Social Motility: Interaction between two sessile soil bacteria leads to emergence of surface motility

Bacteria often live in complex communities in which they interact with other organisms. Consideration of the social environment of bacteria can reveal emergent traits and behaviors that would be overlooked by studying bacteria in isolation. Here we characterize a social trait which emerges upon interaction between the distantly-related soil bacteria Pseudomonas fluorescens Pf0-1 and Pedobacter sp. V48. On hard agar, which is not permissive for motility the mono-culture of either species, co-culture reveals an emergent phenotype we term ‘social motility,’ where the bacteria spread across the hard surface. We show that initiation of social motility requires close association between the two species of bacteria. Both species remain associated throughout the spreading colony, with reproducible and non-homogenous patterns of distribution. The nutritional environment influences social motility; no social behavior is observed under high nutrient conditions, but low nutrient conditions are insufficient to promote social motility without high salt concentrations. This simple two-species consortium is a tractable model system that will facilitate mechanistic investigations of interspecies interactions and provide insight into emergent properties of interacting species. These studies will contribute to the broader knowledge of how bacterial interactions influence the functions of communities they inhabit.


18
Within the soil live a plethora of microbial species that form complex communities 19 responsible for important ecological functions, such as nutrient cycling and plant health. Omics 20 approaches have given us a wealth of information on the composition, diversity, metabolic 21 potential, and ecology of plant-and soil-associated microbial communities (Reviewed in 22 Philippot et al., 2013;Fierer, 2017). However, to get a complete understanding of microbial 23 functions and interactions within these environments, we must look at every layer, from the full 24 community in vivo to the individual microbe in vitro (Reviewed in Abreu and Taga, 2016). 25 Historically, research has focused on the study of single species in pure culture, but bacteria are 26 social organisms, and thus study of the mechanisms and consequences of multi-species 27 interactions is necessary for us to understand the function of microbial communities as a whole. 28 Investigating entire soil communities in situ presents considerable challenges because of 29 fluctuating soil conditions and the wide range of relevant scales, ranging from particulate to 30 ecological levels (Reviewed in Fierer, 2017). Reducing the microbial community to pair-wise 31 interactions or small consortia allows for a detailed mechanistic study and is an essential link 32 between from the study of isolated microbes in the laboratory to understanding the collective 33 activities of natural microbial communities (Blasche et al., 2017). B. subtilis moves away from a Streptomyces competitor across a solid surface (Stubbendieck 48 and Straight, 2015; Liu et al., 2018). Other behaviors appear less competitive, where a motile 49 species will travel with a non-motile species that can degrade antibiotics, allowing the 50 Xanthomonas perforans can even change the behavior of Paenibacillus vortex, producing a 52 signal that induces P. vortex to swarm towards it so it can hitchhike (Hagai et al., 2014). 53 Pseudomonas fluorescens Pf0-1 and Pedobacter sp. V48 are known to interact though 54 diffusible and volatile signals, which induce changes in gene expression and production of an 55 antifungal compound by P. fluorescens (Garbeva et al., 2011a(Garbeva et al., , 2011b(Garbeva et al., , 2014. Previous studies 56 with Pedobacter and a strain closely-related to P. fluorescens Pf0-1 (AD21) found that, in 57 addition to reciprocal gene expression changes and antagonistic behavior toward the plant 58 pathogen Rhizoctonia solani, the mixture of the strains also showed expansion on the plate 59 beyond the initial area of inoculation (de Boer et al., 2007;Garbeva and de Boer, 2009). We 60 further investigated this observed behavior by moving from culturing P. fluorescens Pf0-1 and 61 Pedobacter without contact, as was done in the antagonism assays (Garbeva et al., 2011a), to 62 mixing them together. We hypothesized that, while antibiotic production can be induced at a 63 distance through diffusible or volatile signals, the motility behavior requires close contact and is 64 therefore controlled in a manner distinct from the other two forms of communication. 65 In this study, we describe an interaction between two distantly-related soil bacteria, P. 66 fluorescens Pf0-1 (phylum: Proteobacteria) and Pedobacter sp. V48 (phylum: Bacteroidetes). 67 This interaction produces an emergent behavior, which we term "social motility," in which the 68 bacteria move together across a hard agar surface. When grown in isolation, neither species 69 moves beyond the normal amount of colony expansion. In co-culture, both bacteria are present 70 throughout the motile colony, and fluorescent imaging shows a non-homogenous distribution. 71 We demonstrate that a close association between the colonies of both species is required for 72 motility to initiate and that the levels of nutrients and salts in the media affect the development of 73 the motile phenotype. 74

Social motility arises when mixing two distantly-related bacteria. 76
In previous studies, antifungal activity was observed when P. fluorescens Pf0-1 and 77 Pedobacter sp. V48 were cultured 15 mm apart (Garbeva et al., 2011a). In addition to this 78 interaction-induced trait, the possibility of motility was noted in a mixture of P. fluorescens AD21 79 and Pedobacter (de Boer et al., 2007;Garbeva and de Boer, 2009). When we plated P. 80 fluorescens Pf0-1 and Pedobacter on TSB-NK medium solidified with 2% agar a mixed colony 81 of the two bacteria expanded across the surface of the agar, an environment in which neither 82 monoculture exhibited motility. The emergent social motility is shown in Fig. 1. 83 Figure 1. Mixed colony of P. fluorescens Pf0-1 and Pedobacter sp. V48 spreads across a hard agar surface (2%), a behavior not observed in the mono-culture of either species. a) Diameter of colonies at 24 h intervals. b) Phenotypes of mono-and co-cultures at 24 h intervals. Contrast and brightness levels were adjusted for optimal viewing. Social motility becomes apparent between 24 and 48 h after inoculation, when the 84 colony begins to spread from the edge of the inoculum (Fig. 1b). The diameter of the motile co-85 culture is significantly different from the colony expansion of the mono-cultures starting at the 24 86 h time point (p < 0.001) (Fig. 1a). Once the motility phenotype is fully visible (around 72 h), the 87 average speed of expansion is 1.69 µm/min +/-0.09 (s.e.m). At the onset of movement, the 88 leading edge has a visibly thicker front (Fig. 1b 48 h). As the colony spreads, the thick front 89 disappears and small 'veins' radiating from the center develop. Over time, the 'veins' become 90 more pronounced towards the leading edge, making a 'petal' pattern (Figs. 2a, b). The leading 91 edge is characterized by a distinctive, terraced appearance comprised of three to six layers (Fig.  92 2c). 93 absence of each species was tested by culturing these samples on selective media. We 101 recovered both species from each point in the motile colony (data not shown), showing co-102 migration rather than an escape strategy by Pedobacter. 103 To obtain a more detailed look at the spatial relationships within the motile colony, we 105 tagged P. fluorescens with a cyan fluorescent protein (eCFP [Choi and Schweizer, 2006

]) and 106
Pedobacter with a red fluorescent protein (dsRedEXPRESS [Choi and Schweizer, 2006]), 107 integrated into the chromosome. In P. fluorescens, eCFP carried by miniTni7 was integrated 108 upstream of glmS (Lambertsen et al., 2004), creating Pf0-ecfp. In Pedobacter, dsRedEXPRESS 109 carried by the HimarEm transposon (Braun et al., 2005) was integrated at random locations in 110 the chromosome, resulting in 16 independently-derived mutants with an insert. Each tagged 111 Pedobacter strain (V48-dsRed) was indistinguishable from the wild-type in social assays with P. appear more well-mixed, though the red signal becomes difficult to detect toward the edge of 131 the colony (Fig. 3d). Overall, imaging data show that we can find both species throughout the 132 colony, but the distribution is not homogenous. Rather, we observed reproducible patterns with 133 some well-mixed areas and others of high spatial assortment. 134

Physical association of P. fluorescens Pf0-1 and Pedobacter V48 is required for social 135 motility 136
Previous studies demonstrated interactions between P. fluorescens and Pedobacter 137 were mediated via both diffusible and volatile signals (Garbeva et al., 2011a(Garbeva et al., , 2011b(Garbeva et al., , 2014. We 138 asked whether a close association between the two bacteria was a necessary condition for 139 social motility or whether signaling via diffusible compounds could trigger the movement. To 140 answer this question, we used assays in which the bacterial participants were plated side-by 141 side with no physical barrier and in which they were separated by semi-permeable membranes. 142 When colonies were adjacent, rather than mixed, no social motility was observed while 143 the P. fluorescens and Pedobacter colonies were visibly separate (data not shown). However, 144 once the colonies grew sufficiently to make contact ( Fig. 4 24 h), the colony started to spread 145 out from the point of contact (72 h). The spreading front radiates outward (96 h), first developing 146 around the P. fluorescens colony (144 h), then proceeding to surround the Pedobacter colony 147 (192 h). At this level of resolution, contact between the colonies appears to occur before any 148 spreading can be seen. 149 Samples were collected from the edge of the moving front every 24 hours after contact, 150 both on a y-axis from the point of contact and following the moving front as it wrapped around 151 the P. fluorescens colony (Fig. 4). The presence of each species was tested by culturing these 152  To further evaluate the requirement that P. fluorescens and Pedobacter be physically 157 associated, we inoculated both strains immediately adjacent to each other but separated by 158 either semi-permeable mixed-ester cellulose or PES (polyethersulfone) membranes. When 159 inoculated this way, individual colony growth continued as normal, but these bacteria were 160 unable to trigger social motility despite their close proximity. After six days of growth, no sign of 161 social motility was observed (Fig. 5). 162 Conditions in soil and rhizosphere environments fluctuate, with bacteria subjected to a 164 wide range of environmental stressors, including limited nutrient and water availability (Fierer, 165 2017). Because such fluctuations may influence expression of traits, we examined the effect of 166 nutrient level on social motility. Our standard assay condition, TSB-NK, consists of 10% strength 167 Tryptic Soy (3 g/L) supplemented with NaCl (5 g/L) and KH2PO4 (1 g/L) (Figs. 1b and 6b). 168 We first asked if social motility could initiate under richer nutrient conditions. No social 169 motility was apparent when P. fluorescens and Pedobacter were mixed on full-strength TSB (30 170 g/L) (Fig. 6a), with the co-culture exhibiting the same characteristics and colony expansion as 171 the P. fluorescens mono-culture. We next asked whether the salt amendments to TSB-NK 172 influence social motility, using assays without the addition of salts, and with the addition of NaCl 173 and KH2PO4 individually. When grown on 10% TSB, the co-culture is motile, but the distance 174 moved is modest compared to when the medium is supplemented with both salts (Fig. 6c). The 175 individual P. fluorescens colony expands similarly to the co-culture, suggesting minimal social 176 behavior under these conditions. Growth on TSB-K changes neither pattern nor rate of mono-177 and co-culture expansion compared to 10% TSB (data not shown). On TSB-N, the mixed 178 culture spreads and develops the patterns characteristic of social motility, while the P. 179 fluorescens mono-culture does not expand (Fig. 6d). The phenotype and diameter of the 180 spreading colony are most similar to those observed in TSB-NK conditions (Fig. 6b). 181  (Figs. 6b, d). 197 Based on these results, we conclude that full social motility expansion was only 198 observed in low nutrient medium supplemented with NaCl (Figs. 6b, d, and 7g). We observed 199 reduced social motility on low nutrient media without salt supplementation (Figs 6c and 7h

220
In this study we investigate social motility, a phenomenon that emerges from the 222 interaction of two distantly-related soil bacteria. Neither species moves on its own, but a mixture 223 of the two species can spread across a hard agar surface (2%). Contact between the two 224 bacterial colonies is required for motility to initiate, and this association is maintained as the co- Bacillus subtilis (Lazazzera, 2000), and yeast (Chen and Fink, 2006).  (Hagai et al., 2014). 292 In addition to describing a new mode of motility, this discovery highlights the possibility 293 that many functions and behaviors of bacteria in complex communities may be triggered by 294 interactions between different species or even domains. Studying interactions between two or 295 more microorganisms may lead to the discovery of emergent traits that would be impossible to 296 predict based on the study of each organism in isolation. Alongside approaches that 297 characterize the members and connectedness of microbial communities, tools to decipher the 298 phenotypic outcomes of interactions are needed in order to develop a full appreciation of 299 microbiomes. Studies of this type are important for understanding the role of microbial 300 communities within an ecological context. 301 We have investigated an interaction-dependent trait which emerges under particular 302 nutritional conditions when distantly-related bacteria come into close physical contact. This 303 social motility gives the participating bacteria the ability to spread on a hard agar surface, which 304 neither can do alone. This strategy of co-migration may serve as an additional mechanism by 305 which plant-and soil-associated bacteria can move in their natural environments, when the 306 conditions do not favor the modes of single-species motility previously described. Given the 307 distant and different locations from which these two strains were isolated, we hypothesize this is 308 not a unique interaction between this pair, but rather has evolved between various Pedobacter 309 and Pseudomonas species. To understand the phenomenon, several lines of investigation 310 should be pursued: mechanistic studies which explore the factors each species is contributing to 311 social motility, the process by which contact triggers motility, and the way in which 312 environmental conditions are integrated into the decision to move together. Such studies will 313 enable the application of our findings to the search for new examples of interaction-mediated 314 behaviors among bacteria. 315 Bacterial strains, primers, plasmids, and culture conditions. Bacterial strains and plasmids 317 are described in Table 1 coli S17-1, with transposase being provided by pUX-BF13 introduced from a second E. coli 361 S17-1 donor, as previously described (Monds et al., 2006). Transposon-carrying strains were 362 selected by growth on Gentamicin (50 µg/mL), and transposition of the miniTn7 element into the 363 target site in the P. fluorescens genome was confirmed by PCR using primers Tn7-F and glmS-364 R ( Table 2). Pf0-1 with fluorescent inserts were tested for alteration in social motility by co-365 culturing with Pedobacter, as described above. 366 (ii) dsRedEXPRESS labeling Pedobacter. pUC18T-mini-Tn7T-Gm-dsRedExpress was a gift 367 from Herbert Schweizer (Addgene plasmid #65032). To express dsRedEXPRESS in 368 Pedobacter, a Pedobacter promoter was cloned upstream of the dsRedEXPRESS coding 369 sequence. A highly expressed gene from an unpublished RNAseq experiment was identified 370 (N824_RS25200) and the upstream 320 bp were amplified from Pedobacter genomic DNA 371 using primers PompA and dsRed, designed for splicing-by-overlap extension-PCR (SOE-PCR) 372 (Table 2). The promoter was then spliced with the amplified dsRedEXPRESS coding sequence 373 using SOE-PCR (Horton et al., 1989). Flanking primers were designed with KpnI restriction 374 sites, enabling cloning of the spliced product into a KpnI site in pHimarEm1 (Braun et al., 2005). 375 To join compatible ends between the plasmid and the amplicons, we used T4 DNA ligase (New 376 1 competent cells by electroporation (BioRad Micropulser™, Hercules, CA, U.S.A.). S17-1 378 colonies carrying the plasmid were selected by plating on LB medium containing Kanamycin (50 379 µg/mL), and the presence of the dsRedEXPRESS gene was confirmed by PCR, using pHimar 380 KpnI-flank primers ( Table 2). The resulting plasmid is called pHimarEm1-dsRed. 381 pHimarEm1-dsRed was transferred to Pedobacter by conjugation using a method adapted from 382 Hunnicutt and McBride, 2000. Briefly, 20 hour old cultures of E. coli S17-1 (pHimarEm1-dsRed) 383 and Pedobacter were subcultured 1:100 into fresh LB, and grown to mid-exponential phase (E. 384 coli) or for 7 hours (Pedobacter). Cells were collected by centrifugation, suspended in 100µL of 385 LB, and then mixed in equal amounts on TSB-NK with 100 µL of 1M CaCl2 spread on the 386 surface. Following overnight incubation at 30°C, cells were scraped off the surface of the plate, 387 and dilutions were plated on TSB-NK with Erythromycin (100 µg/mL) to select for strains that 388 received the plasmid (ermF is not expressed in E. coli). Transconjugants were incubated at 389 25°C for 3-4 days. Presence of the transposon in Pedobacter was confirmed using ermF 390 primers (Table 2). Using this software, the levels of some images were adjusted to improve contrast, and pictures 401 were converted to greyscale. 402 For microscopy, motile colonies were examined using an Axio Zoom.V16 microscope (Zeiss, 403 Oberkochen, Germany). To visualize fluorescent strains, filter set 43 HE DsRed was used with a 404 1.5 s exposure, shown with pseudo-color orange, as well as filter set 47 HE Cyan Fluorescent 405 Protein, with a 600 ms exposure, shown with pseudo-color turquoise. Images were captured 406 using Axiocam 503 mono camera, with a native resolution of 1936x1460 pixels. For image 407 acquisition and processing we used Zen 2 Pro software (Zeiss). 408 409 We measured the amount of colony expansion of the mono-cultures of both P. fluorescens and 410

Statistics
Pedobacter and the expansion of social motility in co-culture. Colony diameter of three 411 independent experiments was measured every 24 hours. To compare the diameter of mono-412 cultures and co-cultures at each time point, we performed a two-way ANOVA followed by a 413 Bonferroni post-hoc test. 414 We compared the movement speed between a combination of wild type P. fluorescensand 415 Pedobacter to a combination of fluorescently-tagged Pf0-ecfp and V48-dsRed. Colony diameter 416 of six independent experiments were measured every day, and speed was calculated by 417 dividing the distance traveled by the amount of time elapsed since the last time point. To 418 calculate average speed, we only used time points after social motility phenotype developed. To 419 compare the means of the speed of the wild-type and tagged strains, we conducted an 420 unpaired, two-tailed, Student's t-test. 421