To examine whether genetic adaptation to the plant root had taken place during the EE, single evolved isolates from the evolved populations were saved as frozen stocks and subjected to phenotypic and genotypic characterization. To represent different populations and time points during the EE, three isolates were randomly picked from each of population 3, 4, 6, and 7 at transfer 12, 18, and 30 . To detect possible changes in colony morphology, ON cultures of the ancestor and evolved isolates were spotted on LB agar and colonies inspected after 48-h incubation. On LB agar, the ancestor formed a round colony with a wrinkled periphery, whereas different colony morphologies were observed among the evolved isolates . At T12, some isolates displayed a colony morphology resembling the ancestor, e.g. isolate 3 from population 3 and isolate 2 from population 7 , referred to as the ‘‘Wrinkled’’-type. Several other isolates formed a colony with a white sharp edge along the wrinkled periphery , hereafter referred to as the ‘‘Sharp-Wrinkled’’-type. Additionally, isolate 7.1 formed a hyper-wrinkled, white colony, referred to as the ‘‘Snow’’-type. These distinct colony morphologies were also observed at later time points . Interestingly, the Snow-type was only observed in population 7. Furthermore, the three isolates from population 6 at T30 formed slightly less wrinkled colonies compared with the ancestor. We note that three isolates do not represent the entire population, and isolates with other colony morphologies could be present in the populations. Nonetheless, the appearance of isolates with altered colony morphologies in the four populations indicates the presence of genetic changes. Furthermore, the occurrence of isolates with altered colony morphologies already at T12, and especially the presence of three different types in population 7, at this early time point, suggests rapid diversification of B. subtilis during EE on A. thaliana roots. Such diversification into distinct morphotypes was also observed in our previous study on EE of B. subtilis on plant roots and has additionally been observed in EE of B. subtilis pellicle biofilms,dutch buckets indicating successful adaptation to the selective environment.

The design of the EE employed in this study should enable selection for bacteria that efficiently colonize the root. We, therefore, speculated whether the altered colony morphology of some of the evolved isolates was associated with improved productivity on the root . To test this, the ancestor and evolved isolates from the final time point were tested for individual colonization of A. thaliana seedlings under the same conditions applied during the EE. CFU quantification revealed that most evolved isolates tended to show increased root colonization, with five isolates from three different populations displaying significantly increased productivity on the root, with an up to circa 1.3-fold change relative to the ancestor . To track down when such improved root colonizers emerged during the EE, the randomly selected evolved isolates from T12 and T18 were similarly tested. Three and five evolved isolates at T12 and T18, respectively, displayed significantly increased productivity relative to the ancestor. In addition, a single isolate from T12 was significantly reduced in root colonization. These results confirm that indeed the genetic adaptation of B. subtilis to the plant root took place during the EE. Furthermore, the observation of improved root colonizers already at T12 indicates that B. subtilis rapidly adapted to plant root colonization during the EE.While multiple evolved isolates displayed increased individual root colonization relative to the ancestor , we next wanted to test whether the evolved isolates had a noticeable fitness advantage over the ancestor during competition on the root. For this purpose, two selected evolved isolates from independent populations from T30 were competed against the ancestor on the plant root. Following 48 h of root colonization, CFU quantification revealed that both evolved isolates had out competed the ancestor on the root, and statistical analysis confirmed that the evolved isolates had a significantly higher fitness relative to the ancestor . This result was further supported by CLSM imaging: regardless of the fluorescence labeling combination, the two evolved isolates formed biofilms on the roots, as evidenced by aggregates along the root, whereas the ancestor was scarcely present .

Noticeably, the fluorescent images revealed that Ev6.1 formed fewer and smaller aggregates along the root compared with Ev7.3 , consistent with the individual root colonization of the two isolates . To test whether the fitness advantage of Ev6.1 and Ev7.3 over the ancestor was specific to the plant root environment, the two evolved isolates competed against the ancestor in a non-selective environment, i.e. in LB supplemented with xylan, a plant polysaccharide found among others in the secondary cell walls of A. thaliana , and under well-shaking conditions. In this non-selective environment, neither of the evolved isolates out competed the ancestor but instead seemed to suffer a fitness disadvantage compared with the ancestor, although the difference was statistically not significant . Similar results were obtained for the permuted fluorescent combination . These results demonstrated that the evolved isolates had a fitness advantage over the ancestor specifically in the plant root environment. Furthermore, the loss of fitness in a different, non-selective environment suggests an evolutionary cost of adaptation to the plant roots .To identify the genetic changes contributing to the increased root colonization and fitness advantage over the ancestor during root colonization , the genomes of selected evolved isolates were re-sequenced. To represent independent populations, the isolates from populations 6 and 7 at T30 were included. Furthermore, to track molecular evolution over time, the three isolates from population 7 at T12 and T18 were also re-sequenced. Finally, one isolate from population 1 at T30 was included for re-sequencing owing to its ‘‘Smooth’’ colony morphology and reduced root colonization . In the 13 re-sequenced isolates, we observed in total 51 unique mutations of which 37 were non-synonymous . Isolate Ev1.1 harbored several mutations in gtaB encoding a UTP-glucose-1-phosphateuridylyltransferase that synthesizes a nucleotide sugar precursor essential for the biosynthesis of exopolysaccharides and for the synthesis of wall teichoic acids and lipoteichoic acid .

Two of the three isolates from population six at T30 harbored a nonsynonymous point mutation in the fliM gene, encoding a flagellar motor switch protein, part of the basal body C-ring controlling the direction of flagella rotation . All three isolates in population 7 at T30 harbored a mutation in the intergenic region upstream from the sinR gene encoding a transcriptional repressor of the genes responsible for matrix production . Interestingly, this mutation was also present in the three isolates in population 7 at T18 and in one of the isolates in this population at T12, suggesting that this mutation arose rather early in the EE and rose to a high frequency in population 7. Indeed, sequencing of the seven endpoint populations revealed that this mutation upstream from sinR was fixed in population 7 at the final time point, i.e. the mutation had reached a frequency of 1 in this population at T30 . Furthermore, anon-synonymous mutation within the sinR gene was detected at high frequencies in populations 2 and 3. In addition, non-synonymous mutations in genes related to flagellar motility were besides population 6 also observed in populations 3, 4, and 5 . Finally, mutations in genes related to cell wall metabolism were identified across all seven populations . The detection of mutations within genes related to biofilm formation, motility, and cell wall metabolism across independent populations supports the role of these mutations in the adaptation of B. subtilis to A. thaliana roots.Next, we wanted to elucidate which bacterial traits were altered during such adaptation to the plant root. For this purpose, evolved isolates from the final transfer were subjected to further phenotypic characterization. Given the detected mutations and that both biofilm formation and motility are important for successful root colonization by B. subtilis , we hypothesized that these two bacterial traits would be under selection during the adaptation to the plant roots. To this end, plant polysaccharides including xylan have been shown to induce biofilm formation in B. subtilis in a non-biofilm inducing medium .

One way of adapting to the plant root could thereby be through enhanced biofilm formation in response to such PPs. To test whether the improved productivity on the root by the evolved isolates was associated with more robust biofilm formation in response to PPs, the ancestor and evolved isolates were tested for pellicle biofilm formation, a biofilm formed at the medium-air interface , in LB supplemented with xylan . Importantly, a rich medium rather than the minimal medium was used in this assay to provide the bacteria with plenty of nutrients, allowing us to assess only the ability of the evolved isolates to form biofilm in response to xylan, and not the ability to utilize xylan for growth. We observed that a few isolates from T30 developed a pellicle biofilm similar to the ancestor, i.e. Ev4.1, Ev4.2, and Ev6.1 . In contrast, the remaining isolates developed more robust pellicles with highly structured wrinkles indicative of enhanced matrix production. Especially the three isolates from population 7 developed hyper-robust, white pellicles, consistent with the Snowtype colony morphology observed for these isolates . The biofilms developed in response to xylan by the evolved isolates generally correlated with their productivity on the root. For example, isolates Ev4.1, Ev4.2, and Ev6.1 developing similar pellicles as the ancestor and isolates Ev7.1, Ev7.2, and Ev7.3 forming hyper-wrinkled, robust pellicles in response to xylan were among the ones showing the smallest and largest increase in individual root colonization , respectively. This is in accordance with Chen et al. demonstrating that the ability of B. subtilis mutants to form robust biofilms in vitro correlated with that on the root. These results suggest that improved productivity on the root was associated with robust biofilm formation in response to xylan. To test whether this enhanced biofilm formation was specific to the presence of PPs, grow bucket the ancestor and evolved isolates were tested for the ability to form pellicles in LB in the absence of xylan. In this medium, the pellicles developed by both the ancestor and evolved isolates were less robust . For most isolates, the improved biofilm formation was specific to the presence of PPs, whereas the isolates from population 7 displayed robust biofilms also in the absence of plant compounds suggesting a general improvement in biofilm formation in these isolates.To test whether the evolved isolates were affected in motility, the ancestor and evolved isolates from population 6 and 7 were tested for two types of motility: swimming motility, a single cell movement in aqueous environments powered by flagella rotation and swarming motility which is associated with a rapid multicellular movement of hyper-flagellated cells across a surface facilitated by self-produced surfactin.

Interestingly, most isolates were significantly impaired in both swimming and swarming motility . Swimming motility was observed for the ancestor and evolved isolates after 4 h . However, after 6 h only the ancestor and Ev6.2 had reached the edge of the Petri dish, whereas the remaining isolates reached at the most half of the swimming distance of the ancestor. Swarming was observed for the ancestor after 4 h which continued until the expanding colony almost reached the edge of the Petri dish after 8 h . In contrast, the evolved isolates showed reduced or a complete lack of swarming throughout the experiment. The evolution of motility-impaired isolates in independent populations could indicate that motility is not important for root colonization in the selective environment. Notably, during the EE the 48-well plates were continuously shaking at 90 rpm. We speculated, that these mildly shaking conditions could allow the bacteria to get into contact with the root by chance and thereby reducing the impact of motility on root colonization in the selective environment. To test whether motility is important during root colonization under shaking conditions, the ancestor was competed against a Dhag mutant, deficient in the production of the flagellin protein, for three successive rounds of root colonization under static or shaking conditions.Under static conditions, the Dhag mutant was significantly out competed by the WT.In contrast, under shaking conditions, the Dhag mutant was able to co-colonize the root to similar levels as the WT. These results demonstrate that motility is important for competition on the root under static conditions but is not required under shaking conditions. Thereby, impaired motility of several of the evolved isolates is not expected to negatively influence the fitness of these isolates in the selective environment.