN *ANANANA
  • *

    N, non-adapted; A, adapted.

Staph. aureus RN4220 (pQG1)qacG3030252510158080
Staph. aureus RN4220 (pSK265)307025100281050
Staph. ST94qacG152515158142020
Staph. ST94c603002550<2715200
Figure 3.

Fluorimetric assay of ethidium efflux in adapted Staphylococcus ST94cA. Cells were loaded with ethidium in the presence of CCCP (20 μmol l−1; curve A) and absence of CCCP (curves B and C). At time 0 s, the following were added: A and B, glucose (10 mmol l−1) and CCCP (20 μmol l−1); C: glucose (10 mmol l−1)

Suspension test

Figure 4 shows reduction in cell counts at different concentrations of BC for resistant Staph. aureus RN4220 harbouring qacG compared with sensitive Staph. aureus RN4220 (pSK265) negative control cells and qacG harbouring Staph. aureus RN4220 adapted to growth in BC. Cells harbouring the resistance gene qacG (Staph. aureus RN4220 (pQG1)) were more resistant than cells not harbouring the resistance gene qacG (Staph. aureus RN4220 (pSK265) (P= 0·03). Adapted RN4220 (pQG1) cells showed increased resistance compared with unadapted RN4220 (pQG1) cells (P= 0·03).

Figure 4.

Mean logarithmic reduction in cell numbers after exposure to benzalkonium chloride (BC) plotted against BC concentration for Staphylococcus aureus RN4220 (pSK265) negative control (▴), Staph. aureus RN4220 harbouring qacG (clone pQG1; •) and Staph. aureus RN4220 (pQG1)A (○)


Nucleotide sequence analysis of the 2·3 kb plasmid pST94 revealed an open reading frame (qacG) encoding a putative protein of 107 amino acids. The QacG protein with its 69·2% identity to the Smr protein, and 45% and 41% identity to QacE ( Paulsen et al. 1993 ) and EmrE ( Yerushalmi et al. 1995 ), respectively, is a new member of the SMR family of multi-drug export proteins ( Fig. 5). QacG differed from Smr in 33 of 107 amino acid positions dispersed throughout the protein. The similarity in location of hydrophobic amino acids in QacG and Smr also supports the theory that QacG is folded into the membrane in an Smr-like manner ( Grinius & Goldberg 1994; Paulsen et al. 1995 ). Two regions were totally conserved between the two proteins. These included regions of amino acid numbers 38–47 and 52–71. These two regions are localized to the C-terminal part of transmembrane segment 2 and to the N-terminal part (including some cytoplasmic residues) of transmembrane segment 3 of the Smr structural models proposed.

Figure 5.

Multiple alignment of QacG to efflux protein members of the SMR family.Conserved residues are indicated. Fully conserved residues are indicated below the alignment

Multiple amino acid sequence alignments have shown a number of conserved residues among members of the SMR family ( Paulsen et al. 1996b ). In the QacG protein, the majority of these residues was also found to be conserved. However, there were two exceptions: a substitution from Ala9 to Ser9, and a substitution from Ser32 to Thr32. Approximately half the amino acid substitutions in QacG (relative to Smr) could be regarded as conservative involving Leu-Ile, Val-Ile and Phe-Ile hydrophobic amino acid substitutions. Overall, the 33 amino acid changes did not alter the resistance profile of QacG relative to Smr. Both proteins encoded similar resistance to BC but apparently differed significantly in sensitivity to the intercalating dye, Eb. No resistance was observed to the other representative drugs tested (TPP, rhodamine 6G, proflavine, tetracycline hydrochloride and methyl viologen). We cannot conclude, however, that specific substitution of any of these amino acids has no effect on the resistance profile as a concomitant change of two or more amino acids may reverse the effect observed by substitution of a single amino acid. The QacG and/or Smr protein could, of course, differ in resistance to compounds not tested.

Previous reports ( Grinius & Goldberg 1994; Paulsen et al. 1995 ) have identified a number of key residues essential for Smr protein structure and function. The conservation of these residues in the QacG protein indicates that these residues play a critical structural or functional role and supports the structural and mechanistic model proposed. Identification of a new resistance protein of the SMR family may be of value for identifying and determining the role of other amino acids associated with membrane structures and drug efflux.

Two other staphylococci isolates from the food industry, Staph. ST93 and Staph. ST51, hybridized to an oligonucleotide probe designed to be specific to qacG. Two of the isolates (Staph. ST93 and Staph. ST94) had a similar plasmid profile, only differing in an additional large plasmid present in Staph. ST94, while Staph. ST51 had a distinct plasmid profile. The three isolates all showed essentially similar MIC values to BC and Eb. Staphylococcus ST93 and Staph. ST94 were isolated at the same food processing plant, while Staph. ST51 was from a separate plant. This indicates that qacG may be widespread in the food processing industry.

Only small differences in MIC values were evident for strains harbouring either the qacG or the smr resistance determinant when tested against BC (MICs of 10 and 8 mg l−1, respectively) or Eb (MICs of 80 mg l−1). These results were consistent with efflux experiments showing that qacG and smr harbouring Staph. aureus clones had a very similar efflux activity of Eb. Efflux was inhibited by the membrane protonophore CCCP, while the addition of glucose activated efflux. This suggests that the QacG protein uses the same resistance mechanism as the Smr protein, i.e. active efflux of antimicrobial compounds energized by the proton motive force. The hybrid construct Staph. aureus RN4220 (pQCG1) showed MIC values to BC and Eb of 10 and 180 mg l−1, respectively. The MIC values were consistent with the observed reduced accumulation of Eb during loading by the efflux system. Similarly, the MIC value to Eb (20 mg l−1) of the RN4220 (pQGC) cells was in accordance with the reduced efflux observed in the Eb efflux assay. However, only a slightly lower MIC value to BC was observed compared with cells containing other qacG/smr constructs examined (7 mg l−1 and 10 mg l−1, respectively). At present, the reason for the large difference in MIC values to Eb observed for the RN4220 (pQGC) cells (20 mg l−1) and RN4220 (pQCG1) cells (180 mg l−1), while MIC values to BC for these clones are only slightly different, remains unclear. Provided that the mRNA and QacG/Smr protein stability is comparable, in addition to a comparable QacG/Smr protein level in the cell, these results indicate that the qacG gene product confers higher resistance than the smr gene product, and that the smr promoter is more efficient than the qacG promoter.

Littlejohn et al. (1991) discovered significant homology between a region just downstream of qacC and the palA region of other plasmids. Similar sequences have been found in all characterized smr plasmids, including the pST94 plasmid (nucleotide number 1866–2024); palA has been shown to be required for lagging strand replication of pC194 ( Novick 1989) and in addition, palA, with its palindromic sequences, may be a possible transcription terminator.

Comparison of the pST94 sequence with the GenEmbl databank sequences revealed significant homology to a number of small plasmids. pST94 is highly similar to the Staph. epidermidis smr-harbouring plasmid pSK108 ( Leelaporn et al. 1995 ) including the rep gene, flanking sequences of the rep gene and the origin nick site. However, less similarity (except in the palA regions, see below) is observed in the fragments containing the qac determinants, including the promoter regions and the nucleotide sequence 150 bp downstream of palA. The extensive similarity of these two plasmids, except in the above-mentioned region, suggests that qac (qacG and smr) cassettes (including the qac determinant, palA sequence and a region 3′ to palA) are inserted close to the origin nick site of rolling circle plasmids. The organization of other small resistance plasmids replicating by a rolling circle mechanism supports this hypothesis. This is also consistent with the typical ‘cassette-like’ structure of the pC194 family of plasmids ( Novick 1989). Rep94 was found to be most similar to the Rep protein encoded by Orf334 on pSK108 (94·0% identity; Leelaporn et al. 1995 ) and to the replication protein of the Staph. aureus multi-drug resistance (smr) plasmid pKH8 ( Im et al. 1996 ). In addition, significant homology was also found to replication proteins of other small, putative, rolling circle replicating plasmids from other Gram-positive bacteria. A putative origin of replication was observed upstream of the qacG promoter region. Taken together, these results suggest that pST94 also replicates by a rolling circle mechanism.

By growing resistant (qacG-harbouring) and sensitive staphylococcal strains in media containing increasing concentrations of BC, both categories of bacteria were adapted to tolerate higher concentrations of BC. The qacG-containing strains showed substantial increases in MIC to BC only. In the strains not harbouring qacG, development of cross-resistance to other drugs was observed ( Table 2). A hypothetical explanation for this is that the qacG-harbouring strains already have an active system conferring resistance to QACs, and the signals necessary to induce another resistance mechanism may not be present. The increased MIC values to BC observed for adapted qacG-harbouring strains could be a result of increased expression of the qacG determinant. The strains not harbouring qacG, however, have no active system conferring QAC resistance and to cope with the increased concentrations of BC in the environment, they have to induce new resistance mechanism(s) leading to increased MIC values to a range of antimicrobial compounds. Fluorimetric efflux experiments strongly indicated that the exposure of bacteria to sublethal concentrations of BC could induce other energy-linked efflux mechanisms. Adapted, originally sensitive cells of both cured wild type Staph. ST94cA and Staph. aureus RN4220 (pSK265)A showed activated efflux of Eb in the presence of glucose. Efflux was totally inhibited in the presence of CCCP, indicative of an efflux system energized by the proton motive force. Although these adapted cells conferred efflux of Eb, we have at present no evidence that the resistant phenotype observed is due to efflux of BC or the other antimicrobial compounds tested. Overall, the exposure of sensitive bacteria to sublethal concentrations of BC could induce resistance to compounds not structurally related to BC and therefore, be a factor of cross-resistance development. The results of the susceptibility test showed that Staph. aureus cells have an advantage in expressing qacG for survival in environments containing low concentrations of BC compared with QAC-sensitive control cells. Staphylococcus aureus cells harbouring qacG adapted to growth in BC also showed a significantly increased survival rate compared with non-adapted cells, although a 5 log10 reduction in viable counts was obtained at BC concentrations well below recommended user concentrations. However, survival of staphylococcal strains was obtained when tested against user concentration of a commercial brand of QAC-based disinfectant in the presence of other microbes and organic material ( Sundheim et al. 1992 ). In disinfection practice, adaptation and growth of micro-organisms could occur because sublethal concentrations of QACs may be left on surfaces after disinfection. This could be of practical significance and is to be considered when designing improved protocols for better disinfection procedures, as well as in the development of new disinfectants. Also, possible cross-resistance should be considered when choosing disinfectants for alternation with QAC-based disinfectants used daily.


The authors thank Anita Caspersen and Tove Maugesten for performing the suspension tests.


  • *

    Present address: National Institute of Public Health, PO Box 4404 Torshov, N-0403 Oslo, Norway.