The effect of cationic microbicide exposure against Burkholderia cepacia complex (Bcc); the use of Burkholderia lata strain 383 as a model bacterium

Authors


Abstract

Aim

The extensive use of microbicides in a wide range of applications has been questioned with regard to their role in the development of bacterial resistance to antimicrobials. This study aims to measure the phenotypic and genotypic changes in Burkholderia lata strain 383 exposed to chlorhexidine gluconate (CHG) and benzalkonium chloride (BZC), two commonly used cationic microbicides.

Methods and Results

The susceptibility of B. lata strain 383 to CHG and BZC and a range of antibiotics was determined using standardized MIC, MBC and antibiotic susceptibility testing protocols before and after short-term exposure to a low microbicide concentration. Measurements were performed on four separate occasions over a 1-year period. Changes in gene expression were investigated using quantitative real-time PCR. Although the susceptibility profile to CHG and BZC was not altered, a change in antibiotic susceptibility profile was observed for ceftazidime, and for imipenem and ciprofloxacin in 2/4 repeats. An outer membrane protein and ABC transporter were found to be significantly upregulated following treatment with BZC and CHG, respectively.

Conclusions

The comparison of MIC and MBC results following microbicide exposure with baseline data offered a prospective protocol to quantify any change in bacterial susceptibility profile. However, the use of a standardized antibiotic susceptibility protocol with B. lata strain 383 showed some inconsistencies in results between repeats.

Significance and Impact of the Study

With ever-increasing interest in the impact of microbicides on emerging antimicrobial resistance in bacteria growing, this study demonstrated that comparing susceptibility profile obtained after exposure to microbicides with baseline susceptibility values could play a role in establishing the potential risk of microbicide resistance and cross-resistance development and also in the development of a protocol that allows the prediction of microbicide resistance.

Introduction

The increasing use of microbicides in consumer products is raising concerns related to microbicide-induced resistance in bacteria and potential cross-resistance to antibiotics (Maillard and Denyer 2009; Maillard 2010; Scientific Committee on Consumer Safety, 2010: http://ec.europa.eu/health/scientific_committees/consumer_safety/docs/sccs_o_023.pdf; accessed 9th May 2013; Scientific Committee on Emerging and Newly Identified Health Risks, 2009:

http://ec.europa.eu/health/ph_risk/committees/04_scenihr/docs/scenihr_o_021.pdf, accessed 9th May 2013 and http://ec.europa.eu/health/scientific_committees/emerging/docs/scenihr_o_028.pdf, accessed 9th May 2013). It has also been reported that bacteria exposed to low microbicide concentrations show reduced susceptibility to other antimicrobials including antibiotics due to the development/expression of common resistance mechanisms (Levy 2000; Scientific Committee on Emerging and Newly Identified Health Risks, 2009: http://ec.europa.eu/health/ph_risk/committees/04_scenihr/docs/scenihr_o_021.pdf accessed 9th May 2013; Scientific Committee on Consumer Safety, 2010: http://ec.europa.eu/health/scientific_committees/consumer_safety/docs/sccs_o_023.pdf; accessed 9th May 2013; Maillard 2010). Regular microbicide stress could lead to genetic, biochemical, functional or physiological changes that select for bacteria with greater tolerance to these antimicrobials. Despite the increasing number of reports of microbicide resistance in bacteria, microbicide resistance is poorly characterized in comparison with antibiotic resistance.

Data on microbicide usage are lacking together with an understanding of the microbicides most at risk for the development of bacterial resistance. Such information together with a better characterization of the resistance mechanisms involved is essential for developing strategies to prevent/combat resistance (Scientific Committee on Emerging and Newly Identified Health Risks, 2010: http://ec.europa.eu/health/scientific_committees/emerging/docs/scenihr_o_028.pdf, accessed 9th May 2013). Of a particular interest is the development of a standard protocol that could evaluate the ability of a microbicide to induce/select for microbicide and/or antibiotic resistance.

Here, we explored bacterial phenotypic and genotypic changes (i.e. practical markers) that are most indicative of microbicide and/or antibiotic resistance in Burkholderia lata, as a model Burkholderia cepacia complex (Bcc) species implicated in microbial contamination. Burkholderia lata is a member of the Bcc, a group of at least 17 closely related Gram-negative rods that infect 2–8% of patients with cystic fibrosis (CF) (Lipuma 2010). Antibiotic combination therapy is used to treat early infection, but is often ineffective due to the intrinsic antibiotic resistance present in Bcc members (Aaron et al. 2000). Multiple drug resistance (MDR) in CF isolates is defined as resistance to all agents belonging to at least two of three antibiotic classes (Bazzini et al. 2011). Aside from their role in clinical infectious disease, Bcc along with a number of other bacterial species are contaminants of home and personal care products, with B. cepacia being one of the most frequently isolated bacterial contaminants in pharmaceutical samples around the world (Jimenez 2004), although many of these reports have not accurately determined the Bcc species identity. This impacts on the safety and quality of these products and has resulted in the recall of a large number of cosmetic products. Between 1994 and 1998, B. cepacia contamination was the cause of 33% (19/55) of recalled cosmetic products in the United States (Wong et al. 2000) and 4% (1/24) of recalled contaminated cosmetics in the EU between 2005 and 2008 (Lundov and Zachariae 2008). Recalls of food products due to Bcc species have also been common (Wong et al. 2000).

Microbicides (i.e. disinfectants and preservatives) are all employed to control or destroy micro-organisms either in home and personal care products or at the point of use, but may not always be effective due to the intrinsic resistance to a wide range of antimicrobials that Bcc bacteria possess. Antimicrobial resistance in Burkholderia has frequently been attributed to the presence and expression of efflux pumps of the resistance nodule division (RND) (Bazzini et al. 2011; Biot et al. 2011; Coenye et al. 2011), enzyme modification of the antimicrobial, biofilm formation and cell wall impermeability (Drevinek et al. 2008; George et al. 2009). Furthermore, Bcc species are able to metabolize a wide range of carbon sources that makes their elimination from contaminated pharmaceutical and laboratory products difficult (Torbeck et al. 2011).

In this study, the cationic microbicides chlorhexidine gluconate (CHG) and benzalkonium chloride (BZC) were used as model microbicides as they have a large number of applications. In addition to use in home and personal care products, they are common components of pharmaceutical formulations and appear at different concentrations depending on the application of the product. Examples of these products include anticavity rinse, surgical prep cloth and mouthwash (Torbeck et al. 2011). In addition, CHG and BZC have been implicated in the selection of low-level antibiotic resistance (Mayer et al. 2001; Beier et al. 2011; Lavilla Lerma et al. 2013). The overall aim of this study was to explore the change in susceptibility in B. lata strain 383 when exposed to cationic microbicides using a new efficacy test approach based on measuring both phenotypic and genotypic changes.

Materials and methods

Bacterial isolates and culture conditions

Burkholderia lata strain 383 was obtained from the Cardiff University collection (Mahenthiralingam et al. 2008). Liquid cultures of B. lata were grown in basal salts medium (BSM) modified from that of Hareland et al. (1975) by removal of the nitriloacetic acid. BSM broth comprized the following: di-potassium hydrogen orthophosphate trihydrate (85 g l−1), sodium di-hydrogen orthophosphate monohydrate (20 g l−1), ammonium chloride (40 g l−1), magnesium sulfate heptahydrate (20 g l−1), ferrous sulfate heptahydrate(1·2 g l−1), manganese sulfate monohydrate (0·3 g l−1), zinc sulfate heptahydrate (0·3 g l−1), cobalt sulfate heptahydrate (0·1 g l−1), CAS amino acids (50 g l−1), yeast extract (50 g l−1) and glucose (4 g l−1). Cultures were grown at 30°C with shaking at 150 rev min−1 for 18 h. Bacterial stocks were stored in BSM containing 8% v/v dimethyl sulfoxide (Sigma, Dorset, UK) at −80°C. Test inocula were prepared from harvesting from an overnight BSM culture centrifuged at 5000 g for 10 min and resuspended in tryptone sodium chloride buffer [TSC; 0·4 g tryptone (Oxoid, Basingstoke, UK), 3·4 g sodium chloride (Fisher, Loughborough, UK), 400 ml deionized water].

Microbicides and neutralizer

Chlorhexidine gluconate (CHG) and benzalkonium chloride (BZC) (Sigma) were used and prepared freshly on the day in deionized water. The neutralizer used was composed of Tween 80 (30 g l−1; Fisher Scientific, Loughborough, UK), azolectin (3 g l−1; Sigma) and deionised water. Microbicide and neutralizer were sterilized by autoclaving at 121°C for 15 min under 3 psi.

Neutralizer toxicity and efficacy tests

The toxicity of the neutralizer was tested as follows: a bacterial suspension in deionized water was produced from an overnight culture (as described above: test inocula) and standardized to 1 × 10CFU ml−1. One ml of this suspension was added to 9 ml of neutralizer. The suspension was vortex-mixed and left for 5 min. A control experiment was performed alongside this where 1 ml of bacterial suspension was added to 9 ml of deionized water. Viable counts were performed on test and control suspensions using the drop count method (Miles and Misra 1938). Test and control counts were compared to determine whether exposure to neutralizer caused any significant decrease in CFU ml−1. The neutralizer was considered toxic if ≥1 log10 decrease was observed in the test colony count.

The ability of the neutralizer to quench the activity of chlorhexidine gluconate and benzalkonium chloride was tested as follows: One ml of the biocide at the highest concentration used (CHX – 5%, BZC – 12·5%) was added to 8 ml of neutralizer and vortex-mixed. After 5 min, 1 ml of a bacterial suspension containing 1 × 108 CFU ml−1 was added and vortex-mixed. A control experiment was performed alongside this using 8 ml sterile distilled water (SDW) instead of neutralizer. Viable counts of both control and test suspensions were performed using the drop count method. The neutralizer was considered effective if ≤1 log10 reduction was observed in the neutralized biocide suspension.

Determination of minimum inhibitory concentration (MIC)

The MIC of CHG and BZC was determined for strain 383 following the BS EN ISO: 20776-1 (2006) broth dilution protocol. CHG was tested at concentrations of 0·01–0·000019% w/v, and BZC was tested at concentrations of 0·025–0·00001% w/v in BSM. In brief, a microtitre plate containing doubling dilutions of CHG and BZC in broth was set up. An overnight broth culture of strain 383 (test inocula) was standardized to 1 × 10CFU ml−1, and 50 μl volumes of this were added to the microtitre plate. The plate was incubated in a Bioscreen Microbial Growth Analyser (Oy Growth Curves AB Ltd, Helsinki, Finland) for 24 h at 30°C with shaking at 150 rev min−1. The MIC was defined as the lowest microbicide concentration at which no bacterial growth was observed visually on the microtitre plate.

Determination of minimum bactericidal concentration (MBC)

Twenty μl of a bacterial suspension was removed from each well of the MIC microtitre plate where no bacterial growth was observed and at the two lowest microbicide concentrations at which growth was observed and added to 180 μl of neutralizer. Twenty-five μl of this suspension was then spotted on to basal salts agar and incubated at 30°C for 24 h. The minimum bactericidal concentration was defined as the lowest microbicide concentration where no bacterial growth was observed on the agar plate.

Antibiotic susceptibility testing

The susceptibility of B. lata strain 383 to antibiotics that can be used to treat clinical Burkholderia infections (Rose et al. 2009): ciprofloxacin (1 μg), ceftazidime (30 μg), tobramycin (10 μg), imipenem (10 μg) and meropenem (10 μg) was tested following the standardized British Society for Antimicrobial Chemotherapy (BSAC) disc diffusion method for antimicrobial susceptibility testing (Andrews 2009).

Exposure to a low microbicide concentration

Microbicide exposure experiments were based on the British Standard EN 1276 suspension test protocol (2009). Burkholderia lata strain 383 was exposed to 0·005% w/v CHG and BZC (below the MIC) for 5 min at 20 ± 1°C. Briefly, one ml of an overnight suspension (as described above: test inocula) adjusted to 1 × 10CFU ml−1 was added to 9 ml of microbicide dissolved in deionized water. After 5 min exposure at 20°C, 1 ml of this suspension was removed and added to 9 ml of neutralizer. Viable bacteria were enumerated before and after microbicide exposure using the drop count method (Miles and Misra 1938). After microbicide exposure, the neutralized test inoculum was centrifuged at 5000 g for 10 min and the remaining pellet resuspended in TSC buffer. The MIC and MBC of CHG and BZC were then determined for surviving bacteria as described above. Susceptibility to a selection of clinically relevant antibiotics was also determined as described above.

Phenotype stability testing

The stability of any changes observed in antibiotic susceptibility after 5 min microbicide exposure was determined by continuous broth subculture of surviving bacteria in microbicide-free BSM and BSM supplemented with 0·005% w/v CHG or BZC. Subcultures were performed every 24 h, and antibiotic susceptibility was determined after 1, 5 and 10 passages. A check of culture purity was performed at each stage.

Harvesting bacterial cells

For quantitative real-time PCR, work flasks containing 25 ml of BSM were inoculated with 2 × 10CFU of strain 383 and incubated at 30°C with shaking at 150 rev min−1. In the case of microbicide-exposed bacteria, flasks were supplemented with 0·005% w/v CHG or BZC (a sub-MIC microbicide concentration). Growth was monitored spectrophotometrically. Cultures were harvested in microcentrifuge tubes at mid-exponential phase (optical density of 0·5 at 600 nm) and snap-cooled in liquid nitrogen. Samples were then centrifuged at 20 000 g at 4°C for 1 min, before removal of the supernatant, and the pellet was stored at −80°C. Three separate starting cultures were used to obtain three biological replicates.

RNA purification and preparation for real-time PCR

Total RNA was extracted from bacterial cells using a RiboPure Bacteria Kit (Ambion, Applied Biosystems, Austin, TX, USA) according to the manufacturer's instructions. In brief, this involved cell disruption using zirconia beads and a phenol-based detergent, followed by the use of chloroform to separate the aqueous and organic phases during centrifugation. RNA was then filtered, eluted and treated with DNase for 30 min to remove any genomic DNA. RNA was quantified using a NanoVue Plus spectrophotometer (GE Healthcare, Little Chalfon, UK). cDNA was synthesized from RNA using an Improm-II Reverse Transcription System (Promega, Southampton, UK) according to the manufacturer's instructions.

Quantitative real-time PCR

Real-time PCRs were carried out to identify changes in the expression of specific genes after exposure to a range of CHG and BZC concentrations. The genes selected had previously been identified to be upregulated in B. cenocepacia J2315 or B. lata after exposure to various antimicrobial agents (Table 1). To enable amplification across the range of Bcc species, PCR primers were designed as follows: The sequence of each gene from representative Burkholderia genomes (B. ambifaria, B. cenocepacia, B. cepacia, B. multivorans, B. mallei, B. phymatum, B. pseudomallei and B. lata) was obtained from the Burkholderia Genomes Database (www.burkholderia.com) and aligned using molecular evolutionary analysis software (MEGA 5; http://www.megasoftware.net). Primers were designed to selected regions of homology using the Primer 3 software (http://bioinfo.ut.ee/primer3-0.4.0) and checked to their specificity for each Burkholderia gene target using the In Silico PCR tool (http://insilico.ehu.es/PCR). Primer sequences for real-time PCR are shown in Table 2. The metabolism associated gene, phaC (BCAL1861), was selected as a reference control gene because (i) its expression in B. cenocepacia had been shown to remain stable over a wide range of growth conditions (Sass et al. 2013) and (ii) it had been successfully used in the validation of B. lata global gene expression in relation to preservative exposure (Rushton et al. 2013). Real-time PCRs were performed on an Opticon 2 Real Time PCR system (MJ Research, Waltham, MA, USA) using an Absolute QPCR SYBR Green Mix according to the manufacturer's instructions. Cycling conditions were as follows: 95°C for 15 min (1 cycle), 95°C for 15 s, 30 s at the appropriate primer annealing temperature and 72°C for 30 s (50 cycles). A melting curve analysis was included at the end of each run. The reference gene and control samples without cDNA were also included in each experiment. The comparison of gene expression fold change obtained was assessed by Pearson correlation.

Table 1. Genes investigated using real-time PCR
Gene namePutative functionResponse observed by global gene expression analysis using microarrays or gene mutagenesis studies
Burkholderia cenocepacia antimicrobial resistance target genes
BCAM0925Outer membrane protein that is part of an RND efflux pump12-fold upregulated after exposure to 0·05 mmol l−1 chlorpromazine and mutation of the gene results in increased chlorhexidine susceptibility (Sass et al. 2011); 8-fold upregulated after exposure to chlorhexidine (H. Rose and E. Mahenthiralingam, unpublished data)
BCAS0081ABC transporterUpregulated in an antibiotic-resistant clinical clone of B. cenocepacia J2315 and mutation of the gene results in an increased chlorhexidine susceptibility (Sass et al. 2011); 6·8-fold upregulated after chlorhexidine exposure (H. Rose and E. Mahenthiralingam, unpublished data)
BCAM2551Multidrug efflux transport protein CeoA296-fold upregulated in a derivative adapted to trimethoprim sulfamethoxazole (Sass et al. 2011)
BCAS0167Squalene-hopene cyclaseUpregulated 4·36-fold after exposure to 0·05 m mol l−1 chlorpromazine and mutation of the gene results in increased chlorhexidine susceptibility (Sass et al. 2011)
BCAL1663PrkA family serine protein kinaseUpregulated in stationary phase (8-fold), low oxygen (8-fold), and heat stress (2-fold; Sass et al. 2013)
B. lata 383 antimicrobial target gene
Bcep18194_B1327MFS_1 transporter4·7-fold upregulated in derivative adapted to blend of methylisothiazolinone/chloromethylisothiazolinone (M-CMIT) preservative (L. Rushton and E. Mahenthiralingam, unpublished data)
Table 2. Primer sequences for genes investigated using real-time PCR
Primer namePrimer sequence
  1. a

    The phaC forward primer used was as described (Rushton et al. 2013).

phaC Fa (reference gene)5′-AAGCGTTCGACAAGGTCAAG-3′
phaC R (reference gene)5′-GTTCACCGACGAGATGTTGA-3′
BCAM0925 F5′-CTGGCGCACGATGTTC-3′
BCAM0925 R5′-ATGCCGTACTGCGCTTC-3′
BCAS0081 F5′-TTCGACGGGCTGAACCT-3′
BCAS0081 R5′-GCAGCAGCGAGGTATCCT-3′
BCAM2551 F5′-TCGGTGTCGCCGATCTAC-3′
BCAM2551 R5′-TCGACGACGAACACGAACT-3′
BCAS0167 F5′-CCTGATGATGCATTTCATGGAC-3′
BCAS0167 R5′-ACGCGACCTTGTACATCGAG-3′
BCAL1663 F5′-GTTCAAGGCGCCGATCA-3′
BCAL1663 R5′-TCGTTGTTGCGGTTGTTG-3′
Bcep18194_B1327 F5′-GAGGTGGAGATGACCGAATC-3′
Bcep18194_B1327 R5′-GAGGTGGAGATGACCGAATC-3′

Reproducibility of study and statistical tests

Changes in antimicrobial susceptibility profile following exposure to CHG or BZC were performed on four occasions over a 1-year period. Each experiment consisted of three repeats. For each experiment, the same frozen stock liquid culture was used. The test inocula were rigorously prepared as described above on each occasion. Antibiotic zone of inhibition sizes were compared using a Student's t-test (P ≤ 0·05). A one-way anova test was used to compare the zone of inhibition sizes obtained throughout the passage work (P ≤ 0·05).

Results

MIC, MBC and toxicity tests

Baseline MIC and MBC measurements for both microbicides were repeated every 8 weeks to determine the reproducibility of the baseline data for strain 383. The baseline MICs for strain 383 of CHG and BZC recorded after nine separate experiments were 0·070 ± 0·020 and 0·050 ± 0·050% w/v, respectively. The baseline MBCs for B. lata strain 383 to CHG and BZC recorded after nine separate experiments were 0·400 ± 0·600 and 0·050 ± 0·050% w/v, respectively.

The neutralizer toxicity results showed that the neutralizer was not toxic to strain 383 (data not shown). The effect of short-term exposure to a low concentration of CHG or BZC on microbicide and antibiotic susceptibility of strain 383 was investigated and compared to baseline susceptibility data. A 5-min exposure of strain 383–0·005% w/v CHG or BZC resulted in no statistically significant (P ≤ 0·05) change in MIC and MBC for both microbicides (data not shown).

Short-term exposure of Burkholderia lata strain 383 to 0·005 % w/v CHG and BZC and alterations in susceptibility to selected antibiotics

To assess whether short-term exposure (5 min) to either 0·005% w/v CHG or BZC resulted in alterations in antibiotic susceptibility, suspension test experiments were carried out on four occasions over a 1-year period. On each occasion, three repeats were used. Surviving bacteria from the microbicide exposure after neutralization were assessed for any change in their MIC, MBC and antibiotic susceptibility profile. Following exposure to CHG, a significant (P ≤ 0·05) decrease in the zone of inhibition size was observed in two of the four experimental repeats for ceftazidime (Table 3). Significant reductions (P ≤ 0·05) in the zone of inhibition size were also observed for ciprofloxacin and imipenem on two occasions. Following BZC exposure, a significant (P ≤ 0·05) decrease in the zone of inhibition was observed for ceftazidime in 1/4 occasions, for ciprofloxacin and meropenem on two occasions and imipenem on three occasions (Table 3).There is no clinical interpretation of the antibiotic zone of inhibition for Burkholderia spp. If one considers the clinical interpretation for Pseudomonas aeruginosa (Andrews 2009), the statistically significant reduction in zone of inhibition observed with imipenem and ciprofloxacin would have reached clinical significance on only one occasion (Andrews 2009). No change in CHG or BZG MIC was observed in relation to this altered antibiotic resistance. The surviving B. lata derivatives with altered antibiotic susceptibility were stored and analysed for stability as follows.

Table 3. Assessment of antibiotic susceptibility in bacteria surviving CHG and BZC exposure
Mean zone of inhibition size (mm)a
Antibiotic Separate repeatsbBaseline0·005% CHG (1)0·005% CHG (2)0·005% CHG (3)0·005% CHG (4)0·005% BZC (1)0·005% BZC (2)0·005% BZC (3)0·005% BZC (4)
  1. Bold: statistically significant change from baseline value P ≤ 0·05.

  2. a

    Mean zone of inhibition sizes (mm) of a range of clinically relevant antibiotics after 5 min exposure to low CHG and BZC concentrations.

  3. b

    Number in bracket represents the separate repeats performed over a 1-year period. Each repeat was performed in triplicate.

Ciprofloxacin (1 μg)30·0 ± 0·0 11·3 ± 1·2 24·0 ± 0·0 20·0 ± 2·0 29·0 ± 1·5 12·0 ± 1·5 24·0 ± 1·0 28·5 ± 2·529·0 ± 2·6
Tobramycin (10 μg)7·30 ± 1·19·00 ± 1·04·30 ± 1·1 11·6 ± 0·9 8·0 ± 0·08·00 ± 2·60·00 ± 0·0 8·60 ± 0·8 8·00 ± 0·0
Ceftazidime (30 μg)40·3 ± 0·5 33·3 ± 1·8 30·0 ± 2·0 30·3 ± 3·934·5 ± 3·439·0 ± 4·436·0 ± 1·930·0 ± 5·0 32·5 ± 1·2
Imipenem (10 μg)24·0 ± 0·0 15·0 ± 3·0 27·3 ± 3·725·0 ± 0·0 21·0 ± 1·1 16·0 ± 5·0 30·0 ± 4·0 29·0 ± 3·1 24·0 ± 1·0
Meropenem (15 μg)40·7 ± 1·237·0 ± 1·035·3 ± 3·433·0 ± 2·839·0 ± 4·040·7 ± 2·5 35·5   ± 0·5 34·0 ± 1·8 39·2 ± 3·5

Stability of the observed altered phenotype of microbicide-exposed bacteria

Clinical changes in susceptibility (i.e. a change from sensitive to resistant) to imipenem and ciprofloxacin were observed after microbicide exposure. Phenotype stability testing experiments therefore focused on these two antibiotics. The stability of the changes in antibiotic susceptibility observed after 5 min of microbicide exposure was assessed via the passaging of surviving bacteria through BSM with and without microbicide supplementation. For imipenem, there was no significant difference between the baseline zone of inhibition value and passage value (P ≤ 0·05) after just one passage through microbicide-free broth (Table 4). For ciprofloxacin, there was no significant difference between baseline and passage values (P < 0·05) after one passage in the absence of 0·005% CHG and after five passages in the absence of 0·005% BZC (Table 4). These results indicate that the observed decrease in antibiotic susceptibility in microbicide-treated bacteria was transient. Passaging surviving bacteria through BSM containing 0·005% w/v CHG resulted in a maintained significant difference (P ≤ 0·05) between baseline and passage values for ciprofloxacin and imipenem after 10 passages (Table 5) although there was an upward trend in the data suggesting a return towards baseline. This suggests that the decreased antibiotic susceptibility was maintained in the presence of this biocide following at least 10 passages. In the case of 0·005% w/v BZC, reduced imipenem susceptibility was lost after five passages with no significant difference between baseline and passage values (P ≤ 0·05). A significant difference between baseline and passage values (P < 0·05) for ciprofloxacin was maintained after 10 passages (Table 5). This suggests that reduced ciprofloxacin susceptibility was maintained in the presence of 0·005% w/v BZC.

Table 4. Stability of antibiotic susceptibility changes after growth passage in the absence of microbicides
Mean zone of inhibition size (mm)a
AntibioticBaselineInitial 5 min CHG exposure1 passageb (no CHG)5 passagesb (no CHG)10 passagesb (no CHG)Initial 5 min BZC exposure1 passageb (no BZC)5 passagesb (no BZC)10 passagesb (no BZC)
  1. Bold: statistically significant difference between baseline and passage value P ≤ 0·05.

  2. a

    Mean zone of inhibition sizes for the antibiotics imipenem and ciprofloxacin after 1, 5 and 10 passages through microbicide-free BSM (N = 3).

  3. b

    Growth passage refers to 10 subcultures performed every 24 h in the absence of CHG or BZC.

Imipenem24·0 ± 0·0 15·0 ± 3·0 24·0 ± 1·021·6 ± 2·124·3 ± 1·2 16·0 ± 5·0 24·0 ± 1·524·0 ± 4·025·3 ± 3·1
Ciprofloxacin30·0 ± 0·0 11·3 ± 1·2 28·0 ± 1·031·0 ± 1·030·0 ± 4·4 12·0 ± 1·5 26·0 ± 1·1 29·0 ± 3·537·3 ± 4·0
Table 5. Stability of antibiotic susceptibility after growth passages in the presence of microbicides
Mean zone of inhibition size (mm)a
AntibioticBaselineInitial 5 min CHG exposure1 passageb (CHG)5 passagesb (CHG)10 passagesb (CHG)Initial 5 min BZC exposure1 passageb (BZC)5 passagesb (BZC)10 passagesb (BZC)
  1. Bold: statistically significant difference between baseline and passage value P ≤ 0·05.

  2. a

    Mean zone of inhibition sizes for the antibiotics imipenem and ciprofloxacin after 1, 5 and 10 passages through BSM containing 0·005% CH.

  3. b

    Passage refers to 10 subcultures performed every 24 h in the presence of 0·005% CHG or BZC.

Imipenem24·0 ± 0·0 15·0 ± 3·0 14·3 ± 0·6 18·6 ± 0·2 21·6 ± 0·6 16·0 ± 5·0 13·6 ± 0·5 17·0 ± 3·525·6 ± 1·2
Ciprofloxacin30·0 ± 0·0 11·3 ± 1·2 7·60 ± 1·2 11·3 ± 1·2 11·6 ± 0·6 12·0 ± 1·5 7·30 ± 1·5 9·00 ± 1·0 13·0 ± 3·6

Exposure of Burkholderia lata to 0·005% CHG and BZC and resulting changes in gene expression

The expression of specific genes was observed in nonmicrobicide-exposed bacteria and microbicide-exposed bacteria with a view to identifying marker genes for microbicide resistance. Greater than 2-fold changes in the expression of a range of genes were observed after growing B. lata strain 383 to mid-log phase (c. 6 h) in the presence of 0·005% w/v CHG or BZC. The greatest fold changes in expression were observed in an outer membrane protein (BCAM0925; 43-fold upregulation in presence of 0·005% w/v BZC) and in an ABC transporter (BCAS0081; 102-fold upregulation in the presence of 0·005% w/v CHG). The RND efflux pump gene BCAM2551 was upregulated 2·5-fold and 4-fold, when strain 383 was grown in the presence of 0·005% w/v CHG and BZC, respectively. The PrkA serine protein family kinase gene BCAL1663 also increased in expression, 14·7-fold in the presence of CHG and 3·7-fold in the presence of BZC. There were no significant changes in gene expression observed when strain 383 was subjected to the short-term microbicide exposure of 5 min (data not shown).

Discussion

Burkholderia lata strain 383 was exposed to 0·005% w/v CHG and BZC, and the susceptibility of surviving bacteria to these microbicides and a range of antibiotics was compared to baseline susceptibility data, with the aim of understanding the effect of short-term microbicide exposure and to assess the reproducibility of the data over time. This study did not show a change in microbicide susceptibility profile of B. lata strain 383 when exposed to low concentrations of CHG or BZC. Thomas et al. (2000) observed stable chlorhexidine diacetate resistance in Ps. aeruginosa after a single 24-h exposure to 1 μgl−1 and speculated that exposure length may affect which mechanisms the bacterial cell uses to counteract the effect of the antimicrobial.

This study demonstrated a change in antibiotic susceptibility profile for ceftazidime, imipenem and ciprofloxacin in particular. For ciprofloxacin and imipenem, this change was not observed in all four independent repeats. These results highlight the need to provide an indication of data reproducibility in time as well as the stability of susceptibility profile observed. Here, the BSAC methodology (Andrews 2009) was strictly followed, and one could question the reproducibility of zone of inhibition measurements. Schuurmans et al. (2009) found that antibiotic MIC values could vary by a factor of up to eight if small alterations were made in the method used to determine these values. This again reiterates the need for data reproducibility and a standard protocol for susceptibility measurement. Phenotypic variability has also been observed for Burkholderia cepacia when single colonies have been picked to perform the assay (Larsen et al. 1993). Here, this phenotypic variability could be ruled out with our test inoculum preparation that was based on broth culture and not on picking single colonies.

Although the change in zone of inhibition following biocide exposure was statistically significant for some of the antibiotics tested, the clinical significance is more difficult to ascertain. There are no breakpoints available for Burkholderia cepacia complex. If one uses breakpoints available for Ps. aeruginosa (Andrews 2009), then clinical resistance for ciprofloxacin and imipenem would have been observed on only one occasion. In addition, the reduced antibiotic susceptibility observed was not lost after passaging the bacteria in BSM supplemented with 0·005% w/v CHG or BZC. Considering the mechanisms of action of imipenem and ciprofloxacin; inhibition of bacterial cell wall synthesis via the binding of penicillin binding proteins, preventing peptidoglycan formation for penem antibiotics (Sawasdidoln et al. 2010) and inhibition of DNA synthesis via the inhibition of DNA gyrase that unwinds double-stranded DNA for ciprofloxacin (Lunn et al. 2010), it is likely that the mechanisms conferring the change in susceptibility observed in this study were nonspecific. There have been a number of studies looking at changes in antibiotic susceptibility profile following microbicide exposure, but only one was conducted with Burkholderia spp. Rose et al. (2009) investigated microbicide and antibiotic susceptibility of 12 species of the Bcc complex, including B. lata strain 383 and reported no correlation between chlorhexidine and benzalkonium chloride susceptibility and susceptibility to clinically used antibiotics. They did not however investigate the direct effect of microbicide exposure on antibiotic susceptibility, but looked at antibiotic susceptibility of wild-type Bcc strains with high intrinsic microbicide MICs (CHG > 100 mgl−1 and BZC > 400 mgl−1). Alteration of antibiotic susceptibility with no change in microbicide susceptibility following microbicide exposure has been reported, notably with triclosan. Christensen et al. (2011) reported that exposure of Listeria monocytogenes to sublethal concentrations of triclosan resulted in a 16-fold decrease in gentamicin susceptibility, despite no change in susceptibility to triclosan. Gentamicin-resistant organisms were also resistant to other aminoglycosides. However, other studies using this biocide observed no changes in susceptibility profile for selected antibiotics. For example, Birošová and Mikulásová (2009) investigated the effect of a 30-min short-term exposure of S. enterica to sub-MIC concentrations of triclosan on the selection of antibiotic-resistant strains and found that after plating surviving bacteria on to agar containing triclosan or an antibiotic at 2× the MIC, neither microbicide nor antibiotic susceptibility was altered. Whitehead et al. (2011) observed that a single 5-h exposure of S. enterica serovar Typhimurium to cationic microbicides (including a QAC and ‘Superkill’ – a mix of QACs) or triclosan at the in-use concentration (1%) selected for multiple drug-resistant (MDR) survivors. This MDR phenotype included resistance to drugs from different antibiotic classes. These studies did not compare the change in susceptibility profile observed with baseline susceptibility data; as such the reproducibility of their observations cannot be ascertained.

The stability of a change in a susceptibility profile is also important to consider as it has a practical application and raises questions about the selection effect of a given microbicide. In our study, the decreased zones of inhibition observed for ciprofloxacin did not revert to the baseline measurements after ten passages of the bacteria in 0·005% w/v CHG or BZC. However, the zone of inhibition size did return to baseline in the absence of microbicides. For imipenem, reduced susceptibility was maintained in the presence of CHG, but lost after five passages in the presence of BZC (Table 5).

In this study, the greatest changes in gene expression after bacterial exposure to CHG or BZC were observed in genes BCAM0925 and BCAS0081. These encode an outer membrane protein and an ABC transporter, respectively, both of which could have roles in the efflux of antimicrobial compounds from the bacterial cell (Buroni et al. 2009). Efflux has been described as an important nonspecific mechanism that can reduce the intracellular concentration of unrelated antimicrobials (Maillard and Denyer 2009; Davin-Regli and Pages 2012), and as such is a likely candidate to explain changes in a susceptibility profile. Recent work carried out by Bazzini et al. (2011) looked at the effect of deleting the RND-4 and RND-9 efflux systems in B. cenocepacia. They found that a double mutant had a 4–16-fold increase in susceptibility to several antibiotics tested, one of which was ciprofloxacin. This suggests that these putative efflux pumps play a role in antimicrobial susceptibility, and this may also be the case in what has been observed here with strain 383 after microbicide exposure. It also correlates well with the fact that changes in susceptibility to more than one antibiotic were observed. Sass et al. (2011) looked at global gene expression for B. cenocepacia in relation to spontaneous antibiotic resistance and exposure to the cationic antibiotic potentiator chlorpromazine. They observed upregulation of BCAM0925 in response to exposure to chlorpromazine and found that its deletion resulted in increased susceptibility to azithromycin and chlorhexidine. The BCAS0081 protein is composed of an ATP-binding cassette and transmembrane components and has homology to the Escherichia coli mdlB gene, which has been associated with multiple drug resistance. However, Sass et al. (2011) found that deletion of the BCAS0081 gene resulted in increased susceptibility to tetracycline and chlorhexidine, but not to other antibiotics tested including imipenem or ciprofloxacin. It remains that B. cenocepacia J2315 is very different in terms of resistance profile to B. lata strain 383 and comparisons between them cannot be easily made. Although, in correlation with our findings, Rushton et al. (2013) recently demonstrated that upregulation of efflux plays a major role in the preservative resistance of B. lata, and that exposure to a blend of methylisothiazolinone–chloromethylisothiazolinone also led to stable, elevated fluoroquinolone resistance.

In this study, gene expression was observed across all three biological repeats after the bacteria were grown to mid-log phase (c. 6 h) in the presence of 0·005% w/v CHG or BZC. When changes in the expression of BCAM0925 and BCAS0081 were observed after 5-min microbicide exposure, no significant difference in expression was observed. Only a small range of genes that could have played a role in antibiotic susceptibility of B. lata strain 383 after microbicide exposure were tested, and a wider range of genes should be investigated at the same time as establishing a genuine effect.

With questions about the effect of widespread use of microbicides on emerging antimicrobial resistance (Scientific Committee on Emerging and Newly Identified Health Risks, 2010: http://ec.europa.eu/health/scientific_committees/emerging/docs/scenihr_o_028.pdf, accessed 9th May 2013), it is essential to be able to predict bacterial behaviour in response to microbicide exposure. It is equally important to ensure the reproducibility of the observations made. The development of an appropriate protocol, including the use of a suitable bacterial model is thus important to consider. In this study, our protocol was based on standard assays for the determination of MIC, MBC and antibiotic susceptibility profile. The lack of reproducible resistance phenotype in B. lata strain 383 using this protocol suggests that this bacterium may not be appropriate to use in the generation of predictive markers of microbicide resistance.

Acknowledgement

Laura Knapp is the recipient of a PhD studentship at the University of Cardiff, funded by, Unilever Research and Development, SEAC.

Conflict of interest

No conflict of interest declared.

Ancillary