Spatiotemporal Distribution of Pseudomonas aeruginosa Alkyl Quinolones under Metabolic and Competitive Stress

Surface growth in the presence of E. coli elicits earlier production of PQS in P. aeruginosa.While many reports have annotated PQS as a stress response (16, 24, 28), the triggers for AQ production and PQS response are not clear. Here, we analyzed the spatiotemporal aspects of two AQs: 2-heptyl-3-hydroxy-4(1H) (PQS) and 2-heptyl-4-hydroxyquinoline N-oxide (HQNO). We examined the PQS and HQNO responses of P. aeruginosa growing in proximity to E. coli K-12, a model lab bacterium that we used as a nonspecific competitor of P. aeruginosa. These species were cultured simultaneously by inoculation at a distance of 12 mm from each other on a semisolid agar medium (see Fig. S1 in the supplemental material). At time points (t) of 24 h, 48 h, and 96 h, the area between the inoculated spots was imaged using confocal Raman microscopy (CRM), and the results were analyzed using principal-component analysis (PCA) in order to assess the spatiotemporal development of the distribution of signaling molecules in the region between the P. aeruginosa and E. coli cultures.

Previous studies have demonstrated that P. aeruginosa communities growing on semisolid surfaces tend to produce more PQS and HQNO than planktonic cultures (29). When P. aeruginosa encounters E. coli, we find that PQS appears sooner than when no E. coli is present. At 24 h, as the Raman microscopy and PCA results in Fig. 1 confirm, HQNO, with features at 715, 1,207, 1,359, and 1,511 cm⁻¹, is present, but not PQS, in agreement with observations from a P. aeruginosa monoculture (shown in Fig. S2). We also observed a loading plot indicating the presence of P. aeruginosa cells with features at 746, 1,127, 1,313, and 1,583 cm⁻¹. However, between 24 and 48 h, P. aeruginosa steadily expanded toward E. coli, and by 48 h, PQS features (1,158, 1,372, and 1,466 cm⁻¹) appeared, much earlier than the 96-h appearance in monoculture (Fig. S2).

• Open in new tab
• Download powerpoint

FIG 1

Combined CRM and PCA show that P. aeruginosa exhibits signatures of PQS by 48 h when cocultured with E. coli. Shown are images of the resultant P. aeruginosa and E. coli growth on 0.7% agar at 24 h (top) and 48 h (bottom). Raman measurements were taken in the region of the P. aeruginosa advancing edge at 24 h and at the intersection of the two strains at 48 h (shown as boxed areas on plates). One representative Raman image and one representative score image are shown for integrations over the 1,330-to-1,380-cm⁻¹ (top row) and 1,630-to-1,680-cm⁻¹ (bottom row) spectral windows at both 24 h and 48 h. Spectra were acquired and inspected over at least five locations within the region to validate overall consistency. Loading plots and score images of principal components were generated from principal-component analysis of the CRM microspectra acquired over the same region. HQNO and PQS features are labeled in red and blue, respectively. In the 48-h sample, PCA revealed two principal components with distinct features, which represent PQS and HQNO, respectively. Bars, 10 μm.

Multipoint inoculation of P. aeruginosa shows the same temporal PQS response as that in a single monoculture.Since interaction with E. coli was shown to elicit earlier PQS production, we investigated if PQS production is stimulated by any possible competitor. We rationalized that the least competitive coculture conditions would be observed in assays of P. aeruginosa with itself. We again used a combination of CRM and PCA tools to examine the intersection area between intersecting colonies of the same P. aeruginosa strain. These P. aeruginosa self-cocultures exhibited identical surface growth phenotypes, although they were inoculated as two separate points, as shown in the plate assay images (Fig. S3, plate panel). At 48 h, Raman images and PCA confirmed the presence of HQNO in the region where two separate, expanding P. aeruginosa colonies meet (Fig. S3c), with features at positions identical to those in a single monoculture inoculation (Fig. S2). At 96 h, in addition to HQNO, PQS features were observed at 1,158, 1,372, and 1,466 cm⁻¹, in agreement with the results of the single-inoculum monoculture experiment. Thus, the results of double inoculation of P. aeruginosa (Fig. S3) matched the P. aeruginosa monoculture data shown in Fig. S2, in that spectral signatures of PQS were not observed until 96 h. We contrast these monoculture results with the P. aeruginosa–E. coli coculture results of Fig. 1, in which PQS appears by 48 h. We note that the early appearance of PQS clearly correlates with the presence of E. coli, suggesting that this phenomenon is a result of intensified stress from the competition of interspecies coculturing, which is not apparent if P. aeruginosa is cocultured with itself.

PQS production is spatially and temporally distinct from HQNO production in the presence of E. coli.The results presented above suggest a role for PQS in mediating interactions with a foreign bacterial species. Therefore, we sought information about the detailed spatial distribution of PQS produced in cocultures. CRM imaging was performed at five sequential positions spanning a line along the P. aeruginosa–E. coli coculture plate, as shown in Fig. 2a.

• Open in new tab
• Download powerpoint

FIG 2

P. aeruginosa exhibits PQS signatures only in close proximity to E. coli, while HQNO signatures extend beyond the region of P. aeruginosa growth. (a) Images of P. aeruginosa–E. coli coculture plates at 24 h and 48 h (bars, 10 mm). Black squares indicate imaged areas 1 to 5 (along the dashed red lines). At least three regions of interest within each area were picked for scanning. (b) Raman images (integrated over 1,330 to 1,380 cm⁻¹), Z-score spatial maps, and loading plots for the most significant principal component for 24 h as a function of position. Confocal Raman images were integrated over 1,330 to 1,380 cm⁻¹ for areas 1 to 4 and over 2,800 to 3,000 cm⁻¹ for area 5. Principal-component analysis was performed for all areas to generate score images and loading plots. All loading plots showed regions of Raman shifts from 600 to 1,800 cm⁻¹. (c) CRM and PCA results of a 48-h coculture from area 1 to area 5 (all integrated over 1,330 to 1,380 cm⁻¹), showing the locations of features from HQNO (red lettering) and PQS (blue lettering). PQS was detected within areas 2 and 3 at 48 h. Score image values range from low (dark blue) to high (red) in each plot, although the total range differs from plot to plot. All samples were grown on FAB-glucose (12 mM) medium with 0.7% agar (bars, 10 μm).

At 24 h, only HQNO is detected (Fig. 2b). Unsurprisingly, HQNO signatures are apparent in areas 1 to 3, from the distal edge of the P. aeruginosa colony to its advancing edge. The optical image (Fig. 2a, top) shows that at 24 h, the cells of P. aeruginosa and E. coli are not yet in contact. However, HQNO is clearly observed in area 4, the proximal edge of E. coli facing P. aeruginosa, indicating that HQNO is able to translocate into the E. coli colony ahead of the advancing P. aeruginosa cells and is able to occupy the area of the E. coli community as an exogenous molecule. In Fig. 2b, for area 5, located at the edge of the E. coli colony distal from the P. aeruginosa cells, the dominant features detected by PCA of CRM microspectra are those of E. coli cells. The features apparent at 1,005, 1,333, 1,457, and 1,659 cm⁻¹ are in excellent agreement with the Raman spectrum acquired from an E. coli monoculture acquired separately (Fig. S4a). The lack of AQ features evidenced by the Raman image over 1,330 to 1,380 cm⁻¹ in area 5 (Fig. S5) instead allows E. coli cellular chemical features to dominate the PCA output. Moreover, CRM and PCA results from separate E. coli and P. aeruginosa monoculture samples (shown in Fig. S4a and S4b, respectively) clearly indicate that E. coli monoculture shows a cellular fingerprint distinct from that of P. aeruginosa and specifically shows no evidence of HQNO-related signals (cf. Fig. S2a and S4a).

Interestingly, the chemical profiles of these five areas shift by the 48-h time point. As the plate image (Fig. 2a, 48 h) shows, P. aeruginosa has expanded to contact the E. coli colony by this stage. Features of PQS (at 1,159, 1,372, and 1,466 cm⁻¹) are readily apparent, in addition to features of HQNO, within areas 2 and 3, i.e., at the center of P. aeruginosa inoculation and at the active intersection between the two strains. Comparison of the loading plots from areas 2 and 3 indicates that PQS is most concentrated in area 3, suggesting that the stress response of P. aeruginosa to the proximity of E. coli is highly localized. Another change from the 24-h profile is that HQNO is now apparent in area 5 at 48 h (Fig. 2c). Thus, by 48 h, HQNO has spread far from areas of P. aeruginosa growth, to all areas where E. coli cells are located. We note how the distributions of HQNO and PQS were markedly different, in that PQS was not detected at these areas away from the point of P. aeruginosa–E. coli convergence (even at the distal edge of the P. aeruginosa colony). These spatial analysis experiments not only demonstrate that PQS is produced later than HQNO in these coculture assays; they also suggest that PQS is first produced preferentially at the intersection of the cocultured strains, where the competitive stress is highest, and at the P. aeruginosa colony center, where cells have accumulated to the highest density. These observations clearly indicate that P. aeruginosa has the ability to spatially regulate its PQS response (but not HQNO) within just a subset of the community in response to spatially local stimuli.

Metabolic stress boosts PQS production and alters the PQS production pathway.We further probed AQ responses as a function of nutrient availability. We hypothesized that the faster PQS response resulting from E. coli interaction shown above (Fig. 1 and 2) could be further intensified if nutrient availability were more limited, because competitive pressures would be more severe. For simplicity, we modified only the carbon source and performed assays with reduced glucose concentrations. While the initial experiments utilized a glucose concentration of 12 mM, here, glucose concentrations were limited to 6 mM or 3 mM, again using CRM in combination with PCA to assess the spatiotemporal attributes of AQ expression at 24 h and 48 h.

As described above, P. aeruginosa growing on plates containing 12 mM glucose in a single-species monoculture or self-coculture exhibits PQS at 96 h (Fig. S2 and S3), and this timing was accelerated to 48 h in E. coli competition assays (Fig. 1 and 2). Experiments with reduced levels of glucose show that these PQS responses are conditionally affected by nutrient availability. Figure 3 illustrates a matrix of CRM and PCA results acquired as a function of glucose concentration (3 mM and 6 mM), incubation pairing (P. aeruginosa–P. aeruginosa and P. aeruginosa–E. coli), and time (24 h and 48 h). Under the most extreme metabolic stress conditions, i.e., 3 mM glucose, no features characteristic of HQNO or PQS production were observed. We draw this conclusion by examining spectral windows of both 1,330 to 1,380 cm⁻¹ (quinolone ring stretching region; Fig. 3) and 2,800 to 3,000 cm⁻¹ (C-H stretching region; Fig. S6) based on our prior work (12, 25). The only features detected were those of P. aeruginosa cells at 746, 1,127, 1,313, and 1,583 cm⁻¹, independently of time and incubation pairing. Given the adequacy of 3 mM glucose for supporting bacterial growth, which was apparent by visual inspection (Fig. 3, plate panels), we were surprised by the stark contrast in AQ production for these assays. While PQS production is already known to be cell density dependent, it was surprising that colonies of the size observed exhibited no AQ signature even after 96 h (not shown). In combination, the smaller relative expansion and the lack of an AQ signature at 3 mM glucose are clearly consistent with nutrient limitation. This result is in agreement with the findings of prior investigations of the stringent response, where the absence of (p)ppGpp, which accumulates under starvation conditions, was required for production of PQS (21, 30). This reinforces prior genetics work showing that production of any AQ, not just PQS, would require a threshold metabolic state in addition to a quorum population density (21). In comparison to setups such as biofilm flow cell assays, where extremely low nutrient concentrations are sufficient to promote QS responses, our results suggest that certain surface growth conditions may require a higher nutrient threshold to elicit an equivalent QS response. Thus, we conclude that the starvation “threshold” required to enable AQ production is a relative target that varies greatly depending on the specific growth conditions.

• Open in new tab
• Download powerpoint

FIG 3

Effects of metabolic stress on PQS production with and without interspecies competition. Imaged areas are boxed on the plates, and at least three regions of interest within each area were picked for scanning. The matrices of CRM and PCA results were acquired as a function of glucose concentration (3 mM and 6 mM) and time (24 h and 48 h) for both P. aeruginosa with P. aeruginosa and P. aeruginosa with E. coli. Each panel shows (from left to right) an optical image of the plate, the Raman image, the Z-score image, and the loading plot. For 6 mM glucose samples, Raman images were integrated over the spectral window of 1,330 to 1,380 cm⁻¹, except for the 48-h coculture of P. aeruginosa with E. coli, which also includes intensities integrated from 1,630 to 1,680 cm⁻¹. For 3 mM glucose samples, Raman images were integrated over the range of 2,800 to 3,000 cm⁻¹. For the 48-h sample in 6 mM glucose only, the first two principal-component Z-score images and loading plots are shown. Score image values range from low (dark blue) to high (red) in each plot, although the total range differs from plot to plot. All samples were grown on FAB-glucose (at 6 mM [top] or 3 mM [bottom]). Bars, 10 μm.

In assays with 6 mM glucose, PQS is produced in both noncompetitive and competitive modes, i.e., in both P. aeruginosa–P. aeruginosa cocultures and P. aeruginosa–E. coli cocultures, as early as 48 h (Fig. 3). While only HQNO is observed in either coculture at 24 h, PQS is observed under both coculture conditions at 48 h. These 6 mM glucose assays represent the only condition we tested under which P. aeruginosa alone exhibited a PQS response by 48 h. The collective different HQNO and PQS responses indicate that while competition and nutrient stress can both cue the PQS response, a threshold metabolic state is required for its initiation. This result was apparent for the entire colony with no spatial variation. These results also point to the utility of motility plate assays for conducting these experiments: while ample bacterial growth was exhibited in all assays examined (Fig. 1 to 4; also Fig. S2 to S4), a range of condition-specific AQ responses was observed, mediated by the different scenarios we tested.

• Open in new tab
• Download powerpoint

FIG 4

Optical images, Raman images (1,330 to 1,380 cm⁻¹), and PCA of the advancing edge (boxed areas on plates) of P. aeruginosa exposed to 1 μl of a P. aeruginosa, E. coli, or S. aureus supernatant. Supernatants were spotted directly on the edge at 18 h postinoculation, and the plate was then returned to the incubator for another 6 h before being removed for testing. All samples were grown on FAB-glucose (12 mM) medium with 0.5% agar. Bars, 10 μm.

Interspecies supernatants induce PQS production in P. aeruginosa.Having observed PQS production at the intersection of P. aeruginosa and E. coli growth, we asked whether this secretion process needed direct interspecies contact between P. aeruginosa and E. coli or whether a soluble factor could stimulate PQS production. We tested this by adding supernatants from 2-day E. coli planktonic cultures to P. aeruginosa growing under the same surface conditions used above with ample nutrient levels (12 mM glucose). Above, we noted production of PQS in response to the presence of E. coli within 48 h of coinoculation (Fig. 1). We hypothesized that stimulation of P. aeruginosa using a supernatant from an already-grown high-cell-density E. coli culture would stimulate PQS production earlier. To test this, an E. coli-derived supernatant was added to the growing edge of an 18-h P. aeruginosa monoculture plate, and the plate was incubated for another 6 h, thereby achieving a total of 24 h of P. aeruginosa growth time. Figure 4 shows clear spectral evidence of the presence of both HQNO and PQS at the advancing edge of P. aeruginosa, where the supernatant was pipetted, indicating that soluble factors produced by E. coli contained in the small volume (1 μl) of supernatant are sufficient to induce PQS production. Thus, direct contact between P. aeruginosa and E. coli cells is not necessary to induce P. aeruginosa PQS production. We also conclude that 6 h is sufficient to elicit a PQS response under our surface growth conditions. Subsequently, we expanded our investigation to learn if supernatants from bacteria other than E. coli would elicit a PQS response. We chose Staphylococcus aureus USA300 as a second test strain, because it coexists and competes with P. aeruginosa in many clinical settings (27), and it is also important to understand how the behaviors of the two strains are affected by each other. S. aureus supernatant was added to the P. aeruginosa plates in the same fashion as the E. coli samples (Fig. 4). In response to the S. aureus-derived supernatant, the level of PQS spiked in these samples in a manner equivalent to that observed with the addition of E. coli supernatant. To control for the possible stimulation of PQS by the addition of new nutrients, we tested the responses of P. aeruginosa to its own spent medium (Fig. 4, top) and to the growth medium alone as an uninoculated planktonic control (Fig. S7). As shown in both figures, spotting 1 μl of the P. aeruginosa supernatant or growth medium was not sufficient to stimulate PQS production. Thus, both E. coli and S. aureus supernatants elicited PQS production in P. aeruginosa during surface colonization, and we conclude that the stress imposed by other bacterial species on P. aeruginosa is readily conveyed in soluble factors produced by either species. These results for the addition of supernatants align with the findings of Horspool and Schertzer (31) showing stimulation of the production of P. aeruginosa outer membrane vesicles (OMVs) by E. coli supernatants. OMVs have been shown by the Schertzer group and others to be a primary delivery mechanism for PQS.

Two-factor interaction model.These experiments clearly illustrate that PQS production is cued by multiple factors. Moreover, we find that PQS production is cued differently, both temporally and spatially, from HQNO production. Our group and others have previously established that surface growth greatly stimulates both HQNO and PQS production (13, 23). However, under surface motile conditions, the appearance of HQNO predictably precedes the appearance of PQS, but HQNO is not modulated equivalently to PQS (12, 32). Here, we present evidence that at least two separate factors promote PQS production during surface growth, and we present a working model to describe the onset of PQS activation (Fig. 5). PQS production, which reflects a regulatory response to the presence of an environmental challenge, is plotted as a function of growth time either without (Fig. 5, top) or with (Fig. 5, bottom) the presence of a competitor (e.g., E. coli). PQS levels are represented by three curves, corresponding to the different glucose concentrations used in the present experiments (black for 3 mM, gray for 6 mM, and red for 12 mM). In the absence of E. coli (Fig. 5, top), the production of PQS reaches an effective level (indicated by the dashed horizontal line on the plot) in less time when P. aeruginosa experiences some metabolic stress in the form of nutrient limitation (6 mM glucose) than when nutrient levels are adequate (12 mM glucose). However, when the organism is grown with more-severe nutrient limitation (3 mM glucose or less), production of PQS, and of all AQs, is repressed. In the presence of competition stress (e.g., E. coli or soluble factors from a supernatant of E. coli or S. aureus) (Fig. 5, bottom), PQS is produced earlier under both 6 mM and 12 mM nutrient conditions. The production of HQNO was essentially binary: when P. aeruginosa was growing on surfaces with sufficient nutrients to overcome stringent-response repression, HQNO was produced and detected communitywide. Alternatively, in planktonic culture and/or under starvation conditions, HQNO was absent. While this simple model does not likely capture the overall regulatory response to all possible environmental stressors, it does highlight the manner in which the two principal environmental challenges studied here interact to affect PQS production. Clearly, further research is needed to determine the differential regulation of PQS and HQNO, as well as the spatial scales on which these AQs are produced and disseminated.

• Open in new tab
• Download powerpoint

FIG 5

Schematic illustration of the two-factor interaction model. Shown are plots of PQS production as a function of growth time without (top) and with (bottom) a bacterial competitor (e.g., E. coli) at 6 mM or 12 mM glucose. The metabolically stressed condition (3 mM glucose) never achieves PQS production. Dashed horizontal lines indicate an effective level of PQS production.