Interspecific competition can drive the loss of conjugative plasmids from a focal species in a microbial community

Plasmids are key disseminators of antimicrobial resistance genes and virulence factors, and it is therefore critical to predict and reduce plasmid spread within microbial communities. The cost of plasmid carriage is a key metric that can be used to predict plasmids’ ecological fate, and it is unclear whether plasmid costs are affected by growth partners in a microbial community. We carried out competition experiments and tracked plasmid maintenance using a synthetic and stable 5-species community and a broad host-range plasmid as a model. We report that both the cost of plasmid carriage and its long-term maintenance in a focal strain depended on the presence of competitors, and that these interactions were species-specific. Addition of growth partners increased the plasmid cost to a focal strain, and accordingly plasmid loss from the focal species occurred over a shorter time frame in these species combinations. We propose that the destabilising effect of interspecific competition on plasmid maintenance may be leveraged in clinical and natural environments to cure plasmids from focal strains.

65 Moreover, these costs are strongly dependent on the environment; plasmids that are 66 costly in the absence of antibiotics or heavy metals can in turn become highly 67 beneficial in their presence if they encode resistance genes [8].
68 Theory and data suggest that costly plasmids can be lost readily from bacterial 69 populations or communities due to purifying selection, unless conjugation rates -either 70 within or between species -are sufficiently high to support their maintenance [10][11][12].
71 For example, bacteria that lose a plasmid when cultured on their own may still 72 associate with this plasmid when co-cultured with another species, due to high rates 73 of interspecific plasmid transfer [10]. Hence, even bacteria that are unable to maintain 74 plasmids in monoculture may experience increased plasmid persistence in a microbial 75 community.
76 Recent work has shown that fitness effects of chromosomal mutations in bacteria can 77 depend on the microbial community context [13,14]. We therefore hypothesised that 78 the maintenance of plasmids in focal species may similarly be affected by the microbial  156 Stars indicate treatments with significantly lower GFP+ fraction than in monoculture.
157 See Table S2 for values of these summary statistics. *p<0.05, **p<0.01, ***p<0.  171 Additionally, we found that several species proportions, such as Stenotrophomonas 172 and Achromobacter, co-correlated ( Figure S3). Therefore, this simplistic analysis of 8 174 effect driven by a single monoculture treatment, correlation between individual 175 variables, removal of different subsets of data when excluding absent species, and 176 multiple statistical testing. To address all these issues, we instead fitted a single 177 statistical model to all available data, which we used above to analyse the various 178 treatments in Figure 2. This also revealed that only Stenotrophomonas and 179 Achromobacter proportion had statistically significant effects on Variovorax plasmid 180 maintenance (Table S3). There was a clear negative association between the 181 proportion of Stenotrophomonas and the plasmid-bearing Variovorax fraction. 226 Based on these results, we sought to explore whether the observed effects were 227 unique to Variovorax and pKJK5::gfp PL or if they could be observed in other host-228 plasmid pairs. To exclude the possibility that plasmid maintenance in the presence of 229 growth partners may have been influenced by the additional payload genes inserted 230 into pKJK5::gfp PL ( Figure S1b) and to expand the analysis to a plasmid with smaller 231 predicted fitness cost, we used plasmid pKJK5::gfp which lacks these payload genes 232 (Figure S1c 294 Under culture conditions which favour long-term species coexistence, little to no 295 conjugation took place (Table S1). 296 Using this model community, we found that the fitness cost of plasmid carriage can 297 change as a result of interspecific competition. We found that embedding focal strain 13 299 and more rapid loss of this plasmid. The mechanism of plasmid loss in our system is 300 unknown; but both higher intrinsic plasmid loss rates and higher relative growth of 301 plasmid-free segregants would lead to the same observed outcome of population-level 302 plasmid loss, and both can be driven by increased cost of plasmid carriage. This 303 increased cost may be a result of resource limitation during growth with competitors.
304 Generally, plasmid maintenance is dependent not only on fitness costs and benefits 305 to its host, but also the plasmid transfer and loss rate [20,21]. Of these, plasmid 306 transfer can also depend on the community context: In monoculture, a mercury 307 resistance plasmid was rapidly lost from its host species, but with growth partners, the 308 plasmid was maintained in the focal strain due to reinfection by conjugation [10,22].      434 Monocultures and communities were cultured for three days, vortexed, and 100 µL of 435 each culture transferred into a fresh microcosm and incubated for another two days.
436 Communities were then further passaged into fresh microcosms the same way every 437 two days until 17 total days of co-culture (Community Maintenance experiment only).