Effects of organic acids on the chemotaxis profiles and biocontrol traits of antagonistic bacterial endophytes against root-rot disease in Panax notoginseng

Understanding the role of chemotaxis in ecological interactions between plants and microbes in the rhizosphere is necessary to optimize biocontrol strategies targeting plant soil-borne diseases. Therefore, we examined and profiled the antagonistic endophytic bacteria (AEB) population with chemotaxis potential in the medicinal plant Panax notoginseng using a cheA gene-based approach coupled with 16S rRNA sequencing. Phylogenetic analysis of the chemotactic AEB (CAEB) community in P. notoginseng enabled the identification of 56 CAEB strains affiliated with 30 species of Actinobacteria, Firmicutes, and Proteobacteria; Firmicutes, especially Bacillus, were predominant. We then systematically quantified the chemotactic response profiles of CAEB toward five organic acid (OA) attractants: citric acid, fumaric acid (FA), malic acid, oxalic acid, and succinic acid. Further hierarchical cluster analysis revealed that the chemotaxis of CAEB to the same attractant exhibited different patterns among not only genera but also species and even strains of the same species. Following chemotaxis and hierarchical analysis, we selected the strongest chemoattractant, fumaric acid (FA), as the target for evaluating the effects of OAs on the representative CAEB strain Bacillus amyloliquefaciens subsp. plantarum YP1. Application of FA significantly stimulated the chemotaxis ability and growth of YP1, and increased the transcript levels of cheA and biocontrol-related genes in YP1. This is the first study to characterise the diversity of chemotaxis profiles toward OAs in natural bacterial assemblages of P. notoginseng and to highlight how FA promotes the biocontrol-related traits of P. notoginseng-associated CAEB.


Introduction
Panax notoginseng (Burk.) F. H. Chen (Araliaceae) is a well-known traditional Chinese medicinal herb that has been cultivated in southwestern China for more than 400 years (Guo et al. 2010). Extensive pharmacological studies have shown that P. notoginseng and its active ingredients have many effects on the blood system, cardiovascular system, brain, vascular system, nervous system, metabolism, and immune regulation-their specific effects include anti-arrhythmia, anti-cerebral ischemia, anti-platelet aggregation and thrombosis, anti-oxidation, anti-tumor, anti-atherosclerosis, improving hemodynamics, and regulating blood lipid (Sha et al. 2018;Xu et al. 2019). High planting density, a long growth period, and shade and humid planting conditions provide a favorable environment for a number of soil-borne pathogens to survive (Chen et al. 2001;Wang et al. 1998). The most destructive soil-borne disease in the P. notoginseng-growing areas of China is root-rot disease (RRD), which is caused by bacteria, fungal pathogens, and parasitic nematodes alone or simultaneously, and results in a serious reduction in the yield and quality of its raw active ingredients (Mao et al. 2013;Miao et al. 2006). RRD is very difficult to control, mainly because of the complicated interaction among different pathogens (Luo et al. 1997).Current available measures to manage root-rot disease of P. notoginseng mainly rely on the chemical pesticides, which would cause the serious problem of pesticide residues on product and heavy metals contamination in soil at the same time. Additionally, heavy metal pollution and pesticide residues caused by indiscriminate use of chemicals affect the quality and medicinal value of P. notoginseng severely. Recently, biological control, as an eco-friendly countermeasure, has been considered as a potential feasible alternative. Endophytic microbes isolated from P. notoginseng were demonstrated to be effective in control of root-rot disease (Ma et al. 2013).
Endophytic bacteria are an essential part of the plant microbiome and may have a competitive advantage over rhizospheric bacteria because they live in plant tissue without having any detrimental impact on the host plant. Therefore, their re-introduction does not affect the indigenous bacterial population within the host (Khare et al. 2018). Thus, they have become the most attractive and potentially important BCAs against soil-borne diseases (De Silva et al. 2019). Our lab has demonstrated bacterial endophytes of P. notoginseng have antagonistic properties against multiple pathogens-to be potential BCA candidates against RRD in P. notoginseng (Ma et al. 2013). Nonetheless, the exploitation of endophytic inoculants as BCAs has been hampered by inconsistent performance of endophytic bacteria at the field scale (Barret et al. 2011;Raaijmakers et al. 2009). As a result of such variation, a successful rhizosphere interaction with plants has been recognized as a prerequisite for the biocontrol efficacy of the inoculant strain against soil-borne pathogens (Abdallah et al. 2018;Schreiter et al. 2018). Therefore, it is imperative to attain a better understanding of plantendophytic bacterial interactions in the rhizosphere to create successful and reproducible biological controls in agriculture. The ecological interactions between plants and endophytic bacteria, however, are complex in the rhizosphere and are still far from being fully elucidated.
It is well established that chemotaxis toward root exudates (REs) is a key bacterial trait contributing to successful rhizosphere interactions (Raina et al. 2019) and biocontrol efficacy of inoculants under field conditions (Allardmassicotte et al. 2016;Ling et al. 2011;Tan et al. 2013;Ma et al. 2018). Numerous studies have shown the chemotactic response of endophytes to root exudates of host plants (Compant et al. 2010;Khare et al. 2018;Rosenblueth and Martínez-Romero 2006). Organic acids (OAs) are one of the key components in root exudates (Jones 1998;Liu and Wen 2006). Many studies have emphasized the regulatory role of OAs in plant-microbe interactions in the rhizosphere (el Zahar et al. 2014) and demonstrated that OAs can serve as major chemoattractants in root exudates that induce the recruitment of BCAs to the hosts, thus protecting roots from infection (Ling et al. 2011;Tan et al. 2013;Yuan et al. 2015;Zhang et al. 2013Zhang et al. , 2014. Therefore, chemotaxis toward OAs is putatively considered a bacterial trait that defines biocontrol ability against soil-borne pathogens. At the molecular level, bacterial chemotaxis is mediated by a complex chemosensory pathway system composed of the sensor kinase CheA and response regulator CheY. Chemoreceptors, methyl-accepting proteins (MCPs), are transmembrane signal transducers located in the cytoplasmic membrane (Yang and Briegel 2020). CheA and CheY are the core proteins in the chemotaxis signaling pathway, and microbes possessing the encoding gene cheA are believed to be chemotactic (Bi et al. 2018). Accordingly, a molecular diagnostic tool based on the gene encoding the central regulator of bacterial chemotaxis (cheA) has been developed to characterise and track specific populations of native microbes in the rhizosphere with chemotactic potential (Buchan et al. 2010). The cheAbased approach is important because it allows researchers to easily distinguish chemotactic-competent bacteria in a population, which opens up new opportunities for the enrichment of functional species from plant-associated communities. Increasing evidence suggests that chemotaxis is a prevalent phenotype in the plant microbiome within the context of plant-microbe interactions (Li et al. 2012); however, to date, it remains unclear whether endophytic bacteria with chemotaxis potential are present in the microbiomes of medicinal plants.
Bacterial endophytes are important in soil-borne disease management, and chemotaxis improves rhizospheric interactions and the biocontrol efficacy of inoculants; it is very important to optimize the selection of endophytic sources with chemotactic potential to act as biocontrols against soil-borne disease and establish a basis for efficiently and practically applying antagonistic candidates in the field. Therefore, the objectives of the present study were as follows: (1) profile the abundance and diversity of the chemotactic-competent bacteria in an antagonistic endophytic bacteria (AEB) population of P. notoginseng with a cheA gene-based approach combined with 16S rRNA sequencing; (2) systematically evaluate the chemotaxis capabilities of chemotactic AEB (CAEB) populations in P. notoginseng using five OAs (citric, fumaric, malic, oxalic, and succinic acids) and characterise the diversity of chemotaxis profiles of the CAEB population to tested OAs, emphasizing the genus Bacillus; (3) investigate the effects of OAs on biocontrol-related traits of P. notoginseng-associated CAEB by targeting the main chemoattractant in tested OAs to evaluate its effect on chemotaxis and the growth and transcript levels of genes involved in the biocontrol activity of representative CAEB strains. This study could further the understanding of beneficial bacterial endophyte-plant interactions via chemotaxis toward OAs, facilitating the development of novel plant protection strategies.

Sample collection and isolation of endophytic bacteria
Root, stem, petiole, leaf and seed samples were collected from thirty 3 year old healthy P. notoginseng plants, grown in Luxi country of Yunnan Province, China. Surface disinfection of samples and isolation and purification of cultivable endophytic bacteria were performed according to the methods of Ma et al. (2013): Briefly, the samples were washed in running tap water to remove soil and the leaves, stems, petioles, and roots were separated and weighed individually, then immersed in 70% ethanol for 4 min, washed with fresh 2% sodium hypochlorite solution for 1-2 min, depending on the different tissues, followed by soaking in 70% ethanol for 1 min and finally washed three times with sterile distilled water. The disinfection process was checked by plating aliquots of the sterile distilled water used in the final rinse onto a plates containing KMB (King et al. 1954), NA (L -1 : beef extract 3 g, peptone 10 g, NaCl 5 g, agar 20 g, distilled water 1,000 mL, pH 7.0-7.2) and YSP media (L -1 : peptone 10 g, yeast extract 5 g, glucose 20 g, agar 20 g, distilled water 1,000 mL, pH 7.2) supplemented with 50 lg mL -1 of the fungicide Imazalil and incubating the plates at 32°C for 3 days. Sterile samples from the same tissue were homogenized in a sterilized mortar with 10 mL sterile distilled water under aseptic conditions. After filtration by passing through four layers of lens cleaning tissue (Whatman, Catalog Number: 2105-918), 200 lL of appropriate dilutions was spread on a plates containing KMB, NA and YSP media. Triplicates were conducted for each medium of each sample. After incubation at 32°C for 3 days, bacterial colonies with unique morphotypes were picked and sub-cultured and the pure cultures thus obtained were used for the study.
Screening of AEB against pathogens of RRD in P. notoginseng The fungal pathogen Fusarium oxysporum, the bacterial pathogen Ralstonia sp. and the parasitic nematode Meloidogyne hapla, which have been reported as three major pathogens associated with RRD of P. notoginseng (Miao et al. 2006), were used as the targets for antagonistic screening. The pathogens were isolated from root-rot samples of P. notoginseng. F. oxysporum and Ralstonia sp. were cultured respectively on potato dextrose agar (PDA) and NA media. Juveniles of M. hapla were obtained from root-rot samples by the Baermann-funnel method (Baermann 1917) and used directly. For antibacterial bioassay, 200 lL fresh culture of Ralstonia sp. with concentration of 10 8 -CFU mL -1 was mixed with 250 mL NA and evenly distributed into Petri dishes. On each plate, 6 wells of 5 mm diameter were made. Candidate bacteria were cultured in NB medium at 37°C, 200 rpm for 48 h and the cell concentration was adjusted to 10 7 -CFU mL -1 with NB. 200 lL of endophytic bacterial suspension was added to each well. An equivalent volume of liquid NB was used as control in place of bacterial culture. All treatments were tested in triplicate. After 48 h at 32°C diameters of antibacterial zones (AZ) were measured. Here, AZ was directly used to express the antibacterial efficiency of endophytic bacteria as the AZ value of the control was zero. Antifungal bioassays were performed following the Oxford cup method (Wang et al. 2009). An aliquot of 200 lL culture candidate suspension, prepared as described above, was added into an Oxford cup (diameter 5 mm) which was previously placed into the center of a PDA plate. Two 5 mm diameter of mycelial plugs of F. oxysporum from an actively growing colony were placed on two sides of the cup at a 2 cm distance. An equivalent volume of NB medium in place of the suspension was used as a control. All treatments were performed in triplicate. After incubation at 28°C for 4 days, antifungal efficiencies (AE) was calculated using the formula AE = (DC -DT)/ DC 9 100%, where DC and DT respectively represented the colony diameters of F. oxysporum on the control and the treatments. For nematicidal bioassay in the wells of the 24-well microtitre plate, 200 lL of bacterial culture prepared as above was mixed with 50 lL of M. hapla suspension containing about 100 juveniles. Each treatment was replicated three times. Wells containing NB medium served as controls. After incubation at 28°C for 72 h, the numbers of live and dead nematodes were counted under a stereomicroscope. The nematicidal efficiency (NE) was calculated using formula of NE = DN/SN 9 100%, where DN represents the difference in number of dead nematodes between treatment and control, SN represents the sum of counted nematodes.
Screening of candidate chemotactic AEB (CAEB) via cheA gene detection The bacterial DNA was extracted using a bacterial genomic DNA extraction kit (BioTeke Corporation, China, Cat#: DP2001). The cheA gene was amplified by PCR using primers P4P5.for and P4P5.rev (Buchan et al. 2010). The 25 lL reaction mixture consisted of 2 lL DNA template, 3.75 U Taq DNA polymerase (TaKaRa, Japan), 3 lL 10 9 PCR reaction buffer supplied with the enzyme, 1.25 lL 50 mM MgCl 2 , 1 lL each 10 lM primer, 13 lL 2.5 mM dNTP (Vivantis, Malaysia) and 3 lL nuclease-free water (Promega, WI). The reaction conditions were 95°C for 3 min, followed by 35 cycles of 95°C for 1 min, 55°C for 1 min and 72°C for 2 min, with a final extension step at 72°C for 3 min. After purification using an Agarose gel DNA purification Kit (BioTeke Corporation, China, Cat#: DP1502), the amplification products were detected by horizontal electrophoresis through 1% SIGMA Type II agarose gels. The electrophoresis was performed at 120 V, 300 mA, and the image was photographed under UV illumination using Gel Doc (BIO-RAD) software. Here, it must be pointed out that additional 64 AEB strains from P. notoginseng previously identified from our lab (Ma et al. 2013) also included this assay to screen chemotactic AEB.
Phylogenetic analysis of P. notoginsengassociated CAEB population The 16S rRNA sequence-based phylogentic identification was done as previously described (Ma et al. 2013). The genomic DNA of bacteria was extracted using a bacterial DNA extraction kit (BioTeke Corporation, China, Cat#: DP2001) and 16S rRNA genes were amplified by PCR using the primer pair of 27f and 1492r (Lane 1991). The PCR amplified products were separated by agarose gel electrophoresis, and sequenced on an ABI Prism 3730 sequencer at Beijing Huada Biological Company. The identification of the closest phylogenetic neighbours and calculation of pairwise 16S rRNA gene sequence similarities were performed using the EzTaxon-e server http://www. ezbiocloud.net/ (Kim et al. 2012). Sequences chimera checking were performed by the program CHIMERA CHECK of the Ribosomal Database Project (RDP) (Maidak et al. 1997). Sequences with a potential chimeric structure were excluded from further analyses. The 16S rRNA gene sequences were aligned with representative bacterial using the Clustal W program in the MEGA 7 software package (Kumar et al. 2016). Phylogenetic analysis was done using the neighbourjoining method with kimura-2-parameter as a model of nucleotide substitution and complete deletion option as well as 1000 bootstrap replications in the MEGA 7 software package (Kumar et al. 2016). The partial 16S rDNA sequences of the representatives were deposited in GenBank (Table 1).

Chemotaxis experiments
Qualitative soft agar drop plate assays Bacterial chemotaxis was tested using a soft-agar drop plate assay as described previously (Li et al. 2012;Samanta et al. 2000) with slight modification. Briefly, bacteria were grown in NB medium at 32°C, 180 rpm. At log phase (OD 600 = 0.3), bacterial cells were harvested and washed three times with chemotactic buffer (MM solution: 10 mM potassium phosphate buffer (pH 7.0), 0.1 mM EDTA, 0.05% glycerol 5 mM Sodium-D-L, lactate, 0.14 mM CaCl 2 , and 0.3 mM (NH 4 ) 2 SO 4 , then resuspended in MM solution at * 1 9 10 8 CFU mL -1 , and 5 mL of cell suspension was added into a Petri dish containing drop assay medium (MM solution containing 0.3% agar and 1 mM glucose as an energy source); the same volume of chemotactic buffer was used as a negative control. Then 0.1 g of each organic acid compound (citric acid, fumaric acid, malic acid, oxalic acid, and succinic acid) in crystal form was placed in the center of each plate, and an equivalent weight of aspartic acid served as the positive control. No substrate negative controls were used. After incubation for over 12 h at room temperature (25°C), the chemotactic response was observed as the formation of a chemotactic ring near the center of each Petri dish and scored on a scale as described previously (Greer-Phillips et al. 2003;Repik et al. 2000). Subsequently, based on the diameter of the chemotactic ring and bacterial phylogenetic affiliation, hierarchical clustering was performed to characterise chemotactic response profiles toward the five organic acids across the representative CAEB associated with P. notoginseng using the cluster R package (Maechler et al. 2017). The heatmap was realized with the Pheatmap R package (Kolde R. Pheatmap: Pretty Heatmaps; R package version 1.0.8).

Quantitative capillary assays
Capillary assays were performed according to the previously described protocol for quantitatively measuring the chemotaxis response of the bacteria to the chemoattractants (Adler 1973;Rudrappa et al. 2008). The bacterial strain was grown in culture liquid media until the OD 600 reached 0.4-0.6, at which point the cells were collected by centrifugation and washed twice with chemotactic buffer, then adjusted to an OD 600 of 0.1 for further testing. A 60-mm Petri dish was loaded with 10 mL of the cell suspension prepared above. A standard 1 lL capillary was filled with the OAs at different concentrations (0, 10, 25, 50, and 100 lM) and immersed in the cell suspension in the Petri dishes. After 2 h incubation at room temperature, capillary contents were serially diluted in sterile phosphate buffer and plated on LB plates. The number of bacteria accumulating in the capillaries was determined by CFU counting in LB plates incubated for 48 h. Phosphate buffer served as a control. Each treatment was replicated three times. Responses to the attractants were expressed as the relative chemotactic response (RCR), representing the ratio of the cells accumulated in capillaries containing attractants and phosphate buffer (control). An RCR value of 1 and 2 or greater indicates no chemotaxis and significant chemotaxis, respectively.
Effects of OAs on the growth of the representative P. notoginseng-associated CAEB strain in vitro An assay was designed to evaluate the effect of exogenous OAs at different concentrations on the growth of representative CAEB as previously described (Liu et al. 2019). The bacterial cells were incubated overnight in 10 mL NB medium and then adjusted to an OD 600 of 0.1. Subsequently, a 1 mL aliquot of the resulting suspension was inoculated into 100 mL of minimal medium (MM) with different concentrations of an OA as the sole carbon source. Cultures were incubated at 32°C with 180 rpm shaking for 48 h, and finally, the OD 600 was monitored from each treatment. Each treatment group composed of three replicates, and sterile deionized water was used instead of OAs as a control. Transcription analysis of chemotaxis and biocontrolrelated genes in representative P. notoginsengassociated CAEB strains in the presence of an OA To determine the influence of OAs on the transcript levels of biocontrol-related genes in the P. notoginseng-associated CAEB strain, the following modified experiment was performed: (1) strain grown in NB medium without individual OAs (control); (2) strain grown in NB medium containing individual OAs at an appropriate concentration. Strain cultures were incubated overnight and then adjusted to an OD 600 of 0.  was calculated with the 2 -DDCT method (Livak and Schmittgen 2001). The 16S rRNA gene was employed as the internal control for gene expression analysis. Fold induction was calculated by comparing gene expression for growth in the presence of an OA to growth in the absence of an OA. Three replicates were conducted for each treatment. These genes were selected for transcription analysis based on previous studies that found evidence suggesting that those genes are important for the biocontrol efficacy of bacteria (Bais et al. 2004;Joshi and Gardener 2006;Ongena and Jacques 2008). These included lipid peptide antibiotic synthesis genes (srfAA, bmyB), a biotin synthase gene (bioA), an antibacterial protein synthesis gene (yndJ), the lipid peptide antibiotic family (lpa-14, srfAB), and a surfactin production gene (sft) in the Bacillus sp. The primers of the target genes were used according to previous studies (Wu et al. 2015(Wu et al. , 2017.

Statistical analysis
Differences among treatments were analysed using an ANOVA followed by Fisher's LSD test. Before analysis, data were log-transformed to reach normality. The Kolmogorov-Smirnov test was applied to test for a normal distribution (p [ 0.05). Additionally, differences between the presence and absence of nicotine treatments were analysed by the independentsamples t-test at the 5% level. All statistical analyses were performed with the SPSS BASE ver.17.0 statistical software (SPSS, Inc., Chicago, USA). Rejection level was set at p \ 0.05 for all analyses. Unless stated otherwise, all values reported are means ± standard deviation (SD).

Candidate chemotactic AEB for RRD from P. notoginseng
We previously reported that 104 endophytic strains of P. notoginseng screened from leaf, petiole, stem, root, and seed samples grown in Wenshan region of Yunnan Province, China exhibited antagonistic properties against at least one RRD pathogen (Fusarium oxysporum, Ralstonia sp., and Meloidogyne hapla) (Ma et al. 2013), but previous studies have indicated that the abundance, diversity, and species assemblage of endophytes can be strongly influenced by the geographic regions they occur in (Deng et al. 2011;Yaish et al. 2015). These observations prompted us to further study the antagonistic endophytes associated with P. notoginseng growing in a different geographic area (Luxi County, Yunnan Province, China) to explore novel and more extensive antagonistic endophytes associated with RRD of P. notoginseng. Accordingly, 600 endophytic bacteria were isolated from different tissues (leaf, petiole, stem, and root) of P. notoginseng grown in Luxi County, different ecological region from Wenshan County in Yunnan Province, China, and were further evaluated in vitro for their antagonistic activity towards the pathogens F. oxysporum, Ralstonia sp., and M. hapla. Of the 600 isolates, 118 strains showed antagonistic activity against at least one of the tested pathogens (Table 1). PCR amplification using the P4P5.for and P4P5.rev primer pair revealed that 56 of the 222 isolates (104 previously isolated and 118 obtained from the present study) produced a characteristic cheA band (approximately 500 bp), which could be taken as chemotactic candidates (Table 1). The chemotactic ability of all 56 candidates that produced the cheA band was verified by the drop assay with aspartic acid, which was used as the attractant in a previously described chemotaxis assay (Singh et al. 2010). Results indicated that the 56 candidates could be taken as chemotactic strains because they all exhibited positive chemotaxis toward the tested aspartic acid ( Fig. S1 and Table S1).

Phylogeny of P. notoginseng-associated chemotactic AEB
The 16S rRNA gene phylogenetic analysis was used to assign all 56 chemotactic AEB (CAEB) into three bacterial phyla: Actinobacteria, Firmicutes, and Proteobacteria ( Fig. 1 and Table 1). The most abundant c Fig. 1 Neighbor-joining phylogenetic tree of chemotactic antagonistic endophytic bacteria associated with root-rot disease in Panax notoginseng and their closest relatives based on the 16S rRNA gene sequences. Chemotaxis toward organic acids by P. notoginseng-associated CAEB The chemotactic response toward the five organic acids (OAs; citric acid (CA), fumaric acid (FA), malic acid (MA), oxalic acid (OX), succinic acid (SA)) for CAEB from P. notoginseng was determined in drop medium plates. The substrates tested and chemotactic responses of the representative 32 strains observed are listed in Fig. S1-6 and Table S1 to illustrate the scoring of the chemotactic response. A stronger response indicated that more cells accumulated near the attractants, and that they formed a dense ring. Overall, FA-followed by CA and MA-invoked the strongest positive chemotaxis in tested strains according the chemotactic response scoring (diameter of the chemotactic ring) (Fig. 2). As shown in Fig. 2 (Fig. S2-6 and Table S1), which indicated that these strains could be candidates for further study.
Diverse chemotactic response profiles across P. notoginseng-associated CAEB At the genus level, chemotactic response profiles of the CAEB strains varied greatly across all of the tested OAs, and there was no general trend in strong or weak chemotaxis for any of the phylogenetic relationships. Using hierarchical cluster analysis and a heatmap, CAEB strains belonging to different genera (12 genera) exhibited different chemotactic responses to the same tested OAs, even when the strains were responding to a positive control AS (Fig. 3). For example, strain L22 (affiliated with Acinetobacter calcoaceticus) exhibited the strongest chemotactic response to CA, as evident by the diameter of the chemotactic ring reaching 70 mm (Table S1 and Fig. S2). However, very weak chemotaxis of strain YS3 (Kytococcus sedentarius) was measured in the diameter of the chemotactic ring responding to CA (5 mm; Table S1, Fig. S2). Additionally, within the same genus, different species have diverse chemotactic response profiles, as demonstrated by Bacillus ( Fig. S7 and Table S2). We found considerable diversity in chemotactic response profiles among the 15 strains of Bacillus using hierarchical cluster analysis and a heatmap (Fig. S8). Two strains-L14 and NL2, classified as B. aryabhattai and B. aerophilus, respectively-showed the strongest chemotactic response to CA, while another strain-R7, classified as B. amyloliquefaciens subsp. plantarum-exhibited very weak chemotaxis to CA. Interestingly, in our study, the chemotactic response towards specific chemoattractants was found to be diverse not only among species but also among strains of a single species. For example, we found a striking diversity in the responses to all tested substrates across all five strains of B. amyloliquefaciens subsp. plantarum, as all the strains displayed unique chemotaxis Fig. 2 The relative proportions of species of chemotactic antagonistic endophytic bacteria associated with root-rot disease in Panax notoginseng responding to five organic acids. As such, this figure provides an overview of the chemotactic proportion according to the response score by the community to each chemoattractants. For more details on each taxon responding to individual chemoattractants, see Supplementary  Table S1. The species among the chemotactic antagonistic endophytic bacterial community associated with P. notoginseng are denoted by the colors in the legend. The following organic acids were used: citric acid (CA), fumaric acid (FA), malic acid (MA), oxalic acid (OX), and succinic acid (SA). The strength of the chemotactic response was quantified by measuring the diameter of the chemotactic ring and scored on a scale as described previously (Greer-Phillips et al. 2003;Repik et al. 2000). In this figure, the scale was denoted as 1-5 followed by the name of organic acid on the horizontal x-axis from very strong response ([ 51 mm, denoted as 1), stronger response (31-50 mm, 2), strong response (21-30 mm, 3), weak response (1-20 mm, 4), and ''-'' no response (5) response profiles to tested substrates ( Fig. S7 and Fig. S8 and Table S2).
Assessment in the chemotaxis of the representative P. notoginseng-associated CAEB strain B. amyloliquefaciens subsp. plantarum YP1 in the presence of OAs Taking the antagonism, chemotactic response profiles, strength of chemotaxis, and affiliation to the genus Bacillus into consideration, B. amyloliquefaciens subsp. plantarum YP1 was selected as the representative strain to evaluate the role of OAs-identified previously in root exudates of P. notoginseng , in the biocontrol traits of the P. notoginsengassociated CAEB assemblage-concerning chemotaxis, growth, and antagonistic activity. A modified capillary assay was performed to assess the effects of OAs on the chemotaxis of YP1. The results showed that all tested OAs at concentrations of 10-100 lM significantly induced chemotactic activity in YP1 (Fig. 4). The concentrations of OAs (10-100 lm) tested in quantitative chemotaxis studies were chosen, as several studies have shown that OAs detected in REs at this concentration range lead to positive and apparent chemotaxis responses in Bacillus, and thus this range is commonly employed in capillary assays (Ling et al. 2011;Tan et al. 2013). The effects of OAs on the chemotactic response of YP1 differed depending on the OAs used. In general, CA and FA stimulated chemotaxis in YP1 markedly more than other OA treatments. FA had particularly marked effects on YP1 chemotaxis at various tested concentrations, and the RCR values obtained at concentrations from 10 to 100 lM were 1.1-to 1.3-fold higher in the FA treatment (Fig. 4a) than the CA one (Fig. 4b). Meanwhile, additional comparative analyses showed that FA and CA induced the chemotaxis of YP1 in a concentration-dependent manner within a dose range of 10 to 50 lM, and a 50 lM concentration yielded the highest chemotaxis in both cases (RCR = 6.73 for FA and RCR = 5.20 for CA, Fig. 4a and b, respectively). The quantitative reverse transcription-PCR (qRT-PCR) regarding cheA relative gene expression levels in YP1 demonstrated that FA at 50 lM induced the highest increase in cheA gene transcript expression (Fig. 5).
Effects of FA on the growth of B. amyloliquefaciens subsp. plantarum YP1 in vitro The results of the capillary assay suggested that the most potent chemoattractant among tested OAs for strain YP1 was FA, as evident by the observation that FA invoked the highest chemotaxis in YP1, especially at a concentration of 50 lM (RCR = 6.73) (Fig. 4a). Therefore, we putatively selected FA as the target to evaluate the effects of OAs on the growth and biocontrol activity of P. notoginseng-associated CAEB. Compared to the control (0 lM), exogenous applications of different concentrations of FA had different but stimulating effects on YP1 growth at concentrations of 25-100 lM (Fig. 6). Early in incubation (12 h), YP1 had a similar growth tendency in the presence of FA and CK. Subsequently, the results showed that FA significantly stimulated the growth of YP1 at 25 and 50 lM after 12 h incubation, and the OD 600 of cell biomass reached its highest level at 28 h under a 50 lM FA concentration. Moreover, the stimulation effect at 50 lM was significantly (p \ 0.05) higher than at other concentrations.
The effect of FA on the biocontrol activity of B. amyloliquefaciens subsp. plantarum YP1 YP1 growth in vitro was highest with 50 lM FA, and this was thus chosen as the appropriate concentration for further analysis. Results of in vitro assays showed that treatments with 50 lM of FA significantly enhanced the antagonism of YP1 towards RRD of P. notoginseng compared to the control treatment without the FA (Ma et al. unpublished). Thus, to perform an in-depth analysis on the effects of FA on the antagonistic activity of YP1, the relative expression levels of the genes involved in biocontrol activity were analysed in YP1 grown in the presence of FA. As illustrated in Fig. 7, the QRT-PCR results suggested that FA at 50 lM could increase the transcription levels of all tested biocontrol-related genes, and significantly enhance the transcription levels of srfAA and sft genes in YP1, with increases of 2.4-and 3.6fold up-regulation compared to the control (without FA), respectively.

Discussion
The composition and abundance of endophytic communities with the same host plant varies not only from region to region but also among conditions within the same region (Deng et al. 2011;Yaish et al. 2015). Although our previous studies have shown that a diverse assemblage of bacterial endophytes are antagonistic towards RRD in P. notoginseng planted in Wenshan County, Yunnan Province, China (Ma et al. 2013), more extensive antagonistic resources were needed from endophyte populations of P. notoginseng in different growing regions. Therefore, we re-isolated bacterial endophytes from a distinct geographic area, Luxi County in Yunnan Province, and further evaluated in vitro their antagonism against RRDC, which will provide better insight into the diversity of AEB associated with P. notoginseng and a more comprehensive understanding of the biocontrol potential of endophytic bacteria harbored in P. notoginseng plants.
Chemotaxis is considered an important characteristic of a successful bacterial BCA candidate for soilborne pathogens (Allardmassicotte et al. 2016;Raina Fig. 5 Effects of organic acids on the cheA gene transcript expression in Bacillus amyloliquefaciens subsp. plantarum YP1. Strain YP1 grown in NB medium was amended with citric acid (CA), fumaric acid (FA), malic acid (MA), oxalic acid (OX), and succinic acid (SA) at a final concentration of 50 lM. The data presented are the means ± standard deviation (SD) of three independent replicates. Columns with different letters represent statistically significant differences for each gene according to Duncan's multiple range test (p \ 0.05). The data showed cheA gene expression in OAs (FA, CA, MA, OX, and SA)-treatments with fold changes of 2. 11, 1.32, 1.02, 0.57, 0.77, respectively, vs. mock -treatment et al. 2019;Wu et al. 2015). Thus, identifying chemotactic bacterial populations from plant-associated microbiomes and understanding the underlying biological processes in this association is of the utmost importance for properly exploiting efficient BCA candidates for sustainable agriculture. Our previous work employed a cheA gene-based approach (Buchan et al. 2010), focusing on chemotactic antagonistic bacteria against plant pathogens, revealed that chemotaxis may be an important phenotype among soil-associated antagonistic bacteria (Li et al. 2012). However, little information exists about whether chemotaxis is a prevalent phenotype among plantassociated bacterial endophytes. Therefore, this is the first study to examine and profile the genetic diversity of chemotaxis genes in medicinal plant-associated endophyte communities, and it substantially expands the known diversity of chemotactic-competent plantarum YP1. Strain YP1 grown in NB medium without FA was used as CK to normalize the expressed value. NB medium was amended with FA at a final concentration of 50 lM. The data presented are the means ± standard deviation (SD) of three independent replicates. Columns with different letters represent statistically significant differences for each gene according to Duncan's multiple range test (p \ 0.05) bacterial populations. The most abundant class in this assembly was Firmicutes, with members of Bacillus representing the majority. Here, comparative analysis showed different community compositions between P. notoginsen-associated and rhizosphere-associated chemotactic bacterial communities (Buchan et al. 2010;Li et al. 2012), indicating that the abundance of cheA gene profile might vary among microbial assemblages in different ecological environments. Meanwhile, our results demonstrate that bacteria capable of chemotaxis are present and abundant among AEBassociated communities with P. notoginseng, and we thus propose that chemotaxis plays an important role in establishing and maintaining special P. notoginseng-endophyte associations. Therefore, studies on chemotaxis in CAEB communities are supposed to be necessary for a more comprehensive understanding of the ecological mechanisms and bacterial behaviors underpinning P. notoginseng-endophyte ecological associations.
Organic acids, secreted from plant roots, play important roles in plant-BCA interactions and have been reported to be the most common chemoattractants for many biocontrol microbes (el Zahar et al. 2014;Khare et al. 2018;Ling et al. 2011;Tan et al. 2013;Yuan et al. 2015;Zhang et al. 2014). To achieve a deeper understanding of chemotaxis of CAEB assemblages in P. notoginseng, we systematically quantified the chemotactic responses of these bacteria to organic acids. Thus, five OAs (CA: citric acid, MA: malic acid, FA: fumaric acid, OA: oxalic acid, and SA: succinic acid), identified previously in the root exudates of P. notoginseng , were selected to evaluate the chemotactic response of CAEB assemblages in P. notoginseng in this study. Our data provided the first evaluation of chemotactic behavior among natural populations of P. notoginseng-associated CAEB, with the different chemotaxis patterns observed towards the different OAs (Fig. S2-6 and Table S1), which are highly consistent with the findings that many soil microbes respond differently towards different OAs (Ling et al. 2011;Tan et al. 2013). In this study, citric acid and fumaric acid were relatively stronger chemoattractants for most CAEB assemblages among all tested OAs, indicating that these compounds might be the most effective bioattractants, and they may provide a competitive and selective advantage in plant-microbe interactions. Accordingly, we raised the possibility that CA and FA might serve as major chemoattractants in root exudates of P. notoginseng that induce CAEB recruitment to the host. These data underline the need for future investigations aimed at evaluating the actual role of chemotaxis in response to these OAs in P. notoginseng-CAEB interactions.
To perform an in-depth characterization of the chemotactic response profiles of the CAEB strains toward the tested OAs, an hierarchical cluster analysis was employed, which revealed a striking diversity of responses to all five OAs (Fig. 3 and S8). It is clear that the chemotaxis responses of CAEB to the same tested substrate differ not only among genera but also among species and even within strains of a single species. The presence of multiple chemotaxis signal transduction pathways and a great diversity of chemoreceptors with novel sensory specificities in bacterial chemotaxis is well documented (Miller et al. 2009). The diversity in chemotaxis signal transduction systems and chemoreceptors probably explains the diverse response profiles of the CAEB to different OAs. Many bacterial species possess a large number of different chemoreceptors to mediate responses to different chemical stimuli (Yang and Briegel 2020). Further identification of bacterial chemoreceptors for OAs will provide a more thorough understanding of the OA chemosensory system and allow researchers to compare the diversity of chemotaxis pathways and chemoreceptors identified in silico. The discovery that CAEB respond to five OAs via chemotaxis expands the repertoire of known chemotactic bacteria and is a proof that endophytic bacteria show a behavioral response to OAs in medicinal plants, which had previously not been examined.
In light of our results showing that all 32 representative strains of CAEB in P. notoginseng exhibited positive chemotactic responses to all tested OAs, we then hypothesized that chemotaxis towards OAs might be a biocontrol trait of CAEB assemblages in P. notoginseng. To address this question, we investigated the effects of OAs on the antagonistic activities of endophytic bacteria concerning the relationship between OAs and chemotaxis and biocontrol ability of endophytes in the rhizosphere. It is evident from the results that the effects of OAs on the chemotactic response of YP1 differed depending on the OAs used (Fig. 4, Fig. S2-6). Interestingly, the effects of CA and FA on chemotaxis in YP1 were concentration-dependent ( Fig. 4a and b), and FA at 50 lM exhibited a significant effect on the chemotaxis in YP1 (Fig. 4a), which was further confirmed by the expression levels of cheA gene induced by FA at 50 lM (Fig. 5a). The capacity of a compound to elicit a chemotactic bacterial response might be related to its nutritional properties. Thus, we wondered whether OAs are used as a nutritional basis for CAEB strains to support their growth. Following the observation that FA is the strongest chemoattractant to YP1, we further selected FA as the target to test the effect of OAs on the growth of YP1. As expected, the results obtained in vitro suggested that FA at different concentrations promoted YP1 growth to different degrees (Fig. 6). Importantly, we confirmed that YP1 growth significantly improved when FA was its sole carbon source, and peaked when the concentration of FA was 50 lM (Fig. 6). These findings corroborate previous studies that have shown that OAs as main chemoattractants in REs have significant stimulatory effects on the growth of microbes interacting with plants and can be used as nutrients or signals by some beneficial microbes (Liu et al. 2019;Wu et al. 2017;Zhang et al. 2013).
The observation presented here, showing that FA can stimulate the growth and chemotactic ability of YP1, led us to think that the antagonistic activity of YP1 might also be affected by FA. To verify this hypothesis, we carried out a series of experiments to determine the relative contribution of FA to the antagonistic activity of YP1. It was observed that the addition of FA could enhance the transcriptional expression levels of selected biocontrol-related genes in YP1 (Fig. 7) and increase the antagonistic activity of YP1 in vitro (Ma et al., unpublished). Lipopeptides (LPs) produced by Bacillus are well-known to be important factors in the antagonistic activities against pathogens . It has been demonstrated that the transcription level of the sft gene involved in LP production can be significantly enhanced by adding pectin, a kind of tobacco root extract, as it might help increase the biocontrol activity of B. amyloliquefaciens SQY162 against tobacco bacterial wilt (Wu et al. 2015). Accordingly, in present study, significant increases in the expression of genes (srfAA, sft) involved in LP production and synthesis in YP1 by applying FA (Fig. 7) might enhance the colonization ability of YP1, so we postulated that the biocontrol activity of YP1 against RRD might be enhanced in the presence of FA due to enhanced LP production or synthesis. However, the molecular mechanism underlying this enhancement effect by FA on the biocontrol activity in YP1 needs to be explored further. The findings presented here highlight the promoting effects of FA on the growth, chemotaxis, and antagonism of YP1 and expression of its biocontrol-related genes, which indicate that chemotaxis to OAs in P. notoginseng-associated CAEB may directly contribute to their antagonism against RRD of P. notoginseng.
Taken together, this is the first study to examine and profile the genetic diversity of chemotaxis gene cheA in endophytic populations associated with P. notoginseng. Evidence was presented that there is a diverse assemblage of chemotactic bacteria in endophytic antagonists associated with P. notoginseng, and this lays the foundations for further understanding the community structure and functional characteristics of chemotactic-competent microbes in the P. notoginseng-associated endophytic antagonists. Our data provide the first evaluation of chemotactic behavior among natural populations of P. notoginseng-associated AEB. This finding could increase our understanding of how biocontrol organisms respond to OAs and provide new information to create a more comprehensive understanding of the ecological roles that OAs play in plant-microbe interactions. This work will identify the potential candidates with a competitive advantage via chemotaxis mechanisms for developing new strategies to better control RRD in P. notoginseng from an ecological point of view.