The evolutionary experiment employed Bacillus velezensis SQR9, a well-characterized plant growth-promoting rhizobacterium from our laboratory collection. For reliable strain tracking and contamination control, we chromosomally integrated a GFP-chloramphenicol resistance plasmid prior to experimental procedures, according to the chemogenetic transformation method described by Chen et al. (2016). Initial cultures were prepared by streaking the bacterium onto LLB agar (10 g L⁻¹ tryptone, 5 g L⁻¹ yeast extract, 3 g L⁻¹ NaCl, 20 g L⁻¹ agar) containing 5 μg ml⁻¹ chloramphenicol, followed by 30°C incubation for 16 h. Randomly selected colonies were propagated in chloramphenicol-supplemented LLB broth at 30°C with 200 rpm shaking for 12 h. Cells were pelleted, washed thrice with 10 mM MgSO₄ to eliminate medium residues, and standardized to 10⁵ CFU ml⁻¹ in sterile MgSO₄ for evolutionary cultivation.

Pepper (Capsicum annuum var. conoides) was used as the host plant. Seeds were soaked in deionized water for 12 h, then sterilized in 70% ethanol for 1 min and 6% sodium hypochlorite for 6 min. After sterilization, the seeds were rinsed three times with sterile water. Germination occurred on 0.25 × MS agar medium (pH 7.0) in square petri dishes under controlled conditions (22°C, 16/8-h photoperiod, 200 μmol·m⁻²·s⁻¹ PAR) for 7 days. Seedlings were then transferred to sterile 750 ml vessels containing 100 g autoclaved vermiculite and 90 ml 0.25 × MS solution, with two plants per vessel to establish rhizosphere colonization system.

Two seven-day-old sterile pepper seedlings were transplanted into sterilized vermiculite containers and inoculated with an initial bacterial suspension (10⁵ cells ml⁻¹) in the rhizosphere, using a volume of 1 ml for the inoculum. After 1 week of cultivation, plant roots were transferred to centrifuge tubes containing 5 ml of 6 g L⁻¹ NaCl solution with two sterilized 2-mm glass beads. Root-associated bacteria were dislodged by vortex mixing (1,500 rpm, 10 min). One milliliter of the resulting suspension was transferred to new seedling rhizosphere, and the residue underwent serial dilution for colony quantification. The transfer totally repeated for 20 cycles and maintained five independent replicate lineages throughout the experiment.

After overnight activation on LLB agar, the single colonies of different evolved strains were inoculated into LLB broth supplemented with chloramphenicol. The cultures were incubated at 30°C on a shaking platform at 200 rpm for 12 h. The cells were washed three times with 10 mM sterile MgSO₄ to remove residual culture medium. The final concentration was standardized to 10⁵ CFU/ml in sterile MgSO₄.

Seven-day-old sterile pepper seedlings were transplanted into sterile vermiculite containers. One milliliter of the initial bacterial suspension was inoculated at the root zone. After 1 week of cultivation, plant roots were rinsed with sterile water. They were then transferred to 5 ml of 6 g/L NaCl solution and vortexed at 1,500 rpm for 10 min to recover root-attached bacteria. The resulting bacterial suspension was serially diluted and plated on LLB agar for colony counting.

At the end of the final cultivation cycle, nine colonies were randomly selected from each evolutionary lineage to evaluate the following functions:

IAA production Two μl overnight bacterial culture was inoculated into a 96-well plate containing 1 g L⁻¹ tryptophan in the Landy medium, and incubate at 25°C, 100 rpm for 72 h. The culture was centrifuged at 3,000 rpm for 30 min, and the cell-free supernatant was obtained by filtering through a 0.22 μm membrane. The supernatant was reacted with R1 reagent (FeCl₃ 312 g L⁻¹, H₂SO₄ 7.9 M) in the dark for 30 min, and the absorbance was measured at 530 nm (n = 6).

Siderophore production Two μl overnight culture was inoculated into a 96-well plate containing 190 μl of iron-limited MKB liquid medium. The plate was incubated at 30°C with shaking at 170 rpm for 48 h. After incubation, the cultures were centrifuged at 3,000 rpm for 30 min, and the supernatant was filtered through a 0.22 μm membrane. The sterile supernatant was mixed with an equal volume of CAS detection reagent and incubated for 2 h. The absorbance (A) of the samples was then measured at 630 nm, while the A₆₃₀ (Ar) of the control group was prepared following the same method with un-inoculated medium. The iron carrier units (SU) were calculated using the given formula: SU = 1 − (A/Ar; Schwyn and Neilands, 1987).

Biofilm Formation Two μl overnight culture was inoculated into a 96-well plate containing 198 μl of MSgg medium and incubated at 30°C for 48 h. When sampling, the medium beneath the biofilm was carefully aspirated using a micropipette, and the biofilm was then washed twice with 150 μl of phosphate buffer. Next, 0.1% (w/v) crystal violet solution was added and allowed to stain for 30 min. After discarding the staining solution, the biofilm was rinsed twice with deionized water. Finally, 150 μl of an ethanol/acetone solution (80:20, v/v) was added to dissolve the dye attached to the biofilm, and the absorbance was measured at 570 nm to quantify the biofilm formation.

The ancestral strain or evolved strains were inoculated into LLB liquid medium. After shaking at 30°C and 170 rpm for 24 h, the cells were collected by centrifugation at 6,000 rpm for 10 min. The cells were then resuspended in 0.25 × MS culture medium to adjust the concentration to about 5.0 × 10⁸ CFU ml⁻¹ for inoculation.

Hydroponic Experiment Roots of the 7-day-old pepper seedlings (cultivated as described above) were gently rinsed with sterile water to remove residual agar medium, then the seedlings were transferred to 50 ml Erlenmeyer flasks containing 24 ml 0.25 × MS solution pre-inoculated with 1 ml bacterial suspension (ancestral or evolved strains). Flasks were maintained in greenhouse under 25°C with 16/8-h photoperiod for 7 days. Plant growth parameters including shoot fresh weight, root fresh weight, and lateral root count were subsequently quantified.

Pot experiment 25-day-old pepper seedlings were transplanted into sterilized pots. Each pot contained one plant and 3 kg of unsterilized natural soil. Bacterial suspensions with a final concentration of ≥1.0 × 10⁹ CFU were evenly applied around the seedling roots, using a volume of 10 ml for the inoculum. The soil had the following properties: pH 7.83, organic matter content of 12.08 g/kg, total nitrogen of 71.12 mg/kg, total phosphorus of 0.85 g/kg, total potassium of 16.83 g/kg, available phosphorus of 55.55 mg/kg, and available potassium of 137.33 mg/kg. Plants were maintained at 25°C with a 16/8-h light/dark cycle. The experiment included eight treatments as sterile water irrigation (CK), ancestral strain inoculation (WT), and inoculations with different evolved strains (9P15, 9P16, 9P19, 9P27, 9P38, 9P41); and six biological replicates were included in each treatment. After 25 days of cultivation, samples were collected to measure plant height, root length, shoot fresh weight, root fresh weight, and chlorophyll SPAD values.

The evolved strain 9P41 that showed the best plant growth-promotion performance, was selected for genome resequencing. The frozen bacterial solution was taken out from the refrigerator and spread on LLB plate containing antibiotics. After extracting the genomic DNA of the evolutionary strain and the ancestral strain, resequencing was performed following the processes as: (1) DNA quality assessment; (2) fragmentation of DNA using ultrasonic methods to achieve sizes between 300–500 bp; (3) end repair of DNA fragments using T4 DNA polymerase, adding an A base at the 3′ end to create sticky ends; (4) ligation of DNA adapters containing index sequences to the sticky ends, following the base pairing principle, for Illumina sequencing; (5) selection of specific length fragment sequences using magnetic beads; (6) PCR amplification of the target fragments, adding index sequences at both ends for downstream library preparation and sequencing; (7) fixation of the sequencing library onto a sequencing chip via bridge PCR; (8) paired-end sequencing on the Illumina NovaSeq platform, executed as 2 × 150 bp. The mutation screening criteria are as follows: the total coverage of mutation site detection is at least 2 times (i.e., covered by at least 2 reads). The minimum frequency of the gene mutation in the population is ≥5%. The gene locus mutation is a non-synonymous mutation.

Data analysis and visualization were performed using R (v4.1.3) and GraphPad Prism 9. Statistical significance was set at p < 0.05. Multiple group comparisons employed two-way ANOVA with Tukey’s post hoc tests for pairwise comparisons.

In order to improve the rhizosphere adaptation and probiotic function of a model PGPR strain B. velezensis SQR9, we designed an in situ rhizosphere experimental evolution by continuous domestication of the bacteria in pepper rhizosphere (see Materials and Methods, Figure 1A), resulting in approximately 160 generations of cumulative evolution. Five parallel evolutionary lineages (designated as P1-P5) were established, and the bacterial root colonization capacity was quantified after each transfer via dilution plating. The ancestral strain initially colonized plant roots at approximately 1.7 × 10⁷ CFU g⁻¹ root (Figure 1B), and the average colonization level significantly evolved during the transfer process: colonization slowly increased in the early stage before the fifth transfer (pre-T5), experienced fluctuations during the mid-stage (T10 to T14), and had significant growth in the later stage (T14 to T17). By T18 and T20, bacterial colonization stabilized across all lineages, ranging from 2.6 × 10⁷ to 5.0 × 10⁷ CFU g⁻¹ root. This change marked an improvement of 1.5 to 2.9 times compared to the ancestral strain.

Based on the five evolved B. velezensis SQR9 lineages cultured on LLB agar, nine colonies were randomly selected from each lineage (totaling 45 evolved strains, labeled as 9P1 ~ 9P45). These strains were assessed for three plant-associated traits as indole-3-acetic acid (IAA) production, siderophore production, and biofilm formation (Figure 2A). The ancestral strain SQR9 displayed baseline values of IAA secretion by 12.03 mg L⁻¹, siderophore production of 0.399 units (SU), and biofilm formation ability of OD₅₇₀ as 1.779. Among the picked 45 evolved strains, 22 exhibited enhanced IAA production than the ancestor (6.1% ~ 54.1% increase, maximum production of 18.54 mg L⁻¹ by 9P12), nine showed improved siderophore secretion (6.2% ~ 12.3% increase, peak SU of 0.448 by 9P45), and 26 demonstrated superior biofilm formation (8.2% ~ 53.5% increase, highest OD₅₇₀ as 2.731 by 9P6). As a result, 29 strains showing improvements in at least one trait (with an increase of ≥5%) were selected for further evaluating the plant growth promotion effects in hydroponic experiment.

The selected 29 evolved strains were further evaluated for their plant growth-promotion efficacy using a hydroponic system. Pepper plants inoculated with ancestral strain SQR9 revealed baseline performance with an aboveground biomass of 0.175 g, a root biomass of 0.079 g, and an average of 34 lateral roots per plant. Among the evolved strains, 3 demonstrated improved the plant aboveground biomass compared to the ancestral strain (8.0% ~ 14.9% increase, maximum 0.201 g by 9P41); five strains announced enhanced root biomass (8.3% ~ 29.1% increase, maximum 0.102 g by 9P41), and 24 strains exhibited an increase in lateral root formation (5.8% ~ 79.4% increase, maximum 61 roots per plant by 9P41; Figure 2B). As a result, six evolved strains (9P15, 9P16, 9P19, 9P27, 9P38, and 9P41) each showing an advantage in at least one plant growth parameter than the ancestor (with an increase of ≥5%), were selected for assessing their plant growth-promoting capacity in the following pot experiment.

To better simulate the field production condition, finally we inoculated the 6 selected evolved strains into 25-day-old pepper plants in pots for growth promotion assessment. After 25 days, all treatments with evolved strains outperformed the ancestral strain in at least one trait (Figure 3). Specifically, evolved strain 9P41 pronounced the best plant growth-promotion performance, with an 11.4% increase in plant height than the ancestor-inoculated plants (24.5 vs. 22.0 cm), a 28.7% increase in root length (18.4 vs. 14.3 cm), a 21.0% increase in aboveground biomass (11.31 vs. 9.35 g), a 29.1% increase of underground biomass (1.23 vs. 0.92 g), and a 5.8% increase in chlorophyll SPAD value (38.6 vs. 36.5). Other evolved strains also displayed specific improvements, such as 9P38-inoculated plants showed an increase of 28.7% in root length than the ancestor, while 9P15 demonstrated a 4.3% increase in underground biomass. These results indicate that rhizosphere domestication selects strains with evolved plant-associated functions, with 9P41 emerging as the most effective plant growth promoter.

Given the outstanding plant growth-promotion performance by the evolved strain 9P41, we aimed to identify the genetic determinants of its enhanced traits through whole-genome resequencing. Comparative genomic analysis with its ancestor revealed three mutations in coding regions as mlnD, smc, and fhuC (Table 1). In detail, the mlnD gene encodes a type I polyketide synthase that catalyzes the biosynthesis of polyketides through sequential condensation of carboxylic acid precursors (Liang et al., 2019). In the Bacillus genus, these polyketide derivatives, such as macrolides, exhibit broad-spectrum antibacterial activity against both Gram-positive and Gram-negative pathogens by interfering with cell division and cell wall synthesis (Jun et al., 2012). The product of the smc gene forms a conserved ATP-dependent protein complex essential for chromosome organization, mediating ATP-driven DNA loop extrusion to maintain genomic integrity during replication (Jun et al., 2012). The fhuC gene encodes a key ATPase subunit of an ABC transporter that is crucial for the uptake of iron–siderophore complexes, where ATP hydrolysis provides the energy necessary for transmembrane transport (Athanasios and Wolfgang, 1998; Wenxin and Hongjin, 2021). Collectively, three genes (mlnD, smc, and fhuC) with potential links to plant-microbe interactions, are identified to be mutated during the evolution process by strain 9P41.