Soil samples were diluted to three concentrations (10⁻³, 10⁻⁴, and 10⁻⁵, w/v) with sterile water. A volume of 20 microliters of soil suspension was inoculated onto a Luria–Bertani agar medium (LB medium) and incubated for 24 h at a temperature of 27°C. Individual colonies with different morphological characteristics were picked and inoculated on LB medium; subsequently, the sample was isolated and purified through growth at 28°C for 24 h (Wang et al., 2020). Purified strains were stored at 4°C for backup. The plate standoff method (Naveed et al., 2022) was then performed, and F. oxysporum was used as an indicator. Fusarium oxysporum was inoculated in the center of the PDA medium, and Bacillus was inoculated equidistant around F. oxysporum, while only F. oxysporum was inoculated in the control group and cultured at 27°C for 5 days. The diameter of pathogen colonies was measured using the crossover method to screen for strains with biocontrol effects against F. oxysporum.

Where “d” represents the diameter of the colony, and the subscript “CK” indicates the control fungus (F. oxysporum).

The morphological, physiological, and biochemical characteristics of the isolated and purified strains were identified using sugar fermentation experiments, indole experiments, Voges–Proskauer experiments, methyl red experiments, Congo red experiments, starch hydrolysis experiments, and contact enzyme activity tests specified in the Manual of Systematic Identification of Common Bacteria (Dong and Cai, 2001).

The total DNA of the purified strain was extracted and purified using a bacterial DNA kit (Tian Gen., Beijing, China). The primers used were 16S (27F: AGAGTTTGATCMTGGCTCAG, 1492R: GGCTACCTTGTTACGACTT), gyrB (F: AACGATTTTGCGGGCTGAG, R: GCGTAAACATCACGATTGAAGAC), rpoB (F: AGCCGTTATTTGTAAACACCCTG, R: CGTCGTTGACTTCCGCAC), and recA: (F: TCACGCAATCGCTGAGGTT, R: CAAGACGCACGGAAGAATAGAAT). The PCR system consisted of 12.5 μL of the 2 × San Taq PCR Mix, 1 μL of the DNA template, 1 μL of each primer, and 9.5 μL of ddH₂O. The PCR conditions were pre-denaturation at 94°C for 2 min, denaturation at 94°C for 30 s, annealing at 56°C for 1 min, and extension at 72°C for 1 min, and the total PCR system included 35 cycles and finally extension at 72°C for 10 min. Next, 1% agarose gel electrophoresis was performed, and gel images were obtained using a gel imaging system. The amplified 16S rRNA PCR products were sequenced (all purchased from Shanghai Sangon Biotech, Shanghai, China). After obtaining the sequences, the NCBI BLAST program was used to compare the 16S rRNA sequences with those in the database, and a phylogenetic tree was constructed using the maximum likelihood (ML) method with 1,000 bootstrap replications in MEGA 7.0 software (Sharma and Kumar, 2024).

Employing the plate confrontation method, filter paper disks infused with Bacillus fermentation liquid and each of 11 pathogenic fungi were inoculated onto PDA medium at a temperature of 27°C for a duration of 5 days; the control group was solely inoculated with the pathogenic fungus. The experiment was conducted with three independent replicates. The diameters of the colonies were measured using the criss-cross method, and the inhibition rates were calculated to identify the optimal antagonistic bacteria.

The activated B. subtilis LK−1 was inoculated into 50 mL triangular flasks containing 20 mL of LB medium and incubated at 37°C with a shaking speed of 180 rpm. The bacterial inhibition rate of the culture medium was assessed at various time intervals.

Employing the flat plate confrontation method, the F. oxysporum was inoculated in the central region of a PDA culture medium on a flat plate. Three inoculants (fermentation solution, sterile filtrate, and bacterial suspension) were inoculated equidistantly around F. oxysporum. Each treatment was conducted in triplicate. Following a 5-day incubation period at 27°C, the colony diameter was measured, and the morphological alterations of the hyphae at the interface between diseased and healthy tissues were examined.

An equal volume of F. oxysporum (spore suspension) and B. subtilis LK−1 (fermentation liquid) were combined. The spore germination rate, the inhibition rate of spore germination, and the morphological changes in spores were assessed at 2, 4, 6, 8, 10, and 12 h after combining. In the control, the spore suspension of F. oxysporum was replaced with sterile water.

Once the L. barbarum seedlings developed two true leaves, the fermentation liquid of the biocontrol bacteria and the spore suspension of F. oxysporum were irrigated to the root zone as follows: (1) inoculation with 40 mL of water, set as the control group (CK); (2) inoculation with 20 mL of LK−1 and 20 mL of water (E); (3) inoculation with 20 mL of LK−1 and 20 mL of F. oxysporum (E + P); and (4) inoculation with 20 mL of water and 20 mL of F. oxysporum (P). After 30 days, various physical indexes of Lycium plants were measured, including plant height, root length, leaf number, leaf length, leaf width, and leaf area, as well as fresh and dry weights of both the aboveground and belowground parts.

Based on the one-way test, Design-Expert 12 software was used to design the PB test protocol to screen the main influencing factors from seven influencing factors using the inhibition rate (Y) as the response value (Table 1).

The steepest climb test was performed based on the analysis of the PB test results to determine the center of the response surface optimization.

Based on the variables screened in the PB test and the steepest climb test, the response values were analyzed using regression equations, with the inhibition rate (Y) designated as the response variable. This analysis was conducted using Design-Expert 12 software to design a three-factor, three-level experiment aimed at determining the optimal fermentation conditions for strain LK−1.

Seventy-five soil samples were collected from various regions in Ningxia, from which 136 bacterial strains exhibiting diverse morphologies were isolated. The F. oxysporum was set as the indicator organism for the chosen antagonistic bacteria, and five strains that demonstrated antagonistic activity against F. oxysporum were identified. The inhibition rate of strain LK−1 reached 58.48%, the inhibition rate of strain LK-2 was 53.91%, that of LK-3 was 56.52%, that of LK-4 was 57.39%, while that of LK-5 was the lowest, at 44.13%. Among these, LK−1 showed the best pathogen control ability with a pathogen inhibition rate of 58.48% and has been preserved for further study (Figures 1A,B).

LK−1 exhibited a positive Gram stain reaction and displayed a rod-shaped morphology. The individual colonies cultured on LB medium were characterized by their white coloration and wrinkled margins (Figures 1C,D).

The physiological and biochemical results were referred to “Bergey’s Manual of Systematic Bacteriology” and “Manual of Systematic Identification of Common Bacteria,” and combined with the morphological characteristics, the five strains were initially identified as Bacillus spp. (Table 2).

The sequence of strain LK−1 was amplified by PCR using the bacterial universal primer 16S rRNA, specific primers gyrB, rpoB, and recA, and strain LK−1 DNA as a template. The fragments amplified from strain LK−1 were all around the expected amplified bands (Figure 1E).

The 16S PCR product was sent to Shanghai Sangon Biologic Engineering Co., Ltd. for sequencing, and a sequence of 1,253 bp was obtained and uploaded to NCBI under accession number PP582380.

The sequences obtained from sequencing were subjected to BLAST homologous sequence comparison in NCBI. In total, 10 strains with high homology to the target sequences were selected, and a phylogenetic tree was constructed using MEGA 7.0 software (Figure 1F). Combined with morphological, physiological, and biochemical characteristics, strain LK−1 was identified as B. subtilis.

Co-infection with various Fusarium species often exacerbates the root rot disease affecting L. barbarum. Therefore, selecting a strain of antagonistic bacteria that exhibits solid inhibitory effects against multiple Fusarium strains is essential. The biocontrol effect of strain LK−1 was significantly higher than that of the other four strains, especially against the F. oxysporum and F. chlamydosporum of Ningxia. Consequently, the selection of the antagonist bacterium LK−1 for future experiments is warranted (Table 3; Supplementary Figure 1).

Figure 2A presents the inhibitory effect of LK−1 on Fusarium, where the inhibition rate increases over time, peaking at 48 h (Figure 2A).

Figure 2B compares the emergence or growth of microorganisms under sterile water treatment conditions. The emergence rate under PDB treatment is significantly higher than that under sterile deionized water treatment, indicating that PDB may promote the emergence or growth of microorganisms.

Figure 2C presents a typical growth curve of LK−1, where the increase in OD value reflects the increase in the number of microorganisms. The peak at 24 h and eventual stabilization likely indicate that the microorganisms have reached their maximum growth density. LK−1 exhibits logarithmic growth during the initial 0–24 h. Following this period, it transitions into a stabilization phase. During the logarithmic growth phase, this bacterium demonstrates rapid proliferation and heightened metabolic activity, rendering it particularly suitable for use as a seed fluid.

The most pronounced antagonistic effect was observed in the fermentation broth after 48 h of incubation (Figure 2), following which a slight decrease in this effect occurred, likely attributable to cell lysis. Subsequent experiments were conducted using the fermentation broth of strain LK−1, which had been cultivated for 48 h.

The B. subtilis solution exhibited the highest mycelial inhibition rate of 54.44%, followed by the bacterial suspension with a secondary inhibition rate. At the same time, the sterile filtrate demonstrated the lowest inhibition efficacy (Table 4).

Under normal culture conditions, the mycelium exhibits a consistent morphology characterized by uniform thickness and length, as well as straight growth patterns (Figures 3A,B). In contrast, mycelium subjected to LK−1 treatment demonstrates abnormal branching (Figures 3C–F), expansion (Figure 3D), and even lysogenic effects (Figure 3F). These alterations lead to pathological phenomena such as cell membrane rupture and leakage of intracellular inclusions.

The germination rate of spores cultured in sterile water consistently exceeded that of spores cultured in PDB across all time points. At the 12 h mark, the germination rates of spores cultured in PDB and sterile water were nearly identical (Figure 3H), but the spore size was significantly smaller than that of spores cultured in sterile water. Considering the consumption of experimental materials and lowering the cost, the spores incubated in sterile water for 12 h (90.33% germination rate) were selected to observe their spore morphology.

The bacterial suspension exhibited the most pronounced antagonistic effect against F. oxysporum (Table 4). As with the results of Section 3.2.3, the antagonistic effect of the sterile filtrate was weak, which may be due to the fact that strain LK−1 produces fewer inhibitory substances or has a shorter lifespan.

Under standard culture conditions, the spores exhibit normal germination characterized by elongated bud tubes (Figures 3G,H). Following treatment with strain LK−1, the spores exhibit characteristics such as inflation (Figure 3J) and deformation (Figures 3I,K,L), which adversely impact their normal germination and contribute to antagonistic effects.

The growth index of L. barbarum revealed that strain LK−1 significantly enhanced the growth of L. barbarum plants, primarily evidenced by increases in plant height, root length, aboveground fresh weight, underground dry weight, and leaf area. Compared with the blank control (CK), they increased by 37.25, 33.95, 93.02, 88.14, and 38.13%. There was a significant increase in the dry matter of L. barbarum roots after treatment with strain LK-1 (E). Under the stress of F. oxysporum, LK−1 was able to enhance the growth parameters of L. barbarum as demonstrated by the following results: compared to the pathogenic bacteria control group (P) treated with F. oxysporum alone, treatment (E + P) resulted in increases of 2.45% in plant height, 2.06% in root length, and significant enhancements of 70.49, 25.92, 52.97, 71.36, and 28.99% in aboveground fresh weight, belowground fresh weight, aboveground dry weight, belowground dry weight, and leaf area, respectively (Table 5).

The influence of the selected factors on the rate of inhibition, ranked in descending order, is as follows: pH > speed>time>beef paste>glucose>NaCl>inoculum (Tables 6, 7).

The fermentation broth of LK−1 exhibited the highest inhibition rate in test 3 (Table 8). Therefore, the conditions established in test 3 (fermentation time of 36 h, pH 7, and rotational speed of 180 rpm) should be designated as the central point for the factor levels in the response surface methodology.

Following the outcomes of the steepest ascent test, the Box–Behnken design principle from Design-Expert 12 software was employed to perform a three-factor, three-level experiment on the selected primary factors. Each experiment was replicated three times, and the results of this experimental design are presented in Table 9.

The analysis of variance for the response surface regression model is presented in Table 8. The multivariate equation correlating the inhibition rate (Y) with pH (A), speed (B), and time (C) was derived from the results analyzed using Design-Expert 12 software, expressed as Y = 61.56–0.16A - 0.3912B + 0.0587C - 0.145AB - 0.46 AC - 0.7525 BC - 1.83A²-1.6B²-0.0657C².

The regression model has an F = 62.54 and p < 0.00001 (Table 10), indicating that the model has reached a highly significant level. The coefficient of determination of the model, R² = 0.9877 and R²adj = 0.9719, indicates that the error of the regression model is small, and the model is well fitted with a high degree of confidence.

Following response surface optimization, the optimal fermentation parameters for strain LK−1 were established as 0.5% glucose, 2% beef paste, 2% NaCl, 0.5% yeast powder, a fermentation duration of 46.261 h, an inoculum of 1%, a pH of 6.647, and a speed of 170.541 rpm. At the same time, in order to validate the accuracy and feasibility of the obtained optimal fermentation conditions, the fermentation conditions were adjusted to 0.5% glucose, 2% beef paste, 2% NaCl, 0.5% yeast powder, a fermentation time of 46 h, an inoculum of 1%, a pH of 6.6, and a speed of 170 rpm, and the experiment was carried out for three replications. The inhibition rate obtained was 62.5%. It was 1.43% higher than the theoretical value and 8.06% higher than the unoptimized value (54.44%). It showed that the fermentation conditions of strain LK−1 obtained by response surface optimization were accurate and reliable.