Soil microbe inoculation alters the bacterial communities and promotes root growth of Atractylodes lancea under heat stress

Atractylodes lancea is a medicinal plant widely used in treating rheumatic diseases, digestive disorders, night blindness, and influenza. Microbes greatly impact plant growth and metabolism. However, the microbiome associated with A. lancea remains unclear. Hence, we aimed at assessing the effect of soil microbe inoculation on A. lancea under heat stress from multiple perspectives, including regulation of growth, valuable secondary metabolites, root endophytic and rhizosphere bacterial communities. A. lancea was inoculated with soil microbes, then grown under normal/high temperature. Biomass, chlorophyll contents, production of major medicinal compounds, physiochemical properties of the soil, and in the composition of root bacterial communities of A. lancea were investigated. Soil microbe inoculation promoted root sink strength, accumulation of medicinal compounds, and attenuated damage caused by heat stress. A. lancea showed preference for the endophytic bacterial genera Rhodococcus, Ralstonia, Dongia Paenibacillus and Burkholderia-Caballeronia-Paraburkholderia post-inoculation, the latter four genera playing important roles in protection from heat stress, with abundance of the latter two specifically positively correlated to medicinal compound production. A. lancea enriched the bacterial genera Saccharimonadales, Novosphingobium and excluded Chitinophaga in its rhizosphere post-inoculation. Soil microbes characteristically promoted A. lancea growth, improved heat stress tolerance, and promoted root medicinal compound accumulation. A. lancea selectively enriched particular endophytic and rhizospheric bacterial communities post-inoculation, possibly due to unique aromatic root exudates. The selected bacteria potentially synergistically improved soil available nutrients and uptake by root. Bacterial species selected by A. lancea root have the potential to serve as biological fertilizers for A. lancea farming.


Introduction
Atractylodes lancea has a long history as a medicinal herb and is widely distributed in China, Japan and the Korean peninsula (Tsusaka et al. 2019). A native traditional Chinese medicinal (TCM) herb, its rhizome has been used since ancient times in the treatment of ailments as diverse as rheumatic diseases, digestive disorders, night blindness and influenza (Wang et al. 2008;Nie 2018;Qu et al. 2018). According to ancient TCM writings, A. lancea rhizomes of the best quality, termed 'Dao-di' (Geo-authentic), distributed in the mountainous areas of Jiangsu Province (approximately 30°45′32″ to 35°7′15″ N, 116°21′28″ to ″121°56′38″ W) of China owing to the favorable local climate conditions. However, in modern times, wild Geo-authentic A. lancea are endangered (Zhou et al. 2016). Hence, the plantation and production of highquality A. lancea rhizomes in specific geographical regions of China is of great economic value. However, as a perennial plant, A. lancea plantations are faced with multiple challenges including the impedance of growth by heat stress (Guo et al. 2005;Yan et al. 2010) and soil-borne disease outbreaks resulting from monocropping ). In the natural ecosystem, soil microorganisms play a pivotal role in the geochemical cycle as well as in the growth and development of plants (Schimel and Bennett 2004;Jacoby et al. 2017;Johns 2017). While certain soil microorganisms are pathogenic for plants, others can be neutral or beneficial (Berg et al. 2020). The bacteria that can benefit plant growth are termed plant growth-promoting bacteria (PGPB) (Sousa and Olivares 2016). They can colonize the root or the rhizosphere of the host plants, and in turn promote the plants' growth as well as their resistance to abiotic stress, including salt stress (Motaleb et al. 2020), drought stress (Devanathan et al. 2021), heat stress (Khan et al. 2020), and heavy metal toxicity (Ahemad 2019). Hence, the use of beneficial microbes can be a sustainable solution with the potential to alleviate the economic loss in A. lancea farming.
Beneficial bacteria induce and improve the host plants' resistance to stress through various biological mechanisms. One key mechanism is the bacteria's induction of excessive secondary metabolite production in the host plant. These secondary metabolites function in mitigating the damages caused by abiotic or biotic stresses (Pang et al. 2021). Medicinal plants contain a great variety of bioactive secondary metabolites in high concentrations, endowing them with great medicinal, pharmaceutical and economic value. In A. lancea, a broad range of volatile bioactive compounds with medicinal functions are enriched in its enlarged underground rhizome. Currently, the volatile sesquiterpenoids hinesol, β-eudesmol, atractylon and the polyacetylene-type atractylodin are acknowledged as the four major compounds that endow A. lancea rhizome products with pharmaceutical value (Guo et al. 2002). The biosynthesis of sesquiterpenoids can be induced and promoted by bacteria (Huang et al. 2003), which has also been reported in A. lancea (Zhou et al. 2016). Another key mechanism is that plants can interact with soil bacteria at the microbiota level. Plants can recruit specific microbiota to their root and rhizosphere through the screening of root exudates, and subsequently modulate the microbiota through the selection of the bacteria existing on the root surface. As a result of such recruitment and modulation, the host plants can achieve optimal beneficial effects from the soil microbes (Bulgarelli et al. 2013;Mendes et al. 2014;Reinhold-Hurek et al. 2015;Zhong et al. 2019).
The cultivation area of A. lancea is expanding every year. However, traditional cultivation methods are not conducive to sustainable development of the A. lancea industry. Geo-authentic A. lancea farming has been suffering great losses from increased soil-borne pathogenic microbe populations due to monocropping. Chemical pesticides are not an option because the rhizome products need to be clear of pesticide toxicity for the subsequent pharmaceutical use. Moreover, the use of chemical fertilizers can result in excessive vegetative growth but reduced concentration of secondary metabolites. Taken together, the 1 3 Vol.: (0123456789) exploitation of beneficial soil microbes appears an optimal, if not singular, option to facilitate a highyield, high-quality and sustainable A. lancea industry. However, there is currently a paucity of research on the interaction between A. lancea and soil microbes and the composition of the microbial community associated with the root of A. lancea remains unclear.
In the present study we attempted to shed light on the above-mentioned issues by subjecting A. lancea plantlets to heat stress and investigating the effects of soil microbe inoculation on the growth, heat resistance and medicinal compound production. Subsequently, we investigated the modulating effects of the inoculum soil microbes on the root endophytic and rhizosphere bacterial communities under room temperature or heat stress.

Plant material and soil microbe inoculum
Seeds of wild Geo-authentic A. lancea were collected at Jin-Niu-Dong-Shan (Mount 'Jin-Niu-Dong', 31°46′37″ N, 119°18′52″ W), Jintan City, Jiangsu Province. Surface-sterilized seeds were placed on Murashige & Skoog (MS) medium to germinate surface-sterile seedlings. The aerial part of the approximately 2-3 cm tall seedlings were cut and cultured on solid MS medium (pH = 5.8) containing 30 g/L sucrose, 0.1 mg/L naphthalene acetic acid (NAA) and 1 mg/L 6-benzyladenine (6-BA) (named the tillering medium) for vegetative propagation through tillering. Rooting was performed by culturing four-week-old vegetatively propagated A. lancea plantlets (approximately 4-cm tall) on the rooting medium, which was solid MS with 30 g/L sucrose and 0.5 mg/L NAA, for another four weeks. Subsequently, A. lancea plantlets with approximately 2-cm-long adventitious roots were removed from the rooting medium and planted in the sterile mixture (hereafter referred to as 'soil') of peat soil (Jiffy product, Netherlands) and vermiculite (6:1, v/v) under sterile conditions, then placed in a plant nursery room set at approximately room temperature (23 ± 2 °C, referred to as 'room temperature', RT hereafter) and with a 12 h (h) /12 h light/ dark cycle for nine days before subjected to further treatments.
The soil (hereafter referred to as the 'Geoauthentic soil') approximately 5-10 cm beneath the surface was collected as samples at five random sites in a forest-covered mountainous area (31° 36′ 18″ N, 119° 6′ 48″ E) located in a habitat of wild Geo-authentic A. lancea in Lishui District, Nanjing City, Jiangsu Province. The soil samples were then mixed thoroughly and used in the subsequent experiments. The water suspension of the mixed Geoauthentic soil sample, which contained an entire Geo-authentic microbial community, was used as the soil microbe inoculum. 10 g of the mixed Geoauthentic soil sample was placed in 100 mL of sterile water, then oscillated on a shaker at 220 rpm (rpm) for 10 min (min) to produce the soil microbe inoculum. The soil microbe inoculum that was autoclaved for 1 h at 121 °C was used as the mock inoculum.

Inoculation and heat stress treatment
Inoculation was performed when the A. lancea plantlets had grown on the sterile soil for nine days. For the inoculation, 5 mL of the inoculum was carefully added onto the soil close to the A. lancea plantlet. As high temperatures over 30 °C in the laboratory were likely to cause the death of A. lancea plantlets, the heat stress treatment was set at the critical temperature of 30 °C. A 30-day heat stress treatment was initiated right after the inoculation; while the control plantlets continued their growth under RT (23 ± 2 °C) for 30 days. For each treatment, 15 individual plantlets were used; in order to collect sufficient sample for the subsequent measurements and experiments, samples of multiple individuals were pooled as biological replicates; see the 'Sample collection' section for the pooling details. The inoculation experiments were additionally performed on a number of sterile soil samples without A. lancea plantlets growing in it as the control soil experimental groups (Fig. 1). Nine individual replicates were performed for each group of the soil inoculation experiments. Each three individual soil replicates were pooled as one biological replicate for further soil measurements, while three individual replicates were performed for the blank soil. Each was used as one biological replicate for further measurements.

Sample collection and measurement of biomass
At 40-day old, the A. lancea plantlets had developed multiple adventitious roots and barely recognizable rhizomes. The compartmentalization between rhizome and adventitious root was unclear. Hence, in this study we simply referred to the underground compartment of the 40-day-old A. lancea plantlets as the root, while the aerial compartment was referred to as the shoot (Fig. S1). Firstly, rhizosphere soil samples of all the 15 A. lancea plantlets were collected. The plantlets were very carefully removed from the soil to avoid breaking and loss of root. The soil remaining on the root surface was gently removed. The thin soil that appeared to be adhering on the root surface was then carefully removed and collected as our rhizosphere soil samples; we believe it was the soil within 1 mm from the root. The rhizosphere samples were stored in liquid nitrogen immediately after collection. Due to the scarcity of the rhizosphere soil samples, all the plantlets were used for sample collection; five random samples of each experimental group were pooled as one biological replicate, resulting in three biological replicates in total.
The plantlets were then all rinsed using sterile distilled water and six plantlets were randomly selected from each treatment group for the measurement of biomass. They were subsequently also used for the measurement of volatile compounds. Dry weight data was measured after freeze-drying for approximately 72 h to constant weight. We performed freezedrying instead of heat-drying to maintain the volatile compounds in the dry root samples for subsequent measurements. The root samples of two individual plantlets were pooled as one biological replicate, resulting in three biological replicates in total. The remaining nine plantlets of each group were collected and three individual plantlets were pooled as one biological replicate, resulting in three biological replicates for the root endophyte analyses and realtime quantitative reverse transcription polymerase chain reaction essays (qRT-PCR). The samples used for endophyte analyses were placed in clean 50 mL conical tubes and pre-rinsed three times with sterile distilled water. The washed roots were then treated with 70% ethanol for 10 min, followed by a treatment with 2.5% sodium hypochlorite and sonication for an additional 10 min. The samples were then drained and rinsed with sterile distilled water three times. To check for surface sterility, 100 μL of the final rinsed solution was plated in Potato Dextrose Agar (PDA) and Nutrient agar (NA) and this resulted in zero colonies.

Measurement of volatile compounds
Freeze-dried root samples were ground into fine powder. Approximately 0.1 g of the powder was carefully measured (with the exact weight recorded as 'mg'), placed in an Eppendorf tube, then 400 μL of analytically pure n-hexane was added to the powder and mixed thoroughly. Extraction was performed via ultrasound treatment at 60 Hz for 15 min. The mixture was then centrifuged at 5000×g and 4 °C for 5 min. The supernatant was filtered through 0.22-μm PES membrane filter capsules (Sterivex; Millipore) Fig. 1 Schematic diagram of the workflow of this study. A. lancea plantlets were grown in rooting medium for 4 weeks to generate 2-3 cm adventitious root, then transplanted into soil and cultured under sterile conditions; new roots were visible after nine days' growth in soil. Then, the plantlets were inocu-lated and immediately subjected to stress treatment for 30 days. RT, approximately 23 °C; HT, high-temperature (30 °C); +I, with Geo-authentic soil microbe inoculation; +p, with A. lancea plantlet; −p, without A. lancea plantlet 1 3 Vol.: (0123456789) and subjected to GC-MS analysis. The concentration of hinesol, β-eudesmol, atractylon, and atractylodin in freeze-dried root samples was measured via gas chromatography coupled with mass spectrometry (GC-MS) using a Trace 1310 series GC with a TSQ8000 MS detector (Thermo Fisher Scientific Co. Ltd., Waltham, Massachusetts, USA) and a TR-5 ms capillary column (30 m 3 0.25 mm i.d., DF = 0.25 mm, Thermo Fisher Scientific) according to Vannier et al. (2018) with slight modifications: the injected sample (1 μL) was separated at the Helium flow rate of 1 mL/ min; the temperature program was 2 min at 120 °C followed by a gradient from 120 °C to 240 °C at 5 °C/ min, and held at 240 °C for 5 min; the injector and detector temperatures were set at 240 °C and 350 °C, respectively. The MS operating conditions were as follows: the MS ionization mode indicated the electron impact ion source (EI) at 230 °C, with an acceleration voltage of 70 eV. The interface temperature was 240 °C and the total ion current was recorded for a mass range of 40-500 amu (Vannier et al. 2018;Yang et al. 2019;Yuan et al. 2019). The contents of four volatile oils in each sample were quantitatively determined by the standard curves (see Supplementary Table S1).

Determination of soil physicochemical properties
Soil samples were frozen in liquid nitrogen immediately upon collection for the subsequent pH and available nutrient measurements. Soil pH was measured using the glass electrode method (dry soil and water suspension v/v 1:2.5) (Li et al. 2006). The measurement of nitrate nitrogen, ammonium nitrogen, available phosphorus, and available potassium of soil samples were all measured using assay kits manufactured by Sinobestbio Technology Co., Ltd., (Shanghai, China) according to the manufacturers' instructions.

Endophytic and soil bacterial community analyses
The total bacterial DNA for 16S amplicon sequencing was extracted from 100 mg of A. lancea root sample or150 mg of soil sample using the PowerSoil DNA Isolation Kit (Mo Bio Laboratories, Solana Beach, CA, USA) per the manufacturer's instructions. The quality of the extracted DNA was verified via 1% agarose gel electrophoresis. The amplification and sequencing of the 16S rRNA targeting the variable V3-V4 regions (Xu et al. 2016;Perez-Jaramillo et al. 2019) was subsequently performed using the primers 338F (5′-ACT CCT ACG GGA GGC AGC AG-3′) and 806R (5′-GGA CTA CHVGGG TWT CTAAT-3′) and resulted in amplicons of approximately 460 bp. Error-correcting barcodes were added to both forward and reverse primers (Hamady et al. 2008). The amplification was carried out via PCR using a GeneAmp 9700 PCR system (Applied Biosystems, Foster City, CA, USA). The total reaction volume was 25 μL, including 1 μL DNA template, 0.5 μL forward primer, 0.5 μL reverse primer, 0.25 μL bovine serum albumin, 12.5 μL 2 × DreamTaq Green PCR Master Mix (Thermo Scientific, USA), replenished with ddH 2 O to 25 μL. Setting three technical replicates for each reaction, PCR was carried out as follows: 95 °C for 3 min, followed by 27 cycles of 95 °C for 30 s, 55 °C for 30 s, and 72 °C for 45 s, and a final extension at 72 °C for 10 min. Three technical repeats of one sample were mixed into a single PCR product. The products were separated via 2% agarose gel electrophoresis and purified using a Qiagen PCR purification ki5t (Qiagen, Hilden, Germany). Furthermore, the purified products were quantified with Pico Green using a QuantiFluorTM-ST Fluorometer (Promega Biotech, Beijing, China) and were then pooled at equal concentrations. Thereafter, the amplicons were sequenced in an Illumina MiSeq platform (San Diego, CA, USA) at Shanghai Majorbio Bio-pharm Technology Co., Ltd., Shanghai, China.
The data were analyzed on the online platform of Majorbio Cloud Platform (www. major bio. com). Paired-end (PE) reads (average length 410 bp) obtained by MiSeq sequencing were spliced according to their overlap relations using FLASH (Magoč and Salzberg 2011). Quality control was performed via filtering using Trimmomatic (Bolger et al. 2014). All sequences were clustered into operational taxonomic units (OTUs) with 97% similarity or greater using UPARSE (version 7.0) (Edgar 2013), and a majority consensus taxonomy was obtained for each OTU. Singletons were removed from the datasets to minimize the impact of sequencing artifacts (Dickie 2010). Chimeric sequences were identified and removed using UCHIME (Edgar et al. 2011). In order to obtain the species classification information corresponding to each OTU, the RDP classifier algorithm (https:// sourc eforge. net/ proje cts/ rdp-class ifier/) was applied to compare the OTU representative sequences with the Silva database (SSU138) for taxonomic analysis using the confidence threshold of 70%. Among these, chloroplasts and mitochondrial sequences were removed. The bacterial community diversity and richness were demonstrated using the sobs and invsimpson indexes using Mothur v.1.30.1 (Schloss et al. 2009). The relative abundance bar of bacteria at the phylum level was visualized using R language tools (v.3.3.1). Principal Coordinates Analysis (PCoA) analyses was performed using QIIME (version 1.9.1) based on unweighted UniFrac distance matrix or Bray-Curtis dissimilarity. The Student's t test within STAMP (Parks et al. 2014) was used to identify genus that showed significant differences in abundance between groups (confidence interval method). Spearman's rank correlations and q-values were calculated using the Psych packages (Revelle 2017). These correlations were visualized using the Pheatmap package in R (Kolde and Heatmaps 2019).

Statistical analysis
Data were recorded and processed by Excel (Office 2019). GraphPad Prism 8.0.1 (GraphPad Software Inc., USA) was used for rendering graphics. One-way ANOVA was performed using IBM SPSS Statistics 19.0 (SPSS, Chicago, IL, USA). Significance was calculated by Tukey's test (P < 0.05). Results were expressed as mean ± standard deviation (S.D.). Principle component analysis was performed using the 'Factor Analysis: Extraction' function of SPSS with default parameters and no rotation.

Soil microbe inoculation facilitated A. lancea growth under heat stress
The short A. lancea adventitious roots formed in the medium grew rapidly after being transplanted into soil. All the plantlets had developed roots visible from the bottom of the culturing vessel after nine days when we subjected them to soil microbe inoculation (+I) and heat stress (high temperature, HT) treatments. We also performed inoculation and HT treatments on sterile soil without A. lancea plantlets for comparison (Fig. 1).
The inoculation with Geo-authentic soil microbes (hereafter referred to as the GSM inoculation) significantly improved A. lancea root biomass under normal growth conditions (approximately room temperature, RT) ( Fig. 2A). The heat stress at the critical temperature of 30 °C significantly impaired A. lancea growth. The shoot fresh weight, root fresh weight and total fresh weight all decreased significantly under HT. By contrast, no significant change in the root or total fresh weight was observed under HT when the plantlets were previously inoculated with Fig. 2 Biomass, concentration of leaf chlorophylls, root medicinal compounds and relative expression of the key biosynthetic genes of the medicinal compounds. Data were shown as mean ± SD. A-D: n = 6. E, F: n = 4. G-M: n = 3. Different lower-case letters represent significant differences (one-way ANOVA, P < 0.05). G-J: 'FW', fresh weight; 'nd.', not detected. RT, approximately 23 °C; HT, high-temperature (30 °C); +I, with Geoauthentic soil microbe inoculation soil microbes, demonstrating significant HT damage mitigation facilitated by the soil microbe inoculation. Notably, when the GSM inoculation improved root biomass, it also decreased the shoot biomass of A. lancea under all the treatments in this study, resulting in a profoundly increased root/shoot ratio in the plantlets (Fig. 2B). Moreover, GSM greatly improved root length and root dry weight while decreasing root relative water content (Shivakrishna et al. 2018) under both RT and HT, demonstrating the specific beneficial effects of the GSM on dry mass accumulation in the A. lancea root ( Fig. 2A, C and D). Under HT, the GSM inoculation resulted in the lowest root water content and the highest root dry weight, root length and root/shoot ratio among all treatments, particularly highlighting the great beneficial potential of GSM on the dry A. lancea rhizome yield in plantation practices.
Microbe inoculation usually improves shoot health and the chlorophyll concentration of the leaves (Zhang et al. 2008). However, our shoot fresh weight data suggested impairment of shoot growth inflicted by GSM inoculation, which coincided with the improvement in root growth but indicated that GSM caused alterations in the allocation of photosynthetic products and absorbed soil nutrients. We hypothesized that more energy, organic and mineral matter might have been preferentially invested in the root instead of the shoot in the inoculated A. lancea plantlets. We therefore measured the leaf chlorophyll concentration, which is an important indicator of leaf health (Ling et al. 2011). The results demonstrated that HT and GSM caused the concentrations of both chlorophyll a and b, and thus total chlorophyll concentration, to decrease (Fig. 2E). Such decrease coincided with the impairment of shoot growth. However, the chlorophyll a/b ratio increased due to GSM (Fig. 2F), which indicated improved heat tolerance was due to GSM inoculation (see the discussion section for details).

Soil microbe inoculation improved the biosynthesis of A. lancea medicinal compounds
To evaluate the beneficial potentials of GSM on the medicinal value of A. lancea, we measured the yield of the four major volatile medicinal compounds that represented A lancea rhizome quality, namely hinesol, β-eudesmol, atractylon and atractylodin, in the root of the plantlets in this study (Fig. 2G-J). Due to the young age of the plantlets, hinesol and β-eudesmol were not detected under RT and/or HT. However, the GSM inoculation greatly improved the yield of all the four medicinal compounds, with even higher yields of hinesol, β-eudesmol and atractylon under HT than RT (although this was not statistically significant).
Subsequently, we selected three key rate-limiting genes of the terpenoid backbone biosynthesis pathway according to Vranová et al. (2013) (Fig. S2) and determined the relative expression of these genes in the A. lancea plantlets through qRT-PCR (Fig. 2K-M). The results demonstrated that GSM inoculation significantly induced the expression of FPPS and HMGR under both RT and HT. The expression of DXS was also induced by GSM under HT. Taken together, these results show that GSM induced the production of the major medicinal compounds in A. lancea root.

Soil microbe inoculation decreased soil pH and improved nutrient availability
To reveal the effects of GSM on the availability of soil nutrients, we measured the soil pH and the nitrate nitrogen (NO 3 − ), ammonium nitrogen (NH 4 + ), available phosphorus (P), and available potassium (K) of the inoculated soil after 30 days of co-culturing with (+p) or without (−p) the A. lancea plantlet under RT or HT (Fig. 1). The results showed that the presence of either the A. lancea plantlet or the GSM could decrease the soil pH and improve the concentration of available nitrogen (N) and P compared to the mockinoculated control placed under RT. Particularly, the dual presence of the A. lancea plantlet and soil microbes greatly improved NO 3 − content. Notably, under HT, the available N and P were both improved compared to RT. The presence of the plantlet and/or the microbes specifically improved the available N to a considerable extent. We performed a principal component analysis (PCA) on the soil pH and nutrient data and one component was extracted (PC1 in Fig. 3). The results demonstrated a significant negative correlation (P < 0.05) between the soil pH and available N and P. However, no evident correlation between the available K and any treatments was observed (Fig. 3, Supplementary Table 3 and 4).

Soil microbe inoculation and heat modified community composition and richness of endophytic bacteria
To reveal whether HT could alter the composition and richness of the bacterial communities in A. lancea root and rhizosphere, we performed 16S rRNA amplicon sequencing and comparative analyses of the root, rhizosphere soil and the inoculated soil samples without A. lancea plantlets after the 30 days' treatment under HT or RT. In total, 1.19 million high-quality sequence tags in total were generated for all the sequenced samples.
For the A. lancea root, 342 OTUs of the endophytic bacteria corresponding to 19 phyla and 226 genera were obtained and annotated. The invsimpson index (Fig. 4A) and sobs index (Fig. 4B) demonstrated relatively high bacterial variety and richness in the mock inoculated A. lancea plantlets grown under RT, suggesting that the plantlets were not internally sterile. The diversity and richness of the endophytic bacteria in the mock inoculated A. lancea roots both decreased considerably (P = 0.0537 and P = 0.1224, respectively) when subjected to HT (Fig. 4A and B). In contrast, after the inoculation with GSM, the root endophytic bacteria diversity was well maintained (Fig. 4A); however, their richness decreased while subjected to HT (P = 0.0540) (Fig. 4B).
The endophytic bacteria of A. lancea root mainly consisted of the phyla Proteobacteria, Firmicutes, Actinobacteria, and Bacteroidetes, which accounted for over 95% of the total (Fig. 4C). The composition of the endophytic bacterial community could be affected by both HT and GSM inoculation; to explore which factor(s) exerted the most influence on the modification of A. lancea root endophytic bacteria communities, we performed principal co-ordinates analyses (PCoA) based on the unweighted UniFrac distances between samples. The results showed that GSM was the primary factor affecting the endophytic bacteria community composition (PC1 = 19.04%), followed by HT (PC2 = 14.07%) (Fig. 4D).
A. lancea formed characteristic rhizosphere bacterial communities regardless of the temperature In total, 1522 OTUs annotated to 26 phyla and 496 genera were obtained from the rhizosphere soil and the inoculated soil samples without A. lancea plantlets under RT and HT. The invsimpson index Fig. 3 Soil pH, concentration of available nutrients and results of the principal component analysis (PCA). One component (PC1), which could explain 77.3% of the total variance, was extracted. Data of the loaded component matrix representing the correlation between PC1 and each variable (the soil pH and nutrients) was shown on the top. Measured soil pH and nutrient data were shown as mean (n = 3). Max/Min, the maximum/ minimum result in each set of data. Different lower-case letters represent significant differences between experimental groups (one-way ANOVA, P < 0.05). RT, approximately 23 °C; HT, high-temperature (30 °C); +I, with Geo-authentic soil microbe inoculation; +p, with A. lancea plantlet; −p, without A. lancea plantlet; 'DW', dry weight 1 3 Vol:. (1234567890) (Fig. 5A) demonstrated that the rhizosphere (+p) bacterial diversity under both RT (P < 0.05) and HT (P < 0.05) was lower significantly than that of the inoculated control group (−p). The rhizosphere bacterial diversity under HT was mildly higher than RT (P = 0.1425). This suggested that A. lancea shaped a unique rhizosphere environment, which could also be modulated by heat stress. The sobs index (Fig. 5B) showed that the bacterial richness in the (rhizosphere) soil increased under HT compared to RT, regardless of whether an A. lancea plantlet was present or absent.
Compared with the control soil (−p) samples, the relative abundance of the phyla Proteobacteria (P = 0.0588 and P = 0.0221) and Bacteroidetes (P = 0.0010 and P = 0.1382) in the rhizosphere Invsimpson index (A) and sobs index (B) demonstrated the alpha diversity of the control soil and rhizosphere soil bacteria (n = 3). C: Taxonomic distribution was relative abundance of the control soil and rhizosphere soil bacteria at the phylum level. D: PCoA results of rhizosphere soil bacteria based on the Bray-Curtis dissimilarity of at the genus level (n = 3), the degree of variation explained by the clade (R 2 ) and the significance (P-values) according to permutational multivari-ate analysis of variance (PERMANOVA) are provided, the significance analysis was based on 999 permutation tests. E, F: Differential abundance of the richest 15 bacterial genus in control soil and rhizosphere at room temperature or heat stress (n = 3), respectively. Asterisks represent significant difference by Student's t test: * P < 0.05; ** P < 0.01; *** P < 0.001. RT, approximately 23 °C; HT, high-temperature (30 °C); +I, with Geo-authentic soil microbe inoculation, −p, without A. lancea plantlet; +p, with A. lancea plantlet decreased under RT and HT, whereas that of Patescibacteria (P = 0.0013 and P = 0.0221) increased (Fig. 5C). At the genus level, the bacterial relative abundance of as many as nine genera under RT and three genera under HT were significantly different between the A. lancea rhizosphere and the inoculated control soil (−p) (Fig. 5E-F). Notably, the high temperature significantly suppressed the genera Sphingomonas (P = 0.0241) and Chitinophaga (P = 0.0031) in the control soil samples but resulted in no significant changes in A. lancea rhizosphere (supplementary Fig. S4A, S4B). Based on Bray-Curtis dissimilarity, the PCoA analysis of the bacterial composition showed marked differences between the A. lancea rhizosphere and the control soil bacteria communities (PC1 = 62.74%) (Fig. 5D). These results demonstrated that A. lancea formed characteristic rhizosphere bacterial communities.

The bacterial factors that regulated A. lancea medicinal compound accumulation and soil physiochemical properties
To explore how the bacterial communities affected the growth, medicinal compound production and soil physicochemical properties of the A. lancea plantlets we subsequently performed Spearman correlation analyses and revealed significant positive correlation between the endophytic bacterial genera Paraburkholderia, Paenibacillus, Bradyrhizobium and the four major volatile medicinal compounds in A. lancea root. This suggests that these three genera of endophytic bacteria specifically promoted the biosynthesis of medicinal compounds. By contrast, the relative abundance of the endophytic bacterial genera Escherichia-Shigella, Cutibacterium, Enterococcus, norank_f__Muribaculaceae, and Stenotrophomonas showed negative correlation with the medicinal compounds (Fig. 6A).
In the soil, the relative abundance of the genus Massilia exhibited significant positive correlation with the available P. Notably, multiple rhizosphere bacterial genera, such as Candidatus_Solibacter, Gemmatimonas, etc., showed positive correlation with the NO 3− -N, NH 4+ -N, and available P while being negatively correlated (although not statistically significant) with the soil pH (Fig. 6B). This coincided with our finding of an evident negative correlation between the soil pH and available N and P (Fig. 3).

Discussion
In this study we revealed characteristic promoting effects of soil microbe inoculation on the growth of the plant A. lancea and its production of medicinally valuable secondary metabolites. In inoculated plants underground biomass accumulation was favored whether or not under heat stress, as shown by the maximum root length, root dry weight ( Fig. 2A, C) and the significantly improved fresh root/shoot ratio post-inoculation (Fig. 2B) under both RT and HT. Plants are known to adjust the sink-source relation between the aerial and underground compartments when subjected to abiotic stresses (Prasad et al. 2008). For example, under mild drought stress, when both shoot and root growth is inhibited, root growth is often relatively favored. An increased root/shoot ratio can reflect both the response and tolerance of plants to drought stress (Prasad et al. 2008). Heat and drought stresses often accompany each other in fields. Consistently, an increased root/shoot ratio could also reflect the heat tolerance of plants: compared to a heat-sensitive rice cultivar, a heat-tolerant cultivar suppressed its shoot growth to a higher extent, while promoting its root growth after short-duration heat stress or suffering less inhibition of root growth after long-duration heat stress, all resulting in the higher root/shoot ratio of the heat-tolerant cultivar (Sailaja et al. 2014). Similar to drought stress, heat stress is associated with accelerated water loss, impaired water uptake and water-use efficiency of the plants (Prasad et al. 2008;Lipiec et al. 2013). The relative water content of root decreases under either heat or drought stress Meher et al. 2018), which was also the case for the A. lancea plantlets subjected to HT in our study (Fig. 2D). Plant roots have a relatively narrow and low range of optimum temperature compared to the growth of the other compartments, hence root growth is particularly sensitive to heat stress (Porter and Gawith 1999;Huang et al. 2012). However, the accumulation of root dry matter did not suffer any loss under HT post-inoculation. Taken together, these results led us to conclude that GSM inoculation modified the resource allocation during growth, particularly improving the sink strength of the root and thus promoting the heat tolerance of A. lancea.
The medicinal compounds in the A. lancea rhizome and adventitious roots are secondary metabolites whose biosynthesis and accumulation can be induced under environmental stress. Plants suffer the damage caused by excessive reactive oxygen species (ROS) produced under abiotic or biotic stress (Caverzan et al. 2016;Qi et al. 2018); sesquiterpenoids can serve as ROS scavengers as well as inter-plant signaling molecules owing to their antioxidant and volatile physiochemical characteristics (Bartikova et al. 2014). In particular, the volatile compounds such as sesquiterpenoids in A. lancea rhizome/root have antimicrobial activities and could thus be induced by endophytic microbes (Ren and Dai 2012). Our results suggested GSM inoculation was of fundamental importance in the accumulation of two of the four major medicinal compounds in the root of A. lancea plantlets, namely hinesol and β-eudesmol ( Fig. 2G-J). Notably, β-eudesmol, a sesquiterpenoid that is present in many volatile-oil-bearing plants, increased to a detectable level under HT in uninoculated A. lancea plantlets (Fig. 2H). It was reported that the accumulation of β-eudesmol was highly induced in Parthenium argentatum flowers subjected to drought stress (Nik et al. 2008), indicating a specific function of β-eudesmol in plants' tolerance against abiotic stresses associated with disturbed water use, including heat stress. Hence, our study suggested that GSM inoculation improved the heat tolerance of A. lancea root by promoting the accumulation of volatile compounds, which, combined with the improved root biomass, would improve the medicinal and economic benefits in A. lancea farming.
The GSM inoculation also altered the chlorophyll contents of A. lancea shoot as a strategy of adaptation to heat stress. Lower chlorophyll content can prevent photoinhibition and might result in lower leaf temperatures, which can in turn mitigate heat stress, reduce leaf respiration and water loss across the cuticle (Tambussi et al. 2007). Similarity, a heat-tolerant rice cultivar had the lower total chlorophyll content after both short-or long-duration heat treatments than the heat-sensitive cultivar (Sailaja et al. 2014). Hence, the decreased total chlorophyll content of A. lancea shoot (Fig. 2E) post-inoculation indicated adaptation to the heat stress. Chlorophyll a/b ratio is also an important trait that can reflect leaf/plant health or stress tolerance. For instance, inoculation with the beneficial arbuscular mycorrhizal fungus Glomus mosseae attenuated the damage of NaCl stress on beach plum (Zai et al. 2012); induction with ethylene mitigated the damage of heat stress on rice (Wu and Yang 2019). In both cases, significantly higher chlorophyll a/b ratio was detected in the more stress-tolerant plants, as was the case of A. lancea subjected to HT post-inoculation (Fig. 2F). In sum, GSM inoculation regulated multiple aspects of A. lancea growth and metabolism, improving its heat stress resilience.
Our microbe analyses revealed that HT constitutively increased bacterial diversity and richness in the soil (including the rhizosphere soil), but suppressed the endophytic bacterial diversity and richness in A. lancea root to varied extents (Figs. 4A, B and 5A,  B). Meanwhile, the bacterial diversity in the rhizosphere of A. lancea was constitutively lower than in the control soil samples (Fig. 5A). These results suggested characteristic selectivity of A. lancea root in its endophytic and rhizospheric bacterial communities, and suppression of the endophytic bacterial richness, especially under HT. GSM inoculation could further enhance the selectivity of A. lancea root in the modulation of diversity and suppression of the richness of endophytic bacteria, especially under HT (Fig. 4A, B). Comparing the number of endophytic and rhizospheric bacterial genera with significantly different relative abundance among different experimental groups and the P values, the GSM inoculation resulted in more evident alteration in the composition of bacterial communities than did the temperature. These results suggested that the effect of HT on endophytes of A. lancea was not as strong as GSM, and the effect of HT on rhizosphere bacteria was not as strong as the selectivity of A. lancea root, which is consistent with the results of PCoA (Figs. 4D and 5D). It is well known that plants are selective hosts of their endophytic and rhizospheric microbes (Abedinzadeh et al. 2019;Afzal et al. 2019); the uptake of toxic substances of plants grown in polluted soil can inflict further selective pressure on the root endophytic bacterial communities (Phillips et al. 2008;Qiong et al. 2021). Hence, the metabolic changes in A. lancea root caused by HT might have contributed to the selectivity and repression on the bacterial community composition and abundancy.
Our results suggested complex, interdependent regulation of A lancea root growth, metabolism, soil physiochemical properties, and the bacterial communities. The soil pH constitutively decreased under HT although the soil remained neutral (Fig. 3). Soil pH has a strong influence on soil microbial communities as well as P availability; and soil pH can be greatly influenced by temperature (Siciliano et al. 2014;Yao et al. 2017). Soil fertility can be associated with bacterial richness (Yao et al. 2017). Particularly, our results revealed a positive role for the bacterial genus Massilia in increasing soil available P (Fig. 6B), consistent with its well-documented function of solubilizing phosphate (Zheng et al. 2017;Wan et al. 2020). Notably, while Massilia spp. appeared more enriched in the control soil samples than in the A. lancea rhizosphere (Fig. 5E, F; P = 0.1690 and 0.2269, respectively), it was enriched inside A. lancea root as endophytes under the high selective pressure under HT (Fig. 4F, P = 0.0520). Meanwhile, the GSM appeared to be an important source of Massilia since its relative abundance increased specifically post-inoculation (Fig. 4F).
The endophytic bacteria favored by A. lancea root mainly derived from the soil, as their relative abundance increased post-inoculation (Fig. 4E, F). Our study indicated preference of A. lancea root for the endophytic bacterial genera Rhodococcus (P = 0.0320) and Paenibacillus (P = 0.0640) postinoculation (Fig. 4E), and crucial roles of the genera Burkholderia-Caballeronia-Paraburkholderia, Ralstonia, Paenibacillus, and in particular Dongia, in protecting A. lancea root under HT post-inoculation (Fig. 4F, Supplementary Fig. S3B). There is currently a paucity of data on the interaction of Dongia spp. with A. lancea or other medicinal plants. Our study suggests that this is worthy of further exploration. Paenibacillus spp. are well acknowledged plant-growth-promoting bacteria with antagonistic activity against phytopathogens (Grady et al. 2016;Rybakova et al. 2016). However, the genera Burkholderia-Caballeronia-Paraburkholderia and Ralstonia include both plant-beneficial-environmental bacteria (Chen et al. 2003;Kaur et al. 2017) and vicious phytopathogens (Elphinstone 2005;Lebeau et al. 2011). Remarkably, the genera Burkholderia-Caballeronia-Paraburkholderia, Paenibacillus, and a genus of nitrogen-fixing symbiotic rhizobacteria usually found in legumes (Grady et al. 2016;Kaur et al. 2017;Padukkage et al. 2020), Bradyrhizobium, all had significant positive effects on promoting the production of volatile medicinal compounds (Fig. 6A). However, the increase of endophytic Bradyrhizobium in A. lancea root post-inoculation was not statistically significant (Fig. 4E, F; P = 0.1670 and 0.1450,

3
Vol.: (0123456789) respectively). Ralstonia was reported to exhibit no pathogenic effects on A. lancea  and also showed promoting effects on medicinal compound production (Fig. 6A). The functions of these bacterial genera on A. lancea are worthy of further study.
Plants secrete up to 40% of photosynthetic products into the rhizosphere, resulting in a dense microbe population in the surrounding soil. This is known as the rhizosphere effect (Bais et al. 2006). Our study revealed a particular preference of A. lancea for the genera Saccharimonadales, Novosphingobium and exclusion of Chitinophaga in its rhizosphere postinoculation, especially under HT (Fig. 5E-F). Species of Saccharimonadales are enriched within the range of plant root exudation and contributes greatly to soil phosphorus cycling by enhancing alkaline phosphatase activity in the rhizosphere (Wang et al. 2022). Saccharimonadales was also reported to show synergistic effects on the nitrogen cycling in acidsoil (Shi et al. 2021). Combined with their reported functions in soil fertility and nutrient cycling (Zheng et al. 2017;Wan et al. 2020;Shi et al. 2021;Wang et al. 2022), our findings of the enrichment of endophytic Massilia (Fig. 4F) and rhizospheric Saccharimonadales (Fig. 5E, F) on A. lancea post-inoculation suggest synergistic activities of these microbes promoting plant growth via improving soil fertility and nutrient uptake, a worthy subject for future research. Notably, bacteria of the genus Novosphingobium are prevalent bioagents for the degradation of substrates in lakes, soil, sea, wood and sediments , with outstanding abilities to colonize rhizosphere environments and degrade aromatic compounds (Segura et al. 2021). The volatile exudates of A. lancea rhizome are rich in terpenes and consisted of a high concentration (~20%) of the aromatic hydrocarbon 1,2,4,5-Tetra-methylbenzene . Such characteristic rhizosphere environment might have been specifically selective for Novosphingobium spp. Curiously, the bacteria belonging to the genus Chitinophaga are also well-known as beneficial biocontrol agents that protect plants from fungal pathogens and nematodes (Yin et al. 2013;Sharma et al. 2020). However, its relative abundancy was reduced in A. lancea rhizosphere post-inoculation (Fig. 5E, F). In summary, we revealed the bacterial communities favored by A. lancea root from a mixture of natural soil microbes under heat stress.

Conclusion
In the present study, we revealed that inoculation with soil microbes could alter resource allocation during A. lancea growth and in this way either improve the potential tolerance to or attenuate the damage caused by heat stress. The root of A. lancea was highly selective for its root endophytic bacteria, and favored the genera Rhodococcus, Ralstonia, Dongia Paenibacillus and Burkholderia-Caballeronia-Paraburkholderia. The latter four played important roles in protecting A. lancea from heat stress, especially the genus Dongia. The latter two and the genera Bradyrhizobium and Massilia likely contributed to the production of major medicinal compounds in A. lancea root. A. lancea enriched the bacterial genera Saccharimonadales and Novosphingobium but excluded the beneficial genus Chitinophaga in its rhizosphere, possibly due to characteristic root exudates rich in aromatic compounds. The endophytic and rhizospheric bacteria of A. lancea root improved soil available nutrients, and possibly collaborated in soil phosphate solubilization and uptake. Bacterial species of the genera enriched by A. lancea root can thus serve as potential biological fertilizers for the improvement of both A. lancea yield and quality.