dict.md logo

Swarming populations of Salmonella represent a unique physiological state coupled to multiple mechanisms of antibiotic resistance

Salmonella enterica serovar Typhimurium is capable of swarming over semi-solid surfaces. Although its swarming behavior shares many readily observable similarities with other swarming bacteria, the phenomenon remains somewhat of an enigma in this bacterium since some attributes skew away from the better characterized systems. Swarming is quite distinct from the classic swimming motility, as there is a prerequisite for cells to first undergo a morphological transformation into swarmer cells. In some organisms, swarming is controlled by quorum sensing, and in others, swarming has been shown to be coupled to increased expression of important virulence factors. Swarming in serovar Typhimurium is coupled to elevated resistance to a wide variety of structurally and functionally distinct classes of antimicrobial compounds. As serovar Typhimurium differentiates into swarm cells, the pmrHFIJKLM operon is up-regulated, resulting in a more positively charged LPS core. Furthermore, as swarm cells begin to de-differentiate, the pmr operon expression is down-regulated, rapidly reaching the levels observed in swim cells. This is one potential mechanism which confers swarm cells increased resistance to antibiotics such as the cationic antimicrobial peptides. However, additional mechanisms are likely associated with the cells in the swarm state that confer elevated resistance to such a broad spectrum of antimicrobial agents.

Salmonella enterica serovar Typhimurium (S. typhimurium) is a bacterium capable of infecting a wide range of mammalian hosts. Although it can cause typhoid-like fever in mice, it is usually associated with a self-limiting gastroenteritis in humans (1). In some cases, however, the disease can manifest as a chronic infection in humans where the biliary duct and gall bladder may serve as the potential reservoirs for re-infection (2, 3). In addition to being a clinically significant pathogen, S. typhimurium has also long served as an important bacterial model system for studying many aspects of general microbial physiology and molecular biology. One such aspect is chemotaxis and swimming motility in aqueous environments (4-9). More recently, it was demonstrated that S. typhimurium is also capable of swarming over semi-solid surfaces (10). These two distinct forms of motility are driven by the same physical (i.e. flagella) and chemotactic machineries (10, 11), however, swarming is strictly preceded by a morphological differentiation of short swimmer cells into elongated, multinucleated and hyperflagellated swarmer cells.

Although the mechanisms that drive the surface swarming behavior in S. typhimurium remain largely uncharacterized, several intriguing phenotypes have been reported to be associated with swarming motility in other organisms. Cell-density is a key factor that triggers swarmer differentiation and migration (10, 12), which suggests that swarming is a co-ordinated multicellular behavior. Co-ordinated behaviors can be mediated by intricate communication mechanisms between individual cells within a given population, a phenomenon generally referred to as quorum sensing (13, 14). Indeed, the disruption of the quorum sensing systems in a variety of organisms has been shown to abolish their ability to swarm (15-18). In the uropathogen Proteus mirabilis, the transition from the swimmer to swarmer cells plays an important role during the invasion of uroepithelial cells (19), which is largely due to the increased expression of important virulence factors coupled to swarm-cell differentiation (20). Taken together, swarming is a clinically significant population behavior, which promotes co-ordinated, yet rapid community migration and expansion. These observations warrant studies to unravel other physiological phenomena that are linked to the swarming behavior.

By utilizing a transposon-based mutagenesis approach, we isolated a swarm-defective mutant of S. typhimurium (21). The transposition site was determined to be in an operon associated with resistance to cationic peptides. E-test analyses revealed that swarm-cell differentiation in wild-type S. typhimurium results in elevated resistance to not only cationic peptides, but also against several other classes of antibiotics. In this report, we provide an in-depth methods supplement to the E-test analyses, in addition to other experimental approaches to further validate that swarm cell differentiation in S. typhimurium is coupled to multiple antibiotic resistance.

S. typhimurium 14028 is a wild-type virulent strain (ATCC 14028). All media components were from Difco, or as noted otherwise. For routine culturing, Luria broth (LB) and agar were used, and when necessary, kanamycin (Sigma) was added at 50 μg/ml. For swarm and swim motility assays, 0.5 μl of an overnight culture was spotted in the middle of a swarm plate (Nutrient broth [NB], 0.5% [w/v] glucose, 0.5% [w/v] agar) or a swim plate (NB, 0.5% glucose, 0.25% agar) and allowed to dry for 1 h at room temperature. All plates were incubated at 37°C for 10 h, or as noted otherwise.

The intergenic region between the divergently oriented pmrG gene and the pmrHFIJKLM operon was PCR amplified from S. typhimurium 14028 by primers Pmr1 (5’-AGTCGGATCCACTACCGGATGCTGCTTC-3’) containing a BamH1 restriction site and Pmr2 (5’-AGTCCTCGAGATTGCCAGTTAGCCGACA-3’) with an XhoI restriction site. Plasmid pCS26 is a low-copy vector that contains a kanamycin resistance gene cassette, and a multiple cloning site (MCS) upstream of a promoterless luciferase operon (luxCDABE) (22). Plasmid pU220 is identical to pCS26, except that the orientation of the MCS is reversed (i.e. BamH1-XhoI-lux operon orientation). The PCR product containing the pmr operon promoter (Ppmr) was digested with BamH1 and XhoI, and subsequently cloned into pU220 to generate the Ppmr::lux transcriptional reporter plasmid pPMRH. Plasmid pPMRH was then transformed into strain 14028. Molecular techniques were performed as described by Sambrook et al. (23), and all enzymes were obtained from Invitrogen.

The strain 14028 harbouring the plasmid pPMRH was used to monitor the expression of the pmr operon. 0.5 μL of an overnight culture was spotted onto NBG (NB with 0.5% glucose) plates containing 0, 0.25, 0.5, 0.7, 1.0, or 1.5% agar. Following 10 h of incubation at 37°C, cells were harvested with sterile toothpicks and inoculated into NBG, and the optical density at 620nm (OD620) of all samples were equilibrated to 0.08-0.09 in 200 μl of fresh NBG. The luciferase activity in the cell preparations and the OD were measured in clear-bottom 96 well black plates (Nunc) with the Victor2 (Wallac) luminometer, as described by Bjarnason et al. (22).

To monitor the changes in the pmr operon expression during de-differentiation of swarmer cells into swimmer cells, the same procedures were followed as described above, except that measurements were taken at 30 minute intervals over 24 h at 37°C. Sterile mineral oil (50 μl) was added to individual wells to prevent evaporation during the assay.

All assays were performed with or without the presence of kanamycin in NBG (+/- agar) and no significant differences were observed in luciferase activity. The low-copy plasmid pU220 is therefore stable even in the absence of kanamycin. All assays were repeated four times to ensure reproducibility.

E-test strips (AB Biodisk, Solna, Sweden) were used to compare the relative resistances of swarming and non-swarming cells of strain 14028 to a variety of antibiotics. Non-swarming (vegetative) cells were analyzed on either solid (NBG, 1.5% agar) or swim plates. For solid plate analysis, 200 μl inocula containing 104-105 cells were evenly spread and allowed to dry at room temperature for 30 min, and the E-test strip was placed in the middle of each plate. On swim plates, 0.5 μl of the seed culture was inoculated at two spots and allowed to dry, and the E-test strip was placed between the two spots. Swarm plates were inoculated in the same manner as the swim plates. All plates were incubated overnight at 37°C, and the E-test results were interpreted as recommended by the manufacturer, or as noted otherwise. All assays were repeated at least two times to ensure reproducibility.

To determine the swimmer cells’ minimum inhibitory concentration (MIC) for polymyxin and kanamycin, an overnight LB culture of strain 14028 was diluted in fresh LB medium and 104-105 cells were inoculated into 96 well plates, with each well containing 100 μl LB medium supplemented with a series of two-fold dilutions of polymyxin and kanamycin at concentrations ranging from 500 to 0.000125 μg/ml. The plates were incubated with shaking at 37°C overnight, and the minimum concentration that inhibited visible growth was declared as the MIC (kanamycin 31.25 μg/ml, polymyxin 8 μg/ml).

To monitor the changes in resistance during de-differentiation, cells were harvested from swarm plates with 3 ml LB broth by gentle shaking on a rotating platform and inoculated into a flask containing 250 ml LB at a final OD600 of 0.01. The flask was incubated with shaking (200 rpm) at 37°C, and starting at the time of initial transfer, 1-5 ml aliquots were removed at 30 minute intervals over a period of 8 h. At each interval, the aliquot was diluted back to the initial OD600 with fresh LB. In three 1.5 ml microfuge tubes each containing 1 ml LB, cells were exposed to either 62.5 μg/ml kanamycin or 15 μg/ml polymyxin for 30 min at room temperature. For the reference culture, three additional aliquots were incubated at each time point in the same manner as above, but without exposure to antibiotics. Following exposure, all samples were immediately diluted by a series of 10-fold dilutions in 96 well plates with fresh LB to dilute out the antibiotic, and 5 μl aliquots from each well were spot-plated onto Luria agar plates, allowed to dry, and incubated at 37°C overnight. Colonies were counted and averaged from the three independent exposures to each antibiotic, and divided by the number of colonies obtained from unexposed samples for each time point to generate the survival curve.

Lipopolysaccharide (LPS) plays an important role in swarming, as mutations in numerous loci involved in LPS biosynthesis and modification result in a defective swarming phenotype (24). The pmrHFIJKLM operon in S. typhimurium encodes enzymes that add 4-amino-4-deoxy-L-arabinose (L-AraN) residues to the lipid A portion of LPS (25, 26). This modification results in a more positively charged outer membrane, reducing the binding potential of positively charged antimicrobial peptides such as polymyxin. A mutation in the pmr operon also abolishes swarming motility in S. typhimurium 14028 (21). To determine whether swarm-cell differentiation leads to altered regulation of the pmr operon, we constructed a pmr operon promoter (Ppmr)-luciferase transcriptional fusion to compare its expression under swim, swarm, and non-motile conditions. When grown in NBG medium containing 0.5-1.0% (w/v) agar, S. typhimurium undergoes swarmer differentiation and exhibits surface motility (Fig. 1). The pmr operon is significantly up-regulated under conditions that promote swarming, and the highest expression was observed in the presence of 0.5% agar, which coincides with the most proficient swarming motility. Although the pmr operon is also expressed in non-swarming (vegetative) populations, its expression was consistently lower than in those conditions which promote swarmer differentiation (Fig. 1). The expression profile is in agreement with β-galactosidase activities measured under similar conditions in a strain containing a lacZ insertion (MudJ) in the pmrK gene (21). The up-regulation of the pmr operon in swarm cells suggests that cells in the swarm state should be more resistant to polymyxin than those in the vegetative state.

It was not feasible to compare polymyxin resistance between vegetative and swarm cells by the conventional overnight minimum inhibitory concentration (MIC) assays, since swarm cells de-differentiate into vegetative cells when grown in liquid media. To overcome this issue, we utilized polymyxin-coated E-test strips (AB Biodisk) directly on swarm plates and compared relative MIC differences against swim and solid plates. Unlike the standard antibiotic diffusion disks, E-test strips contain a defined immobilized antibiotic gradient, generating a continuous antibiotic gradient when placed on an agar surface (27). In agreement with the pmr operon expression data, swarm cells exhibited greater tolerance to polymyxin at least an eight-fold higher concentration in comparison to the cells in the vegetative state (Fig. 2). Remarkably, the E-test strips generated very similar results between the swim and solid plates, indicating that the relative differences in the agar concentrations (i.e. wetness of the plates) did not significantly distort the immobilized antibiotic gradient.

Given the relative simplicity of utilizing the E-test strips, we compared the MICs of 25 other antibiotics against the swarm and vegetative cells of strain 14028 (21). Detailed results are presented in Kim et al. (21) and are summarized below. Against penicillin G, oxacillin, bacitracin, clarithromycin, erythromycin, clindamycin, and trimethoprim, both populations of cells grew at concentrations beyond the maximum range of the strips. Ampicillin, amoxicillin, amoxicillin-clavulanic acid, imipenem, fosfomycin, tetracycline, and doxycycline, representing different generations of β-lactams and tetracyclines, yielded identical results between the swarm and vegetative cells. These results further support the notion that the MIC differences observed between swarm and vegetative cells are genuinely the result of the swarm-cell differentiation. Swarm cells exhibited elevated resistance to 11 antibiotics, with cellular targets extending beyond the cell envelope, including those involved in protein synthesis, DNA replication, and transcription. Greater than five-fold increase in resistance was observed against colistin (polymyxin E) and three of the newer generation β-lactams. Swarm cells were greater than 20-fold more resistant against the aminoglycosides, quinolones and rifampin, and with the exception of ciprofloxacin (93-fold increase), the swarm front migrated beyond the maximum range of the strips (Fig. 2). Distinct waves of swarm fronts were observed against increasing levels of several antibiotics (Fig. 3), suggesting that we may be selecting for resistant mutants. To address this issue, cells were harvested from the swarm fronts at the highest concentrations from seven different antibiotics, including those shown in Fig. 3, and incubated overnight in fresh nutrient broth supplemented with glucose (NBG). Swarm cells de-differentiate into vegetative cells when transferred from swarm plates to liquid media. The cell preparations were spread onto fresh solid NBG plates and spotted onto fresh swim plates, then re-exposed to a respective set of E-test strips. As represented by polymyxin and kanamycin strips in Fig. 2, identical MIC values were generated in comparison to the initial inocula, for all seven antibiotics tested. Therefore, the elevated antibiotic resistance phenotype observed in swarm cells is reflective of adaptive changes associated with this transient physiological state rather than selective mutations. The distinct waves of swarm fronts (Fig. 3) are somewhat similar in appearance to the concentric rings produced by Proteus spp., which represent phases of consolidation, where swarm cells repeatedly de-differentiate and re-differentiate in spatially defined cycles (12). However, Salmonella spp. do not typically exhibit phases of consolidation (10), thus, the distinct fronts may indeed represent phases of adaptation that are unique to the cells in the swarm state in response to increasing levels of antibiotics. Hence, we hypothesize that distinct fronts will form against the aminoglycosides at concentrations higher than the maximum range of the E-test strips.

In order to obtain further insight into the de-differentiation phase, we monitored the activity of the pmr operon during the transition from swarm to swim cells. Similar to the experiment shown in Fig. 1, the Ppmr-luciferase reporter strain was grown under swim, swarm and non-motile conditions. Cells from individual growth conditions were harvested and inoculated into fresh liquid NBG, and luciferase activities in each cell types were measured over time. The expression of the pmr operon in cells grown on 0.5% agar (i.e. swarm plates used in E-test analyses) dropped approximately 50% just thirty minutes following transfer, and continued to decrease rapidly until 60 minutes post-transfer (Fig. 4). These results suggest that de-differentiation begins rapidly, and the protective L-Ara4N LPS-modification is most likely diluted away in the population at a similar rate. In agreement with the data shown in Fig. 1, the pmr operon expression profile was highest at the moment of transfer (t=0), under growth conditions that favor swarm-cell differentiation (Fig. 4).

Albeit at a much reduced amplitude, the same expression profile can be seen up to 90 minutes post-transfer. Regardless of the agar concentration, pmr activity continues to drop until 210 minutes post-transfer, when the activity from all cell types reach a similar level to that detected in NBG without agar. Similar expression profiles were seen as those at 210 and 240 minutes post-transfer throughout the period of 24 hours.

Only the cationic peptide strips generated sharp and pronounced zones of growth inhibition on the swarm plates (Fig. 2). With the exception of E-test strips that permitted the swarm front to migrate beyond their maximum range, and those that produced no difference between the swarm and vegetative cells, multiple swarm fronts were observed (Fig. 3). These observations, in addition to the wide range of structural differences and the cellular targets represented by the wide variety of antimicrobial compounds, suggest that swarm-cell differentiation is likely coupled to multiple resistance mechanisms. In addition to the LPS modification resulting from the up-regulation of the pmr operon, the dramatic morphological changes associated with swarm-cell differentiation may be indicative of an altered cell envelope structure. This could decrease the effectiveness of the cell wall-targeting compounds and the influx-rate of those that target the components within the cytoplasm (28, 29).

The rapid decrease in the expression of the pmr operon (Fig. 4) suggests that the swarm cells’ susceptibility to polymyxin should be rapidly restored to that of vegetative cells when forced into the de-differentiation phase. To test this hypothesis, swarm cells were inoculated into fresh LB to trigger de-differentiation and the transferred cells were exposed to polymyxin at a concentration that was previously determined to be lethal to cells grown in LB. Immediately following transfer, less than 10% of the population survived the polymyxin treatment and 100% death was observed when cells were exposed after 30 minutes post-transfer (Fig. 5). The rapid killing rate is in agreement with the rapid decrease in the pmr operon expression. The post-transfer time indicated on the x-axes in Figs. 4 and 5 are not in frame with one another, since cells sampled at each time point was exposed to polymyxin for an additional 30 minutes prior to plating out the cells. The sharp zone of growth inhibition generated by the polymyxin E-test strips against the swarm front (Fig. 2) may be reflective of its rapid killing mechanism, as evidenced by the sharp killing curve (Fig. 5). There are at least five other genetic loci in Salmonella that have been implicated in conferring resistance to polymyxin and other antimicrobial peptides, by unknown mechanisms that are independent of the L-Ara4N LPS modification (30, 31). Whether or not these genes play physical and/or regulatory roles in polymyxin resistance, some may also behave differently during the transition into the de-differentiation phase. In contrast, virtually the entire population survived lethal dose exposures to kanamycin shortly after transfer (Fig. 5).

There was more significant killing at 30 minutes post-transfer, where only about 50% of the population survived the exposure. Between 30 to 150 minute post-transfer, there were no dramatic changes in the survival of the population to kanamycin exposure, but viability dropped down to less than 25% after 180 and 240 minutes post-transfer. The kanamycin survival profile may reflect the presence of a biphasic, or at least two independent resistance mechanisms in the swarm population of S. typhimurium. The first significant killing at 30 minutes post-transfer may be related to the dramatic reduction in the pmr operon expression, which would lead to a less positively charged cell surface. Since aminoglycosides are also positively charged, the pmr-induced modification may provide the initial protection. Numerous other changes may occur in the outer membrane, since the influx rate of aminoglycosides have been shown to increase in the presence of other cell-wall targeting antibiotics (32) and other alterations in the outer membrane stability (28). Alternatively, extracellular components have been shown to bind and block the entry of the another aminoglycoside antibiotic, tobramycin (29). The second significant killing observed at 180 minutes post-transfer may reflect the loss of these and/or other yet-to-be-characterized resistance mechanisms. Variability was quite high at certain time points, but this is due to the fact that the results were averaged from three independent exposures for each time point.

The use of E-test strips on non-traditional media revealed that swarm-cell differentiation in S. typhimurium is intimately associated with elevated antibiotic resistance, whereby at least one mechanism was traced down to the level of gene regulation. In the instruction manuals provided with the E-test strips, the manufacturer (AB Biodisk) instructs the user to disregard the swarm fronts of Proteus spp. when declaring the MIC. Unlike most bacteria that are capable of swarming, Proteus spp. represent a rather unique class in that they undergo differentiation and exhibit surface motility even on standard solid (i.e. 1.5% agar) plates (12) that are the standard medium for E-test strip applications. Interestingly, the absence of L-Ara4N modification of the LPS was shown to affect swarming motility and to increase sensitivity to cationic antimicrobial peptides in P. mirabilis (33). Thus, elevated antibiotic resistance may be a general phenotype associated with other swarming bacteria and may share similar resistance mechanisms.

The secondary swarm fronts that were observed in response to some of the E-test strips add another level of interest. In addition to comparing swarm-cells and vegetative cells, such subpopulations can be separately harvested and compared to the remaining population for subtractive analyses of differential gene regulation and biochemical modifications. This approach may provide insight into novel resistance mechanisms and antimicrobial targets. Biofilm cells are believed to possess a variety of antibiotic resistance mechanisms, however, there are some convincing evidence that the major resistance mechanism stems from the fact that biofilm cells grow very slowly, and most antibiotics are most effective against rapidly growing cells (34, 35). Unlike in a biofilm, S. typhimurium grows exponentially in the swarm state, which suggests that swarming S. typhimurium may possess a distinct set of resistance mechanisms. Conversely, these two populations of cells may also share a similar set of resistance mechanisms, most likely those associated with alterations in the cell envelope and extracellular components.

This study supports the view that bacteria are highly adaptive, and their robustness may be achieved by numerous transient physiological states. A swarming population of bacteria represents a unique physiological state that promotes rapid population expansion and migration. Swarming in Salmonella may potentially be a clinically significant state, since the cells are well equipped to withstand high levels antimicrobial compounds that are either artificially introduced or naturally generated by the hosts. In addition to the utility of E-test strips, the luciferase-based reporters provide a simple and sensitive means for monitoring gene regulation in real-time. The methods described here can be easily implemented into studying other swarming bacteria in addition to other interesting phenomena, where a bacterium undergoes a transition from one physiological state to another.