Pacing across the membrane: the novel PACE family of efflux pumps is widespread in Gram-negative pathogens

The proteobacterial antimicrobial compound efflux (PACE) family of transport proteins was only recently described. PACE family transport proteins can confer resistance to a range of biocides used as disinfectants and antiseptics, and are encoded by many important Gram-negative human pathogens. However, we are only just beginning to appreciate the range of functions and the mechanism(s) of transport operating in these proteins. Genes encoding PACE family proteins are typically conserved in the core genomes of bacterial species rather than on recently acquired mobile genetic elements, suggesting that they confer important core functions in addition to biocide resistance. Three-dimensional structural information is not yet available for PACE family proteins. However, PACE proteins have several very highly conserved amino acid sequence motifs that are likely to be important for substrate transport. PACE proteins also display strong amino acid sequence conservation between their N— and C-terminal halves, suggesting that they evolved by duplication of an ancestral protein comprised of two transmembrane helices. In light of their drug resistance functions in Gram-negative pathogens, PACE proteins should be the subject of detailed future investigation.


Introduction
In the broadest sense, drug resistance may arise in actively growing bacterial cells in two distinct ways, either the drug target site is protected from the toxic activities of the drug by modification or bypass, or the drug can't reach the target site due to degradation, sequestration, reduced cellular entry, or active efflux.Efflux is a major mechanism of drug resistance, and due to the high promiscuity in substrate recognition by the transport proteins involved, efflux mediated resistance is found for a wide range of different antimicrobial compounds.
Bacterial drug efflux proteins from five distinct families of transport proteins were described between the 1970s and 2000 and have been studied extensively at both the functional and structural levels [7].These families include the ATP-binding cassette (ABC) superfamily, the major facilitator superfamily (MFS), the resistance-nodulation-cell division (RND) superfamily, the multidrug/oligosaccharidyl-lipid/polysaccharide (MOP) flippase superfamily, and the drug/metabolite transporter (DMT) superfamily.In the last five years two new transporter families that include bacterial drug efflux systems have been identified; these are the Proteobacterial Antimicrobial Compound Efflux (PACE) family and the p-Aminobenzoylglutamate transporter (AbgT) family [12,13,22].Proteins from the PACE family transport biocides such as chlorhexidine and acriflavine, whereas AbgT family transporters transport sulphonamides.

The Acinetobacter baumannii AceI protein is a prototype for the novel PACE family of transport proteins
Drug efflux systems, and drug resistance factors in general, are frequently controlled by regulators that can sense the transported drug substrates or their downstream effects in the cell.
For example, TetR controls expression of the tetB tetracycline transporter gene in response to tetracyclines, and QacR controls expression of the multidrug efflux pump gene qacA in response to cationic antimicrobials [10].For bacterial cells, this regulatory control means that efflux pump gene expression will proceed only when the pumps are required, saving cellular resources and preventing the potential toxic effects of constitutive high-level efflux pump expression [1].From a research perspective, this tight regulatory control of drug efflux pump genes means that transcriptional changes may be used to highlight efflux pumps that might recognise substrates of interest, or to identify novel factors that may be involved in drug resistance.
The Acinetobacter chlorhexidine efflux protein (AceI) was identified by analysing the transcriptomic response of Acinetobacter baumannii to the membrane active biocide chlorhexidine [12,11].Chlorhexidine is listed as an essential medicine by the World Health Organisation, and is commonly used as an antiseptic in wound dressings, hand washes and mouthwashes.The transcriptome of A. baumannii ATCC 17978 cells exposed to a subinhibitory concentration of chlorhexidine, equivalent to half the minimum inhibitory concentration, was compared to control cells.The major gene expression changes were to genes encoding the AdeAB components of the AdeABC multidrug efflux pump and a gene annotated as encoding a hypothetical protein, A1S_2063 [12].
From its sequence the A1S_2063 gene was predicted to encode an inner-membrane protein with four transmembrane helices (Figure 1A).The gene was cloned into an E. coli expression vector and was shown to confer significant levels of resistance to chlorhexidine when overexpressed in E. coli.Deletion of the A1S_2063 gene in A. baumannii ATCC 17978 and its ortholog in Acinetobacter baylyi ADP1, resulted in a halving of chlorhexidine resistance in the host strain, demonstrating the genes had a resistance function in native hosts [12,23].The Biolog Phenotype Microarray system was used to determine whether the A1S_2063 gene could confer resistance to more than 200 other antimicrobials.These assays demonstrated an apparent specificity of A1S_2063 for chlorhexidine [12].
When overexpressed in E. coli the A1S_2063 protein product was identified in the inner membrane.The protein could be readily extracted from the membrane by detergent solubilisation and purified.The detergent solubilised protein was found to bind to chlorhexidine with high affinity (Kd in the low M range) as determined by tryptophan fluorescence quenching and near-UV synchrotron radiation circular dichroism [12].Transport experiments using [ 14 C]-chlorhexidine demonstrated that the A1S_2063 protein prevented the high-level accumulation of chlorhexidine when expressed in E. coli, until the cells were de-energised using a protonophore, and could mediate the efflux of chlorhexidine from E. coli cells pre-loaded with chlorhexidine [12].Together these results suggested that the A1S_2063 protein was a novel chlorhexidine efflux protein, which was named AceI (Acinetobacter chlorhexidine efflux protein I).

PACE proteins are a family of multidrug efflux systems conserved across many Gramnegative pathogens
Genes encoding proteins homologous to AceI are found in the genomes of many bacterial species.These genes are particularly common among Proteobacteria, but can be found in some Actinobacteria and in a limited number of other unrelated bacterial species.To determine whether, like AceI, these proteins can mediate chlorhexidine resistance, more than 20 phylogenetically diverse homologs were cloned into an E. coli expression system and examined by routine minimum inhibitory concentration analyses.Most of the cloned proteins were expressed at detectable levels, and about half could confer resistance to chlorhexidine [13].
Notably, at least two of the aceI homolgs found to confer chlorhexidine resistance, are also highly expressed in their native hosts, Pseudomonas aeruginosa and Burkholderia cenocepacia, in response to chlorhexidine treatment [17,5].
Additional resistance tests were performed to determine whether the antimicrobial recognition profiles of these homologs might extend beyond chlorhexidine.Many of the proteins were able to confer resistance to several additional biocides, including acriflavine, proflavine, benzalkonium and dequalinium [13].The substrate profile of one pump, VP1155 encoded by Vibrio parahaemolyticus, was investigated using the Biolog phenotype microarray system.In addition to chlorhexidine, benzalkonium, proflavine and acriflavine, this analysis suggested that VP1155 could confer resistance to 9-aminoacridine, domiphen bromide, guanazole and plumbagin [13].
The demonstration that many AceI homologs are able to confer resistance to compounds such as proflavine and acriflavine presented the possibility of assaying transport by measuring their fluorescence in real time [24].These compounds intercalate into nucleic acids, which leads to a quenching of their fluorescence.This property facilitates a convenient assay for their transport in cells expressing an efflux pump [2].Cells expressing the protein of interest can be loaded with proflavine or acriflavine in the presence of a protonophore, such as carbonyl cyanide mchlorophenylhydrazone (CCCP), then washed and re-energised by the addition of an energy source, such as D-glucose.Fluorescence can be monitored before and after energisation to examine transport [24].These transport experiments have been performed for a number of AceI homologs and identified proteins that mediate transport of these compounds.For example, the B. cenocepacia HI2424 homolog Bcen2424_2356 is able to transport acriflavine, whereas at least one other homolog encoded by this strain, Bcen2424_5347, does not (Figure 2).
Bcen2424_2356 has been previously shown to confer resistance to chlorhexidine, benzalkonium, proflavine and acriflavine.The Biolog phenotype microarray antimicrobial resistance tests confirmed several of these phenotypes and suggested that Bcen2424_2356 also confers resistance to benzethonium, 9-aminoacridine, methyl viologen, guanazole and plumbagin (Supplemental Figure S1).
The observation that several AceI homologs can confer resistance to multiple biocides, and can mediate transport of the fluorescent substrates, proflavine and acriflavine, lead to their designation as a new family of efflux pumps.This family was called the Proteobacterial Antimicrobial Compound Efflux (PACE) family, due to their abundance in Proteobacteria [13].
Proteins from this family have been incorporated into the Transporter Classification Database [20] under the original family title, the Proteobacterial chlorhexidine efflux (CHX) family (TCDB number: 2.A.117), and are captured in the TransportDB 2.0 database [8], which catalogues all putative transport proteins from sequenced genomes in the NCBI RefSeq database.

Predicted topology and sequence conservation in PACE pumps
All PACE family proteins analysed to date are predicted to contain four transmembrane helices, organised into two tandem bacterial transmembrane pair (BTP) domains (Figure 1; pfam: PF05232) [9].Given their small size, it seems very likely that PACE proteins function as oligomers.However, the oligomeric state of PACE family proteins remains unresolved.
Several PACE family proteins have been experimentally characterised by overexpression and purification (Henderson et al., unpublished).When expressed in E. coli these proteins localise to the inner-membrane and can be readily purified by extraction with a mild detergent such as n-dodecyl--D-maltoside [12], or using styrene maleic acid co-polymer (Supplemental Figure S2) [15].Analysis of the purified detergent-solubilised proteins by far-UV circular dichroism has confirmed their high -helical content and demonstrated that they typically show structural stability to around 50-60 C.
A high level of amino acid sequence conservation is apparent between members of the PACE family (Supplemental Figure S3).Two amino acid residues appear to be universally conserved across these proteins, a glutamic acid residue within transmembrane helix 1 and an alanine residue at the periplasmic/membrane boundary of transmembrane helix 4 (bold upper case font in Figure 1A).The functional importance of the conserved alanine has not yet been investigated, but neutralisation of the glutamic acid residue in the prototypical PACE family member AceI by substitution with a glutamine abolished chlorhexidine resistance and transport [12].
However, this mutant (E15Q) was still able to bind chlorhexidine with only slightly reduced affinity compared to the parental protein.Furthermore, the mutant protein was less thermostable than the parental protein in the absence of substrate, but was significantly more stable than the parental protein in the presence of a molar excess of chlorhexidine.These results suggested that the glutamic acid residue is involved in an aspect of transport unrelated to substrate binding, possibly a proton coupling reaction.PACE family proteins contain several highly conserved amino acid residues in addition to the two universally conserved residues.The amino acid sequence conservation is particularly strong close to the predicted cytoplasmic boundaries of the transmembrane helices, where four amino acid sequence motifs have been identified (Figure 1).In line with the PACE proteins containing tandem BTP domains, the amino acid sequence motif in transmembrane helix 1 (motif 1A; RxxhaxxfE, where upper case residues are conserved in more than 90% of proteins and lower case residues in at least 65% of proteins) is very similar to that in transmembrane helix 3 (motif 1B; RxxHaxxFe) (Figures 1B and 1C), and the motif in transmembrane helix 2 (motif 2A, WNxxy/fNxxFd) is very similar to that in transmembrane helix 4 (motif 2B; Yxxxf/ynwxyD) (Figures 1D and 1E).The notable features of the sequence motifs in helices 1 and 3 are the membrane embedded glutamic acid residue (universally conserved in helix 1), and histidine and arginine residues at the membrane boundary.The motifs found in helices 2 and 4 notably contain several aromatic residues along one helical face adjacent to polar asparagine residues, and an aspartate residue at the membrane boundary (Figure 1A).
Based on the distribution of charged residues within the loop regions, the N-and C-termini of most PACE family proteins are predicted to lie within the cytoplasm (Figure 1).However, some PACE family homologs, primarily from Acetobacter, contain predicted N-terminal signal sequences, suggesting that the N-terminus is moved across the cytoplasmic membrane, and that they may exist in an alternative topology, e.g., APA01_04520 and APO_1949 from A. pasteurianus IFO 3283-01 and A. pomorum DM001, respectively.Representatives of these proteins have been expressed in E. coli, but as yet, no resistance or transport functions have been identified (Hassan et al., unpublished).These proteins may be defined in a separate protein sub-family from those that mediate drug resistance in the future.

Conservation of PACE family genes
PACE family proteins are typically highly conserved in the genome of their encoding bacterial species.For example, genes encoding three different PACE proteins have been identified in the A. baumannii pan-genome (based on the genomes of 623 strains) [11].Of these two were conserved in 100% or close to 100% of the strains and can be considered to be part of the core genome.The third gene was found in only two strains and is part of the accessory genome.

Similar to A. baumannii, Pseudomonas aeruginosa isolates have two PACE proteins encoded
in the core genome and one in the accessory genome, which is found in only a few strains, and B. cenocepacia strains encode three PACE pumps in their core genome [11].This high level of conservation suggests that PACE pumps are acquired vertically and have been maintained in their host species since their divergence from related organisms.As such they are likely to have an important core function that may be unrelated to drug resistance.Indeed, the biocides that are recognised by PACE family pumps have only been present in the environment for 50-100 years, and are thus very unlikely to be the physiological substrates of these proteins.
In contrast to the species described above E. coli do not encode PACE pumps in their core genomes; four different genes encoding PACE homologs were found among the genomes of 1986 sequenced E. coli strains, but these were each found in 0.2% of strains or less [11].These accessory genes are likely to move between related species on mobile genetic elements.However, there is as yet no strong evidence for how these genes are mobilised.

Evolution of the PACE family
The conservation of sequence motifs between the N-and C-terminal halves of PACE proteins suggests that these proteins may have evolved by a duplication event of an ancestral single BTP domain protein.To investigate this further, the N-and C-terminal BTP domains were compared between 48 diverse PACE family proteins (Supplemental Figure S4).The level of amino acid sequence similarity between the N-and C-terminal BTP domains in these proteins ranged from 26% to 57% (mean 47%).The presence of such high levels of sequence conservation between the N-and C-terminal BTP domains across diverse PACE family proteins suggests that these proteins have not diverged significantly since the occurrence of the duplication event(s).Along with the distribution of these proteins almost exclusively within the Proteobacteria, and their likely vertical acquisition, due to their presence on the core genome, this may suggest that this protein family is relatively young compared to other families of transport proteins, which show lower levels of sequence conservation between domains that are thought to have arisen via duplication [19].
To examine further the evolution of PACE family proteins, the levels of sequence similarity between the N-terminal and C-terminal BTP domains of different PACE proteins were determined.It was found that the N-terminal BTP domains of PACE family proteins are almost always more similar to the N-terminal BTP domains of other PACE proteins than they are to their own C-terminal BTP domain, or the C-terminal BTP domain of other PACE family pumps (Supplemental Figure S4).This suggests that a BTP domain duplication event occurred only once in an ancestral gene, and that there is little or no recombination between the N-and Cterminal BTP domains in individual strains.The C-terminal BTP domains of different PACE pumps typically show even higher levels of sequence similarity than the N-terminal domains (Supplemental Figure S4).The high conservation of sequence within the C-terminal domain of different proteins may reflect the involvement of the C-terminal domain in a core part of the functional mechanism.

Concluding remarks
The PACE family of transport proteins is one of two transporter families discovered only recently to mediate drug efflux.From currently available analyses PACE family proteins display somewhat restricted drug substrate recognition profiles, which include primarily synthetic biocides such as chlorhexidine and acriflavine, rather than the multitudes of diverse antibiotics and biocides recognised by transport proteins from families such as the RND superfamily.This may be a primary reason for the family being only recently identified, 15 years after the first descriptions of MATE family pumps [16,3].However, PACE proteins are highly conserved in a range of opportunistic Gram-negative pathogens, including A. baumannii, P. aeruginosa, B. cenocepacia and Klebsiella pneumoniae, and in serious human pathogens such as Yersinia pestis, Francisella tularensis, and Burkholderia pseudomallei.Therefore, the role of these proteins in drug resistance warrants future investigation.
As mentioned above the drug recognition profile of PACE pumps includes primarily synthetic biocides, most of which have only been in the environment for 50-100 years.However, genes encoding homologous PACE family proteins are found in the core genomes of bacterial genera that diverged much earlier than this, hundreds of millions of years ago.Therefore, these proteins are likely to mediate an important core function and may have common physiological substrates that are yet to be described.The importance of PACE family proteins is likely to extend beyond an apparently fortuitous role in drug resistance.Sequence logos, made using WebLogo [6], showing conservation of amino acid residues in the four sequence motifs identified in PACE proteins.Fbal_3166 [13], using styrene maleic acid co-polymer.Fbal_3166 protein was overexpressed in E. coli BL21 cells grown in a 30 L fermenter using the pTTQ18 expression system [25,21].
Styrene maleic acid co-polymer preparation, membrane solubilisation and Ni-affinity purification were performed as previously described [15].Samples consisting of solubilised membrane proteins (lane A), proteins that did not bind to the Ni-affinity column (lane B), and purified Fbal_3166 (lane C) were run on a 15% SDS-PAGE gel and stained with coomassie brillian blue R-250.The size (KDa) of co-migrated soluble molecular weight markers is show on the left side of the gel.Figure S3.Amino acid sequence alignment of 48 diverse PACE family proteins.Sequences were obtained from the NCBI genomes database and aligned using ClustalX [14].The alignment is coloured according to the level of amino acid sequence conservation at each position, colours were added using the UGENE toolkit [18].

Figure 1 .
Figure 1.Predicted transmembane topology and conserved amino acid sequent motifs present

Figure 2 .
Figure 2. Acriflavine transport mediated by PACE family proteins encoded by the human

Figure S4 .
Figure S4.Pairwise comparisons of the individual BTP domains in 48 diverse PACE family