Interspecies interactions induce exploratory motility in Pseudomonas aeruginosa

Microbes often live in multispecies communities where interactions among community members impact both the individual constituents and the surrounding environment. Here, we developed a system to visualize interspecies behaviors at initial encounters. By imaging two prevalent pathogens known to be coisolated from chronic illnesses, Pseudomonas aeruginosa and Staphylococcus aureus, we observed P. aeruginosa can modify surface motility in response to secreted factors from S. aureus. Upon sensing S. aureus, P. aeruginosa transitioned from collective to single-cell motility with an associated increase in speed and directedness – a behavior we refer to as ‘exploratory motility’. Explorer cells moved preferentially towards S. aureus and invaded S. aureus colonies through the action of the type IV pili. These studies reveal previously undescribed motility behaviors and lend insight into how P. aeruginosa senses and responds to other species. Identifying strategies to harness these interactions may open avenues for new antimicrobial strategies.


Introduction
While it is clear that many microbial infections do not occur with a single species, we have only recently begun to understand the profound impacts microbial species have on each other and patients (Nguyen and Oglesby-Sherrouse, 2016). Studies of dental biofilms, intestinal communities, chronic wounds, and respiratory infections in patients with cystic fibrosis (CF) demonstrate that community interactions influence microbial survival and disease progression (Limoli and Hoffman, 2019;Gabrilska and Rumbaugh, 2015). For example, we and others find an association between coisolation of Pseudomonas aeruginosa and Staphylococcus aureus from the CF airway or chronic wounds and poor patient outcomes, including decreased lung function and shortened life-spans (Limoli et al., 2016;Maliniak et al., 2016;Hubert et al., 2013). Laboratory studies also reveal interactions between these two pathogens can alter virulence factor production by one or both species, potentially influencing pathogenesis, persistence, and/or antibiotic susceptibility (Hotterbeekx et al., 2017). For example, in a model of coinfection on CF-derived bronchial epithelial cells, we observed P. aeruginosa and S. aureus form mixed microcolonies, which promotes the survival of S. aureus in the presence of vancomycin (Orazi and O'Toole, 2017). One strategy to improve outcomes for these coinfected patients may be to block harmful interspecies interactions before they begin.
Here, we designed a system to visualize early interactions between P. aeruginosa and S. aureus and follow single-cell behaviors over time with live-imaging. We show that P. aeruginosa can sense S. aureus secreted products from a distance and in turn, dramatically alter the motility behaviors of this Gram-negative bacterium. In response to S. aureus, individual P. aeruginosa cells transition from collective to single-cell movement, allowing exploration of the surrounding environment and directional movement towards S. aureus. We find that such 'exploratory motility' is driven primarily by the P. aeruginosa type IV pili (TFP) and modulated, in part, by the CheY-like response regulator, PilG. Importantly, we find P. aeruginosa is capable of responding to an array of bacterial species and strains recovered from a variety of sources, including CF patients, demonstrating a broad capacity for P. aeruginosa to sense other microbial species and modulate motility in response. Thus, we provide a new means to study polymicrobial interactions at the single-cell level and reveal that P. aeruginosa can sense the presence of other microbial species and dramatically, yet specifically, modify its behavior in response to such interspecies signals.

S. aureus promotes exploratory motility in P. aeruginosa
To understand early microbial interactions, we established an in vitro coculture system to monitor P. aeruginosa and S. aureus at first encounters. Bacteria were inoculated at low cell densities between a coverslip and an agarose pad in minimal medium, supplemented with glucose and tryptone, and imaged with phase contrast time-lapse microscopy every 15 min for 8 hr. Alone, P. aeruginosa cells replicate and expand outward as raft-like groups, as previously described for P. aeruginosa surfacebased motility (Anyan et al., 2014;Burrows, 2012) (Video 1; Figure 1, still montage, top row). In comparison, coincubation with S. aureus resulted in dramatically altered behavior (Video 2, Figure 1, still montage, bottom row). After two to three rounds of cell division, instead of remaining as a group, individual P. aeruginosa cells began to increase motility and moved as single-cells, suggesting that P. aeruginosa responds to the presence of S. aureus by altering motility behaviors. P. aeruginosa significantly inhibited S. aureus growth, as previously reported (see Figure 1-video 1 for representative time-lapse video of S. aureus alone).
To visualize P. aeruginosa motility in the presence of S. aureus in more detail, the inoculating cell density was reduced (2-3 cells of each species per field of view), and images were taken at 5 s intervals for 8 hr. Video 3 shows images taken during hours 4-6 of coculture, when P. aeruginosa initiates single-cell movement under these conditions ( Figure 2A, still montage). We observed a number of surprising behaviors by P. aeruginosa in the presence of S. aureus. P. aeruginosa cells initially replicated and remained in a raft (t = 4 h:28 m), as we and others have observed for P. aeruginosa in monoculture, but as the community approached S. aureus, individual cells: (1) exited the raft (t = 4 h:34 m), (2) moved with increased speed, and (3) moved towards and surrounded S. aureus, 'explored' the surface of the colony until ( Figure 2A-B), (4) P. aeruginosa cells entered the S. aureus colony ( Figure 2C, t = 6 hr), and finally, (5) by 8 hr, P. aeruginosa completely dismantled the S. aureus community ( Figure 2D, t = 8 hr). P. aeruginosa was also observed to adopt a swift motion, beginning between 4.5-6 hr, moving in and out of the plane of focus during imaging (Video 4, yellow circle and Figure 2D, red arrows).
To quantitate the movement of P. aeruginosa single-cell motility in comparison to collective motility in rafts, the movement of individual cells and the leading edge of the rafts was tracked over time. In comparison to cells moving in rafts, individual cells moved with increased speed (mm/s), acceleration (mm/s 2 ), and mean squared displacement (MSD, mm 2 ) ( Figure 2E, F and G, respectively). MSD represents a combined measure of both the speed and directional persistence of the cell, thus an increased MSD in single-cells over a change in time suggests single-cells exhibit directed motion, followed by a decreased MSD when P. aeruginosa cells reach the S. aureus colony.

P. aeruginosa type IV pili drive exploratory motility
How is P. aeruginosa motility generated in response to S. aureus? When grown in monoculture, P. aeruginosa performs cellular movement through the action of a single polar flagellum and the type IV pili (TFP) (Conrad et al., 2011;Gibiansky et al., 2010;Merritt et al., 2010). To determine how the observed P. aeruginosa 'exploratory motility' is generated, P. aeruginosa strains deficient in the production of either TFP (DpilA; encoding the pilin monomers), flagella (DflgK; encoding the flagella hook protein), or both (DpilA DflgK) were examined. Time-lapse images were taken as described for Figure 2, except they were acquired at 50 ms intervals for visualization of specific motility patterns. Representative videos and snap-shots ( Figure 3A) were chosen at the time-point where P. Video 1. WT P. aeruginosa in monoculture. Duration 8 hr. Acquisition interval 15 m. Playback speed 3054x. https://elifesciences.org/articles/47365#video1 Figure 1. S. aureus increases P. aeruginosa motility. Live-imaging of polymicrobial interactions. P. aeruginosa (rod-shaped) was inoculated between a coverslip and an agarose pad, either in monoculture (top) or in coculture with equal numbers of S. aureus (cocci-shaped, bottom). Images were acquired every 15 m for 9 hr. Representative snap-shots of Video 1 (top) and Video 2 (bottom) are shown. Founding cells identified in the first frame are indicated with green rods (P. aeruginosa) or yellow circles (S. aureus). The location of the founding cell is indicated in each subsequent frame for positional reference. At t = 9 h:00 m, the founding P. aeruginosa cell has moved outside the field of view.
The online version of this article includes the following video for figure 1: Figure 1 aeruginosa was found to exhibit both slow and swift single-cell movements. The DpilA mutant was unable to move away from the group as single-cells (Video 5), as seen for the parental P. aeruginosa (Video 4), suggesting the TFP are required for P. aeruginosa exploratory motility. However, the swift movement observed in the WT ( Figure 3A, see boxed inset with red arrows), was maintained in DpilA, demonstrating that TFP are not required for this behavior.
We next examined the DflgK mutant (Figure 3-video 1), which was seen to adopt the single-cell behaviors of the WT, except the swift movements were not observed, supporting the hypothesis that this movement is generated by the flagellum. These data suggest that while S. aureus modulates both TFP and flagella-mediated motility, the early events necessary for exploration (initiation of single-cell movement and directional movement towards S. aureus) require the TFP.
We also examined the response of a double DpilA DflgK mutant in the presence of S. aureus ( Figure 3-video 2). Surprisingly, this mutant exhibited a phenotype distinct from either the WT, or the individual DpilA and DflgK mutants. P. aeruginosa cells deficient in both TFP and flagella were not only unable to produce the respective movements characteristic of these motility motors, but also were unable to remain within a raft-like group ( Figure 3A). This behavior was not dependent upon the presence of S. aureus, as a similar pattern was observed for DpilA DflgK when visualized in the absence of S. aureus (Figure 3-figure supplement 1).
The influence of P. aeruginosa exploratory motility on S. aureus growth was also examined by measuring the area of the S. aureus colonies at 4.5 hr post-inoculation. S. aureus colonies were significantly smaller in the presence of WT P. aeruginosa or the DflgK mutant in comparison to S. aureus grown in monoculture ( Figure 3B). However, in the presence the DpilA mutant, which was deficient in exploratory motility, colonies were significantly larger in comparison to WT and the DflgK mutant. Moreover, when grown in the presence of the double mutant, the area of the S. aureus colonies was not significantly different from S. aureus monoculture. These data suggest that ability of P. aeruginosa to perform exploratory motility influences P. aeruginosa inhibition of S. aureus growth.
While visualizing P. aeruginosa-S. aureus interactions at 50 ms intervals, we observed an additional P. aeruginosa behavior. When P. aeruginosa first encounters the S. aureus colony, the cell body orients perpendicular to the surface of the colony and moves back-and-forth (Video 4, blue circle and Figure 3C) and appears to move P. aeruginosa cells into the S. aureus colony. This behavior was only observed in the WT and DflgK mutant, suggesting that it is driven by the TFP.
Agr-regulated secreted S. aureus factors promote P. aeruginosa motility Live-imaging suggests P. aeruginosa is capable of sensing S. aureus and initiating exploratory motility from a distance. Thus, we hypothesized that P. aeruginosa responds to S. aureus secreted factors. Since we observed that TFP were required for exploratory motility, we sought to develop a macroscopic assay where TFP motility could be monitored in the presence of S. aureus secreted factors only. Macroscopically, population-scale TFP-mediated twitching motility can be visualized by inoculating P. aeruginosa cells onto the bottom of a plate through 1.5% agar (sub-surface); cells move on the surface of the plate under the agar and can be stained with crystal violet for visualization (Turnbull and Whitchurch, 2014). To test the hypothesis that P. aeruginosa responds to S. aureus secreted products, cell-free supernatant derived from overnight cultures of WT S. aureus, (normalized to OD 600 = 5.0) was spread on the plate, prior to pouring molten agar. S. aureus supernatant significantly increased the motility diameter of P. aeruginosa in a dose-dependent manner ( Figure 4A), supporting the hypothesis that S. aureus secreted factors increase P. aeruginosa twitching motility.
Two primary regulators of secreted factors in S. aureus are the alternative stress sigma factor, sigma B (SigB) (Baek et al., 2013) and the accessory gene regulator (Agr) quorum sensing system (B) Snap-shot at 5 hr, zoomed in to visualize single-cells. (C) Snap-shot at 6 hr of P. aeruginosa (mKO, red) and S. aureus (GFP, green) illustrating P. aeruginosa invasion into S. aureus colonies. (D) Snap-shot of coculture at 8 hr, showing disruption of S. aureus colonies (green arrows) and swift-moving P. aeruginosa cells out of the plane of focus (red arrows). Single P. aeruginosa cells and the leading edge of rafts were tracked in the presence of S. aureus and the speed (mm/s), acceleration (mm/s 2 ), and mean squared displacement (mm 2 ) for four independent videos are indicated, respectively, in (E -G). (Nair et al., 2011). S. aureus strains with transposon insertions in either sigB or agrB (Fey et al., 2013) were examined for their ability to induce P. aeruginosa twitching motility ( Figure 4B). While the sigB::Tn strain phenocopied the WT, the agrB::Tn mutant lost all ability to induce P. aeruginosa twitching motility, suggesting that Agr regulates the production of the factors promoting motility in P. aeruginosa. The agr operon is organized around two divergent promoters, P2 and P3, and generates two primary transcripts, RNAII and RNAIII, respectively. RNAII encodes AgrB, AgrD, AgrC, and AgrA (Le and Otto, 2015). To confirm a role for the Agr quorum sensing system, an unmarked deletion of agrBDCA was generated and complemented with WT agrBDCA. P. aeruginosa motility was examined in the presence of supernatant derived from these strains. As predicted, the DagrBDCA mutant was unable to enhance P. aeruginosa motility, and complementation restored activity to WT levels ( Figure 4C).
To confirm that the TFP are necessary for increased P. aeruginosa motility in this assay, we examined the response of pilA-deficient P. aeruginosa to S. aureus supernatant. In the absence of S. aureus supernatant, the DpilA mutant was unable to perform twitching motility, as previously reported (Darzins, 1994). However, while motility was reduced in the presence of S. aureus supernatant, surprisingly, the DpilA mutant retained some ability to respond to S. aureus secreted factors ( Figure 4D). Since we observed during live-imaging that S. aureus can increase P. aeruginosa flagella-mediated motility, in addition to TFP-mediated motility, we hypothesized that the response retained in the DpilA mutant was due to increased flagella-mediated motility. To test this hypothesis, we examined the response of DflgK and a double DpilA DflgK mutant to S. aureus supernatant. The single DflgK mutant phenotype trended lower, but was not significantly different from WT, while the motility of the DpilA DflgK mutant was reduced to levels not significantly different from DpilA or DpilA DflgK in the absence of supernatant ( Figure 4D). These data support our observation from live-imaging that, while TFP were the primary contributors to increased motility, S. aureus secreted factors could increase both TFP-and flagella-mediated motility.
To formally examine the response of P. aeruginosa flagella-mediated motility to S. aureus supernatant, traditional macroscopic low-percentage agar assays that measure the contribution of flagella motility and chemotaxis were performed. Cell-free S. aureus supernatant was mixed into 0.3% agar prior to inoculating P. aeruginosa cells into the agar. The diameter of the P. aeruginosa motility zone showed a modest, but not significant increase in the presence of S. aureus supernatant, compared to medium alone ( Figure 4E). To examine the role for flagella in response to S. aureus under swim assay conditions, the DflgK mutant was tested. As expected, flagella-mediated motility, both with and without S. aureus supernatant, was significantly reduced. To determine if TFP contribute under these assay conditions, the DpilA and DpilA DflgK mutants were also examined. The motility diameter of DpilA was not significantly different from WT, and the double mutant phenocopied the single DflgK mutant, suggesting that flagella are primarily responsible for the motility observed here.

P. aeruginosa biases the directionality of movement up a concentration gradient of S. aureus-secreted factors
Plate-based macroscopic motility assays measure both an absolute increase in motility and directional movement up a self-generated gradient (chemotaxis) as bacterial populations metabolize the available substrates and expand outward radially (Shapiro, 1984). While flagella-based movement aureus per frame in monoculture or in the presence of the indicated P. aeruginosa strain was calculated at t = 5 hr by dividing the total area occupied by S. aureus in a single frame by the number of S. aureus colonies. A minimum of four videos were analyzed per condition. The mean and standard deviation are indicated. Statistical significance was determined by one-way ANOVA followed by Tukey's Multiple Comparisons Test -a indicates a statistically significant difference (p 0.05) between S. aureus in monoculture and in the presence of either WT P. aeruginosa or DflgK; b indicates a statistically significant difference between DpilA and DpilA DflgK. (C) Representative snap-shots of Video 4, WT P. aeruginosa and S. aureus beginning at 4 hr with 50 ms intervals, showing back-and-forth motion. Red arrows indicate when a P. aeruginosa cell is moving in towards the S. aureus colony and green arrows indicate when a cell is moving away. The online version of this article includes the following video and figure supplement(s) for figure 3:  through liquid has been extensively described in P. aeruginosa for a variety of chemoattractants, directional movement on a surface is poorly understood. Kearns et al. previously reported P. aeruginosa pili-mediated biased movement up a gradient of phosphatidylethanolamine (PE) on the surface of an agar plate (Kearns et al., 2001). To determine if P. aeruginosa is capable of moving up a previously established concentration gradient of S. aureus supernatant, supernatant derived from either WT or the DagrBDCA mutant of S. aureus was spotted onto the surface of 1.5% agar (containing buffered medium only, no carbon source). The secreted factors were allowed to diffuse and establish In D and E, the motility of the indicated P. aeruginosa mutants was analyzed in the presence of medium alone or supernatant derived from WT S. aureus (0.5 dilution) under 1.5% agar (D) or within 0.3% agar (E). The mean and standard deviation are indicated for at least three biological replicates. Statistical significance was determined by one-way ANOVA followed by Tukey's Multiple Comparisons Test -a indicates a statistically significant difference (p 0.05) between the motility observed in the presence of S. aureus supernatant compared to medium alone, and b indicates a statistically significant difference (p 0.05) between the motility observed in the mutant strain (S. aureus mutants in B and C, P. aeruginosa mutants in D and E) compared to the parental. a gradient for approximately 24 hr, prior to inoculating P. aeruginosa onto the surface, as previously described (Kearns et al., 2001). Preferential movement of P. aeruginosa towards supernatant derived from WT S. aureus was observed, but not for medium alone or supernatant derived from the DagrBDCA mutant ( Figure 5A). The response for the DpilA mutant was also examined and all motility was abrogated in this mutant, demonstrating motility observed in this assay is entirely pili-mediated. These data support the hypothesis that P. aeruginosa biases the directionality of TFPmediated motility up a concentration gradient of S. aureus secreted factors.
We next asked if P. aeruginosa would also fail to migrate towards S. aureus deficient in Agr activity at the single-cell level. WT P. aeruginosa was visualized in coculture with DagrBDCA, as previously performed in the presence of WT S. aureus -with images acquired every 5 s for 8 hr (Video 3 for WT: Figure 2A and Video 6 for the DagrBDCA mutant: Figure 5B). In comparison to WT S. aureus, P. aeruginosa behavior was significantly altered in the presence of DagrBDCA. P. aeruginosa remained capable of initiating single-cell movement; however, once cells initiated single-cell movement, the path of their movement did not appear as directed towards the S. aureus colonies. In fact, some cells seemed to actively avoid the S. aureus colony all together.
To quantify the directedness of P. aeruginosa movement in the presence of WT and the DagrBDCA mutant, single P. aeruginosa cells were tracked from the first frame a cell exited the raft to the frame where it first encounters S. aureus ( Figure 5B). The accumulated track distance, D (A) , was measured and compared to the Euclidean distance, D (E) , between the position of the cell in the first and last frame tracked. The directness of P. aeruginosa towards S. aureus was calculated as a ratio of D (E) /D (A) . P. aeruginosa exhibited a significantly higher directedness ratio towards WT S. aureus compared to the DagrBDCA mutant ( Figure 5C). These data support the hypothesis that Agr regulates the production of S. aureus secreted factors driving the directionality of P. aeruginosa motility.
The CheY-like response regulator, PilG, modulates P. aeruginosa response to S. aureus How does P. aeruginosa sense the presence of S. aureus secreted products and initiate TFP-driven exploratory motility? P. aeruginosa encodes three known chemotaxis pathways: two flagella-mediated (che and che2) and a putative TFP-mediated system (pil-chp) (Darzins, 1994;Whitchurch et al., 2004). While several proteins encoded in the Pil-Chp pathway comprise a signal transduction pathway very similar (by gene homology) to the flagella-mediated chemotaxis pathway, their role in directional twitching motility remains unclear. A significant challenge lies in that many of these proteins are required for pilus assembly, thus teasing apart their requirement for motility per se versus regulation of a chemotactic response is challenging. Nonetheless, pilJ is predicted to encode the TFP methyl-accepting chemoreceptor protein (MCP) and by homology to the flagella MCPs, PilJ is expected to detect changes in the concentration of an attractant or repellent and initiate a signaling cascade to control twitching motility. To determine if PilJ is necessary for P. aeruginosa to sense S. aureus, we examined the response of a DpilJ mutant to S. aureus (WT) in the macroscopic sub-surface inoculation assay and by microscopic live-imaging. In the macroscopic assay, the DpilJ mutant exhibited low twitching motility, as previously described (Darzins, 1994) and responded to S. aureus supernatant to a similar degree as the DpilA mutant ( Figure 6-figure supplement 1). However, examination by live-imaging revealed that DpilJ remained capable of performing surface motility and phenocopied the behavior of WT P. aeruginosa in the presence of S. aureus; moving as single-cells and with preferred directionality towards S. aureus ( Figure 6A), suggesting that PilJ is not necessary for exploratory motility. A second protein encoded within the Pil-Chp chemosensory system that has been implicated in TFP-mediated chemotaxis is the CheY-like response regulator PilG, which regulates the ATPase necessary for TFP extension (PilB). Prior work demonstrates that while a DpilG mutant is non-motile in a macroscopic sub-surface inoculation assay (as previously described, Bertrand et al., 2010), in a microfluidic assay, individual surface attached P. aeruginosa cells were capable of performing twitching motility on a single-cell level; yet, they were unable to respond to a chemotactic gradient (Oliveira et al., 2016). We therefore examined a role for PilG in P. aeruginosa response to S. aureus. Similar to DpilJ we found that DpilG phenocopied the DpilA mutant in the macroscopic assay and was unable to generate twitching motility ( Figure 6-figure supplement 1). However, during liveimaging, in comparison to DpilJ, DpilG was diminished in its capacity to increase motility in response Interestingly, we did note that in a few circumstances (3 of 9 videos), when a sufficient number of S. aureus cells were present (five or more founding cells per field of view) DpilG would eventually respond to the presence of S. aureus (Figure 6-figure supplement 3).
To confirm that DpilG exhibits a diminished response to S. aureus in comparison to WT, we imaged both WT P. aeruginosa and the isogeneic DpilG mutant in coculture with S. aureus simultaneously to account for variability in the number of S. aureus cells in the initial inoculum. P. aeruginosa strains were engineered to differentially express GFP (WT) or mKate (red, DpilG). Here the DpilG mutant displayed a diminished response to S. aureus in comparison to WT P. aeruginosa ( Figure 6C). These data support the previous observations suggesting that a DpilG mutant is capable of performing TFP-mediated motility, but has a diminished response to chemotactic signals (Oliveira et al., 2016).

P. aeruginosa responds to a broad range of model organisms and CF pathogens
Our data thus far demonstrate that P. aeruginosa can sense S. aureus from a distance and modulate both flagella and TFP-mediated motility. To begin to examine if these behaviors might be relevant during airway infection in CF patients, we examined the capacity of a panel of P. aeruginosa isolates from CF patients to respond to S. aureus. Phenotypic loss of twitching motility in monoculture is often observed in chronic P. aeruginosa isolates from CF patients (Mayer-Hamblett et al., 2014), thus it was not surprising that two of the four isolates examined exhibited no detectable twitching motility in the absence of S. aureus. Nonetheless, each P. aeruginosa isolate was able to respond to S. aureus supernatant, although to varying degrees ( Figure 7A). Importantly, even mucoid P. aeruginosa isolates (CFPBPA38, 43, 37), exhibited at least a two-fold increase in motility in the presence of S. aureus supernatant, above medium alone.
Next, we asked if S. aureus isolates from CF patients are capable of inducing P. aeruginosa motility in the macroscopic twitching assay. Supernatant derived from two out of three clinical CF S. aureus isolates (each from different patients) was capable of inducing motility of the WT P. aeruginosa laboratory strain, to an extent not significantly different from that previously observed with WT S. aureus ( Figure 7B). Since we previously observed Agr was necessary for S. aureus to promote motility, we hypothesized that CFBRSA48 was unable to induce P. aeruginosa due to reduced activity of the Agr quorum sensing system -an adaptation previously reported for S. aureus CF isolates (Nair et al., 2011;Goerke and Wolz, 2010). Since Agr also positively regulates the production of S. aureus b-hemolysin, hemolysis on sheep blood agar plates is often used as an indicator of Agr activity (Peng et al., 1988). However, all three clinical isolates maintain WT levels of hemolysis, suggesting Agr is active in these strains and an unknown mechanism accounts for reduced activity towards S. aureus in CFBRSA isolate 48 (Figure 7-figure supplement 1).
Can P. aeruginosa also respond to other CF pathogens, or is this behavior specific to S. aureus? To investigate this question, we examined clinical isolates from CF patients of three additional CF pathogens: Haemophilus influenzae, Burkholderia cepacia, and Achromobacter xylosoxidans. When possible, we examined multiple isolates for the ability of cell-free supernatant to enhance twitching motility in P. aeruginosa. We observed that one of two B. cepacia isolates tested enhanced P. aeruginosa motility to a similar degree as S. aureus, while none of those tested from H. influenzae or A. xylosoxidans significantly increased motility under these conditions ( Figure 7C).
We then asked if P. aeruginosa is able to sense and respond more broadly to the presence of other bacterial species. Here we tested three additional non-CF organisms: Bacillus subtilis, Escherichia coli, and Salmonella Typhimurium. While the amount of P. aeruginosa motility observed varied between species and laboratory strains utilized, we observed a significant increase in P. aeruginosa motility in response to each organism tested, demonstrating that P. aeruginosa responds to a variety of bacterial species ( Figure 7D).

Figure 5 continued ratio of D (E) /D (A) (Directedness) is shown in (C) for P. aeruginosa moving towards WT S. aureus compared to
DagrBDCA with the mean indicated. Statistical significance was determined by an unpaired Student's t-test (a = P 0.05).
To gain insight into the secreted microbial factors that promote P. aeruginosa motility, we performed preliminary biochemical analysis on S. aureus supernatant. First, cell-free supernatant was subjected to heat treatment at 95˚C for 30 min. Heat treatment did not influence supernatant motility-inducing activity. However, protease treatment of the supernatant, with either proteinase k or trypsin, almost entirely eliminated activity ( Figure 8A). These data suggest that the active factors are heat-stable proteins. Supporting this conclusion, we also found that the active factors were soluble in the aqueous fraction following hydrophobic extraction ( Figure 8B). These data suggest the S. aureus factors responsible for promoting P. aeruginosa surface motility are proteinaceous in nature, but the activity is insensitive to denaturation by heat.
Together our biochemical and genetic evidence predict that heat-stable, Agr-regulated proteins and/or peptides promote P. aeruginosa motility. S. aureus produces such peptides referred to as phenol-soluble modulins (PSM). PSMs are amphipathic, multifunctional peptides that lyse many human cell types, are pro-inflammatory and chemoattractant to neutrophils, and influence S. aureus biofilm morphology and dispersal (Cheung et al., 2014). S. aureus produces five apeptides (PSMa1-4 and d-toxin) and two b-peptides (PSMb1 and 2). PSMa are encoded by the psma operon, PSMb by the psmb operon, and the d-toxin (also called d-hemolysin, Hld) is encoded within the RNAIII, regulatory RNA encoded from the agr operon. To determine if S. aureus PSMs might influence P. aeruginosa twitching motility, we collected supernatant from S. aureus deficient in the production of all seven PSMs, Dpsma1-4 Dpsmb1-2 and d-toxin start (ATG) to stop (ATT) mutation, to preserve expression of RNAIII (referred to as Dpsmabd, from this point forward) (Syed et al., 2015). Supernatant derived from the Dpsmabd mutant showed a significantly reduced ability to promote P. aeruginosa motility in comparison to WT; however, the activity remained higher than DagrBDCA, suggesting one or more additional unidentified S. aureus factors also promote P. aeruginosa motility ( Figure 8C).

Discussion
Current massive sequencing efforts yield unprecedented information regarding microbial community composition during infection, but fail to provide the necessary information required to functionally understand microbial behaviors in mixed species communities. Moreover, while the bulk assays most frequently utilized to study microbial interactions have provided insight into how bacterial species can influence community survival, metabolism, or virulence factor production (Hotterbeekx et al., 2017), we lack detailed information regarding how single-cells respond to the presence of another species at initial encounters. Here, we developed quantitative live-imaging methods to visualize the behaviors of two clinically important organisms and track their behavior over time. Through these studies, we uncovered microbial behaviors that could not have been predicted from bulk assays alone. We observed that P. aeruginosa is able to alter its behavior in the presence of S. aureus from a distance by responding to secreted factors. P. aeruginosa is then able to tune its motility patterns, adopting a behavior we refer to as 'exploratory motility'.
What interspecies factors do P. aeruginosa sense and respond to? Our data thus far reveal that P. aeruginosa can not only respond to secreted products from S. aureus, but also more broadly to a range of model bacteria and CF pathogens, both Gram-positive and negative. P. aeruginosa increased surface motility to varying extents depending on the species and strain utilized -a feature we predict will inform the nature and breadth of factors to which P. aeruginosa is capable of responding. For S. aureus, we determined the factors are heat stable proteins regulated by the Agr quorum sensing system. This insight led to the determination that S. aureus PSMs contribute to the ability of S. aureus to induce surface motility in P. aeruginosa, but do not fully account for the observed phenotype, suggesting that additional unidentified factors are necessary. Moreover, S. aureus may also produce Agr-independent factors that influence P. aeruginosa motility, given that an Agr-deficient strain is unable to induce motility in the macroscopic sub-surface inoculation motility assay or directional motility on either the macroscopic or single-cell level, but remains capable of increasing single-cell motility during live-imaging. Thus, it is formally possible that separate factors are required for initiating single-cell motility and directional motion. Live-imaging also suggests that P. aeruginosa cells may actively avoid Agr-deficient S. aureus, raising the possibility that the Agr system negatively regulates a secreted factor that is unfavorable for P. aeruginosa.
How do PSMs promote P. aeruginosa motility? PSMs are amphipathic peptides that are able to reduce interfacial tension (IFT), that is, the cohesive or excess energy between molecules at an interface. While such surfactants have been extensively characterized to play an essential role in flagella-   LB: B. cepacia, A. xylosoxidans, B. subtilis, E. coli, and S. Typhimurium). The mean and standard deviation are indicated for at least three biological replicates. Statistical significance was determined by one-way ANOVA followed by Tukey's Multiple Comparisons Test -a indicates a statistically significant difference (p 0.05) between the motility observed in the presence of S. aureus supernatant compared to medium alone (A-D), and b indicates a statistically significant difference (p 0.05) between the motility observed in the CF isolates, in comparison to laboratory strains (P. aeruginosa laboratory strain PA14 in (A) and S. aureus USA300 LAC in (B)). The online version of this article includes the following figure supplement(s) for figure 7: mediated bacterial surface motility, referred to as swarming (Kearns, 2010), their role in TFP-mediated motility is not well described. TFP generate bacterial movement on surfaces though polymerization and extension of pilus fibers, followed by pilus retraction, which pulls the cell body forward. Reduced IFT between the cell body and the surface, without affecting interaction between the pilus tip and the surface, may reduce drag on the cell body and reduce the force necessary to retract the pilus. TFP-mediated motility is most often described as a behavior where cells are primarily motile only in large groups. The behaviors for WT P. aeruginosa in monoculture observed here are consistent with previous descriptions of twitching motility, including the movement of cells in rafts, preferentially aligned along their long axis, and group tendril formation at the edges of the expanding community (Burrows, 2012). In contrast, in the presence of S. aureus, P. aeruginosa cells were seen to exit an expanding raft and transition to single-cell motility. Perhaps PSMs also reduce the IFT between neighboring P. aeruginosa cells, disrupting collective movement within rafts.
PSMs are known to promote S. aureus spreading on wet surfaces, in particular PSMa3 and dtoxin, which possess high surfactant activity and low hydrophobicity (Tsompanidou et al., 2013). In the current study, we examined a mutant deficient in all seven PSMs, thus it is unknown if each S. aureus PSM possesses similar ability to promote motility and/or if their mechanism of action toward P. aeruginosa will also depend on their specific surfactant activity. Motility experiments in the presence of B. subtilis support the hypothesis that surfactants may increase TFP-mediated motility in P. aeruginosa. Two strains were examined, PY79 and 3610. 3610 induced P. aeruginosa motility similar to S. aureus. In comparison to PY79, which did not induce motility and is deficient in surfactant production, 3610 is an undomesticated strain that secretes a potent lipopeptide surfactant, called surfactin (Kearns and Losick, 2003). Thus, it is interesting to speculate that the different levels of Figure 8. Phenol soluble modulins contribute to S. aureus-induced P. aeruginosa motility. Motility of WT P. aeruginosa was monitored by macroscopic sub-surface inoculation assays in the presence of medium alone or cell-free supernatant derived from WT S. aureus treated with either heat, proteinase K (PK), or trypsin (tryp) in (A), methanol/chloroform extraction of the supernatant and subsequent analysis of the organic and aqueous fraction (methanol-chloroform (2:1, Me/Chlor) was included as the vehicle control) (B), or in the presence of untreated supernatant derived from WT or the indicated S. aureus mutants (0.5 dilution) in (C). The mean and standard deviation are indicated for at least three biological replicates. Statistical significance was determined by one-way ANOVA followed by Tukey's Multiple Comparisons Test -a indicates a statistically significant difference (p 0.05) between the motility observed in the presence of S. aureus supernatant compared to medium alone, b between the motility observed in the mutant strains compared to the parental, and c between the DagrBDCA and Dpsmabd mutants.
surfactin account for the differential ability of these B. subtilis strains to promote P. aeruginosa motility.
PSMs can also form small transient pores in lipid membranes resulting in cell lysis of eukaryotic and some prokaryotic cells (Cheung et al., 2014). Thus, it is possible that interactions of PSMs with the P. aeruginosa outer membrane influence P. aeruginosa surface motility, independent of reduced IFT at the surface. For example, surfactin has been shown to influence B. subtilis quorum sensing and biofilm formation through the formation of membrane pores, resulting in potassium leakage. The resulting low cellular potassium activates the protein kinase, KinC, which regulates biofilm formation (Ló pez et al., 2009). Future studies will investigate if PSMs influence P. aeruginosa surface motility by reduction of IFT, interactions with the P. aeruginosa outer membrane, and/or alternative mechanism(s). Future studies are also necessary to determine if PSMs are sufficient to promote directional P. aeruginosa motility or if PSMs function in concert with additional S. aureus factors to increase surface motility per se in order to facilitate movement towards an additional stimulus.
While flagella-mediated chemotaxis has been extensively studied in P. aeruginosa, little is understood regarding how bacteria direct movement on a surface utilizing the TFP. While the P. aeruginosa Pil-Chp system harbors similarity to the well-studied Che system in E. coli, initial studies uncover significant differences in the regulation of this pathway. PilJ is a predicted chemoreceptor, by sequence similarity to the flagella chemotaxis system (Darzins, 1994); however, the functional role in chemotaxis remains unclear. In initial studies of TFP-mediated chemotaxis, P. aeruginosa was shown to direct motility on agar surfaces up a gradient of phosphatidylethanolamine (PE) (Kearns et al., 2001). While a DpilJ mutant did not exhibit preferential movement up a PE gradient, the lack of twitching motility in this mutant prohibited interpretation of a role for PilJ. Here, despite the lack of development of a macroscopic twitching motility zone, individual cells were capable of initiating motility in the presence of S. aureus when monitored by live-imaging, suggesting that P. aeruginosa is capable of sensing S. aureus and performing twitching motility in the absence of functional PilJ.
How does P. aeruginosa sense an external stimulus and transmit this into a mechanical response driving twitching motility? Twitching cells move by pulling themselves along by pili clustered at one pole. Cells can reverse direction by extending the pili from the opposite pole, changing the direction of movement along the long axis. These reversals have recently been suggested to be important for P. aeruginosa to direct movement up a chemotactic gradient while on a surface (Oliveira et al., 2016). In a DpilG mutant (the response regulator for pilus extension), cells were observed to perform fewer reversals in comparison to WT in response to a chemotactic gradient and thus were unable to bias the directionality of their movement. Here we examined a role for PilG in modulating a response to secreted factors from S. aureus and found a DpilG mutant was significantly diminished in the ability to respond. Both DpilJ and DpilG mutants have been previously reported to be deficient in producing traditional twitch rings in the macroscopic sub-surface inoculation assay (Luo et al., 2015;Bertrand et al., 2010;Darzins, 1994). Our observations here for PilJ and PilG, combined with those from Oliveira et al. suggest the collective movements of these mutants as measured in bulk assays may not be reflected on the single-cell level. However, in comparison to studies by Oliveira, we observed very little movement for P. aeruginosa in the absence of PilG. These studies utilized the P. aeruginosa parental strain PAO1 (in comparison to PA14 used here). We hypothesized that differences in parental strain background (PAO1 verses PA14) may account for the differences observed; however, we acquired the PAO1 parental and PAO1 DpilG mutant utilized in these studies and obtained similar results as observed for PA14 in coculture with S. aureus ( Figure 6-figure supplement 2). Therefore, it is likely that experimental conditions influence to what extent P. aeruginosa is capable of performing TFP-mediated motility in the absence of PilG.
Myxococcus xanthus is one of the most notorious predatory bacteria -elaborating an array of social behaviors reminiscent of what we observe here for P. aeruginosa. M. xanthus coordinates a cooperative, density-dependent feeding behavior, resulting in propulsion of the cells rapidly through the colony of prey, leading to prey lysis, and nutrient acquisition. When M. xanthus contacts prey cells, pili-dependent reversals are stimulated, which keeps the cells 'trapped' near the prey colony, promoting contact-dependent prey killing and increased local concentration of antimicrobials (Muñoz-Dorado et al., 2016). Similarly, we observed that when P. aeruginosa encounters S. aureus, the P. aeruginosa cells appear to mount a coordinated response whereby P. aeruginosa surrounds the S. aureus colony and eventually invades the colony -a behavior referred to as the 'wolf pack' strategy. P. aeruginosa produces an arsenal of secreted antimicrobials shown to inhibit the growth of S. aureus, yet these secreted factors alone are insufficient for cellular lysis (Limoli et al., 2017). Thus, it is interesting to speculate that these behaviors function to synergistically increase cellular contacts necessary to kill S. aureus by a contact-dependent mechanism and/or to locally increase the concentration of secreted products.
By acquiring a fundamental understanding of how bacteria sense and respond to life with each other, we move closer to learning how to rationally manipulate interspecies behaviors during infection and in the environment. While for CF patients, this may mean preventing P. aeruginosa and S. aureus physical interactions, in other instances, we might bring species together who synergize to produce a beneficial compound.

Materials and methods
Key resources S. aureus genetic manipulation. In-frame deletion of the agrBDCA operon in a S. aureus USA300 LAC strain (JE2) was generated using pMAD-mediated allelic replacement (Arnaud et al., 2004). Briefly, a pMAD deletion vector was created by Gibson assembly (Gibson et al., 2009) using EcoRI/ BamHI-linearized pMAD and PCR amplicons of 1 kb regions up-and downstream of agrBDCA (primer sets agr-a/agr-b and agr-c/agr-d, respectively, Table 1). The vector to chromosomally complement this deletion strain was similarly produced using Gibson assembly on the EcoRI/BamHI-lineararized pMAD vector and the PCR product of the agr-a/agr-d primers. Following the standard protocol of heat shift on selective media, strains with successful deletions and restorations of agrBDCA were verified by PCR analysis and chromosomal DNA sequencing.

Live-imaging of interspecies interactions
Bacteria were inoculated between a coverslip and an agarose pad and imaged with time-lapse microscopy. Pads were made by adding 900 ml of M8 medium with 1% glucose, 1% tryptone, and 2% molten agarose to a 10 mm diameter silicone mold on a coverslip. Pads were allowed to dry for 1 hr at room temperature, followed by 1 hr at 37˚C. Meanwhile, bacteria were prepared by growing to mid-log phase in M8 medium with 1% glucose and 1% tryptone, diluting to OD 600 = 0.15 in warm media, and mixing P. aeruginosa and S. aureus 1:1. 2 ml of inoculum was spotted evenly onto the center of a warm 35 mm glass bottom dish, #1.5 mm coverglass (Cellvis). The agarose pad was removed from the mold and placed on top of the bacterial inoculum. Bacteria were immediately imaged with an inverted Nikon TiE or Ti2 at 37˚C for 8 hr. Images were acquired with a 100x oil objective (1.45NA) with phase contrast and an ORCA Flash4.0 Digital CMOS camera (Hammamatsu).
Videos were generated and cells tracked and analyzed in Nikon Elements AR. For single-cells, binary images were generated and single P. aeruginosa cells were identified and tracking began when the first P. aeruginosa cell exited a raft. Rafts were tracked up until the time point where single-cell tracking began. Rafts edges were manually identified. The speed (mm/s), acceleration (mm/ s 2 ), mean squared displacement (mm 2 ), accumulated track distance, D (A) , and the Euclidean distance, D (E) , were measured for at least 30 tracks (for single-cells) in four independent videos. The area of S. aureus was determined by dividing the total area occupied in a field of view divided by the number of colonies.
Macroscopic coculture assays P. aeruginosa sub-surface twitch (Turnbull and Whitchurch, 2014) and 0.3% soft agar motility assays (Ha et al., 2014) were performed as previously described with modifications for treatment with bacterial supernatant. Supernatant was derived by growing the indicated species overnight centrifuging to pellet cells, and supernatant filtered through a 0.22 mm filter. S. aureus was grown in TSB and strains normalized to OD 600 = 5.0 (and subsequently diluted to the indicated concentrations). For soft agar assays, cell-free supernatant was mixed at the indicated concentrations with cooled, but still molten agar prior to pouring. For subsurface inoculation assays, 100 ml of supernatant was spread onto the bottom of a petri plate before pouring media (1.5% agar). P. aeruginosa was grown to OD 600 = 2.0 in TSB and inoculated into motility plates by either stabbing with a toothpick halfway through the agar (0.3%), all the way though (sub-surface). Plates were incubated at 37˚C for 24 hr, followed by 24 hr at room temperature. For twitch plates, agar was dropped out and the P. aeruginosa biomass stained with 1% crystal violet for visualization. The diameter of the motility zones was measured in mm.
Biochemical analysis of S. aureus supernatant Supernatant was collected as described above and exposed to the following conditions: 1. Heat: 95C for 30 min. 2. Protease treatment: Proteinase K (400 mg/ml) at 55˚C for 30 min or trypsin (2.5 mg/ ml) at 37˚C for four hours with gentle shaking, followed by heat inactivation at 95˚C for 20 min. 3. Hydrophobic extraction: Performed according to the guidelines of Bligh and Dyer (1959). In brief, a volume of 3.75 ml of methanol-chloroform (2:1) was added to 1.6 ml of S. aureus supernatant and vortexed for 1 hr. The suspension was centrifuged at 10,000 x g for 5 min and the aqueous phase was saved. The organic phase was then reextracted with 4.75 ml of methanol-chloroform-water (2:1:0.8) and centrifuged again, and the extracted organic phases were combined.

Directional twitching motility assay
Experiments were performed as previously described (Miller et al., 2008). In brief, buffered agar plates (10 mM Tris, pH 7.6; 8 mM MgSO 4 ; 1 mM NaPO 4 , pH 7.6; and 1.5% agar) were poured and dried at room temperature for 24 hr. 4 ml of supernatant or medium (TSB) was spotted on the plates, and gradients were established by incubating the plates at 30˚C for 24 hr. Supernatants were prepared as described above. P. aeruginosa strains were grown to early stationary phase, normalized to an OD 600 of 1.2 and 2 mls pelleted and resuspended in 100 ml of MOPS buffer (10 mM MOPS, pH 7.6, and 8 mM MgSO 4 ). 2 ml of this cell suspension was spotted approximately 6 mm from the center of the supernatant spot. Plates were incubated for 24 hr at 37˚C followed by 24 hr at room temperature before images were taken of the twitching zones.