Rhizobial migration toward roots mediated by FadL-ExoFQP modulation of extracellular long-chain AHLs

Identification of broad-host-range rhizoplane colonization genes by Tn-seq

This work was focused on SF2 harboring a typical multipartite genome of Sinorhizobium (chromosome, chromid, and symbiosis plasmid) [59]. To perform genome-wide survey of rhizoplane colonization genes of SF2 (Fig. 1), the input mutant library was inoculated on filter paper of plant culture dish, and output mutant libraries were collected from filter papers at 1 h post inoculation (F1h) and 7 days post inoculation (dpi; F7d), and from rhizoplane of cultivated soybean (CS7d), wild soybean (WS7d), rice (R7d), and maize (Z7d) at 7 dpi. To facilitate Tn-seq library construction, all output mutant libraries were subject to 32 h cultivation in the TY rich medium, with input libraries cultivated at the same condition as control (TY). Tn-seq revealed that transposon insertion density in three input and 21 output samples ranged from 57.03 to 86.99% (Table S3), which are above the threshold of 50% insertion density for a good Tn-seq dataset [49]. A reproducible rhizosphere effect was observed in three independent experiments (Fig. S1), i.e., rhizoplane samples (CS7d, WS7d, R7d, and Z7d) consistently formed distinct clusters compared to those of TY, F1h, and F7d. A considerable signature of three independent input libraries was also identified (Data S1, Data S2, and Fig. S1). These results highlight that stochastic variations among multiple independent input libraries should be considered before making conclusions on gene fitness, which has been largely overlooked in earlier studies based on just one input library [49].

Based on gene fitness scores of rhizoplane samples (CS7d, WS7d, R7d and Z7d) compared to corresponding F1h datasets (Fig. S2A; Data S2), 93, 91, 127, and 206 genes were identified as rhizoplane colonization genes for test plants of cultivated soybean, wild soybean, maize, and rice, respectively, accounting for 1.4–3.1% of the SF2 genome (p values < 0.01). This range is similar to those reported for rhizoplane colonization genes in Sinorhizobium meliloti (2%) [60], Rhizobium leguminosarum bv. viciae (2.9%) [17], Agrobacterium tumefaciens (2.9–4%), [18], Pseudomonas aeruginosa (1.6%) [15], Pseudomonas simiae (2%) [13], Pseudomonas sp. WCS365 (3.8%) [14], Azoarcus olearius (2.2%), and Herbaspirillum seropedicae (2.7%) [16]. Among these rhizoplane fitness genes, 8, 8, 21, and 82 genes were specific to cultivated soybean, wild soybean, maize, and rice, respectively (Fig. S2A and Data S2.1). Moreover, 43 genes were reproducibly identified as a rhizoplane colonization gene in three independent experiments for at least one test plant species (Fig. S2B, and Data S2.2), which were therefore considered as a robust list of rhizoplane colonization genes for further investigation. The control samples from filter papers (F7d) formed a distinct cluster compared to those rhizoplane samples in the hierarchical clustering analysis (Fig. S2B). Since a notable noise was observed for c09590 in R7d.3 and Z7d.3 datasets, a c09590 mutant was constructed and co-inoculated with SF2 at 1:1 ratio under the same condition used for Tn-seq sample collection. Moreover, a mutant for c04390 was also constructed and used as a positive control in the same experiment. The c09590 and c04390 mutants exhibited significantly impaired broad-host-range rhizoplane colonization ability compared to SF2 (p values < 0.001; Fig. S3), supporting the Tn-seq results (Fig. S2B).

There were 41 out of 43 genes making positive contribution to rhizoplane colonization. These genes are involved in repair of DNA (UvrA) and protein (Pcm), translation (YchF and LepA), respiration (seven proteins related to cytochrome C), phosphonate metabolism (PhnP), antioxidant activity (SodA), uptake of ferric iron (c03990 and c04080) and oligopeptide (b57060, b57070, and b57080), lipoprotein attachment (c19900), secretion of EPS (ExoF, ExoP, and ExoQ), anabolism for tryptophan (TrpA, TrpB, TrpD, and TrpE), glutamate (GltA, GltB, and ArgD), leucine (LeuA and LeuB), methionine (MetA), IMP (PurC and PurF), AMP (PurA) and dUMP (Dcd). Within this gene list, homologs of 22 genes were also identified as rhizoplane colonization players (Data S2.2) in S. meliloti-alfalfa [60], P. simiae–Arabidopsis thaliana [13], Pseudomonas sp. WCS365-A. thaliana [14], P. aeruginosa-maize [15], A. olearius/H. seropedicae–Setaria viridis [16], R. leguminosarum-pea [17], or Agrobacterium tumefaciens-tomato pairs [18]. In contrast, genes encoding a putative anti-sigma factor (RsbU) and the only known outer membrane hydrophobics transporter FadL [61,62,63,64] negatively regulated rhizoplane colonization of SF2, with fadL having a stronger effect (Fig. S2B). In a recent Tn-seq survey of colonization genes of Aquitalea magnusonii on the floating whole plant of duckweed, the fadL mutants were enriched in the output pool indicating a negative impact on colonization [65].

These 43 rhizoplane colonization genes have a biased replicon distribution in SF2 multipartite genome: 81.4% on chromosome, 18.6% on chromid, and 0% on symbiosis plasmid. This is consistent with earlier metabolic modeling evidence from S. meliloti showing that Sinorhizobium chromosome and chromid contribute to rhizosphere fitness [66], while the symbiosis plasmid encoding genes mainly involved in establishing and maintaining nitrogen-fixing symbiosis with legume plants [59]. Nevertheless, symbiotic optimization by chromosomal and chromid genes has been reported [34, 67, 68], among which EPS biosynthesis genes on chromid have been intensively studied regarding their role in optimizing both infection process of Sinorhizobium on legume plants [69, 70] and biofilm formation in vitro [71]. In this work, among 8 chromid genes involved in rhizoplane colonization (Fig. S2B), exoF, exoP, and exoQ encode membrane proteins directing EPS polymerization and secretion (Fig. 2A) [72,73,74,75,76,77]. It was intuitively unexpected that various EPS biosynthesis genes within the same gene cluster of exoF, exoP, and exoQ were not essential for rhizoplane colonization (Fig. 2B, C). This implies a working model that other unknown physiological function of ExoF, ExoP, and ExoQ play a more important role than EPS production in rhizoplane colonization.

Opposite contribution to rhizoplane colonization by FadL and ExoFQP

As described above, among broad-host-range rhizoplane fitness genes (Data S2.2 and Fig. S2), mutations of fadL led to outstanding high rhizoplane colonization rates, while mutations in exoFQP involved in EPS polymerization and secretion, but not those in exo genes required for EPS biosynthesis, led to severely impaired colonization rates. To verify the opposite role of FadL and ExoFQP in rhizoplane colonization, mutants of fadL, exoF, exoP, and exoQ were constructed. Moreover, mutants of five exo genes involved in different steps of EPS biosynthesis and degradation were constructed for comparison: exoB (precursor synthesis), exoY (priming glycosyltransferase), exoA (glycosyltransferase), exoZ (substitution), and exoK (polysaccharide degradation) (Fig. 2C). These mutants were individually co-inoculated with SF2 at 1:1 ratio under the same condition for Tn-seq (Fig. 1). Their competitive ability in rhizoplane colonization was generally consistent with the Tn-seq results. The fadL mutant was more competitive than SF2 in the CS7d, WS7d, and R7d treatments (Fig. 3A; p values < 0.001) but indistinguishable from SF2 in Z7d. By contrast, the exoF, exoP, and exoQ mutants were outcompeted by SF2 on four plant species (Fig. 3A; p values < 0.001).

The exoY, exoA, exoZ, and exoK mutants were indistinguishable from SF2 in the CS7d, WS7d, and R7d treatments, while the exoY, exoZ, and exoK mutants showed an impaired competition ability in Z7d (33–43%; p values < 0.001), though to a lesser extent than the exoF, exoP, and exoQ mutants (27–28%). The notable exception was the exoB mutant which was outcompeted by SF2 herein (Fig. 3A) while no depletion of exoB mutants was observed in Tn-seq data from rhizoplane samples (Fig. 2B). ExoB, encoding a UDP-glucose 4-epimerase, synthesizes UDP-galactose [78] which might be provided by other mutants within the mixture of mutant library. This hypothesis is supported by several lines of evidence: (1) a cryptic uxe gene in Sinorhizobium encodes a bifunctional UDP-sugar 4-epimerase, catalyzing UDP-xylose/UDP-arabinose and UDP-glucose/UDP-galactose interconversions, and its overexpression can functionally replace exoB [79]; (2) this uxe gene is however generally repressed by the global silencer MucR in wild-type Sinorhizobium strains [79,80,81]; (3) a considerable number of transposon insertions were observed in two mucR copies (c08920 and a45250; 377-758 insertions per gene) within Tn-seq data of rhizoplane samples (Data S1). The drastic rhizoplane colonization defect of the exoB mutant compared to the exoY, exoA, exoZ, and exoK mutants (Fig. 4A) may be due to the fact that UDP-galactose, synthesized by ExoB, is a precursor for multiple galactose-containing polysaccharides (lipopolysaccharides, succinoglycan and galactoglucan) [82].

Opposite contribution to competitive nodulation by FadL and ExoFQP

Since competitive nodulation between commercial rhizobial inoculants and indigenous rhizobia is a long-lasting issue in agriculture practices [83], these mutants were further tested for their nodule occupancy ability on plants grown in vermiculite. The wild-type SF2 can form effective nodules on many wild soybean accessions and some soybean landraces but not on certain modern soybean cultivars [67, 84]. When these mutants were inoculated on wild soybean plants, only the exoB mutant showed significant defects in nodulation and symbiotic performance regarding shoot dry weight (Table S4). Moreover, the exoB mutant was totally outcompeted by SF2 in competitive nodulation experiment (1:1 inoculation ratio; Fig. 3B). The exoF, exoP, and exoQ mutants also showed a significant defect in nodule occupancy (13–17%; Fig. 3B), while only exoK, among the other exo mutants, showed an impaired nodule occupancy to a lesser extent (36%). The fadL (78%) and exoZ (63%) mutants occupied more nodules than SF2. As the exoY, exoA, exoZ, and exoK mutants were all indistinguishable from SF2 regarding rhizoplane colonization on wild soybean plants (Fig. 3A), the role of ExoZ and ExoK in competitive nodulation may be not due to their predicted functions in EPS modification and degradation [73, 85].

We have previously reported that mutations in rhcV, encoding an essential component of type three secretion system in SF2, allow an improved nodulation and symbiotic performance on certain commercial soybean cultivars such as JD17 [84, 86]. When fadL was further mutated in the rhcV mutant background, the rchV fadL double mutant exhibited a higher nodule occupancy (61%) than the rhcV mutant on JD17 (Fig. 3B; p < 0.01). We further tested the effects of fadL mutation in a model strain SM2011 [87] and an inoculant strain SM01290 [88] of S. meliloti associated with alfalfa. These two fadL mutants occupied more nodules than their wild-type strains SM2011 and SM01290 (Fig. 3B; 74–78%; p values < 0.001). Collectively, contrasting nodule occupancy ability between the fadL mutant and the exoF, exoP, and exoQ mutants were in line with their opposite rhizoplane colonization ability.

Variation in surface motility of the fadL, exoF, exoP, and exoQ mutants

In order to uncover potential traits associated with the contrasting rhizoplane colonization ability of test mutants (Fig. 2 and Fig. 3A), their growth curves (Fig. S4A), EPS production (Fig. S4B), biofilm formation (Fig. S4C), swimming and surface motility (Fig. 4A–C) were compared. The exo and fadL mutants were not significantly different from each other on their growth curves in the TY rich medium (Fig. S4A). All exo mutants except exoZ produced less EPS (both high- and low- molecular weight compositions; Fig. S4B), which were in line with their biofilm formation ability (Fig. S4C). These results support the view that EPS is important matrix components of biofilm [25]. Notably, the exoF, exoP, and exoQ mutants were not distinguishable from the other exo mutants except exoZ, and the fadL mutant had similar EPS production and biofilm formation ability as SF2 (Fig. S4B, C). Apparently, these characteristics of growth curves, EPS and biofilm production could not explain the observed variation in rhizoplane colonization competence of the fadL and exo mutants (Fig. 2 and Fig. 3A). This is likely due to the fact that biofilm is the “home” for multiple cells [25], among which some family members (such as certain exo mutants) can effectively co-colonize rhizoplane with other EPS producers in the root microbiota.

Given the important role of motility in migration toward roots, swimming and surface motility of test mutants were compared (Fig. 4A–C). The fadL mutant had a higher surface motility ability than SF2 (Fig. 4C; ANOVA followed by Duncan’s test, alpha = 0.05). All exo mutants exhibited a lower surface motility regarding the diameter of colonies on the semisolid plate (0.5% agar with Congo red; Fig. 4C) while the exoF, exoP, and exoQ mutants had notable radial cracks in their colonies (marked by * in Fig. 4A), which were not observed for the other exo mutants (Fig. 4A). The variation in the swimming ability of test mutants were however not consistent with their rhizoplane colonization ability (Fig. 4B). To further verify the contrasting surface motility phenotypes, the GFP-labeled SF2 and mCherry-labeled fadL or exoF mutants were mixed at different ratios 1:9, 3:7, 5:5: 7:3 and 9:1 (Fig. 4D), and the higher and lower surface motility ability of the fadL and exoF mutant, respectively, was demonstrated across all inoculation ratios. The surface motility ability of the fadL, exoF, exoP, and exoQ mutants can be fully restored in their corresponding complementary mutant strains (Fig. S5).

Available literature suggests that surface motility on the test semisolid plate (0.5% agar) can be flagellum-dependent swarming and/or flagellum-independent motility processes e.g., pilus-dependent twitching, adhesin-dependent gliding, and sliding driven by the pressure of growing cells and facilitated by additional factors (e.g., surfactant, EPS, and surface protein) in a species- and resource-dependent manner [89, 90]. In this work, no known genes on flagellum, pilus or adhesin were found in the list of broad-host rhizoplane fitness genes (Fig. S2 and Data S2.2). For example, when four flaA genes encoding flagellar filament subunits were deleted (Fig. S6A), the resultant flaA mutant showed an impaired flagellum-dependent swimming ability as expected (TY plates with 0.3% agar; Fig. S6B–D; p < 0.01), but was indistinguishable with SF2 regarding surface motility (TY plates with 0.5% agar; Fig. S6B, D) and in the rhizoplane colonization ability assay (Fig. S7). In short, the variation in surface motility, likely sliding and/or other unknown movement mechanisms, was in line with the opposite rhizoplane colonization ability between the fadL mutant and the exoF, exoP, and exoQ mutants.

Variation in migration ability toward roots

The evidences mentioned above (Figs. 2–4) suggest a correlation between surface motility ability and competitive rhizoplane colonization. Here we further evaluated the rhizoplane colonization ability of individual mutants of fadL, exoF, exoP, and exoQ, with the exoY, exoA, and SF2 strains as controls. The fadL mutant showed a significantly higher rhizoplane colonization ability on wild soybean, cultivated soybean, rice, and maize, while the exoF, exoP, and exoQ mutants, but not the exoA and exoY mutants, showed significant impairment in rhizoplane colonization (Fig. 5A; p values < 0.001). By contrast, these mutants showed similar survivability on the filter paper of culture dish in the absence of plants (Fig. 5A; p > 0.05). The number of colony-forming units (CFUs) on rhizoplane showed a global host-dependent pattern (Fig. 5A), consistent with the variation in root exudates of different plants species [20] and their effects on root microbiome assembly [21]. Nevertheless, the opposite rhizoplane colonization ability between the fadL mutant and the exoF, exoP, and exoQ mutants was not host-dependent. Therefore, the following experiments on rhizosphere effects were only performed on one plant species (wild soybean).

The observed variation in rhizoplane colonization rates (Fig. 2, Fig. 3, and Fig. 5A) can involve survivability in rhizosphere and on rhizoplane, and migration from rhizosphere to rhizoplane, which however has not been well addressed in published Tn-seq studies of rhizoplane colonization genes [13,14,15,16,17,18]. The test exo and fadL mutants didn’t show growth defect in the TY rich medium (Fig. S4A) and survivability defect on filter paper of plant culture dishes (Fig. 5A), but it remained unknow if this would hold true on rhizoplane. To answer this question, test strains were individually inoculated on seedling roots before planting in culture dishes (Fig. 5B). The number of CFUs on rhizoplane at 7 dpi was significantly higher for the fadL mutant compared to the other test strains (SF2, exoY, exoA, exoQ, exoF, and exoP; p values < 0.001), while the exo mutants and SF2 were not significantly different from each other (Fig. 5B). These results suggest that rhizoplane survivability at least partially accounts for the superior rhizoplane colonization ability of the fadL mutant, but does not correlate with colonization defects of the exoF, exoP, and exoQ mutants.

These results further raised a question whether both survivability and surface motility would be the rhizosphere processes underlying the contrasting rhizoplane colonization abilities among the fadL, exoF, exoP, and exoQ mutants. To answer this question, the fadL or exoF mutant, and SF2 was individually spotted at 1/2/3/4 cm away from seedlings in a symmetrical way (Fig. 6A). CFUs at the inoculation spot and on rhizoplane were determined at 7 dpi. For all strains, the farther the inoculation site was away from seedlings, the lower the number of CFUs on rhizoplane (Fig. 6B–E; grey bars), which indicated a distance-decay pattern for rhizoplane colonization rates. This provides direct evidence for a spatial pattern of bacteria-root interaction efficiency, which is taken for granted in various models of rhizosphere effects but rarely tested [19, 20, 91]. Across 1–4 cm sites, these two mutants showed no significant difference in survivability compared to SF2 (Fig. 6B, C; p values > 0.05). Regardless of survivability across these sites, the fadL and exoF mutants had higher (153–223% more CFUs than SF2; p values < 0.001; grey bars on the left panel of Fig. 6B) and lower (75–98% lesser CFUs than SF2; p values < 0.001; grey bars on the left panel of Fig. 6C) rhizoplane colonization rates, respectively, compared to SF2 (grey bars on the right panels of Fig. 6B, C). All rhizosphere phenotypes associated with these two mutants can be restored in their corresponding complementary mutant strains (Fig. 6D, E). These results are consistent with their contrasting surface motility ability (Fig. 4 and Fig. S5). Collectively, the variation in rhizoplane colonization ability among the fadL and exoF mutants, and SF2 is mainly modulated by their surface motility in the migration process from rhizosphere to rhizoplane.

Extracellular quorum-sensing long-chain AHLs enhance migration toward roots

The available evidence suggests that the homolog of long-chain fatty acid transporter FadL in S. meliloti may promote long-chain AHL (with acyl chains containing at least 12 carbons) uptake [92], and long-chain AHLs can function as biosurfactants facilitating surface motility of Rhizobium etli on the semisolid plate (yeast extract mannitol medium with 0.75% agar) [93]. However, direct measurement of AHL mixture from the fadL mutant was not available yet. We therefore examined the extracellular content of AHLs in SF2, mutants of fadL and exo, and their complementary mutant strains by using an AHL indicator strain A. tumefaciens KYC55(pJZ372)(pJZ384)(pJZ410) [46] (Fig. 7A). This indicator strain, lacking the AHL biosynthesis ability, overexpresses the AHL receptor TraR which can activate the expression of PtraI–lacZ fusion upon interaction with AHL molecules of various chain lengths, saturation levels, and oxidation states [46]. On the surface motility plate, the fadL mutant made A. tumefaciens KYC55(pJZ372)(pJZ384)(pJZ410) expressing more β-galactosidase than SF2 (Fig. 7B) and thus cleaving more X-gal into the blue product (Fig. 7A). By contrast, the exoQ, exoF, and exoP mutants had a reduced level of extracellular AHLs and the exoY and exoA mutants were similar to SF2 (Fig. 7A, B). The increased and reduced levels of extracellular AHLs from the fadL and exoF/exoP/exoQ mutants can be restored in their complementary mutant strains under test conditions (Fig. 7A, B). To our knowledge [72,73,74,75,76,77, 94], this may represent the first report of the involvement of EPS secretion system in modulating extracellular AHLs.

To get further insight into the mixture of AHLs associated with the fadL and exoF mutants, intra- and extracellular AHLs were determined using the HPLC-Q-Exactive-PRM-MS approach (Fig. 7C). A small amount of a short-chain AHL 3-OXO-C8-HSL was detected inside the exoF mutant, but undetectable in cells of the other test strains. By contrast, no extracellular 3-OXO-C8-HSL was detected for the exoF mutant, but detectable for the other strains, among which the complementary mutant strain of the exoF mutant had more extracellular 3-OXO-C8-HSL than SF2 (ANOVA followed by Duncan’s test, alpha = 0.05). The exoF mutant was also characterized by its lower intracellular (3-OXO-C12-HSL, 3-OXO-C14-HSL, and C14-HSL) and extracellular long-chain AHLs (3-OXO-C12-HSL, C12-HSL, 3-OXO-C14-HSL, and C14-HSL; ANOVA followed by Duncan’s test, alpha = 0.05) than SF2. Its complementary mutant strain had higher extracellular long-chain AHLs than SF2 (C12-HSL, 3-OXO-C14-HSL, and C14-HSL; ANOVA followed by Duncan’s test, alpha = 0.05), while being indistinguishable from SF2 in their intracellular long-chain AHLs. Collectively, these results suggest a model where ExoF is essential for secretion of the short-chain AHL 3-OXO-C8-HSL, intracellular accumulation of which in the exoF mutant can result in down-regulation of the biosynthesis and extracellular level of long-chain AHLs (Fig. 7D). Earlier efforts show that a multidrug efflux system MexAB-OprM from P. aeruginosa is involved in active secretion of long-chain AHLs, and that the BpeAB-OprB efflux system from Burkholderia pseudomallei is required for secretion of AHLs of various length [95, 96]. The short-chain AHLs in other bacteria have been hypothesized to diffuse freely across membranes [97]. Apparently this long-last hypothesis on passive diffusion of AHLs, particularly for those short-chain AHLs, deserves more extensive investigations.

In S. fredii, the traI and sinI genes are essential for the biosynthesis of short- and long-chain AHLs, respectively [98]. Indeed, no 3-OXO-C8-HSL was detected from the traI and traI exoF mutants of SF2 (Fig. 8A). The depleted levels of intracellular (1.7–4.6% of SF2) and extracellular (25.9–57.4% of SF2) long-chain AHLs of the exoF mutant (Fig. 7C) were considerably restored by further deletion of traI (Fig. 8A; 58.9–87.7% and 79.6–85.4% for intracellular and extracellular long-chain AHLs, respectively). Moreover, the traI sinI double mutant, unable to produce endogenous AHLs (Fig. 8A), accumulated lesser intracellular 3-OXO-C8-HSL than the traI sinI exoF triple mutant in the presence of 100 ng/ml exogenous 3-OXO-C8-HSL (Fig. 8B). Collectively, these results support the important role of ExoF in secreting short-chain AHLs, and that intracellular accumulation of short-chain AHLs negatively modulates the biosynthesis of long-chain AHLs (Fig. 7D).

In contrast, the fadL mutant had a similar intracellular level of AHLs compared to SF2, while characterized by the highest extracellular long-chain AHLs (Fig. 7C; ANOVA followed by Duncan’s test, alpha = 0.05). The intra- and extracellular levels of short-chain AHL 3-OXO-C8-HSL were similar between the fadL mutant and SF2. The extracellular level of long-chain AHLs of the fadL mutant can be restored in its complementary mutant strain (Fig. 7C). Moreover, the traI sinI fadL triple mutant showed a significant defect in 3-OXO-C14-HSL uptake compared to the traI sinI double mutant in the presence of 100 ng/ml exogenous 3-OXO-C14-HSL (Fig. 8C). These findings provide more insightful evidences to support the hypothesis of FadL as an uptake system for long-chain AHLs [92] (Fig. 7D).

Apparently, the contrasting rhizoplane colonization ability and surface motility among the fadL and exoF mutants, and SF2 are in concord with their variations in the mixture of extracellular AHLs. It has been demonstrated that long-chain AHLs can function as biosurfactants promoting surface motility of R. etli [93]. We wondered if a synthetic mixture mimicking the extracellular composition of AHLs from the fadL mutant would improve surface motility and rhizoplane colonization rates of SF2 and its various mutants. Thus 3-OXO-C12-HSL (4.0%), 3-OXO-C14-HSL (91.3%), and C14-HSL 3 (4.7%) were synthesized and mixed as a mixture of AHLs. On the semisolid TY plate (0.5% agar), this mixture of AHLs significantly improved surface motility of SF2, flaA, exoF, and traI sinI mutants (Fig. S8; p values < 0.001). The exoF, traI sinI, and sinI mutants (p values < 0.001), but not the flaA and traI mutants, showed impaired root colonization rates in the symmetrical inoculation experiments (Fig. S7 and Fig. 6). When the synthesized mixture of long-chain AHLs was symmetrically inoculated as SF2, traI sinI, and exoF at the same distance to roots (Fig. 8E, F), rhizoplane colonization rates of these strains were significantly enhanced compared to the 0.8% NaCl control (p values < 0.05). This diffusible effect of long-chain AHLs was detectable at 1 cm, 2 cm, 3 cm or 4 cm away from roots (Fig. 8E, F). Moreover, this AHL mixture can also significantly improve the survivability of SF2 at 1 cm and 2 cm sites (Fig. S9; p values < 0.001). Similarly, the fadL mutant can also enhance the survivability of SF2 when these two strains were symmetrically inoculated on either side of seedlings (Fig. 6B), though the fadL mutant, being an overproducer of extracellular long-chain AHLs, was superior in rhizoplane colonization. Notably, the sinI mutant unable to produce long-chain AHLs (Fig. 8A) showed an impaired rhizoplane colonization rate in the symmetrical inoculation experiment (Fig. S7), whereas abundant sinI mutants were identified on the rhizoplane of test plants in the Tn-seq screen (Fig. S10). The Tn-seq result was consistent with that the sinI mutant was as competitive as SF2 in terms of rhizoplane colonization rate (sinI vs SF2 = 52% vs 48%; p > 0.05) when they were inoculated as a 1:1 mixture. In line with these rhizoplane colonization experiments, the sinI or traI sinI mutants showed impaired surface motility compared to SF2 and the traI mutant (Fig. S10B; p values < 0.001), but this surface motility defect can be recovered when they were mixed with SF2 (Fig. S10C). Therefore, the rhizoplane colonization defect of the sinI mutant was not as significant as that of the mutant lacking a multifunctional ExoF in an inoculant mixture containing the wild-type SF2.

Source: Ecology - nature.com