Journal of Antimicrobial Chemotherapy Advance Access published October 31, 2013
J Antimicrob Chemother doi:10.1093/jac/dkt420
Inhibition of multidrug efflux as a strategy to prevent biofilm formation Stephanie Baugh, Charlotte R. Phillips, Aruna S. Ekanayaka, Laura J. V. Piddock and Mark A. Webber* School of Immunity and Infection, Institute for Microbiology and Infection, University of Birmingham, Edgbaston, Birmingham B15 2TT, UK *Corresponding author. Tel: +0121-414-2859; Fax: +0121-414-6819; E-mail: [email protected]
Received 25 June 2013; returned 2 August 2013; revised 24 September 2013; accepted 1 October 2013
Methods: Mutants lacking components of the AcrAB-TolC system in Salmonella enterica serovar Typhimurium were investigated for their ability to aggregate, produce biofilm matrix components and form a biofilm. The potential for export of a biofilm-relevant substrate via efflux pumps was investigated and expression of genes that regulate multidrug efflux and production of biofilm matrix components was measured. The ability of efflux inhibitors carbonyl cyanide m-chlorophenylhydrazone, chlorpromazine and phenyl-arginine-b-naphthylamide to prevent biofilm formation by Escherichia coli, Pseudomonas aeruginosa and Staphylococcus aureus under static and flow conditions was assessed. Results: Mutants of Salmonella Typhimurium that lack TolC or AcrB, but surprisingly not AcrA, were compromised in their ability to form biofilms. This defect was not related to changes in cellular hydrophobicity, aggregative ability or export of any biofilm-specific factor. The biofilm defect resulted from transcriptional repression of curli biosynthesis genes and consequent inhibition of production of curli. All three efflux inhibitors significantly reduced biofilm production in both static and flow biofilm assays, although different concentrations of each inhibitor were most active against each species. Conclusions: This work shows that both genetic inactivation and chemical inhibition of efflux pumps results in transcriptional repression of biofilm matrix components and a lack of biofilm formation. Therefore, inhibition of efflux is a promising antibiofilm strategy. Keywords: curli, AcrAB-TolC, efflux inhibitors
Introduction Bacterial biofilms are a major clinical and industrial problem, and eradication of biofilms presents a challenge for antimicrobial chemotherapy.1 – 3 We recently described an inability of Salmonella enterica serovar Typhimurium (hereafter Salmonella Typhimurium) to form a competent biofilm that resulted from inactivation of any of the multidrug resistance (MDR) efflux systems of Salmonella.4 MDR efflux systems have also previously been suggested to be important in the intrinsic antibiotic resistance of biofilms and have been shown to be involved in the ability of various species to form a biofilm: Escherichia coli, Pseudomonas putida, Klebsiella pneumoniae 5 and Pseudomonas aeruginosa.5,6 As well as various bacterial species, efflux has also been shown to be important in biofilm formation of the fungus Candida albicans.7 We found that production of curli, a major component of the Salmonella biofilm extracellular matrix, was defective in all these strains, suggesting a common mechanism for the lack of biofilm formation in all
mutants.4 Additionally, we found that the addition of three inhibitors of efflux with different mechanisms of action also resulted in prevention of biofilm formation and that this was also associated with loss of curli production.4 Here, using AcrAB-TolC as a paradigm MDR efflux system, we investigated the mechanism by which loss of efflux activity results in a lack of curli production. We ruled out export of a factor crucial for biofilm development via AcrAB-TolC and also showed that inactivation of components of AcrAB-TolC did not alter cellular hydrophobicity. However, inactivation of efflux components was found to significantly alter the expression of biofilmrelated genes. We demonstrate that the biofilm defect of mutants lacking a functional AcrB or TolC is due to transcriptional repression of curli biosynthesis. Curli is an essential proteinaceous component of the biofilm matrix for various bacterial species, including Salmonella. The major curli transcriptional regulator, CsgD, controls transcription of csgBA (the operon encoding the two structural
# The Author 2013. Published by Oxford University Press on behalf of the British Society for Antimicrobial Chemotherapy. All rights reserved. For Permissions, please e-mail: [email protected]
1 of 9
Downloaded from http://jac.oxfordjournals.org/ at Georgia State University on October 31, 2013
Objectives: We have recently shown that inactivation of any of the multidrug efflux systems of Salmonella results in loss of the ability to form a competent biofilm. The aim of this study was to determine the mechanism linking multidrug efflux and biofilm formation, and to determine whether inhibition of efflux is a viable antibiofilm strategy.
Baugh et al.
proteins comprising curli), as well as genes responsible for cellulose biosynthesis. csgD is itself controlled by a complex network of both positive and negative regulation.8 Secondly, we determined whether inhibition of efflux is effective against a range of pathogens for which biofilm formation is problematic under flow conditions, as well as in static biofilm assays. All three inhibitors tested were able to reduce biofilm formation by a range of pathogens in both static and flow biofilm assays.
Materials and methods Strains and growth media All strains used in this study and their origins are shown in Table 1. Salmonella Typhimurium ATCC 14028S (L828) was used as a control strain throughout. Isogenic derivatives, L829 (tolC::cat) and L830 (acrB::aph) have been described previously.9 The wild-type strains of E. coli, P. aeruginosa and Staphylococcus aureus were MG1655, PA01 and NCTC strain 8532, reTM spectively. Strains were stored at 2208C on Protect beads and routinely cultured on Luria–Bertani (LB) agar or broth unless stated otherwise.
Creation of recombinant strains by P22 transduction New mutants required for this study were created by P22 transduction of mutant alleles into L828 (Salmonella Typhimurium 14028S wild-type).10 Briefly, phage stocks were prepared by inoculation of 5 mL of fresh LB broth (supplemented with 10 mM MgSO4/5 mM CaCl2) with 50 mL from an overnight culture of the mutant strain with the desired alleles to be packaged and incubated at 378C with shaking for 30 min. A 5 mL aliquot of a P22 phage stock was added to the culture and incubated at 378C overnight. After incubation, phages were isolated by adding 1 mL of chloroform to the culture in a Falcon tube and vortexing for 15 s; the cells were harvested by centrifugation at 3000 g for 10 min at 48C and the resulting phage-rich supernatant was removed to a glass bijou and stored at 48C. To transfer mutant alleles, overnight cultures (5 mL) of recipient strains were pelleted by centrifugation at room temperature for 10 min. The pellets were resuspended in 1 mL of LB broth supplemented with 10 mM MgSO4/5 mM CaCl2. Transduction reactions were set up by dispensing 100 mL aliquots of cell suspensions and adding varying volumes of P22 phage (containing DNA packaged from donor strains) before incubating at 378C for 15 min. Divalent cations were chelated upon addition of 1 mL of 100 mM sodium citrate (in LB broth) and then reactions were incubated at 378C for 45 min. An aliquot of 100 mL of each reaction was spread onto LB plates supplemented with 50 mg/L kanamycin or 25 mg/L chloramphenicol and incubated overnight at 378C. The transfer of each mutant allele was verified by PCR and DNA sequencing of putative mutants.
Table 1. List of strains used in this study Strain
Genotype
L828
wild-type
L829 L830 L1271 F77 G1 I364
14028S tolC::cat 14028S acrB::aph 14028S acrA::aph wild-type wild-type wild-type
2 of 9
Description Salmonella Typhimurium ATCC 14028S mutant lacking TolC mutant lacking AcrB mutant lacking AcrA S. aureus NCTC 8532 P. aeruginosa PA01 E. coli MG1655
Reference ATCC 9 9
this study NCTC ATCC ATCC
Biofilm formation assays Various models were used to analyse biofilm formation in this study.
Static biofilm models For crystal violet biofilm assays with Salmonella, overnight cultures of strains were diluted in fresh LB broth without salt to an optical density (OD) of 0.1 at 600 nm. Then, 96-well untreated polystyrene microtitre trays (Sterilin, UK) were inoculated with 200 mL of this suspension and incubated at 308C for 48 h with gentle agitation. Assays for other species were identical apart from the media used: Mueller– Hinton broth (Sigma– Aldrich, UK) was used for S. aureus and LB broth with salt (Sigma– Aldrich, UK) for P. aeruginosa and E. coli. After incubation, liquid was removed from all wells and wells were washed with sterile distilled water to remove any unbound cells. Biofilms were stained by adding 200 mL of 1% crystal violet (Sigma– Aldrich, UK) to appropriate wells for 15 min. Crystal violet was removed and each well was washed with sterile distilled water to remove unbound dye. The stained biofilm was solubilized by adding 200 mL of 70% ethanol and the OD measured at 600 nm using a FLUOstar OPTIMA (BMG Labtech, Germany). All biofilm assays were repeated three times with two biological and four technical replicates per repeat. The impact of efflux inhibitors in static assays was assessed using the same conditions with a range of concentrations of chlorpromazine, carbonyl cyanide m-chlorophenylhydrazone (CCCP) or phenyl-arginine-bnaphthylamide (PAbN) added to the relevant media. To determine whether biofilm formation by L829 (tolC::cat) and L830 (acrB::aph) could be rescued by coincubation with L828 (wild-type), strains were grown separated by a 0.45 mm membrane and biofilms formed as in the crystal violet assay, but in 500 mLvolumes in 24-well transwell plates. Assays were repeated with and without the presence of L828 (wild-type) in the upper ‘insert’ chamber with liquid contiguous between the upper and lower chambers. Biofilms were stained with crystal violet and quantified as above. Assays were repeated with addition of either a mid-logarithmic- or stationary-phase culture of L828 (wild-type) to assess whether the growth phase had an impact upon the production of any soluble biofilm-promoting factor. Statistical significance in both static biofilm assays was calculated using Student’s t-test and any differences between datasets with a P value of ,0.05 were considered significant.
Flow biofilm assays Biofilms under flow conditions were formed and visualized using a Bioflux microfluidic system (Fluxion, USA) and phase contrast microscopy (Labtech, USA).11 Outlet wells of 48-well microfluidic flow cell plates (Fluxion, USA) were inoculated with 50 mL of bacterial suspension diluted to an OD of 0.8 at 600 nm in LB broth without salt and bacteria introduced into the flow cell by pulsing with 3 dynes of force for 5 s. The flow cell plates were then incubated at 308C for 3 h to allow the bacteria to adhere to the untreated glass flow cells. Fresh LB broth without salt was added to the inlet wells and pumped through the flow cells at a force of 0.3 dynes at room temperature for 48 h. Phase contrast microscopy (LabtechScope LTS1– 1000, Labtech, UK) was used to visualize the biofilms formed at ×10, ×20 and ×40 magnification (numerical apertures 0.25, 0.4 and 0.55, respectively). Images were captured using a Micropublisher 3.3 CCD camera and QCapture Pro 7 software (Qimaging, USA). Assays were repeated in the presence and absence of efflux inhibitors (added to the inlet well media reservoir) and the resulting impact upon the maturation of a biofilm was evaluated. Estimation of areas of flow cells covered used Bioflux EZ software (Labtech, UK) and triplicate flow cells were measured multiple times to give average values for coverage in defined areas, which were then compared between strains or treatments.
JAC
Efflux and biofilm formation
Aggregation assays To examine whether loss of acrB or tolC led to alteration in the cellular hydrophobicity or aggregative ability, two different assays were used. To measure the time taken for strains to settle, strains were incubated overnight in 10 mL of LB broth without salt with shaking (150 rpm) before being placed statically on a bench. Samples (100 mL) were taken periodically from immediately below the surface of the liquid and the OD at 600 nm measured and recorded. Any changes in the rate at which cells settled were analysed for statistical significance using Student’s t-test, where any P value ,0.05 was considered significant. To determine whether there were any intrinsic differences in the aggregative ability of each strain, ammonium sulphate was used to induce aggregation of bacterial cells. A 4 M stock of (NH4)2SO4 was made in 1× PBS and adjusted to pH 6.8. This stock was then serially diluted and mixed 1: 1 (in 100 mL final volume) with bacterial suspensions (adjusted from an overnight culture to an OD at 570 nm of 0.8) for each strain. These suspensions were immediately added to a microscope slide and rocked gently for 30 min before aggregation was scored visually under a microscope as the presence of a precipitate. The lowest concentration of (NH4)2SO4 required to induce aggregation was recorded for each strain.
used and, again, colonies were incubated for 48 h at 308C; plates were then visualized under UV light, with fluorescence indicating the presence of cellulose.
Quantitative curli assay Strains were streaked onto LB agar plates without salt and incubated at 308C overnight. Bacterial suspensions were then made by harvesting a 1 cm patch of bacterial growth from a densely inoculated section of the plate in 5 mL of PBS and the OD at 600 nm measured. All samples were corrected to the lowest OD by diluting with PBS. An aliquot of 500 mL of each bacterial suspension was added to 500 mL of a 20 000 mg/mL Congo red solution and left to stain for 20 min. Pellets were obtained by centrifugation at 10000 g for 5 min, washed with 1 mL of sterile distilled water and harvested again. The pellets were then resuspended in 200 mL of PBS and the OD at 492 nm and fluorescence (excitation 492, emission 600) measured in a FLUOstar OPTIMA.
Measuring gene expression via RT– PCR Staining of biofilm matrix components Phenotypic differences in the production of bacterial matrix components were visualized by growing strains on agar supplemented with various compounds. To visualize the presence of curli fibres, strains were grown on LB agar without salt containing 40 mg/L Congo red (Sigma– Aldrich, UK) and incubated for 48 h at 308C. A red, dry and rough phenotype indicated production of curli, whereas a pink and smooth phenotype indicated absence of curli. For visualizing cellulose production, LB agar without salt containing 200 mg/L calcofluor (Sigma– Aldrich, UK) was
RNA preparation Overnight cultures were used to inoculate 10 mL of fresh LB broth without salt (1:100) and incubated at 308C with shaking (150 rpm) to stationary phase. An aliquot of 4 mL of this culture was added to 1 mL of 5% phenol/95% ethanol and incubated on ice for 30 min. The SV40 total RNA kit (Promega, UK) was then used to isolate RNA according to the manufacturer’s recommendations. RNA was quantified using a NanoDrop spectrophotometer (NanoDrop Technologies, USA).
Table 2. Primers used in this study Primer acrB-checkF acrB-checkR tolC-checkF tolC-checkF acrA-checkF acrA-checkF 16S-RTF 16S-RTR csgA-RTF csgA-RTR csgB-RTF csgB-RTR csgC-RTF csgC-RTF csgD-RTF csgD-RTR csgE-RTF csgE-RTR csgF-RTF csgF-RTR csgG-RTF csgG-RTR
Sequence
Description
GGATCACACCTTATTGCCAG TTAACAGTGATCGTCGGTCG CTTCTATCATGCCGGCGACC CGCTTGCTGGCACTGACCTT ACATCCAGGATGTGTTGTCG CAATCGTCGGATATTGCGCT CCTCAGCACATTGACGTTAC TTCCTCCAGATCTCTACGCA AGCATTCGCAGCAATCGTAG TTAGCGTTCCACTGGTCGAT ATCAGGCGGCCATTATTGGT TACTGGCATCGTTGGCATTG AATTCTCTCTGTGCGCGACG GCAGTGATTGTCCGTCCGAA GGTATTCTGCGTGGCGAATG AGTAATGCGGACTCGGTGCT ACGCTATCTGACCTGGATTG CGTTATGGTGATCCAGCTTC GACGTTCCAGTTCGCTAATC ATCGTTGGTCACCATACGTC CTGGAACGACAAGGCTTACA TGATCCAGCTGATACTGCGT
acrB mutant check forward primer acrB mutant check reverse primer tolC mutant check forward primer tolC mutant check reverse primer acrA mutant check forward primer acrA mutant check reverse primer 16S qRT– PCR forward primer 16S qRT– PCR reverse primer csgA qRT–PCR forward primer csgA qRT–PCR reverse primer csgB qRT– PCR forward primer csgB qRT– PCR reverse primer csgC qRT–PCR forward primer csgC qRT–PCR reverse primer csgD qRT–PCR forward primer csgD qRT–PCR reverse primer csgE qRT–PCR forward primer csgE qRT–PCR reverse primer csgF qRT–PCR forward primer csgF qRT–PCR reverse primer csgG qRT– PCR forward primer csgG qRT– PCR reverse primer
3 of 9
Baugh et al.
cDNA synthesis cDNA was synthesized by the addition of 500 ng of RNA to 1 mL of 50 ng/mL random primers (Invitrogen, UK), 1 mL of 10 mM dNTP mix (Invitrogen, UK) and sterile distilled water up to 13 mL. This mixture was heated to 658C for 5 min and incubated on ice for 1 min. Then, 4 mL of first-strand buffer (Invitrogen, UK), 1 mL of 0.1 M DTT (Invitrogen, UK), 1 mL of RNaseOUT (Invitrogen, UK) and 1 mL of Superscript III reverse transcriptase (Invitrogen, UK) were added to the 13 mL reaction above, mixed and incubated at 258C for 5 min followed by 508C for 1 h. The enzyme was then inactivated by heating to 708C for 15 min. cDNA was stored at 2208C until required.
Primer design Primers were designed using the Primer v 2.0 software package (Science and Educational Software) and produced by Invitrogen (UK). They were diluted with sterile distilled water to a concentration of 100 mM and stored at 2208C. When required, primers were diluted with sterile distilled water to a 25 mM working stock and stored at 2208C. All primers used in this study are shown in Table 2. Comparative RT–PCR was used to compare expression of genes representing both curli operons as previously described. Briefly, reactions
1.6 Biofilm formation (OD600)
1.4 1.2 1 0.8 0.6 *
0.4 *
0.2 0 L828 (wild-type)
L829 (tolC::cat)
L830 (acrB::aph)
L1271 (acrA::aph)
Strain Figure 1. Crystal violet biofilm assay quantifying biofilm formation of L828 (wild-type), L829 (tolC::cat), L830 (acrB::aph) and L1271 (acrA::aph), and showing that genetic inactivation of tolC or acrB creates an inability to form a biofilm. Asterisks indicate significantly different average biofilm values when compared with L828 (wild-type) using Student’s t-test (P,0.05).
(a)
(b)
consisted of 45 mL of Reddymix (Abgene, UK), 1.5 mL of forward primer (25 mM), 1.5 mL of reverse primer (25 mM) and 2 mL of cDNA (diluted to appropriate concentrations where saturation of reactions will not occur). Expression of 16S rRNA was measured and used as a control gene to normalize the data obtained for the genes of interest. PCR amplimers were quantified using a denaturing HPLC WAVE system (Transgenomic Inc.).
Results Mutants lacking a functional acrB or tolC do not form competent biofilms, whereas a mutant lacking acrA is able to form a biofilm A high-throughput biofilm assay using crystal violet to stain cells adhered to a 96-well plate showed a significant decrease in the biofilm formation ability of L829, which lacks the outer membrane protein TolC (tolC::cat), and L830, which lacks the inner membrane pump AcrB (acrB::aph) (Figure 1). However, genetic inactivation of acrA [L1271 (acrA::aph)], the periplasmic adapter protein, had no effect on biofilm formation. The biofilm defect of efflux mutants seen in static assays was replicated when tested under flow conditions; all mutants tested were unable to form a mature biofilm under flow. Phase contrast microscopy images from biofilms formed in a flow cell under shear stress showed that L828 (wildtype) formed a mature biofilm and L829 (tolC::cat) and L830 (acrB::aph) were able to adhere as individual cells to the flow cell, but unable to form a mature, three-dimensional biofilm, with both mutants covering ,5% of the surface of the flow cells on average, whereas L828 (wild-type) had covered 99% of the area of the flow cell after 48 h; the difference in area coverage between the wild-type and mutants was statistically significant (Figure 2).
The biofilm deficit of mutants lacking a functional acrB or tolC is not associated with altered aggregative ability or a secretion deficit of any biofilm-promoting factor We hypothesized that the efflux mutants may have a greater ability to aggregate than the wild-type due to increased cell surface hydrophobicity. To determine whether inactivation of AcrAB-TolC had altered the intrinsic aggregative nature of the strains lacking acrB or tolC, a settle assay was used; this showed no significant decrease in the aggregative ability of L829 (tolC::cat) or L830
(c)
Figure 2. Phase contrast microscopy images of (a) L828 (wild-type), (b) L830 (acrB::aph) and (c) L829 (tolC::cat) at ×40 magnification after 24 h of incubation under flow conditions. Bar¼5 mm.
4 of 9
JAC
Efflux and biofilm formation
(a) 140%
L828 (wild-type) L829 (tolC::cat)
120%
L830 (acrB::aph)
* *
Percentage of initial OD600
*
*
*
100%
80% * * 60%
40%
20%
0% 0
1
2
3
4
5
6
24
Time (h) (b)
L828 (wild-type)
L829 (tolC::cat)
L830 (acrB::aph)
1 M (NH4)2SO4
2 M (NH4)2SO4
Figure 3. (a) Settle assay of L828 (wild-type), L829 (tolC::cat) and L830 (acrB::aph). E. coli O42 was used as a positive, aggregative control. Values indicate the percentage of an initial absorbance from readings taken immediately below the surface of the liquid of a broth that was incubated statically over a 24 h period. Asterisks indicate average values significantly different to the wild-type under the same conditions. (b) Salt aggregation test images of L828 (wild-type), L829 (tolC::cat) and L830 (acrB::aph) in 1 and 2 M ammonium sulphate. Aggregation was recorded as formation of a visible precipitate and the lowest concentration of ammonium sulphate to prompt precipitation was recorded for each strain.
(acrB::aph) (Figure 3a). Salt aggregation tests also showed no defect in the mutants’ ability to aggregate. In fact, L829 (tolC::cat) cells aggregated in a lower concentration of ammonium sulphate than L828 (wild-type), showing a slightly greater tendency for cells to aggregate than the wild-type (Figure 3b). We also hypothesized that a soluble biofilm-promoting factor was exported by AcrAB-TolC; therefore, addition of culture supernatant conditioned by growth with L828 (wild-type) should ‘rescue’ the ability of the tolC and acrB mutant strains to form a
biofilm. However, two coincubation assays with wild-type and mutant strains suggested that there is no ‘biofilm factor’ exported by AcrAB-TolC. Transwell assays showed the same poor ability to form a biofilm of the acrB and tolC mutants when incubated alone or coincubated with L828 (wild-type) (Figure 4). In addition, no rescue of the biofilm defect was observed when coincubated with logarithmic- or stationary-phase cultures of L828 (wild-type) (Figure 4). Similarly, biofilm mat assays coinoculated with an equal ratio of wild-type and mutant showed that mutant cells were not
5 of 9
Baugh et al.
Biofilm formation (relative to L828)
1.4 1.2 1 0.8 0.6 0.4 0.2 0
(b)
1.4
Biofilm formation (relative to L828)
(a)
1.2
The lack of curli production in mutants lacking acrB and tolC is due to transcriptional repression * * *
*
1 *
0.8
*
0.6 0.4
*
0.2
*
0 L828
L830 (acrB::aph)
L829 (tolC::cat)
L829 (tolC::cat)
L828 Strain
L830 (acrB::aph)
+ wild-type supernatant
Figure 4. Transwell assays showed no rescue of biofilm formation by mutants when coincubated with wild-type (L828). (a) Strains incubated with and without the presence of L828 inoculated at the same density as the mutants. (b) The same experiment, but with coincubation with an overnight, undiluted culture of L828. Asterisks indicate average values significantly different to the wild-type under the same conditions.
L828 (wild-type)
L829 (tolC::cat)
L830 (acrB::aph)
L1271 (acrA::aph)
csgG
csgF
csgE
csgD
csgB
csgA
csgC
Key (average expression relative to wild-type): 76%–100% of wild-type expression
51%–75% of wild-type expression
26%–50% of wild-type expression
0%–25% of wild-type expression
Figure 5. Transcriptional expression of curli genes. A schematic of the curli biosynthesis genes, with average expression values determined by comparative RT–PCR used to shade each gene. Where expression in the tolC and acrB mutant was ≤50% of the wild-type, all these values were statistically significantly different using Student’s t-test (P,0.05).
present in any of the biofilm mats formed, whereas the corresponding planktonic culture comprised an equal mixture of mutant and wild-type cells (data not shown).
6 of 9
Congo red-supplemented agar and Congo red staining of bacterial suspensions showed qualitatively and quantitatively that curli expression is repressed in the tolC and acrB mutants and produced at wild-type levels in the acrA mutant.4 This lack of curli production was found to result from significant transcriptional repression of all functional genes in the curli biosynthetic loci, as measured by comparative RT– PCR. The structural curli genes were both downregulated in the acrB and tolC mutants; csgA was down-regulated 2.5- and 4-fold, respectively, and csgB was down-regulated 3-fold in both mutants. The master regulator, csgD, was down-regulated 10-fold in both mutants. The three genes required for efficient assembly and translocation of the structural subunits were also all down-regulated to varying degrees: csgE was down-regulated 1.5fold in both mutants; csgF was down-regulated by 2- and 4-fold in the acrB and tolC mutants, respectively; and csgG was also down-regulated by 5- and 10-fold in the acrB and tolC mutants, respectively. Repression of these genes (bar csgD) was not observed, however, in the acrA (Figure 5) mutant, correlating with the phenotypic curli morphologies on Congo red.
Inhibition of efflux is effective in preventing biofilm formation by various species under static and flow conditions We previously showed that inhibitors of efflux were able to prevent Salmonella Typhimurium 14028S from forming a biofilm under static conditions.4 To assess whether inhibition of efflux has impact as an antibiofilm approach in other bacterial species, we evaluated the ability of the same three inhibitors as in our previous study4 (chlorpromazine, CCCPand PAbN) to prevent biofilm formation by representative strains of E. coli (MG1655), P. aeruginosa (PA01) and S. aureus (NCTC 8532), species where biofilm-associated infections are problematic. These inhibitors act in different ways; therefore, we hypothesize that any antibiofilm effect common to all species investigated is due to efflux inhibition rather than drug-specific effects. In the static crystal violet assay, all three inhibitors significantly reduced biofilm formation (P, 0.05) by all three species, although the concentration of each inhibitor that provoked the greatest effect varied between species (Figure 6). S. aureus was most susceptible to CCCP, with 2 mg/L causing a 5-fold reduction in biofilm formation. PAbN was the inhibitor with the greatest impact on E. coli, with 16 mg/L preventing formation of a normal biofilm. P. aeruginosa was least susceptible to the inhibitors, with 128 mg/L CCCP and 512 mg/L of the other inhibitors needed to prevent biofilm formation. Irrespective of the species, biofilm inhibitory concentrations were all below the MIC of each compound under the test conditions, demonstrating a specific antibiofilm effect rather than a generalized growth inhibition. Under flow conditions, after an initial attachment period, the inhibitors also prevented biofilm formation. Bacterial coverage of the flow cell by S. aureus and biofilm formation was severely reduced or absent after addition of CCCP (4 mg/L, resulted in a 6-fold reduction in coverage of the flow cell), chlorpromazine (64 mg/L, resulted in a 6-fold reduction in coverage of the flow cell) and PAbN (64 mg/L, resulted in a 500-fold reduction in coverage of the flow cell) (Figure 7). Similar results were obtained for both E. coli and P. aeruginosa (data not shown).
JAC
Efflux and biofilm formation
S. aureus
E. coli
150
P. aeruginosa
150
100
150
100 *
CCCP
* 50
*
* *
50
*
*
100
*
50 *
PAbN
CPZ
Biofilm formed (relative to no drug control)
* 0
0 0
0.5
1
2
4
8
150
0
1
2
4
8
16
*
*
*
8
16
32
64
128
100
100 *
* *
50
0
150
150
100
0
*
*
50
*
50
* *
0
0
0 0
16
32
64
128 256
150
0
4
8
16
32
64
150 *
100
*
0
*
*
16
32
64
256
128 256
512
*
100
* 50
50
*
* 0
0 0
128
* *
50
64
150
100
*
0
0
8
16
32
64
128
0
64
128
256
512
Drug concentration (mg/L) Figure 6. Impact of efflux inhibitors on the ability of S. aureus, P. aeruginosa and E. coli to form a biofilm in the crystal violet assay. Various efflux inhibitor concentration ranges were used, depending on the efficacy of the inhibitor against the species. Bars show the average biofilm formed by ≥24 replicate samples at each drug concentration after 48 h relative to untreated controls. Error bars indicate standard deviations. Asterisks indicate a significant decrease in biofilm formation when compared with the efflux inhibitor-free control using Student’s t-test (P, 0.05). CPZ, chlorpromazine.
Discussion Multidrug efflux pumps have a central role in the biology of bacteria, with roles in drug resistance, cell division, pathogenicity and, as recently described, the formation of biofilms.4,5 Here, we investigated the mechanism for the inability of mutants lacking AcrB and TolC, constituents of the major AcrAB-TolC system of Enterobacteriaceae, to form a competent biofilm. Mutants of Salmonella lacking a functional TolC or AcrB were unable to form biofilms under various conditions and this was not related to any defect in growth, cellular hydrophobicity/aggregative ability or export of a biofilm-promoting substrate. Surprisingly, a mutant lacking a functional AcrA (but still expressing AcrB)12 was not defective in its ability to form a biofilm. Loss of acrA has previously been shown to result in hypersusceptibility to various drugs and a decreased ability to attach to and invade epithelial cells in tissue culture.12 The biofilm phenotype of an acrA mutant was,
however, distinct from that of mutants lacking acrB.12 One major difference between acrA and acrB/tolC mutants is the expression of curli genes, repressed in the acrB and tolC mutants, but not the acrA mutant.13 This led us to hypothesize that curli repression was the reason that the efflux mutants did not form a biofilm and we show here the absence of curli production in mutants lacking tolC or acrB. These data suggest that in the absence of efflux there is strict repression of all curli biosynthetic genes. The regulation of curli expression is extremely complex, with multiple pathways known to impact curli production. Amongst these pathways are two-component systems that respond to membrane stress (cpxRA, rcsCB and envZ/ompR), rpoE and the lytic transglycosylases mltC and mltE. 14 All these systems have membrane-bound or periplasmic components, suggesting that curli synthesis is sensitive to changes in the membrane. To determine whether a lack of biofilm formation in mutants lacking efflux pumps was Salmonella specific or a general
7 of 9
Baugh et al.
(a)
(b)
(c)
(d)
Figure 7. Phase contrast microscopy images of glass flow cells after incubation with S. aureus for 16 h under flow conditions at×40 magnification. (a) No inhibitor, (b) CCCP, (c) chlorpromazine and (d) PAbN. Bar¼5 mm.
phenomenon when efflux was reduced, we examined biofilm formation for various different strains representing both Gramnegative and Gram-positive species. We examined biofilm formation under flow and static assays for E. coli, P. aeruginosa and S. aureus with and without efflux inhibitors previously found to be active against Salmonella.4,5 All the inhibitors tested prevented biofilm formation in both static and flow assays. Importantly, in the flow assays, cells were allowed to adhere to the flow cell for 3 h before the addition of inhibitor, and whilst individual cells did attach they were not able to mature and form a threedimensional biofilm in the presence of inhibitors. All concentrations tested were below the MIC of each inhibitor for the specific strains; therefore, the efflux inhibitors are acting as antibiofilm agents and reduction in biofilm formation is not due to an antimicrobial action preventing growth. The prevention of formation of a mature biofilm in all species indicates that repression of matrix components essential for biofilm maturation is likely to be the mechanism by which efflux inhibition prevents biofilm formation, supporting the Salmonella model where curli repression appears to be central to the mechanism. It has recently been shown that curli genes are more conserved across species than first thought; therefore, curli
8 of 9
is important not just for Enterobacteriaceae, but also for a variety of different bacteria.15 Efflux inhibitors have previously been shown to decrease biofilm formation in P. putida and S. aureus; the efflux inhibitors used in that study were thioridazine, NMP and PAbN.5 The one direct comparison to this study that can be drawn relates to the effect of PAbN on the formation of S. aureus biofilms: Kvist et al. 5 showed that 20 mg/L PAbN reduced biofilm formation to 60%. Our study corroborates this finding. Although the concentrations are slightly different, we found that both 16 and 32 mg/L PAbN also reduced biofilm formation of S. aureus to 60% in a static assay. A greater reduction was, however, seen in assays under flow conditions, indicating that different conditions may influence the antibiofilm efficacy of efflux inhibitors. The efflux inhibitors used in this study are not clinically useful for systemic treatment and some have relatively high toxicity;16 however, there are a number of efflux inhibitors that belong to classes of successful drugs used in human medicine and there is no reason why inhibitors of bacterial multidrug efflux systems should be generically toxic to humans. Whilst the most effective inhibitor varied with the species tested, the inhibition of efflux is a promising novel antibiofilm
Efflux and biofilm formation
strategy and is effective under a range of conditions. The potential for the impregnation of medical devices and other surfaces with efflux inhibitors that can be colonized by biofilms and represent a source of infection warrants further evaluation.
Funding This work was supported by the Medical Research Council via a doctoral training award to M. A. W. used to support S. B.
Transparency declarations None to declare.
References 1 Costerton JW, Cheng KJ, Geesey GG et al. Bacterial biofilms in nature and disease. Ann Rev Microbiol 1987; 41: 435–64. 2 Lewis K. Persister cells, dormancy and infectious disease. Nat Rev Microbiol 2007; 5: 48– 56. 3 Watnick P, Kolter R. Biofilm, city of microbes. J Bacteriol 2000; 182: 2675– 9. 4 Baugh S, Ekanayaka AS, Piddock LJV et al. Loss of any of the multidrug efflux pumps of Salmonella Typhimurium results in a loss of ability to form a biofilm. J Antimicrob Chemother 2012; 67: 2409 –17. 5 Kvist M, Hancock V, Klemm P. Inactivation of efflux pumps abolishes bacterial biofilm formation. Appl Environ Microbiol 2008; 74: 7376–82. 6 De Kievit TR, Parkins MD, Gillis RJ et al. Multidrug efflux pumps: expression patterns and contribution to antibiotic resistance in Pseudomonas aeruginosa biofilms. Antimicrob Agents Chemother 2001; 45: 1761– 70.
JAC 7 Ramage G, Bachmann S, Patterson TF et al. Investigation of multidrug efflux pumps in relation to fluconazole resistance in Candida albicans biofilms. J Antimicrob Chemother 2002; 49: 973– 80. 8 Gerstel U, Park C, Romling U. Complex regulation of csgD promoter activity by global regulatory proteins. Mol Microbiol 2003; 49: 639– 54. 9 Nishino K, Latifi T, Groisman EA. Virulence and drug resistance roles of multidrug efflux systems of Salmonella enterica serovar Typhimurium. Mol Microbiol 2006; 59: 126– 41. 10 Ebel-Tsipis J, Fox MS, Botstein D. Generalized transduction by bacteriophage P22 in Salmonella typhimurium. II. Mechanism of integration of transducing DNA. J Mol Biol 1972; 71: 449–69. 11 Benoit MR, Conant CG, Ionescu-Zanetti C et al. New device for high-throughput viability screening of flow biofilms. Appl Environ Microbiol 2010; 76: 4136– 42. 12 Blair JM, La Ragione RM, Woodward MJ et al. Periplasmic adaptor protein AcrA has a distinct role in the antibiotic resistance and virulence of Salmonella enterica serovar Typhimurium. J Antimicrob Chemother 2009; 64: 965– 72. 13 Bailey AM, Paulsen IT, Piddock LJ. RamA confers multidrug resistance in Salmonella enterica via increased expression of acrB, which is inhibited by chlorpromazine. Antimicrob Agents Chemother 2008; 52: 3604– 11. 14 Monteiro C, Fang X, Ahmad I et al. Regulation of biofilm components in Salmonella enterica serovar Typhimurium by lytic transglycosylases involved in cell wall turnover. J Bacteriol 2011; 193: 6443– 51. 15 Dueholm MS, Albertsen M, Otzen D et al. Curli functional amyloid systems are phylogenetically widespread and display large diversity in operon and protein structure. PLoS One 2012; 7: e51274. 16 Lomovskaya O, Warren MS, Lee A et al. Identification and characterization of inhibitors of multidrug resistance efflux pumps in Pseudomonas aeruginosa: novel agents for combination therapy. Antimicrob Agents Chemother 2001; 45: 105–16.
9 of 9
J Antimicrob Chemother doi:10.1093/jac/dkt420
Inhibition of multidrug efflux as a strategy to prevent biofilm formation Stephanie Baugh, Charlotte R. Phillips, Aruna S. Ekanayaka, Laura J. V. Piddock and Mark A. Webber* School of Immunity and Infection, Institute for Microbiology and Infection, University of Birmingham, Edgbaston, Birmingham B15 2TT, UK *Corresponding author. Tel: +0121-414-2859; Fax: +0121-414-6819; E-mail: [email protected]
Received 25 June 2013; returned 2 August 2013; revised 24 September 2013; accepted 1 October 2013
Methods: Mutants lacking components of the AcrAB-TolC system in Salmonella enterica serovar Typhimurium were investigated for their ability to aggregate, produce biofilm matrix components and form a biofilm. The potential for export of a biofilm-relevant substrate via efflux pumps was investigated and expression of genes that regulate multidrug efflux and production of biofilm matrix components was measured. The ability of efflux inhibitors carbonyl cyanide m-chlorophenylhydrazone, chlorpromazine and phenyl-arginine-b-naphthylamide to prevent biofilm formation by Escherichia coli, Pseudomonas aeruginosa and Staphylococcus aureus under static and flow conditions was assessed. Results: Mutants of Salmonella Typhimurium that lack TolC or AcrB, but surprisingly not AcrA, were compromised in their ability to form biofilms. This defect was not related to changes in cellular hydrophobicity, aggregative ability or export of any biofilm-specific factor. The biofilm defect resulted from transcriptional repression of curli biosynthesis genes and consequent inhibition of production of curli. All three efflux inhibitors significantly reduced biofilm production in both static and flow biofilm assays, although different concentrations of each inhibitor were most active against each species. Conclusions: This work shows that both genetic inactivation and chemical inhibition of efflux pumps results in transcriptional repression of biofilm matrix components and a lack of biofilm formation. Therefore, inhibition of efflux is a promising antibiofilm strategy. Keywords: curli, AcrAB-TolC, efflux inhibitors
Introduction Bacterial biofilms are a major clinical and industrial problem, and eradication of biofilms presents a challenge for antimicrobial chemotherapy.1 – 3 We recently described an inability of Salmonella enterica serovar Typhimurium (hereafter Salmonella Typhimurium) to form a competent biofilm that resulted from inactivation of any of the multidrug resistance (MDR) efflux systems of Salmonella.4 MDR efflux systems have also previously been suggested to be important in the intrinsic antibiotic resistance of biofilms and have been shown to be involved in the ability of various species to form a biofilm: Escherichia coli, Pseudomonas putida, Klebsiella pneumoniae 5 and Pseudomonas aeruginosa.5,6 As well as various bacterial species, efflux has also been shown to be important in biofilm formation of the fungus Candida albicans.7 We found that production of curli, a major component of the Salmonella biofilm extracellular matrix, was defective in all these strains, suggesting a common mechanism for the lack of biofilm formation in all
mutants.4 Additionally, we found that the addition of three inhibitors of efflux with different mechanisms of action also resulted in prevention of biofilm formation and that this was also associated with loss of curli production.4 Here, using AcrAB-TolC as a paradigm MDR efflux system, we investigated the mechanism by which loss of efflux activity results in a lack of curli production. We ruled out export of a factor crucial for biofilm development via AcrAB-TolC and also showed that inactivation of components of AcrAB-TolC did not alter cellular hydrophobicity. However, inactivation of efflux components was found to significantly alter the expression of biofilmrelated genes. We demonstrate that the biofilm defect of mutants lacking a functional AcrB or TolC is due to transcriptional repression of curli biosynthesis. Curli is an essential proteinaceous component of the biofilm matrix for various bacterial species, including Salmonella. The major curli transcriptional regulator, CsgD, controls transcription of csgBA (the operon encoding the two structural
# The Author 2013. Published by Oxford University Press on behalf of the British Society for Antimicrobial Chemotherapy. All rights reserved. For Permissions, please e-mail: [email protected]
1 of 9
Downloaded from http://jac.oxfordjournals.org/ at Georgia State University on October 31, 2013
Objectives: We have recently shown that inactivation of any of the multidrug efflux systems of Salmonella results in loss of the ability to form a competent biofilm. The aim of this study was to determine the mechanism linking multidrug efflux and biofilm formation, and to determine whether inhibition of efflux is a viable antibiofilm strategy.
Baugh et al.
proteins comprising curli), as well as genes responsible for cellulose biosynthesis. csgD is itself controlled by a complex network of both positive and negative regulation.8 Secondly, we determined whether inhibition of efflux is effective against a range of pathogens for which biofilm formation is problematic under flow conditions, as well as in static biofilm assays. All three inhibitors tested were able to reduce biofilm formation by a range of pathogens in both static and flow biofilm assays.
Materials and methods Strains and growth media All strains used in this study and their origins are shown in Table 1. Salmonella Typhimurium ATCC 14028S (L828) was used as a control strain throughout. Isogenic derivatives, L829 (tolC::cat) and L830 (acrB::aph) have been described previously.9 The wild-type strains of E. coli, P. aeruginosa and Staphylococcus aureus were MG1655, PA01 and NCTC strain 8532, reTM spectively. Strains were stored at 2208C on Protect beads and routinely cultured on Luria–Bertani (LB) agar or broth unless stated otherwise.
Creation of recombinant strains by P22 transduction New mutants required for this study were created by P22 transduction of mutant alleles into L828 (Salmonella Typhimurium 14028S wild-type).10 Briefly, phage stocks were prepared by inoculation of 5 mL of fresh LB broth (supplemented with 10 mM MgSO4/5 mM CaCl2) with 50 mL from an overnight culture of the mutant strain with the desired alleles to be packaged and incubated at 378C with shaking for 30 min. A 5 mL aliquot of a P22 phage stock was added to the culture and incubated at 378C overnight. After incubation, phages were isolated by adding 1 mL of chloroform to the culture in a Falcon tube and vortexing for 15 s; the cells were harvested by centrifugation at 3000 g for 10 min at 48C and the resulting phage-rich supernatant was removed to a glass bijou and stored at 48C. To transfer mutant alleles, overnight cultures (5 mL) of recipient strains were pelleted by centrifugation at room temperature for 10 min. The pellets were resuspended in 1 mL of LB broth supplemented with 10 mM MgSO4/5 mM CaCl2. Transduction reactions were set up by dispensing 100 mL aliquots of cell suspensions and adding varying volumes of P22 phage (containing DNA packaged from donor strains) before incubating at 378C for 15 min. Divalent cations were chelated upon addition of 1 mL of 100 mM sodium citrate (in LB broth) and then reactions were incubated at 378C for 45 min. An aliquot of 100 mL of each reaction was spread onto LB plates supplemented with 50 mg/L kanamycin or 25 mg/L chloramphenicol and incubated overnight at 378C. The transfer of each mutant allele was verified by PCR and DNA sequencing of putative mutants.
Table 1. List of strains used in this study Strain
Genotype
L828
wild-type
L829 L830 L1271 F77 G1 I364
14028S tolC::cat 14028S acrB::aph 14028S acrA::aph wild-type wild-type wild-type
2 of 9
Description Salmonella Typhimurium ATCC 14028S mutant lacking TolC mutant lacking AcrB mutant lacking AcrA S. aureus NCTC 8532 P. aeruginosa PA01 E. coli MG1655
Reference ATCC 9 9
this study NCTC ATCC ATCC
Biofilm formation assays Various models were used to analyse biofilm formation in this study.
Static biofilm models For crystal violet biofilm assays with Salmonella, overnight cultures of strains were diluted in fresh LB broth without salt to an optical density (OD) of 0.1 at 600 nm. Then, 96-well untreated polystyrene microtitre trays (Sterilin, UK) were inoculated with 200 mL of this suspension and incubated at 308C for 48 h with gentle agitation. Assays for other species were identical apart from the media used: Mueller– Hinton broth (Sigma– Aldrich, UK) was used for S. aureus and LB broth with salt (Sigma– Aldrich, UK) for P. aeruginosa and E. coli. After incubation, liquid was removed from all wells and wells were washed with sterile distilled water to remove any unbound cells. Biofilms were stained by adding 200 mL of 1% crystal violet (Sigma– Aldrich, UK) to appropriate wells for 15 min. Crystal violet was removed and each well was washed with sterile distilled water to remove unbound dye. The stained biofilm was solubilized by adding 200 mL of 70% ethanol and the OD measured at 600 nm using a FLUOstar OPTIMA (BMG Labtech, Germany). All biofilm assays were repeated three times with two biological and four technical replicates per repeat. The impact of efflux inhibitors in static assays was assessed using the same conditions with a range of concentrations of chlorpromazine, carbonyl cyanide m-chlorophenylhydrazone (CCCP) or phenyl-arginine-bnaphthylamide (PAbN) added to the relevant media. To determine whether biofilm formation by L829 (tolC::cat) and L830 (acrB::aph) could be rescued by coincubation with L828 (wild-type), strains were grown separated by a 0.45 mm membrane and biofilms formed as in the crystal violet assay, but in 500 mLvolumes in 24-well transwell plates. Assays were repeated with and without the presence of L828 (wild-type) in the upper ‘insert’ chamber with liquid contiguous between the upper and lower chambers. Biofilms were stained with crystal violet and quantified as above. Assays were repeated with addition of either a mid-logarithmic- or stationary-phase culture of L828 (wild-type) to assess whether the growth phase had an impact upon the production of any soluble biofilm-promoting factor. Statistical significance in both static biofilm assays was calculated using Student’s t-test and any differences between datasets with a P value of ,0.05 were considered significant.
Flow biofilm assays Biofilms under flow conditions were formed and visualized using a Bioflux microfluidic system (Fluxion, USA) and phase contrast microscopy (Labtech, USA).11 Outlet wells of 48-well microfluidic flow cell plates (Fluxion, USA) were inoculated with 50 mL of bacterial suspension diluted to an OD of 0.8 at 600 nm in LB broth without salt and bacteria introduced into the flow cell by pulsing with 3 dynes of force for 5 s. The flow cell plates were then incubated at 308C for 3 h to allow the bacteria to adhere to the untreated glass flow cells. Fresh LB broth without salt was added to the inlet wells and pumped through the flow cells at a force of 0.3 dynes at room temperature for 48 h. Phase contrast microscopy (LabtechScope LTS1– 1000, Labtech, UK) was used to visualize the biofilms formed at ×10, ×20 and ×40 magnification (numerical apertures 0.25, 0.4 and 0.55, respectively). Images were captured using a Micropublisher 3.3 CCD camera and QCapture Pro 7 software (Qimaging, USA). Assays were repeated in the presence and absence of efflux inhibitors (added to the inlet well media reservoir) and the resulting impact upon the maturation of a biofilm was evaluated. Estimation of areas of flow cells covered used Bioflux EZ software (Labtech, UK) and triplicate flow cells were measured multiple times to give average values for coverage in defined areas, which were then compared between strains or treatments.
JAC
Efflux and biofilm formation
Aggregation assays To examine whether loss of acrB or tolC led to alteration in the cellular hydrophobicity or aggregative ability, two different assays were used. To measure the time taken for strains to settle, strains were incubated overnight in 10 mL of LB broth without salt with shaking (150 rpm) before being placed statically on a bench. Samples (100 mL) were taken periodically from immediately below the surface of the liquid and the OD at 600 nm measured and recorded. Any changes in the rate at which cells settled were analysed for statistical significance using Student’s t-test, where any P value ,0.05 was considered significant. To determine whether there were any intrinsic differences in the aggregative ability of each strain, ammonium sulphate was used to induce aggregation of bacterial cells. A 4 M stock of (NH4)2SO4 was made in 1× PBS and adjusted to pH 6.8. This stock was then serially diluted and mixed 1: 1 (in 100 mL final volume) with bacterial suspensions (adjusted from an overnight culture to an OD at 570 nm of 0.8) for each strain. These suspensions were immediately added to a microscope slide and rocked gently for 30 min before aggregation was scored visually under a microscope as the presence of a precipitate. The lowest concentration of (NH4)2SO4 required to induce aggregation was recorded for each strain.
used and, again, colonies were incubated for 48 h at 308C; plates were then visualized under UV light, with fluorescence indicating the presence of cellulose.
Quantitative curli assay Strains were streaked onto LB agar plates without salt and incubated at 308C overnight. Bacterial suspensions were then made by harvesting a 1 cm patch of bacterial growth from a densely inoculated section of the plate in 5 mL of PBS and the OD at 600 nm measured. All samples were corrected to the lowest OD by diluting with PBS. An aliquot of 500 mL of each bacterial suspension was added to 500 mL of a 20 000 mg/mL Congo red solution and left to stain for 20 min. Pellets were obtained by centrifugation at 10000 g for 5 min, washed with 1 mL of sterile distilled water and harvested again. The pellets were then resuspended in 200 mL of PBS and the OD at 492 nm and fluorescence (excitation 492, emission 600) measured in a FLUOstar OPTIMA.
Measuring gene expression via RT– PCR Staining of biofilm matrix components Phenotypic differences in the production of bacterial matrix components were visualized by growing strains on agar supplemented with various compounds. To visualize the presence of curli fibres, strains were grown on LB agar without salt containing 40 mg/L Congo red (Sigma– Aldrich, UK) and incubated for 48 h at 308C. A red, dry and rough phenotype indicated production of curli, whereas a pink and smooth phenotype indicated absence of curli. For visualizing cellulose production, LB agar without salt containing 200 mg/L calcofluor (Sigma– Aldrich, UK) was
RNA preparation Overnight cultures were used to inoculate 10 mL of fresh LB broth without salt (1:100) and incubated at 308C with shaking (150 rpm) to stationary phase. An aliquot of 4 mL of this culture was added to 1 mL of 5% phenol/95% ethanol and incubated on ice for 30 min. The SV40 total RNA kit (Promega, UK) was then used to isolate RNA according to the manufacturer’s recommendations. RNA was quantified using a NanoDrop spectrophotometer (NanoDrop Technologies, USA).
Table 2. Primers used in this study Primer acrB-checkF acrB-checkR tolC-checkF tolC-checkF acrA-checkF acrA-checkF 16S-RTF 16S-RTR csgA-RTF csgA-RTR csgB-RTF csgB-RTR csgC-RTF csgC-RTF csgD-RTF csgD-RTR csgE-RTF csgE-RTR csgF-RTF csgF-RTR csgG-RTF csgG-RTR
Sequence
Description
GGATCACACCTTATTGCCAG TTAACAGTGATCGTCGGTCG CTTCTATCATGCCGGCGACC CGCTTGCTGGCACTGACCTT ACATCCAGGATGTGTTGTCG CAATCGTCGGATATTGCGCT CCTCAGCACATTGACGTTAC TTCCTCCAGATCTCTACGCA AGCATTCGCAGCAATCGTAG TTAGCGTTCCACTGGTCGAT ATCAGGCGGCCATTATTGGT TACTGGCATCGTTGGCATTG AATTCTCTCTGTGCGCGACG GCAGTGATTGTCCGTCCGAA GGTATTCTGCGTGGCGAATG AGTAATGCGGACTCGGTGCT ACGCTATCTGACCTGGATTG CGTTATGGTGATCCAGCTTC GACGTTCCAGTTCGCTAATC ATCGTTGGTCACCATACGTC CTGGAACGACAAGGCTTACA TGATCCAGCTGATACTGCGT
acrB mutant check forward primer acrB mutant check reverse primer tolC mutant check forward primer tolC mutant check reverse primer acrA mutant check forward primer acrA mutant check reverse primer 16S qRT– PCR forward primer 16S qRT– PCR reverse primer csgA qRT–PCR forward primer csgA qRT–PCR reverse primer csgB qRT– PCR forward primer csgB qRT– PCR reverse primer csgC qRT–PCR forward primer csgC qRT–PCR reverse primer csgD qRT–PCR forward primer csgD qRT–PCR reverse primer csgE qRT–PCR forward primer csgE qRT–PCR reverse primer csgF qRT–PCR forward primer csgF qRT–PCR reverse primer csgG qRT– PCR forward primer csgG qRT– PCR reverse primer
3 of 9
Baugh et al.
cDNA synthesis cDNA was synthesized by the addition of 500 ng of RNA to 1 mL of 50 ng/mL random primers (Invitrogen, UK), 1 mL of 10 mM dNTP mix (Invitrogen, UK) and sterile distilled water up to 13 mL. This mixture was heated to 658C for 5 min and incubated on ice for 1 min. Then, 4 mL of first-strand buffer (Invitrogen, UK), 1 mL of 0.1 M DTT (Invitrogen, UK), 1 mL of RNaseOUT (Invitrogen, UK) and 1 mL of Superscript III reverse transcriptase (Invitrogen, UK) were added to the 13 mL reaction above, mixed and incubated at 258C for 5 min followed by 508C for 1 h. The enzyme was then inactivated by heating to 708C for 15 min. cDNA was stored at 2208C until required.
Primer design Primers were designed using the Primer v 2.0 software package (Science and Educational Software) and produced by Invitrogen (UK). They were diluted with sterile distilled water to a concentration of 100 mM and stored at 2208C. When required, primers were diluted with sterile distilled water to a 25 mM working stock and stored at 2208C. All primers used in this study are shown in Table 2. Comparative RT–PCR was used to compare expression of genes representing both curli operons as previously described. Briefly, reactions
1.6 Biofilm formation (OD600)
1.4 1.2 1 0.8 0.6 *
0.4 *
0.2 0 L828 (wild-type)
L829 (tolC::cat)
L830 (acrB::aph)
L1271 (acrA::aph)
Strain Figure 1. Crystal violet biofilm assay quantifying biofilm formation of L828 (wild-type), L829 (tolC::cat), L830 (acrB::aph) and L1271 (acrA::aph), and showing that genetic inactivation of tolC or acrB creates an inability to form a biofilm. Asterisks indicate significantly different average biofilm values when compared with L828 (wild-type) using Student’s t-test (P,0.05).
(a)
(b)
consisted of 45 mL of Reddymix (Abgene, UK), 1.5 mL of forward primer (25 mM), 1.5 mL of reverse primer (25 mM) and 2 mL of cDNA (diluted to appropriate concentrations where saturation of reactions will not occur). Expression of 16S rRNA was measured and used as a control gene to normalize the data obtained for the genes of interest. PCR amplimers were quantified using a denaturing HPLC WAVE system (Transgenomic Inc.).
Results Mutants lacking a functional acrB or tolC do not form competent biofilms, whereas a mutant lacking acrA is able to form a biofilm A high-throughput biofilm assay using crystal violet to stain cells adhered to a 96-well plate showed a significant decrease in the biofilm formation ability of L829, which lacks the outer membrane protein TolC (tolC::cat), and L830, which lacks the inner membrane pump AcrB (acrB::aph) (Figure 1). However, genetic inactivation of acrA [L1271 (acrA::aph)], the periplasmic adapter protein, had no effect on biofilm formation. The biofilm defect of efflux mutants seen in static assays was replicated when tested under flow conditions; all mutants tested were unable to form a mature biofilm under flow. Phase contrast microscopy images from biofilms formed in a flow cell under shear stress showed that L828 (wildtype) formed a mature biofilm and L829 (tolC::cat) and L830 (acrB::aph) were able to adhere as individual cells to the flow cell, but unable to form a mature, three-dimensional biofilm, with both mutants covering ,5% of the surface of the flow cells on average, whereas L828 (wild-type) had covered 99% of the area of the flow cell after 48 h; the difference in area coverage between the wild-type and mutants was statistically significant (Figure 2).
The biofilm deficit of mutants lacking a functional acrB or tolC is not associated with altered aggregative ability or a secretion deficit of any biofilm-promoting factor We hypothesized that the efflux mutants may have a greater ability to aggregate than the wild-type due to increased cell surface hydrophobicity. To determine whether inactivation of AcrAB-TolC had altered the intrinsic aggregative nature of the strains lacking acrB or tolC, a settle assay was used; this showed no significant decrease in the aggregative ability of L829 (tolC::cat) or L830
(c)
Figure 2. Phase contrast microscopy images of (a) L828 (wild-type), (b) L830 (acrB::aph) and (c) L829 (tolC::cat) at ×40 magnification after 24 h of incubation under flow conditions. Bar¼5 mm.
4 of 9
JAC
Efflux and biofilm formation
(a) 140%
L828 (wild-type) L829 (tolC::cat)
120%
L830 (acrB::aph)
* *
Percentage of initial OD600
*
*
*
100%
80% * * 60%
40%
20%
0% 0
1
2
3
4
5
6
24
Time (h) (b)
L828 (wild-type)
L829 (tolC::cat)
L830 (acrB::aph)
1 M (NH4)2SO4
2 M (NH4)2SO4
Figure 3. (a) Settle assay of L828 (wild-type), L829 (tolC::cat) and L830 (acrB::aph). E. coli O42 was used as a positive, aggregative control. Values indicate the percentage of an initial absorbance from readings taken immediately below the surface of the liquid of a broth that was incubated statically over a 24 h period. Asterisks indicate average values significantly different to the wild-type under the same conditions. (b) Salt aggregation test images of L828 (wild-type), L829 (tolC::cat) and L830 (acrB::aph) in 1 and 2 M ammonium sulphate. Aggregation was recorded as formation of a visible precipitate and the lowest concentration of ammonium sulphate to prompt precipitation was recorded for each strain.
(acrB::aph) (Figure 3a). Salt aggregation tests also showed no defect in the mutants’ ability to aggregate. In fact, L829 (tolC::cat) cells aggregated in a lower concentration of ammonium sulphate than L828 (wild-type), showing a slightly greater tendency for cells to aggregate than the wild-type (Figure 3b). We also hypothesized that a soluble biofilm-promoting factor was exported by AcrAB-TolC; therefore, addition of culture supernatant conditioned by growth with L828 (wild-type) should ‘rescue’ the ability of the tolC and acrB mutant strains to form a
biofilm. However, two coincubation assays with wild-type and mutant strains suggested that there is no ‘biofilm factor’ exported by AcrAB-TolC. Transwell assays showed the same poor ability to form a biofilm of the acrB and tolC mutants when incubated alone or coincubated with L828 (wild-type) (Figure 4). In addition, no rescue of the biofilm defect was observed when coincubated with logarithmic- or stationary-phase cultures of L828 (wild-type) (Figure 4). Similarly, biofilm mat assays coinoculated with an equal ratio of wild-type and mutant showed that mutant cells were not
5 of 9
Baugh et al.
Biofilm formation (relative to L828)
1.4 1.2 1 0.8 0.6 0.4 0.2 0
(b)
1.4
Biofilm formation (relative to L828)
(a)
1.2
The lack of curli production in mutants lacking acrB and tolC is due to transcriptional repression * * *
*
1 *
0.8
*
0.6 0.4
*
0.2
*
0 L828
L830 (acrB::aph)
L829 (tolC::cat)
L829 (tolC::cat)
L828 Strain
L830 (acrB::aph)
+ wild-type supernatant
Figure 4. Transwell assays showed no rescue of biofilm formation by mutants when coincubated with wild-type (L828). (a) Strains incubated with and without the presence of L828 inoculated at the same density as the mutants. (b) The same experiment, but with coincubation with an overnight, undiluted culture of L828. Asterisks indicate average values significantly different to the wild-type under the same conditions.
L828 (wild-type)
L829 (tolC::cat)
L830 (acrB::aph)
L1271 (acrA::aph)
csgG
csgF
csgE
csgD
csgB
csgA
csgC
Key (average expression relative to wild-type): 76%–100% of wild-type expression
51%–75% of wild-type expression
26%–50% of wild-type expression
0%–25% of wild-type expression
Figure 5. Transcriptional expression of curli genes. A schematic of the curli biosynthesis genes, with average expression values determined by comparative RT–PCR used to shade each gene. Where expression in the tolC and acrB mutant was ≤50% of the wild-type, all these values were statistically significantly different using Student’s t-test (P,0.05).
present in any of the biofilm mats formed, whereas the corresponding planktonic culture comprised an equal mixture of mutant and wild-type cells (data not shown).
6 of 9
Congo red-supplemented agar and Congo red staining of bacterial suspensions showed qualitatively and quantitatively that curli expression is repressed in the tolC and acrB mutants and produced at wild-type levels in the acrA mutant.4 This lack of curli production was found to result from significant transcriptional repression of all functional genes in the curli biosynthetic loci, as measured by comparative RT– PCR. The structural curli genes were both downregulated in the acrB and tolC mutants; csgA was down-regulated 2.5- and 4-fold, respectively, and csgB was down-regulated 3-fold in both mutants. The master regulator, csgD, was down-regulated 10-fold in both mutants. The three genes required for efficient assembly and translocation of the structural subunits were also all down-regulated to varying degrees: csgE was down-regulated 1.5fold in both mutants; csgF was down-regulated by 2- and 4-fold in the acrB and tolC mutants, respectively; and csgG was also down-regulated by 5- and 10-fold in the acrB and tolC mutants, respectively. Repression of these genes (bar csgD) was not observed, however, in the acrA (Figure 5) mutant, correlating with the phenotypic curli morphologies on Congo red.
Inhibition of efflux is effective in preventing biofilm formation by various species under static and flow conditions We previously showed that inhibitors of efflux were able to prevent Salmonella Typhimurium 14028S from forming a biofilm under static conditions.4 To assess whether inhibition of efflux has impact as an antibiofilm approach in other bacterial species, we evaluated the ability of the same three inhibitors as in our previous study4 (chlorpromazine, CCCPand PAbN) to prevent biofilm formation by representative strains of E. coli (MG1655), P. aeruginosa (PA01) and S. aureus (NCTC 8532), species where biofilm-associated infections are problematic. These inhibitors act in different ways; therefore, we hypothesize that any antibiofilm effect common to all species investigated is due to efflux inhibition rather than drug-specific effects. In the static crystal violet assay, all three inhibitors significantly reduced biofilm formation (P, 0.05) by all three species, although the concentration of each inhibitor that provoked the greatest effect varied between species (Figure 6). S. aureus was most susceptible to CCCP, with 2 mg/L causing a 5-fold reduction in biofilm formation. PAbN was the inhibitor with the greatest impact on E. coli, with 16 mg/L preventing formation of a normal biofilm. P. aeruginosa was least susceptible to the inhibitors, with 128 mg/L CCCP and 512 mg/L of the other inhibitors needed to prevent biofilm formation. Irrespective of the species, biofilm inhibitory concentrations were all below the MIC of each compound under the test conditions, demonstrating a specific antibiofilm effect rather than a generalized growth inhibition. Under flow conditions, after an initial attachment period, the inhibitors also prevented biofilm formation. Bacterial coverage of the flow cell by S. aureus and biofilm formation was severely reduced or absent after addition of CCCP (4 mg/L, resulted in a 6-fold reduction in coverage of the flow cell), chlorpromazine (64 mg/L, resulted in a 6-fold reduction in coverage of the flow cell) and PAbN (64 mg/L, resulted in a 500-fold reduction in coverage of the flow cell) (Figure 7). Similar results were obtained for both E. coli and P. aeruginosa (data not shown).
JAC
Efflux and biofilm formation
S. aureus
E. coli
150
P. aeruginosa
150
100
150
100 *
CCCP
* 50
*
* *
50
*
*
100
*
50 *
PAbN
CPZ
Biofilm formed (relative to no drug control)
* 0
0 0
0.5
1
2
4
8
150
0
1
2
4
8
16
*
*
*
8
16
32
64
128
100
100 *
* *
50
0
150
150
100
0
*
*
50
*
50
* *
0
0
0 0
16
32
64
128 256
150
0
4
8
16
32
64
150 *
100
*
0
*
*
16
32
64
256
128 256
512
*
100
* 50
50
*
* 0
0 0
128
* *
50
64
150
100
*
0
0
8
16
32
64
128
0
64
128
256
512
Drug concentration (mg/L) Figure 6. Impact of efflux inhibitors on the ability of S. aureus, P. aeruginosa and E. coli to form a biofilm in the crystal violet assay. Various efflux inhibitor concentration ranges were used, depending on the efficacy of the inhibitor against the species. Bars show the average biofilm formed by ≥24 replicate samples at each drug concentration after 48 h relative to untreated controls. Error bars indicate standard deviations. Asterisks indicate a significant decrease in biofilm formation when compared with the efflux inhibitor-free control using Student’s t-test (P, 0.05). CPZ, chlorpromazine.
Discussion Multidrug efflux pumps have a central role in the biology of bacteria, with roles in drug resistance, cell division, pathogenicity and, as recently described, the formation of biofilms.4,5 Here, we investigated the mechanism for the inability of mutants lacking AcrB and TolC, constituents of the major AcrAB-TolC system of Enterobacteriaceae, to form a competent biofilm. Mutants of Salmonella lacking a functional TolC or AcrB were unable to form biofilms under various conditions and this was not related to any defect in growth, cellular hydrophobicity/aggregative ability or export of a biofilm-promoting substrate. Surprisingly, a mutant lacking a functional AcrA (but still expressing AcrB)12 was not defective in its ability to form a biofilm. Loss of acrA has previously been shown to result in hypersusceptibility to various drugs and a decreased ability to attach to and invade epithelial cells in tissue culture.12 The biofilm phenotype of an acrA mutant was,
however, distinct from that of mutants lacking acrB.12 One major difference between acrA and acrB/tolC mutants is the expression of curli genes, repressed in the acrB and tolC mutants, but not the acrA mutant.13 This led us to hypothesize that curli repression was the reason that the efflux mutants did not form a biofilm and we show here the absence of curli production in mutants lacking tolC or acrB. These data suggest that in the absence of efflux there is strict repression of all curli biosynthetic genes. The regulation of curli expression is extremely complex, with multiple pathways known to impact curli production. Amongst these pathways are two-component systems that respond to membrane stress (cpxRA, rcsCB and envZ/ompR), rpoE and the lytic transglycosylases mltC and mltE. 14 All these systems have membrane-bound or periplasmic components, suggesting that curli synthesis is sensitive to changes in the membrane. To determine whether a lack of biofilm formation in mutants lacking efflux pumps was Salmonella specific or a general
7 of 9
Baugh et al.
(a)
(b)
(c)
(d)
Figure 7. Phase contrast microscopy images of glass flow cells after incubation with S. aureus for 16 h under flow conditions at×40 magnification. (a) No inhibitor, (b) CCCP, (c) chlorpromazine and (d) PAbN. Bar¼5 mm.
phenomenon when efflux was reduced, we examined biofilm formation for various different strains representing both Gramnegative and Gram-positive species. We examined biofilm formation under flow and static assays for E. coli, P. aeruginosa and S. aureus with and without efflux inhibitors previously found to be active against Salmonella.4,5 All the inhibitors tested prevented biofilm formation in both static and flow assays. Importantly, in the flow assays, cells were allowed to adhere to the flow cell for 3 h before the addition of inhibitor, and whilst individual cells did attach they were not able to mature and form a threedimensional biofilm in the presence of inhibitors. All concentrations tested were below the MIC of each inhibitor for the specific strains; therefore, the efflux inhibitors are acting as antibiofilm agents and reduction in biofilm formation is not due to an antimicrobial action preventing growth. The prevention of formation of a mature biofilm in all species indicates that repression of matrix components essential for biofilm maturation is likely to be the mechanism by which efflux inhibition prevents biofilm formation, supporting the Salmonella model where curli repression appears to be central to the mechanism. It has recently been shown that curli genes are more conserved across species than first thought; therefore, curli
8 of 9
is important not just for Enterobacteriaceae, but also for a variety of different bacteria.15 Efflux inhibitors have previously been shown to decrease biofilm formation in P. putida and S. aureus; the efflux inhibitors used in that study were thioridazine, NMP and PAbN.5 The one direct comparison to this study that can be drawn relates to the effect of PAbN on the formation of S. aureus biofilms: Kvist et al. 5 showed that 20 mg/L PAbN reduced biofilm formation to 60%. Our study corroborates this finding. Although the concentrations are slightly different, we found that both 16 and 32 mg/L PAbN also reduced biofilm formation of S. aureus to 60% in a static assay. A greater reduction was, however, seen in assays under flow conditions, indicating that different conditions may influence the antibiofilm efficacy of efflux inhibitors. The efflux inhibitors used in this study are not clinically useful for systemic treatment and some have relatively high toxicity;16 however, there are a number of efflux inhibitors that belong to classes of successful drugs used in human medicine and there is no reason why inhibitors of bacterial multidrug efflux systems should be generically toxic to humans. Whilst the most effective inhibitor varied with the species tested, the inhibition of efflux is a promising novel antibiofilm
Efflux and biofilm formation
strategy and is effective under a range of conditions. The potential for the impregnation of medical devices and other surfaces with efflux inhibitors that can be colonized by biofilms and represent a source of infection warrants further evaluation.
Funding This work was supported by the Medical Research Council via a doctoral training award to M. A. W. used to support S. B.
Transparency declarations None to declare.
References 1 Costerton JW, Cheng KJ, Geesey GG et al. Bacterial biofilms in nature and disease. Ann Rev Microbiol 1987; 41: 435–64. 2 Lewis K. Persister cells, dormancy and infectious disease. Nat Rev Microbiol 2007; 5: 48– 56. 3 Watnick P, Kolter R. Biofilm, city of microbes. J Bacteriol 2000; 182: 2675– 9. 4 Baugh S, Ekanayaka AS, Piddock LJV et al. Loss of any of the multidrug efflux pumps of Salmonella Typhimurium results in a loss of ability to form a biofilm. J Antimicrob Chemother 2012; 67: 2409 –17. 5 Kvist M, Hancock V, Klemm P. Inactivation of efflux pumps abolishes bacterial biofilm formation. Appl Environ Microbiol 2008; 74: 7376–82. 6 De Kievit TR, Parkins MD, Gillis RJ et al. Multidrug efflux pumps: expression patterns and contribution to antibiotic resistance in Pseudomonas aeruginosa biofilms. Antimicrob Agents Chemother 2001; 45: 1761– 70.
JAC 7 Ramage G, Bachmann S, Patterson TF et al. Investigation of multidrug efflux pumps in relation to fluconazole resistance in Candida albicans biofilms. J Antimicrob Chemother 2002; 49: 973– 80. 8 Gerstel U, Park C, Romling U. Complex regulation of csgD promoter activity by global regulatory proteins. Mol Microbiol 2003; 49: 639– 54. 9 Nishino K, Latifi T, Groisman EA. Virulence and drug resistance roles of multidrug efflux systems of Salmonella enterica serovar Typhimurium. Mol Microbiol 2006; 59: 126– 41. 10 Ebel-Tsipis J, Fox MS, Botstein D. Generalized transduction by bacteriophage P22 in Salmonella typhimurium. II. Mechanism of integration of transducing DNA. J Mol Biol 1972; 71: 449–69. 11 Benoit MR, Conant CG, Ionescu-Zanetti C et al. New device for high-throughput viability screening of flow biofilms. Appl Environ Microbiol 2010; 76: 4136– 42. 12 Blair JM, La Ragione RM, Woodward MJ et al. Periplasmic adaptor protein AcrA has a distinct role in the antibiotic resistance and virulence of Salmonella enterica serovar Typhimurium. J Antimicrob Chemother 2009; 64: 965– 72. 13 Bailey AM, Paulsen IT, Piddock LJ. RamA confers multidrug resistance in Salmonella enterica via increased expression of acrB, which is inhibited by chlorpromazine. Antimicrob Agents Chemother 2008; 52: 3604– 11. 14 Monteiro C, Fang X, Ahmad I et al. Regulation of biofilm components in Salmonella enterica serovar Typhimurium by lytic transglycosylases involved in cell wall turnover. J Bacteriol 2011; 193: 6443– 51. 15 Dueholm MS, Albertsen M, Otzen D et al. Curli functional amyloid systems are phylogenetically widespread and display large diversity in operon and protein structure. PLoS One 2012; 7: e51274. 16 Lomovskaya O, Warren MS, Lee A et al. Identification and characterization of inhibitors of multidrug resistance efflux pumps in Pseudomonas aeruginosa: novel agents for combination therapy. Antimicrob Agents Chemother 2001; 45: 105–16.
9 of 9
