|Home | About | Journals | Submit | Contact Us | Français|
Argyrins are natural products with antibacterial activity against Gram-negative pathogens, such as Pseudomonas aeruginosa, Burkholderia multivorans, and Stenotrophomonas maltophilia. We previously showed that argyrin B targets elongation factor G (FusA). Here, we show that argyrin B activity against P. aeruginosa PAO1 (MIC = 8 μg/ml) was not affected by deletion of the MexAB-OprM, MexXY-OprM, MexCD-OprJ, or MexEF-OprN efflux pump. However, argyrin B induced expression of MexXY, causing slight but reproducible antagonism with the MexXY substrate antibiotic ciprofloxacin. Argyrin B activity against Escherichia coli increased in a strain with nine tolC efflux pump partner genes deleted. Complementation experiments showed that argyrin was effluxed by AcrAB, AcrEF, and MdtFX. Argyrin B was inactive against Acinetobacter baumannii. Differences between A. baumannii and P. aeruginosa FusA proteins at key residues for argyrin B interaction implied that natural target sequence variation impacted antibacterial activity. Consistent with this, expression of the sensitive P. aeruginosa FusA1 protein in A. baumannii conferred argyrin susceptibility, whereas resistant variants did not. Argyrin B was active against S. maltophilia (MIC = 4 μg/ml). Spontaneous resistance occurred at high frequency in the bacterium (circa 10−7), mediated by mutational inactivation of fusA1 rather than by amino acid substitutions in the target binding region. This strongly suggested that resistance occurred at high frequency through loss of the sensitive FusA1, leaving an alternate argyrin-insensitive elongation factor. Supporting this, an additional fusA-like gene (fusA2) is present in S. maltophilia that was strongly upregulated in response to mutational loss of fusA1.
Natural products have been and will continue to be an important resource for the identification of new antibacterial compounds and, correspondingly, the bacterial targets of the compounds (1,–3). The argyrins are natural cyclic peptides produced by myxobacteria and actinomycetes (Fig. 1) (4,–7). They have an intriguing antibacterial spectrum of activity, with modest potency against the intrinsically drug-resistant organisms Pseudomonas aeruginosa, Burkholderia multivorans, and Stenotrophomonas maltophilia, but not against other Gram-negative bacteria tested, such as Escherichia coli and Salmonella enterica serovar Typhimurium, unless the cells' outer membrane (OM) permeability barriers are compromised (6,–8). In P. aeruginosa, cells defective in the OM permeability barrier were also much more susceptible to the antibacterial effects of argyrins (8), consistent with the OM permeability barrier being an important factor in determining intrinsic susceptibility to this class of natural-product antibacterial. Argyrins were known for some time to inhibit bacterial protein synthesis and were also shown to have activity against eukaryotic cells. They were reported to be immunosuppressive and antitumorigenic, and it was suggested that argyrin A inhibits the proteasome, induces apoptosis, and blocks angiogenesis by a p27-dependent mechanism (4, 9). More recently, we have shown that the specific cellular target of argyrin in P. aeruginosa cells is elongation factor G (EF-G) (FusA1) (8), a result confirmed by other researchers (10). We also showed that a cellular target of argyrins in eukaryotic cells is mitochondrial elongation factor G (8). Elongation factor G is also inhibited by the antibiotic fusidic acid used clinically; however, the P. aeruginosa argyrin B costructure showed that argyrin B binds FusA1 in a binding pocket distinct from that of fusidic acid and therefore inhibits FusA1 by a novel mechanism (8). Some bacteria possess more than one gene with homology to elongation factor G. For example, the P. aeruginosa genome possesses genes designated fusA1 and fusA2 (PA4266 and PA2071, respectively) (11). Similarly, B. multivorans and S. maltophilia possess two genes designated elongation factor G, whereas Acinetobacter baumannii has only one fusA, underscoring potentially significant differences between various bacteria as regards this particular target. Although natural products can be a rich resource for the discovery of new antibacterials, it is also important to evaluate and understand determinants of the antibacterial spectrum and resistance potential for these compounds. Our recent identification of the specific cellular target of argyrins has allowed us to more fully explore these factors for these novel antibiotics across different Gram-negative pathogens. Here, we have investigated their susceptibility to efflux in P. aeruginosa and E. coli, induction of the MexXY efflux pump expression through the inhibition of protein synthesis, and the impact of natural FusA target sequence variability and genetic organization in determining the spectrum and resistance propensity.
P. aeruginosa is noted for its intrinsic resistance to antimicrobial compounds, mediated in large part by a combination of outer membrane impermeability and active efflux (12, 13). We previously showed that the drug-hypersensitive P. aeruginosa Z61 mutant strain, which has a defective outer membrane barrier and a mutation in oprM (encoding the outer membrane channel for MexAB-OprM and MexXY-OprM), was much more susceptible to argyrin B than its parent (8). However, P. aeruginosa strain NB52023 (Table 1), which lacks both MexAB-OprM and MexXY-OprM functions, was not substantially more susceptibility to argyrins A and B than the parent strain, NB52019 (Table 2). Therefore, neither the MexAB-OprM nor the MexXY-OprM pump, which is inducible by inhibition of protein synthesis (14), appeared to impact intrinsic resistance to argyrin alone or together. As expected, the pump substrate control antibiotic tetracycline was much more active against strain NB52023, confirming the lack of efflux in the strain. To further examine argyrin recognition by efflux pumps, a strain lacking MexAB-OprM, MexXY-OprM, and MexCD-OprJ and with mexEF-oprN silent (NB52245), was used to make strains overexpressing each of the pumps individually. The lack of these pumps in strain NB52245 did not appreciably affect susceptibility to argyrin B, nor did the individual overexpression of MexAB-OprM, MexXY-OprM, MexCD-OprJ, or MexEF-OprN (Table 2). This suggests that argyrin B is not well recognized by any of the main RND pumps. Since the genome of P. aeruginosa encodes at least 12 RND family efflux pumps (including the four examined here), it is possible that one or more of the less well-characterized efflux pumps in P. aeruginosa extrudes argyrin B and therefore would need to also be inactivated to detect efflux. Nonetheless, and unlike a majority of antibacterials (13), argyrin B is not obviously impacted by the four pumps examined here, which may partly account for the ability of argyrin B to accumulate in P. aeruginosa cells in sufficient amounts to engage its target and inhibit growth. This also further supports the role of the outer membrane in limiting the intrinsic activity of argyrin B against P. aeruginosa based on the high susceptibility of the Z61 strain, which has an additional membrane defect (8).
MexXY efflux pump expression is induced by inhibition of protein synthesis (15, 16). We asked whether argyrin, which has a novel mechanism of action for inhibition of protein synthesis (8), would induce MexXY pump expression. Argyrins induced light emission from the P. aeruginosa mexX::luxCDABE reporter fusion strain CDR4 (17), similar to the control antibiotic tetracycline (Fig. 2). There were circa 23-fold and 8-fold increases in mexX and mexY transcript titers, respectively, in cells exposed to argyrin B using Affymetrix gene chips (data not shown). Furthermore, there were 16- and 11-fold increases in transcripts for open reading frames (ORFs) PA5470 and PA5471, respectively, consistent with their expression being responsive to ribosome disruption and the involvement of PA5471 (now designated ArmZ) in modulating MexZ repressor function to control MexXY efflux pump expression (18,–21). Therefore, inhibition of FusA1 is another mode of protein synthesis inhibition that causes PA5471-dependent induction of MexXY. As discussed above, deletion of the mexXY genes (NB52023) or individual overexpression of MexXY did not alter susceptibility to argyrin B, suggesting that, although argyrin B strongly induced the pump, the intrinsic susceptibility of P. aeruginosa to argyrin B is not affected by the pump expression status. Interestingly, we noticed a small but reproducible one-dilution-step antagonism between argyrin B and ciprofloxacin, as tested by checkerboard MIC (data not shown). The antagonism was lost in strains lacking MexXY- but not MexAB-mediated efflux, indicating that the induction of MexXY by argyrin B slightly reduced susceptibility to the MexXY substrate ciprofloxacin. This supported the notion that the pump proteins themselves are induced upon exposure of cells to argyrin and that the effectiveness of a MexXY pump substrate antibiotic may be reduced if the antibiotic is used in combination with a MexXY-inducing compound, such as argyrin B.
Argyrin B did not have potent activity against E. coli (MIC > 64 µg/ml), consistent with previous reports that argyrin was not active against E. coli or Shigella unless the outer membrane was defective, presumably allowing greater access of argyrin to its cytoplasmic target (4, 7). Intriguingly, an E. coli mutant lacking nine genes encoding RND efflux partners of TolC (strain NB27079-CDY0099 [Table 1]) was more susceptible to argyrin B (MIC = 16 μg/ml), indicating that efflux plays a role in determining susceptibility to argyrin in E. coli. Supporting this, complementation of NB27079-CDY0099 in trans with the plasmid pAK1900 harboring acrB, acrF, or mdtF (Table 1) restored nonsusceptibility to argyrin B (MIC ≥ 64 μg/ml), indicating that at least three RND efflux pumps in E. coli appear to extrude argyrin. Although the remaining pump genes provided in trans (acrD, emrB, emrY, entS, macB, and mdtBC) did not restore resistance to strain NB27079-CDY0099, their genomic partner pump genes may not be normally expressed under these growth conditions, so the ability of one or more of them to extrude argyrin B cannot be ruled out. This underscores the fact that a combination of membrane impermeability and efflux by multiple pumps contributes to the low intrinsic susceptibility of E. coli to argyrin B.
Resistant-mutant selection and characterization coupled with cocrystal information from argyrin B-bound FusA1 provided a detailed view of relevant regions of the P. aeruginosa PAO1 FusA1 target vis a vis argyrin B binding (8). The positions of important amino acid substitutions mediating resistance in P. aeruginosa PAO1 FusA1, determined by single-step resistance selection experiments (8), are shown in Table 3 and Fig. S2 in the supplemental material. In addition to these positions, other researchers identified additional residues important for argyrin resistance using P. aeruginosa strain PA14 (10). They were isoleucine 457 (resistance conferred by I457T) and methionine 685 (resistance conferred by M685R) (shown in Fig. S2 in the supplemental material). Both of these amino acids are conserved in FusA1 from PAO1 and are therefore likely important for argyrin binding in either strain. Argyrin B also had moderate activity against another typically antibiotic-resistant Gram-negative pathogen, B. multivorans (8). Similar to P. aeruginosa, the B. multivorans genome has two fusA-like genes, designated fusA1 and fusA2. The B. multivorans FusA1 amino acid sequence is conserved at the residues corresponding to the key interaction sites previously identified in strain PAO1 by mutational studies (Table 3). Consistent with this, argyrin B resistance mutations in B. multivorans were also previously selected in the fusA1 gene at certain of these positions (Table 3) (8). This suggested that, similar to the case with P. aeruginosa, argyrin B can gain entry into B. multivorans and that the conservation of target sequence (susceptibility to argyrin binding) contributes to the cellular activity of argyrin against B. multivorans. In contrast, argyrin B lacked activity against A. baumannii strain NB48062 (MIC > 64 μg/ml) (Table 4). Natural variation in FusA amino acid sequences across different bacteria did not appear to correlate with the antibacterial spectrum of argyrin B (10). However, the genome of A. baumannii NB48062 (ATCC 19606) (Table 1) differs from that of P. aeruginosa in that it encodes only one clearly identifiable FusA protein (22), which has two natural differences at amino acid residues corresponding to those mediating single-step argyrin B resistance in P. aeruginosa (amino acids 417 and 459) (Table 3). We postulated that this contributed to the lack of activity of argyrin B against A. baumannii. To explore this further, we expressed the argyrin-sensitive P. aeruginosa FusA1 protein (FusA1Pa) in A. baumannii (NB48062) using an inducible plasmid to determine if this would confer sensitivity to argyrin B. Induction of FusA1Pa rendered the cells highly susceptible to argyrin B (MIC, 1 μg/ml versus >64 μg/ml) (Table 4). Cells did not become more susceptible to the mechanistically different antibiotics tetracycline, fusidic acid, and ciprofloxacin upon expression of FusA1Pa. Furthermore, expressing the resistant FusA1PaS459S or FusA1PaL663Q variant (previously identified by resistance selection experiments with P. aeruginosa PAO1 ) did not increase susceptibility to argyrin, supporting the notion that the effect on susceptibility engendered by the wild-type FusA1Pa was dependent on argyrin B binding and inhibition. This is further supported by our previous finding that binding of argyrin B to the P. aeruginosa FusA1S459F protein was not detectable using isothermal titration calorimetry or surface plasmon resonance (8). The ability of wild-type, sensitive P. aeruginosa FusA1 protein expression to confer susceptibility to argyrin B in A. baumannii suggests that argyrin B can accumulate sufficiently in A. baumannii cells to engage a sensitive FusA target at levels required to inhibit growth and that natural target sequence variation contributes at least in part to the nonsusceptibility of A. baumannii to argyrin B.
As described above and previously (4, 7, 8), defects in outer membrane permeability increase susceptibility to argyrin B in Gram-negative bacteria. A. baumannii NB48062-LMD0007 (Table 1), which is hyperpermeable due to inactivation of lpxC and the corresponding loss of lipopolysaccharide (LPS) (23), became moderately susceptible to argyrin B (MIC = 8 μg/ml), indicating the target was somewhat susceptible if enough argyrin B could enter the cells. These membrane-defective cells were still much less susceptible than the membrane-defective P. aeruginosa Z61 strain (MIC = 0.25 μg/ml ), perhaps further reflecting the difference in FusA target susceptibility between these two permeability-defective strains. Correspondingly, the argyrin B susceptibilities of mutants selected from the Z61 strain that had an FusA1 S417L or S459F amino acid substitution (MIC = 64 and 8 μg/ml, respectively ) were more similar to that of the permeability-defective A. baumannii NB48062-LMD0007 strain, which has natural differences at these positions. Overall, these observations show that natural target sequence variability, in combination with cell permeability, plays a role in defining the spectrum of activity of argyrins. This is reminiscent of the recently reported spectrum limitation of the natural signal peptidase inhibitor arylomicin, which was also attributed to target sequence variability (24, 25). Intriguingly, the Gram-positive organism Staphylococcus aureus (strain NB01001) (Table 1), which lacks the additional permeability barrier of an outer membrane, was comparatively nonsusceptible to argyrin B (MIC = 128 μg/ml). Similar to the case of A. baumannii, the S. aureus FusA protein has two naturally occurring differences at positions 417 and 459 and an overall identity to P. aeruginosa FusA1 of 58% (26), likely accounting at least partly for this nonsusceptibility. Expression of P. aeruginosa FusA1 in E. coli NB27004 (MG1655) (Table 1) also increased susceptibility to argyrin B (Table 4), although there were no naturally occurring amino acid differences in the E. coli FusA protein from this strain at any of the positions currently known to be resistance determinants (Table 3). Nonetheless, as seen for A. baumannii, expression of resistant variant FusA1PaS459S or FusA1PaL663Q did not alter susceptibility. This suggests that sequence variability at as yet uncharacterized amino acid positions may render E. coli FusA less susceptible to argyrin B than P. aeruginosa FusA1, but a full understanding of this awaits further exploration.
As mentioned above, the genome of P. aeruginosa harbors two fusA genes: PA4266, designated fusA1, and PA2071, designated fusA2 (11). Historic data within our laboratory from multiple transcriptional-profiling experiments using P. aeruginosa PAO1 (NB52019 [Table 1]) revealed that transcript titers of fusA1 were high, typical of many genes encoding translational proteins, whereas the titers of fusA2 were low (data not shown). This would be expected if FusA1 is the primary elongation factor expressed during typical laboratory growth. We and others showed that argyrin B binds the protein and that point mutations in fusA1 encoding amino acid substitutions in the argyrin B binding pocket of FusA1 conferred resistance (8, 10). Recently reported in vitro biochemical data suggested that FusA2 may in fact be the specific elongation factor in P. aeruginosa, while FusA1 functions as the ribosome-recycling factor, and that by extension, the latter function is targeted by argyrin B (27). Given the low level of fusA2 expression, we asked if FusA2 was essential for growth, as would be expected for an elongation factor. We were able to delete fusA2 from the genome of P. aeruginosa PAO1 (NB52019-CDY0245 [Table 1]), so fusA2 appears to be dispensable for cell growth under typical laboratory conditions. Genetic inactivation of fusA2 did not alter susceptibility to argyrin B (data not shown), consistent with FusA1 being the target of argyrin. Overall, this supports the notion that FusA1 is the P. aeruginosa elongation factor during typical laboratory growth or can perform the function in the absence of FusA2 and that it is targeted by argyrin B. However, further work would be required to determine the impact of any argyrin B interaction with FusA2 under conditions where FusA2 might be more highly expressed.
Argyrins have modest activity against S. maltophilia strain NB55011 (MIC = 4 μg/ml) (Table 5). The genome of the organism, like those of P. aeruginosa and B. multivorans, harbors fusA1 and fusA2 (fusA-like) genes (28). Sequencing of S. maltophilia strain NB55011 revealed that FusA1 was conserved at the amino acid residues known to be important for argyrin B binding based on P. aeruginosa resistance selection studies (Table 3), with one exception, a conservative Y683F substitution. In contrast, the FusA-like protein (28), called FusA2 here, had a nonconservative Y683S, as well as two differences at amino acids 417 and 459, known to be important for argyrin B binding (Table 3). Furthermore, the protein also differed at the two additional amino acids I457 and M685 (I457L and M685A substitutions, respectively) found to be important for argyrin B resistance in P. aeruginosa strain PA14 (10). This suggested that the FusA2 protein was nonsusceptible to argyrin B and that FusA1 was the target. Supporting this, S. maltophilia spontaneous single-step resistant mutants selected on argyrin B all had mutations in fusA1 (Table 5). Intriguingly, and in contrast to the point mutations encoding amino acid substitutions identified for P. aeruginosa (8, 10) and B. multivorans (8), there was a preponderance of premature stops or frameshifts (Table 5). Consistent with this, the frequency of mutant selection in S. maltophilia was consistently 10-fold higher (circa 10−7) than for P. aeruginosa or B. multivorans, typical of loss-of-function mutations. Therefore, in S. maltophilia, there had to be an alternative, argyrin B-insensitive FusA protein present and apparently able to function in protein synthesis in the absence of FusA1, allowing the high-frequency mutational loss of FusA1. Overall, this suggested that protein expression in the absence of FusA1 was mediated by FusA2, which was nonsusceptible to argyrin B based on natural amino acid sequence variation. Transcripts of fusA2 were detectable by reverse transcription (RT)-quantitative PCR (qPCR) in wild-type S. maltophilia, suggesting there was some FusA2 present, but at much lower levels than the transcripts for fusA1 (data not shown). However, the fusA2 transcript levels increased dramatically (circa 40- to 70-fold) in fusA1 loss-of-function mutants (Fig. 3). This indicated that FusA2 expression is strongly upregulated in response to an absence of FusA1 function, presumably to replace FusA1. Although one argyrin B-resistant mutant (CDA0071) (Table 5) did possess an amino acid substitution in FusA1 rather than a frameshift or early stop, it was also strongly upregulated for fusA2 (Fig. 3), suggesting it was also a loss-of-function mutation. Therefore, in addition to the naturally occurring resistance of some Gram-negative bacteria caused by target sequence variation, the regulated expression of functionally redundant FusA target proteins and sequence variation between them within S. maltophilia provided a novel, and high-frequency, mechanism by which S. maltophilia populations could escape growth inhibition by argyrins.
The argyrins, in particular argyrin B, have a unique spectrum of activity, with modest potency against certain Gram-negative bacteria that are known to be intrinsically drug resistant. Here, we show that the spectrum of argyrin B across different Gram-negative bacteria is impacted by cell permeability and, to various extents, efflux, as is the case with many antibacterial compounds. However, argyrin B can clearly gain access to some Gram-negative cells at levels necessary to inhibit growth if the cells contain a highly argyrin-sensitive FusA target protein. Correspondingly, reduced sensitivity to argyrin in some bacteria (e.g., A. baumannii) was also mediated by natural amino acid sequence variation in the FusA target. Since argyrins are produced naturally and the FusA protein seemingly tolerates multiple amino acid substitutions in the argyrin binding region, it stands to reason that some environmental selection would occur for resistance to argyrin via target sequence variability. This is reminiscent of the scenario with another class of natural-product antibacterials, the arylomycins, where variability in key residues of the target, signal peptidase, played a role in determining the spectrum of antibacterial activity (24). We also show here that S. maltophilia can become resistant to argyrin B at high frequency via mutational loss of its argyrin-sensitive FusA1, leaving an alternative, functionally redundant, argyrin-insensitive elongation factor (FusA2). Intriguingly, this mechanism, as well as the ability of the sensitive P. aeruginosa FusA1 to confer susceptibility to A. baumannii in the presence of its resistant FusA, strongly suggests that argyrin binds the sensitive FusA and locks it on ribosomes, stalling them and blocking overall protein synthesis, but this awaits further biochemical characterization. We also demonstrate here that the fusA2 gene, and by extension the FusA2 protein, of P. aeruginosa is dispensable for growth, perhaps consistent with its low level of expression under typical growth conditions, suggesting it does not play a role as a primary elongation factor G in the organism under these conditions. Consistent with this, we did not observe loss-of-function mutations in fusA1 conferring resistance to argyrin B (8), as was seen here in S. maltophilia. This suggests that, unlike fusA2, fusA1 is not dispensable in P. aeruginosa and that FusA2 cannot functionally replace FusA1 or, alternatively, is not upregulated if FusA1 is lost in P. aeruginosa. Further work will be required to unravel additional functional relationships between these related genes. Finally, the antibacterial activity of argyrin B allowed us to consider some important differences between bacteria regarding antibacterial targets, even relatively well-validated ones, such as elongation factor G, which is also the target of the natural-product antibiotic fusidic acid used clinically. Clearly, interference with elongation factor G per se is strongly growth inhibitory, and as such, it represents a potentially viable target. However, the novel inhibitory mode of argyrin B, based on a binding pocket that tolerates multiple amino acid substitutions, increases target-based resistance potential specific to the compound-target pair. Tolerance of variation in this pocket also allows natural target variability and/or organism-specific spontaneous resistance potential. These factors present clear challenges regarding optimization of a scaffold such as argyrin B to minimize target-based resistance and/or ensure a broader spectrum. This underscores the notion that antibacterial targets are best evaluated in the context of particular inhibitory scaffolds, and argyrin B was a useful probe to illuminate these considerations vis a vis elongation factor G. Despite the obvious challenges to further optimization of argyrin B, it is nonetheless intriguing that these compounds do gain intracellular access to inhibit the growth of some problematic Gram-negative bacteria, such as P. aeruginosa and B. multivorans, which may be interesting to understand in view of compound penetration and emerging support for narrower-spectrum antibiotics. Whether other chemical matter to exploit this new inhibitory mode, while circumventing these resistance issues, can be identified remains to be determined.
The bacterial strains and plasmids used in this study are listed in Table 1. Bacteria were routinely grown in lysogeny broth (LB) (tryptone, 10 g/liter; yeast extract, 5 g/liter; NaCl, 10 g/liter), cation-adjusted Mueller-Hinton broth (CAMHB) (Becton Dickinson, Franklin Lakes, NJ), or agar. Plasmids were maintained with 100 μg/ml ampicillin or 10 μg/ml gentamicin (E. coli); 100 to 200 μg/ml carbenicillin, 100 μg/ml tetracycline, or 300 μg/ml gentamicin (P. aeruginosa); or 30 μg/ml gentamicin (A. baumannii) unless otherwise indicated. Isolation of argyrin B was as described previously (8). For single-step isolation of mutants with decreased susceptibility to argyrin B, a colony of S. maltophilia NB55011 was used to inoculate 50 ml Mueller-Hinton broth, and the culture was grown to an optical density at 600 nm (OD600) of 0.5 to 0.6 at 37°C with shaking. Cells were harvested by centrifugation at 5,000 rpm for 10 min, and the pellet was suspended in fresh Mueller-Hinton broth to an OD600 of 10. One hundred and 200 μl of this suspension was plated on Mueller-Hinton agar plates containing 16 or 64 μg/ml argyrin B, respectively. Isolated colonies were counted, and several were saved for further analysis. Serial dilutions were also plated on Mueller-Hinton agar without compound for enumeration. Resistance frequencies were calculated as the number of CFU on drug-containing plates divided by the number of CFU plated.
Standard protocols were employed for restriction endonuclease digestion, gel electrophoresis, ligations, and plasmid isolation (29). The oligonucleotide primers used in this study are listed in Table S1 in the supplemental material. PCR was carried out using an Accuprime GC-rich DNA polymerase kit (Invitrogen, Carlsbad, CA) or the Phusion High-Fidelity PCR master mix with GC buffer (New England BioLabs, Ipswich, MA), in accordance with the supplied instructions. PCR fragments were isolated from agarose gels by using a QIAquick gel extraction kit (Qiagen, Inc., Valencia, CA) in accordance with the supplied instructions. Vectors for the P. aeruginosa efflux strain panel were constructed using both Gateway (Thermo Fisher Scientific, Waltham, MA) and In-fusion (TaKaRa Bio USA, Inc., Mountain View, CA) cloning technologies. E. coli individual pump gene complementation constructs (see below) were made using GeneArt (Thermo Fisher Scientific, Waltham, MA). DNA sequencing was done by Agencourt (Beverly, MA); Genewiz, Inc. (South Plainfield, NJ); Elim Biopharmaceuticals, Inc. (Hayward, CA); or Quintara Biosciences (South San Francisco, CA). Oligonucleotide primers were purchased from Integrated DNA Technologies (Coralville, IA).
Strain NB52024 (Table 1) is a nalB mutant of NB52019 (Table 1) in which MexAB-OprM is overexpressed due to mutation in the mexR gene encoding a repressor of MexAB-OprM expression (30). The mexR-mexA-mexB region from NB52024 was amplified in three segments using primers IF RAB 1 to 6. The individual segments were cloned into HindIII/KpnI-digested pEX18ApGW to create pEX18Ap-mexRAB. pEX18Ap-mexRAB was introduced onto the chromosome of strain NB52245 (ΔmexB ΔmexXY ΔmexCD-oprJ) via conjugation, as previously described (17), and plated on cetrimide agar with carbenicillin to select for merodiploids. Merodiploids were grown without selection in 5 ml LB for 2 h, and then levofloxacin (a MexAB-OprM substrate antibiotic) was added at 0.5 μg/ml to select for resolvants that had acquired the mexR-mexA-mexB fragment. This culture was incubated for a further 2 h and then plated onto LB agar with 8% sucrose. Sucrose-resistant colonies were screened for loss of the pEX18Ap-mexRAB vector backbone and successful mexB gene replacement via PCR, and the presence of the nalB-type mutation in the mexR gene was confirmed by sequencing, yielding strain NB52245-CDJ0015 (Table 1).
An overnight culture of the efflux deletion strain NB52023 (ΔmexB ΔmexXY) (Table 1) was diluted 1:100 in LB and grown to an OD600 of 0.5 in a shaking incubator at 37°C. Cells were harvested and plated on LB agar plates containing 0.4 μg/ml ciprofloxacin (a MexCD-OprJ substrate antibiotic). The plates were incubated at room temperature for approximately 72 h, after which isolated colonies were screened for mutations in nfxB using primers nfxB1 and nfxB2. One such strain, NB52023-CDJ0014, harbored an nfxB TGA (stop)→TGC (cysteine) alteration (Table 1).
Strain NB52109 is a pan-aminoglycoside-resistant clinical isolate that constitutively overexpresses MexXY-OprM (31) (Table 1). The mexZ-mexX-mexY region from the strain was amplified in three segments using primers IF ZXY 1 to 6. They were cloned together into HindIII/KpnI-digested pEX18ApGW to create pEX18Ap-mexZXY. Since strain NB52109 (the source of template DNA for the construct) overexpresses MexXY-OprM via a mexZ-independent, unknown mechanism (31) we introduced a mexZ mutation into pEX18Ap-mexZXY as follows. The mexZ region from a sequence-confirmed pEX18Ap-mexZXY vector was amplified in two segments using the IF ZXY 1/mexZ QC R and mexZ QC F/MexX int For primer pairs. The mexZ QC primers introduce a 1-nucleotide deletion in mexZ (ΔT386, i.e., frameshift) that causes MexX-MexY overexpression (32). The two PCR fragments were spliced together in a second round of PCR with the IF ZXY 1/mexX int F primer pair. The PCR product was purified and digested with HindIII/PacI and then ligated into pEX18Ap-mexZXY digested with the same enzymes to yield pEX18Ap-mexZXY QC. This vector was then introduced into the NB52245 (K2896 ΔmexB ΔmexXY ΔmexCDoprJ) (Table 1) chromosome via conjugation, as previously described (17), and cells were plated on LB-irgasan (5 μg/ml) agar plates with carbenicillin (10 μg/ml, since the cells lacked MexAB-OprM efflux) to select for merodiploids. Merodiploids were resolved by selection on LB containing sucrose and levofloxacin (0.1 μg/ml), and the resolvants were screened for loss of carbenicillin resistance. Sequence analysis revealed that the MexXY genes were restored but the mexZ mutation did not transfer. Therefore, mexZ was deleted as follows. The mexZ region was amplified from NB52019 using the mexZ 5′ GW/mexZ SOE 3′ and mexZ SOE 5′/mexZ 3′ GW primer pairs. These fragments were spliced together in a second round of PCR with the mexZ 5′ GW/mexZ 3′ GW primer pair to create the mexZ deletion fragment. Full Gateway adapters were added to the fragment using the attB1/attB2 primer pair, and the fragment was then introduced into pDONR221 via Gateway technology (Thermo Fisher Scientific, Waltham, MA). A sequence-confirmed pDONR221-mexZ knockout (KO) vector was used to create a pEX18Ap-mexZ KO vector via Gateway technology. The pEX18Ap-mexZ inactivation vector was then introduced into the K2896 mexXY restored strain, as previously described (17), and resolvants were selected on sucrose and screened for loss of mexZ by PCR. One confirmed mutant was designated NB52245-CDJ0021 (Table 1).
Strain NB52019 has an inactive mexT gene (a positive activator of MexEF-OprN expression), and therefore, to create a strain overexpressing MexEF-OprN in this background required introduction of an active mexT allele combined with inactivation of the upstream mexS gene (33) to cause MexEF-OprN overexpression. The mexS region was amplified from NB52019 using the mexS 5′ GW SmaI/mexS SOE 3′ and mexS SOE 5′/mexS 3′ GW SmaI primer pairs. These fragments were spliced together in a second round of PCR with the mexS 5′ GW SmaI/mexS 3′ GW SmaI primer pair to create the mexS KO fragment. Full Gateway adapters were added to the mexS KO fragment with the attB1/attB2 primer pair, and the fragment was introduced into pDONR221 via Gateway technology (Thermo Fisher Scientific, Waltham, MA) to yield the pDONR221-mexS KO vector. The vector was then used to create a pEX18Ap-mexS KO vector via Gateway technology. The pEX18Ap-mexS KO vector was introduced onto the chromosome of NB52245 (ΔmexB ΔmexXY ΔmexCD-oprJ) (Table 1) via conjugation, as previously described (17), and plated on LB-irgasan (5 μg/ml) plates with carbenicillin (30 μg/ml). Sucrose-sensitive, carbenicillin-resistant merodiploids were grown without selection in 5 ml LB for 2 h and then plated onto LB agar with 8% sucrose. Carbenicillin-sensitive, sucrose-resistant colonies were screened for inactivation of the mexS gene via PCR. Then, to introduce an active mexT gene into the genome of NB52245 ΔmexS, the mexS-mexT region from NB52109-CDR0036 (a MexEF-OprN overexpressor) (Table 1) was amplified in two segments using primers ST 1 to 4. These fragments were cloned together into HindIII/KpnI-digested pEX18ApGW to create the pEX18Ap-mexST vector. The mexT fragment was digested out of a sequence-confirmed pEX18Ap-mexST vector with PstI and KpnI and ligated into pEX18ApGW digested with the same enzymes to produce the pEX18Ap-mexT vector. pEX18Ap-mexT was sequence confirmed and then introduced into the chromosome of NB52245 ΔmexS as previously described (17). Merodiploids were selected on LB-irgasan (5 μg/ml) plates containing carbenicillin (30 μg/ml). Sucrose-sensitive, carbenicillin-resistant merodiploids were resolved on LB agar containing 8% sucrose and the MexEF-OprN substrate antibiotic chloramphenicol (100 μg/ml). Carbenicillin-sensitive, sucrose-resistant, and chloramphenicol-resistant colonies were isolated, and the mexT gene was sequenced to confirm the presence of the active mexT. One such isolate was designated NB52245-CDJ0054 (Table 1).
Pump overexpression was confirmed in all cases by susceptibility testing with known pump substrate antibiotics (Table 2) and by RT-qPCR (data not shown).
The reporter fusion strain CDR4 (Table 1) inoculum was prepared using the BBL prompt inoculation system (Becton, Dickinson and Company, Franklin Lakes, NJ) according to the manufacturer's instructions. One hundred twenty-five microliters of the prepared inoculum was evenly spread over the entire surface of a 12-mm by 12-mm LB agar assay plate (Greiner Bio-One, Monroe, NC). One microliter of each compound (12.8-mg/ml stock) was pipetted directly onto the surface of the assay plate. Following overnight incubation at 37°C, the plate was visualized using the Bio-Rad ChemiDoc MP imager with the Chemi protocol and a 10-s exposure time.
The outer membrane efflux channel TolC is known to function with at least 9 RND family efflux pumps in E. coli (34, 35). To examine pump substrate specificity for argyrin B, genes required for the function of these pumps (acrB, emrB, entS, acrF, emrY, mdtF, acrD, macB, and mdtBC) were all deleted from the chromosome of E. coli strain NB27079 (BW25113) (Table 1). Briefly, the genes were sequentially deleted by recombineering (36) using DNA fragments flanking the kanamycin marker (aph) with homology sequences to the targeted genes. In the case of entS and mdtBC, aph fragments were directly amplified from plasmid pKD13 (36) using primers that introduced short flanking sequences necessary for recombineering into the correct locations on the genome. For the rest of the pump genes, the DNA fragments were PCR amplified from individual deletion strains from the Keio library (37). The kanamycin resistance marker replacing each pump gene was flanked by FLP recombination target (FRT) sites and was removed by FLP recombinase using plasmid pFLP2 (38). This nonapump deletion strain (NB27079-CDY0099) (Table 1) was then used as the basis for a panel of strains complemented for each deleted pump gene. For this, individual pump genes were PCR amplified using genomic DNA from E. coli K-12 strain NB27074 (BW25113 ΔtolC) (Table 1) as the template and PCR primers introducing homologous flanking sequences for cloning. Plasmid pAK1900 (Table 1) was digested and linearized with KpnI and HindIII and ligated individually with each pump gene using GeneArt reactions (Thermo-Fisher). The reaction mixtures were then transformed into TOP10 competent cells (Thermo-Fisher) and selected on LB agar containing 100 μg/ml ampicillin. Pump gene inserts were sequence confirmed, and the plasmids (Table 1) were transformed into strain NB27079-CDY0099.
To delete fusA2 from the genome of P. aeruginosa, an in vitro deletion construct was created using the pEX18-Tc gene replacement vector (38). The aacC1 gentamicin resistance cassette from plasmid pUCGm (39) was cloned into the PstI/BamHI sites in pEX18Tc to generate pEX18Tc-Gm. Then, regions upstream and downstream of fusA2 were each PCR amplified from the genome of P. aeruginosa NB52019 using primer pairs E507/E508 and E510/E511, respectively. The upstream fragment was then inserted into PstI/HindIII-digested pEX18Tc-Gm, followed by insertion of the downstream flank into the resulting construct digested with BamHI/KpnI to generate pEX18Tc-fusA2::Gm. The fragments were assembled using GeneArt reactions (Thermo-Fisher) as described by the manufacturer. Plasmid pEX18Tc-fusA2::Gm was mobilized into P. aeruginosa NB52019 from the E. coli mating strain S17-1 as previously described (17). Merodiploids were selected on Pseudomonas isolation agar (PIA) with gentamicin and resolved on PIA with gentamicin and 5% (wt/vol) sucrose. Replacement of fusA2 with aacC1 on the genome was confirmed by colony PCR, and the strain was designated NB52019-CDY0245 (Table 1).
We first constructed pNOV052 by cloning P. aeruginosa fusA1 (EF-G) into an E. coli-Acinetobacter shuttle vector under regulation by the Ptac promoter. The map of pNOV052 is shown in Fig. S1 in the supplemental material. For expression of the argyrin-resistant FusA variants, plasmids pNOV076 and pNOV077, derivatives of pNOV052 encoding P. aeruginosa FusA1 S459F and L663Q, respectively, were constructed using plasmid pNOV052 as the template. Plasmids pNOV052, pNOV076, and pNOV077 were transformed into E. coli NB27004 (MG1655) (Table 1) (selected on LB agar plates with 10 μg/ml gentamicin) or A. baumannii ATCC 19606 (selected on LB agar plates with 30 μg/ml gentamicin). Antibiotic susceptibility testing of these expression strains is described under “Antimicrobial Susceptibility Testing” below.
S. maltophilia strains were inoculated onto Mueller-Hinton agar plates and grown overnight at 37°C. Confluent bacterial growth was harvested, resuspended in 4 ml CAMHB, and diluted to an OD600 of 0.05 in 15 ml CAMHB in 50-ml tubes. Biological replicates were prepared for each strain. The strains were incubated at 37°C with shaking until the cultures reached an OD600 of ~0.5 to 0.6. Ten milliliters of culture was harvested from each flask, pelleted, decanted, and stored at −80°C. The cell pellets were thawed on ice, resuspended in 0.5 ml ice-cold Tris-EDTA (TE), and transferred to matrix B lysis tubes (MP Biomedicals, Burlingame, CA). The cells were lysed on a FastPrep-24 machine (speed, 6.0; time, 20 s; MP Biomedicals, Burlingame, CA) twice, with 2-min cooling on ice between runs. The tubes were then spun to pellet the lysis matrix and cellular debris. After the spin, 200 μl of lysate was removed and processed using the RNeasy kit with on-column DNase digestion (Qiagen, Hilden, Germany). RNA yields were determined using the NanoDrop 2000 spectrophotometer (Thermo Fisher Scientific, Waltham, MA). Then, the RNA samples were diluted to 200 ng/μl and serially diluted 5 more times (10-fold dilutions in nuclease-free water) to prepare a 6-point dilution series. The RT-qPCR mixtures were prepared using qScript XLT One-Step RT-qPCR ToughMix (Quanta Biosciences, Beverly, MA) following the manufacturer's instructions and run using the Bio-Rad CFX96 optics module on a C1000 thermal cycler (Bio-Rad, Hercules, CA) with the CFX_RT_qPCR.prcl protocol. The second putative elongation factor G gene (GeneID 6394560; called fusA2 here) was identified using the S. maltophilia K279a genome sequence (28). This sequence was used to design primers to sequence the corresponding gene in our strain NB55011, from which primers were designed for RT-qPCR (see Table S1 in the supplemental material). A similar approach was used to generate RT-qPCR primers for the NB55011 16S housekeeping gene, except that the universal 16S primers (see Table S1 in the supplemental material) were initially used to amplify and sequence the NB55011 16S gene. The undiluted, 10−1, and 10−2 samples were used for fusA2 expression analysis, and the remaining samples were used for control 16S gene expression analysis. Data analysis was carried out using the CFX Manager software (version 3.1) included with the Bio-Rad CFX96 optics module/C1000 thermal cycler system. For fusA2, the dilution that gave a cycle threshold (CT) value between 15 and 25 cycles in NB55011 was selected for data analysis. For the 16S housekeeping gene, the dilution that gave a CT value between 15 and 25 cycles and showed similar CT values across all strains was selected for data analysis.
Antimicrobial susceptibility testing was done using broth microdilution methodology, as described by the Clinical And Laboratory Standards Institute (40). For testing the effect of isopropyl-β-d-thiogalactopyranoside (IPTG) induced expression of heterologous FusA on susceptibility to antibacterial compounds, the following microdilution plate-based protocol was employed. Cells were streaked from frozen stocks onto LB agar with gentamicin (10 μg/ml for E. coli and 100 μg/ml for A. baumannii). The compounds (argyrin, ciprofloxacin, tetracycline, and fusidic acid) were individually serially diluted in LB with gentamicin at 10 μg/ml. IPTG solutions were prepared at 0.5 mM (for A. baumannii) or 0.05 mM (for E. coli) in LB with gentamicin at 10 μg/ml. Bacterial suspensions were prepared using the BBL prompt system (Becton, Dickinson and Company, Franklin Lakes, NJ) according to the supplied instructions, and the cell suspensions were diluted 1:500 in LB medium with gentamicin (10 μg/ml for E. coli and 100 μg/ml for A. baumannii). The compounds (argyrin, ciprofloxacin, tetracycline, and fusidic acid), IPTG, and bacterial suspension were added to 96-well plates at a ratio of 25 μl (compounds):25 μl (IPTG):50 μl (bacterial suspension), for a total volume of 100 μl. The plates were incubated for 18 to 24 h, and then compound MICs were determined by measuring the OD600.
We thank Keith Poole and Andrew Kropinski (Queen's University) and Herbert Schweizer (University of Florida) for P. aeruginosa strains and plasmids. We thank Tsuyoshi Uehara for making Fig. S2 in the supplemental material; Anthony Casarez, David Six, and Peter Skewes-Cox for helpful discussions; and Katherine Thompson and Cindy Li for assistance with MIC determinations.
Supplemental material for this article may be found at https://doi.org/10.1128/AAC.02400-16.