Plant secondary metabolites (SMs), produced from secondary metabolic pathways in plants, comprise numerous structurally diverse compounds that can be classified into several large families: phenolics, terpenes, steroids, alkaloids, and flavonoids (Pang et al., 2021). Plant SMs play a crucial role in plant defense against pathogens and pests and adaptation to environmental stresses, also acting as symbiotic signals for plants in association with microbes (Pang et al., 2021). In addition, some plant SMs have been used as ingredients or additives of pharmaceuticals and nutraceuticals, making enormous contributions to public health (Li et al., 2020). However, as a material basis of clinical efficacy and vital indicators for quality evaluation of the corresponding medicinal materials, the SMs of medicinal plants can vary among individuals (Li et al., 2020). Genetic, ontogenetic, and environmental factors are recognized as the sources of variation for the intraspecific polymorphisms of plant SMs (Moore et al., 2014).

Terrestrial plants host taxonomically diverse microorganisms, which exert beneficial, harmful, or neutral effects on plant growth and resistance to biotic and abiotic stresses (Müller et al., 2016). Using next-generation sequencing and bioinformatic analysis, the phylogenetic composition and functional diversities of microbial communities associated with various host plants, such as Arabidopsis, several corps, and tree species, have been extensively investigated, revealing associations between plant microbiota and plant nutrient uptake, production, and stress tolerance (Bulgarelli et al., 2012, 2015; Lundberg et al., 2012; Kembel et al., 2014; Edwards et al., 2015; Jin et al., 2017; Zhang et al., 2018a, 2019; Fan et al., 2020; Lee et al., 2021). The community structure and functional capabilities of microbiota associated with medicinal plants have received some research attention recently (Chen et al., 2018; Kui et al., 2021; Liu et al., 2021; Na et al., 2021). It has been proposed that the intraspecific variation in SMs of medicinal plants could be partly ascribed to the shift in the composition of the microbial community associated with the plant host (Huang et al., 2018). However, the associations between plant microbiota and accumulation of SMs in medicinal plants remain largely unknown.

Sophora flavescens, a slow-growing Fabaceae shrub, has been planted as an important medicinal plant in China due to its widespread use in pharmacological drugs (Song et al., 2019). The wide-reaching biological activities of SMs of S. flavescens mainly include antitumor, antiviral, anti-inflammatory, antianaphylactic, antiarrhythmic, and hepatoprotective effects (He et al., 2015). The root of S. flavescens contains extensive bioactive constituents, of which alkaloids are characteristic compounds of the plant and have provoked great interest for their pharmacological effects on cancers, viral hepatitis, and cardiac diseases (He et al., 2015). With the increasing demand for products of medicinal plants and limited arable land, sustainable efforts to improve the production of medicinal plants are required. However, this cannot be achieved by the input of chemical fertilizers and pesticides, which can degrade the quality of medicinal plants (Song et al., 2019). Under these circumstances, the management of favorable microbes is expected to be a potential strategy for facilitating the biosynthesis and accumulation of bioactive components in medicinal plants. Thus, comprehensive studies regarding the association between plant microbiota and bioactive components in medicinal plants are required for the identification of potentially beneficial microbes (Jin et al., 2017).

The plant rhizosphere is a specific microhabitat around the root, where complex soil-microbe-plant interactions occur (Zhang et al., 2017; de Vries et al., 2020). The rhizosphere microbiota, whose genome is referred to as the second genome of plants, contributes to plant phenotype and is vitally important for plant growth and health (Berendsen et al., 2012; Zhang et al., 2017). In addition, the rhizosphere is a hotspot for investigating the interactions between plant-associated microbes and plant SMs (Pang et al., 2021). Therefore, we proposed to address the association between the rhizosphere microbiota and the accumulation of alkaloids in S. flavescens.

In the present research, we sequenced the rhizosphere bacterial communities of 72 S. flavescens individuals grown in four well-separated areas in China. The key drivers shaping the rhizosphere bacterial communities were evaluated. Moreover, the association of the rhizosphere bacterial microbiota with the accumulation of alkaloids in roots of S. flavescens was explored. Potentially beneficial microbes correlated with the total accumulation of matrine and oxymatrine were identified. Finally, the direct and indirect effects of spatial, climatic, and edaphic factors and the preponderant rhizosphere phyla on the accumulation of alkaloids were estimated. We expect that our work will provide a basis for the quality improvement and sustainable utilization of this medicinal plant by rhizosphere microbiota manipulation.

Alkaloids, one of the major classes of plant SMs, are critical bioactive constituents in the medicinal plant S. flavescens (He et al., 2015). Although plant SMs can benefit plant fitness and human health, the accumulation of SMs can vary among individuals (Moore et al., 2014; Pang et al., 2021). In the present study, we also discovered variations in the accumulation of alkaloids among root samples of S. flavescens, with geographic location and plant age being important sources of variation (Figure 4). It is recognized that the intraspecific variation in plant SMs is not only dependent on plant genotype and ontogeny but also influenced by various environmental factors involving biotic and abiotic factors (Moore et al., 2014; Yang et al., 2018). The importance of plant-associated microbes, as vital biotic factors, for explaining plant SM polymorphisms in medicinal plants has been highlighted (Huang et al., 2018). In addition, plant rhizosphere is a hotspot for studying the impacts of plant-associated microbes on the accumulation of SMs in plants. Therefore, we determined the taxonomic structure of rhizosphere bacterial communities of S. flavescens and addressed the association between the rhizosphere bacterial microbiota and the accumulation of alkaloids in S. flavescens.

Plants host a taxonomically diverse microbiota but only a few dominant taxa (Müller et al., 2016). The dominance of the bacterial phyla Proteobacteria, Actinobacteria, Bacteroidetes, Acidobacteria, and Firmicutes was discovered in previous studies (Bulgarelli et al., 2012, 2015; Lundberg et al., 2012; Zarraonaindia et al., 2015; Fan et al., 2017; Zhang et al., 2019). In the present research, we found that the rhizosphere bacterial communities of the legume S. flavescens were dominated by Proteobacteria (mostly from the Rhizobiales order), followed by Actinobacteria, Acidobacteria, Firmicutes, Gemmatimonadetes, and Bacteroidetes (Figure 2), revealing similarly dominant phyla with previous studies.

Then, we explored the dissimilarities in the rhizosphere bacterial communities of S. flavescens. The results revealed distinct community compositions across the geographic locations and plant ages and indicated that the geographic location accounted for most of the variation (Figure 3A). Previous reports on maize and soybean plants also indicated that the largest proportion of variance in rhizosphere microbiota could be attributed to the geographic location (Peiffer et al., 2013; Zhang et al., 2021). The large spatial distance of sampling employed in the present study may result in the dominant contribution of geographic location in structuring the rhizosphere bacterial communities. The results of VPA showed that the spatial, climatic, and edaphic factors contributed to the variation in the rhizosphere bacterial community composition (Figure 3C), suggesting that dispersal limitation and environmental filtering could account for the turnover of the rhizosphere microbial communities of S. flavescens across the geographic locations (Zhang et al., 2018a, 2018b). We also conducted RDA ordination to identify the key spatial, climatic, and edaphic drivers responsible for the variation in the rhizosphere bacterial communities of S. flavescens. The results showed that two spatial variables (PCNM 1 and 3), two climatic variables (MAT and MAP), and three edaphic variables (pH, AN, and AP) were significant in driving bacterial community turnover (Figure 3B). Besides the geographic location, plant age could influence the assembly of rhizosphere microbiota (Micallef et al., 2009; Marques et al., 2014; Huang et al., 2020; Wang et al., 2020). Plants can select different microbes at the different stages of development, presumably for specific benefits (Chaparro et al., 2014). Overall, in the present study, the geographic location was a larger source of variation for shaping the rhizosphere bacterial communities than the plant age (Figure 3A). Similar results were obtained with grapevine plants (Manici et al., 2017).

We evaluated the correlation between the rhizosphere bacterial microbiota of S. flavescens and the accumulation of alkaloids in roots of the plant by using the Mantel test. The results elucidated significant correlations between the rhizosphere bacterial communities and the contents of oxymatrine, sophoridine, and matrine + oxymatrine (Supplementary Table S5). Accordingly, we used the random forest model to regress the relative abundance of the rhizosphere bacterial OTUs against the total content of matrine and oxymatrine to identify the potential beneficial bacteria for enhancing the total accumulation of these two molecules in roots of S. flavescens, which is used as a standard for the quality evaluation of this medicinal plant. We found that 59.2% of the variation in the matrine + oxymatrine content could be explained by the random forest model (Figure 5), which reveals the important influence of rhizosphere bacterial microbiota on the accumulation of these two molecules in the plant. Within the top 30 important bacteria, positive markers were mostly from Actinobacteria and Chloroflexi, while negative markers predominantly belonged to Gemmatimonadetes and Acidobacteria.

Actinobacteria members, ubiquitous in terrestrial habitats, are known to produce extensive SMs, many of which are of great importance for medicine and agriculture (Barka et al., 2016; Choudoir et al., 2019). Host plants establishing symbiotic associations with some members of Actinobacteria were reported to have improved resistance to biotic stresses through activation of defense pathways or upregulation of certain SMs (Conn et al., 2008; Kurth et al., 2014). In addition, Yan et al. (2014) found that Streptomyces pactum Act12, a member of Actinobacteria, could improve the growth of hairy roots of Salvia miltiorrhiza and promote the synthesis of tanshinone in hairy roots by upregulating the expression of corresponding biosynthetic genes. As shown in Figure 6, our SEM also indicated a positive and direct effect of Actinobacteria on the total accumulation of matrine and oxymatrine. Acidobacteria is also widespread and abundant in nearly all ecosystems (Kalam et al., 2020). However, owing to difficulties in cultivation, the phylum is rudimentarily explored (Kielak et al., 2016; Kalam et al., 2020). Most of the capabilities and ecological functions of Acidobacteria members, such as decomposition of various biopolymers and plant growth promotion, are predicted from their genomic and metagenomic information (Kalam et al., 2020). Although our knowledge in terms of their influence on plant SM accumulation was limited, the covariance analysis of our SEM revealed a negative correlation between the relative abundance of this phylum and the phylum Actinobacteria (P = 0.002; R = −0.525). This negative correlation could also be indicated by the Spearman’s rank correlation (P = 0.000; R = −0.778) and may allude to a competitive relationship between these two phyla, which may account for the negative association between the phylum Acidobacteria and the total accumulation of matrine and oxymatrine in the outcome of the random forest model (Figure 5). The phylum Chloroflexi has been found in anaerobic habitats, such as sediments and hot springs, exerting a role in biogeochemical carbon and nitrogen cycling (Hug et al., 2013; Speirs et al., 2019; Spieck et al., 2020). A previous study by Hu et al. (2018) demonstrated that Chloroflexi bacteria were attracted by root exudates of maize and were involved in host-medicated rhizosphere microbiota assembly to promote the growth and defense capability of the host. The phylum Gemmatimonadetes, containing phototrophic species, is also a nearly unexplored bacterial group (Zeng et al., 2020). The impacts of these microbes on the accumulation of bioactive components in S. flavescens should be illustrated in further studies based on isolation of the identified bacteria and cocultivation of them with the host plant.

Variations in abiotic factors, such as temperature, light, nutrients, and water, could also cause changes in the type and amount of plant SMs, which may be due to the regulation of the biosynthesis of SMs or a consequence of passive dilution or concentration due to the altered plant growth rate (Moore et al., 2014; Isah, 2019). Therefore, we applied SEM to evaluate the direct and indirect effects of the spatial, climatic, and edaphic factors and the relative abundance of the three dominant rhizosphere phyla on the accumulation of alkaloids in the roots of S. flavescens. The results showed that the spatial, climatic, and edaphic factors had both direct and indirect effects on the accumulation of alkaloids (Figure 6). The direct and indirect (through alteration of the relative abundance of Actinobacteria) effects of edaphic factors on the accumulation of alkaloids implies the impact of comprehensive soil-microbe-root interactions in the rhizosphere on bioactive component accumulation in medicinal plants.

In conclusion, the present study indicated that the accumulation of alkaloids in S. flavescens could be influenced by the rhizosphere bacterial communities of the plant. We identified the potentially beneficial bacteria associated with the total accumulation of matrine and oxymatrine with a random forest machine learning algorithm. In addition, the causal relationship among the spatial, climatic, and edaphic factors, the three most abundant rhizosphere phyla (Proteobacteria, Actinobacteria, and Acidobacteria) and the accumulation of alkaloids was indicated by SEM. We proposed rhizosphere engineering of specific microbial members for the control and improvement of the growth and quality of medicinal plants in the future.

The original contributions presented in the study are publicly available. This data can be found here: PRJNA780464 accession is not yet publicly available.

JChe and GY conceived the study. JChe and NL conducted the experiments, analyzed the data, and drafted the manuscript. JCha and KR participated in the experiments. JChe, GY, and JZ reviewed and edited the manuscript. All authors contributed to the article and approved the submitted version.

This study was supported by the Plan of Shanxi Province Science and Technology Research (201901D211341), the Shanxi Higher Education Innovation Project (2019L0417), the Central Government Guides Local Scientific and Technological Development Fund Projects (YDZX 20201400001443), the Shanxi International Science and Technology Cooperation Project (201803D421065), the Science Research Start-up Fund for Doctor of Shanxi Province (SD1815), the Taiyuan City Science and Technology Project Special Talents Star Project (120247-08), and the Science Research Start-up Fund for Doctor of Shanxi Medical University (XD1816).

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.