Pathogen invasion enhances the abundance of predatory protists and their prey associations in the plant microbiome
Min Gao1,2,3, Chao Xiong3, Clement K. M. Tsui4,5,6, Lei Cai1,2*
1 State Key Laboratory of Mycology, Institute of Microbiology, Chinese Academy of Sciences, Beijing 100101, China.
2College of Life Sciences, University of Chinese Academy of Sciences, Beijing 100049, China.
3 Hawkesbury Institute for the Environment, Western Sydney University, Penrith, NSW 2753, Australia.
4 Division of Infectious Diseases, Faculty of Medicine, University of British Columbia, Vancouver, BC, V6T 1Z3, Canada.
5 National Center for Infectious Diseases, Tan Tock Seng Hospital, Singapore.
6 LKC School of Medicine, Nanyang Technological University, Singapore.
* Author for correspondence: Lei Cai
Corresponding author e-mail: cail@im.ac.cn.

Abstract

Untangling the responses of protistan communities associated with soil and plant compartments and their associations with bacterial and fungal communities to pathogen invasion are critical for understanding the ecological processes governing plant microbiome assembly. Here we examined the protistan communities across the soil–plant continuum of healthy chili peppers and those with Fusarium wilt disease (FWD) and integrated the bacterial and fungal microbiome data from our previous investigation in China. We found that FWD was associated with a significant enrichment of phagotrophic protists in roots and an increase in the proportion and connectivity of these phagotrophic protists in intra- and interkingdom networks. Specifically, FWD increased the negative correlations between phagotrophic protists (especially Cercozoa and Ciliophora) and several members of Actinobacteria, Alphaproteobacteria, and Gammaproteobacteria in the interkingdom networks. Furthermore, the microbiomes of diseased plants not only exhibit a higher relative abundance of functional genes related to bacterial anti-predator responses compared to healthy plants, but also contained a greater abundance of metagenome-assembled genomes possessing functional traits involved in this response. The increased microbial interkingdom correlations among bacteria, fungi, and protists, coupled with the enhanced effects of protists on bacteria and fungi, as well as the notable bacterial anti-predator feedback in the diseased plant microbiome, all suggest that FWD catalyzes the associations between different groups of microbiomes. These findings highlight the role of predatory protists in shaping microbial assembly and functionality through top-down forces during pathogenic stress, potentially contributing to co-evolution within these soil and plant microbiomes.
KEYWORDS: Fusarium wilt disease, protists, plant microbiome, prey defense, metagenomic

1 INTRODUCTION

Soil and plant associated microbiomes are essential in plant growth and health, including nutrient acquisition, abiotic stress tolerance, and disease suppression (D’Hondt et al., 2021; Trivedi et al., 2020; van der Heijden et al., 2016). The assembly of the bacterial and fungal microbiome in soils and plants is determined by certain crucial bottom-up controls, mainly driven by environmental conditions and resource availability, such as the concentration and diversity of root-secreted exudates or volatile organic compounds (Gao et al., 2019; Liu et al., 2019; Zhalnina et al., 2018). In addition to bacteria and fungi, protists are another widely recognized soil and plant microbial components (Bates et al., 2013; Gao et al., 2019; Xiong et al., 2018), and their communities are important in microbiome assembly, as microorganisms dwell with myriad other microbes and are co-embedded in complex food webs (Petters et al., 2021). However, studies on the protistan communities associated with plants and soils are limited and far less than bacterial and fungal communities, despite the fact that protists are main top-down force in shaping bacterial and fungal communities. The top-down control of predator-prey interactions, largely dominate by protists, is considered to play a critical role in boosting microbial turnover in the soil and improving plant performance (Asiloglu et al., 2021; Gao et al., 2019; Geisen et al., 2018; Xiong et al., 2018). Deciphering the ecological processes underlying the protistan community assembly and uncovering the dynamics of protistan–bacterial–fungal interactions is critical to understand the ecological mechanism governing plant microbiome assembly.
Unicellular eukaryotic protists, including plant-, fungus- and animal-like microorganisms, are highly diverse and ubiquitous members of the biosphere (Milner et al., 2021; Sibbald et al., 2017; Whittaker et al., 1978). Based on their life cycles and the mode of nutrient-uptake, protists display parasitic, phagotrophic, phototrophic, plant pathogenic, and saprotrophic behaviors and are thus grouped accordingly (Geisen et al., 2018). Phagotrophic protists, also known as predatory protists, are major predators of ambient microbes, especially bacteria and arguably fungi. Protists predation could increase both species turnover and nutrient release, which fuels microbial activity and shifts in microbiome composition (Bahroun et al., 2021; Friman et al., 2013; Gao et al., 2019; Rosenberg et al., 2009; Weidner et al., 2016). Bacteria respond to predation through the stimulation of prey defense traits that are morphological or structural modifications (Queck et al., 2006; Wildschutte et al., 2004); some defense are chemicals or secretions of defensive secondary metabolites (Jousset, 2012; Alexandre Jousset et al., 2010; Jousset et al., 2009; Mazzola et al., 2009; Weidner et al., 2016). These prey defense traits in response to predation could be pertinent for plant health; for instance, several secondary metabolites conferring resistance to predation are known to be involved in plant pathogen suppression or plant immunity priming (Jousset et al., 2006; A. Jousset et al., 2010). Given the influence of protists on the structure and functioning of microbiomes, and the defense traits from their prey items, it has been postulated that the protist–prey interaction can steer beneficial plant–microbe interactions, leading to better plant growth and health (Bonkowski, 2004; Guo et al., 2021; Krome et al., 2009; Xiong et al., 2020).
Protists are key bioindicators in agroecosystems given their ecological importance (Foissner, 1997), and they can be served as the main predictor/determinants of the performance and health of some crops (Guo et al., 2021; Xiong et al., 2020). Several investigations have been performed to explore the influence of abiotic factors, including aridity, nitrogen fertilization, and seasonal factors, on the assembly of protistan communities (Caracciolo et al., 2022; Chen et al., 2022; Nolte et al., 2010; Oliverio et al., 2020; Ren et al., 2023; Zhao et al., 2019). However, existing literature rarely focused on how pathogen invasion affects the assembly of soil and plant associated protistan communities and the complex interactions among protists, bacteria, and fungi.
In this study, firstly, we investigated how pathogen invasion affected the assembly of soil and plant-associated protistan communities by performing high-throughput amplicon sequencing for multiple compartments (e.g., soils, roots, stems, and fruits) of healthy and Fusariumwilt diseased (FWD) chili peppers from two locations in China. Secondly, in order to understand the interaction among different microbial communities, we integrated the bacterial and fungal microbiome data from our previous investigation on the same pepper–FWD system (Gao et al., 2021) to decipher the potential role of protists in shaping bacterial and fungal communities under FWD. We hypothesized that FWD would influence the protistan community structures especially for those groups that have predatory functions and alter the protists-prey associations which would drive a change in the plant microbiome assembly and functions. Previous studies showed that pathogen invasion influences the soil and plant associated bacterial and fungal communities (Carrión et al., 2019; Gao et al., 2021; Liu et al., 2020), and the functions of the microbiome; however, the role of protists compared with other microbial groups in driving the plant microbiome assembly under pathogen invasion remains poorly understood. We expect our integration of protists, bacteria, and fungi, efforts to provide a new insight of the plant microbiome assembly and function from protists perspective under a fungus pathogen invasion.

2 MATERIALS AND METHODS

2.1 Sampling

Mature pepper plants were collected from the major chili pepper production fields of Huishui (25°48’41”N, 106°31’24”E) and Guiyang (26° 29′31″N, 106° 39′16″E), in Guizhou province, China in August 2018. The sampling and handling of soils, roots, stems and fruits samples, as well as the distinguishing symptoms of diseased versus healthy peppers, have been described in details previously (Gao et al., 2021). In brief, healthy and diseased pepper plants and their corresponding bulk and rhizosphere soils were collected, and the root and fruit samples were divided into the episphere and endosphere by the methods of filtering wash liquid and surface sterilization, respectively. Also, the stem samples were fractionated into upper stem, middle stem, and bottom stem section, with each section further divided into the xylem and epidermis, accordingly. Altogether, each plant sample was divided into 12 compartments (144 samples = 2 sites × 12 compartments × 2 (healthy and diseased) × 3 replicates). Total DNA of each compartment was extracted with the FastDNA SPIN Kit for soil (MP Biomedicals, Solon, USA) following the manufacturer’s instructions.

2.2 Sequencing for protistan communities and bioinformatics analysis

The V9 region of the eukaryotic 18S rRNA gene was amplified using primers: 1389F (TTGTACACACCGCCC) and 1510R (CCTTCYGCAGGTTCACCTAC) (Amaral-Zettler et al., 2009; Bahram et al., 2018; Mahe et al., 2017). Amplicon libraries were sequenced using an Illumina HiSeq2500 platform (Guangdong Magigene Biotechnology Co., Ltd., Guangzhou, China) to generate 250 bp paired-end reads.
The amplicon data were processed using USEARCH v10 and QIIME v1.91 (Caporaso et al., 2010; Edgar, 2010). Raw sequences were quality-trimmed, and the paired-end reads were merged into a single sequence. Biological reads were identified at 100% sequence similarity by using UNOISE3 (Edgar, 2016), set to its default parameters in USEARCH v10. Then, representative sequences for protistan zero-radius operational taxonomic units (ZOTUs) were aligned against the Protist Ribosomal Reference database (PR2) (Guillou et al., 2013), from which these ZOTUs assigned to Rhodophyta, Metazoa, Fungi, and unassigned Opisthokonta sequences were discarded (Guo et al., 2022; Guo et al., 2021). The ZOTU table was rarefied based on the lowest reads among all samples, to estimate the alpha diversity (Shannon and Chao1 indices); for beta diversity analyses, METAGENOMESEQ’s cumulative sum scaling (CSS) was used as the normalization method. Core taxa of healthy and diseased plants protistan microbiomes were defined as those ZOTUs occurring in at least 80% samples of healthy and diseased plants, respectively. In addition, we categorized the protistan OTUs into distinct functional groups based on their mode of nutrient-uptake using relevant literature as a reference (Nguyen et al., 2020; Xiong et al., 2020). These functional groups include parasites, phagotrophs, phototrophs, plant pathogens, and saprotrophs.

2.3 Statistical analysis

The beta diversity of protistan communities was calculated using Bray–Curtis distance matrices and visualized by non-metric multi-dimensional scaling (NMDS) ordinations. The relative contribution of FWD and sampling site to the protistan community composition was tested with permutational multivariate analysis of variance (PERMANOVA), using the “adonis” function in the ‘vegan’ package in R (Oksanen et al., 2007). Differential abundance analysis between protistan ZOTUs of healthy and diseased plants was implemented using EdgeR’s generalized linear model (GLM) approach (Robinson et al., 2010).
To assess the importance of deterministic and stochastic processes in the microbiome assembly, beta Nearest Taxon Index (βNTI) and Bray–Curtis-based Raup-Crick Index (RCI) were calculated by using a null model with 999 randomizations (Stegen et al., 2013). Deterministic dominant processes were quantified as |βNTI|≥2, in which βNTI<−2 indicates heterogeneous selection, and βNTI > +2 indicates homogeneous selection. Stochastic dominant processes were quantified as |βNTI| < 2. Pairwise comparisons with βNTI and RCI values were used to recognize homogenizing dispersal (|βNTI| < 2 and RCI < –0.95), dispersal limitation (|βNTI| < 2 and RCI > 0.95), and undominated (|βNTI| < 2 and |RCI| < 0.95) processes.

2.4 Microbial co-occurrence network analysis

Co-occurrence networks were constructed using the CoNet app in Cytoscape v3.7.1 (Faust, 2016; Shannon et al., 2003), and visualized in the Gephi platform (Bastian et al., 2009). All 1844 protistan ZOTUs were used in the intra-kingdom network analysis, for which only those with correlations having a Spearman’s coefficient (ρ) > 0.70 and P < 0.05 were retained. The interkingdom network analysis integrated 1152 protistan ZOTUs, 2549 bacterial ZOTUs and 1030 fungal ZOTUs, for which a relative abundance > 0.1% was criterion for inclusion in each group. The ZOTUs for bacteria and fungi had been generated from a previous study (Gao et al., 2021). Topological characteristics of each network, including the numbers of positive and negative correlations, average degree, clustering coefficient, and closeness centrality were calculated in Cytoscape v3.7.1 software.

2.5 Structural equation modeling (SEM) analysis

To evaluate the effects of phagotrophic protists on the communities of bacteria, fungi, and protists, we utilized SEM to determine the direction and strength of relationships among different microbial groups (Delgado-Baquerizo et al., 2013; Grace, 2006), using IBM SPSS Amos 21 (Chicago, IL: Amos Development Corporation). SEM was performed independently to healthy and diseased pepper plant data. Variable of Phagotrophs was represented by the composition of phagotrophic protists, and the communities of bacteria, fungi, and protists were represented by NMDS2, respectively. A maximum likelihood evaluation method was used to fit the covariance matrix into the model, and the model fitness was evaluated according to several diagnostic parameters: a non-significant chi-square test, goodness of fit index (GFI), comparative fit index (CFI), root mean square residual (RMR), normed fit index (NFI), and the root mean square error of approximation (RMSEA).

2.6 Metagenomic binning and functional analyses

As phagotrophic protists were significantly enriched in the roots of the diseased plants, healthy and diseased root endosphere samples were selected for additional metagenomic sequencing and binning. Six samples were sequenced on an Illumina NovaSeq 6000 instrument (Majorbio Bio-Pharm Technology Co. Ltd., Shanghai, China), yielding ca. 20 GB of clean data for each DNA sample. The reads were quality-filtered and assembled, using Trimmomatic v0.39 and Megahit v1.2.9 (Bolger et al., 2014; Li et al., 2015), respectively; the assembled sequences were predicted and functional annotated using Prokka v1.14.5 (Seemann, 2014), and eggNOG-mapper v1.0.3 (Huerta-Cepas et al., 2017; Huerta-Cepas et al., 2019), respectively, as described in the previous study (Gao et al., 2021).
Furthermore, the metagenome-assembled genomes (MAGs) were recovered from the assembled contigs, using MetaBAT2 (Kang et al., 2019) and MaxBin2 (Wu et al., 2016) according to the MetaWRAP pipeline (Uritskiy et al., 2018). To refine the ensuing bins, the Bin_refinement module was applied, and the consolidated bin sets were then quality assessed using CheckM v1.1.3 (Parks et al., 2015). Overall, 56 MAGs of sufficient quality (i.e., completeness > 50% and contamination < 10%) were obtained, all of which met currently accepted standards for medium- to high-quality MAGs (Bowers et al., 2017). These MAGs were then classified with GTDB-Tk v1.7.0, on the basis of 120 bacterial marker genes, to generate the taxonomic information (Parks et al., 2018). A phylogenetic tree of 56 MAGs was built using FastTree (Price et al., 2010), and visualized in iTOL v6.4 (Letunic et al., 2019). Also, genes of MAGs were predicted by Prokka v1.14.5 (Seemann, 2014), and the functional annotations organized using DIAMOND (Buchfink et al., 2015) in combination with the KEGG database (Kanehisa et al., 2016). KEGG Orthology (KO) profiles were generated by summing up the abundance of genes affiliated to the same KO. The abundance of MAGs on the microbiomes of healthy and diseased plants was calculated by coverM v 0.6.1 (https://github.com/wwood/CoverM).

3 RESULTS

3.1 FWD affects the assembly of pepper-associated protistan communities

After the removal of amplicon sequences that were of low-quality and/or chimeras, as well as those belonged to non-protists and singletons, a total of 1,765,274 high-quality protistan reads were retrieved, and they were sorted into 1,844 protistan ZOTUs. NMDS ordinations and PERMANOVA showed that protistan communities were most affected by plant compartment (R 2 = 0.48, P < 0.001), and significantly affected by sampling site (R 2 = 0.029, P < 0.001) and FWD (R 2 = 0.012, P = 0.003) (Figure 1a). Protistan community structures in the bottom, middle, and upper stem epidermis and fruit episphere were significantly affected by FWD, with the strongest effect found in upper stem epidermis (Figure 1b and Figure S1).
Protistan Shannon index was significantly reduced in the rhizosphere soil, bottom stem epidermis, and upper stem epidermis by FWD, but increased that in the root endosphere (P < 0.05, Figure S2). Differential abundance analysis indicated that several members within the protistan groups Cercozoa, Ciliophora, Lobosa, Conosa, and Pseudofungi were significantly enriched in the root of the diseased plants (Figure 1c and Figure S4). Most of these enriched protistan taxa belonged to the phagotrophic functional group, except for Pseudofungi. (Figure S3). Although 97.2% (1793 of 1844) protistan ZOTUs were shared between the healthy and diseased plants (Figure S5a), only 15 and 9 ZOTUs were recognized as core ZOTUs (i.e., in at least 80% of samples) in each group, respectively (Figure S5b and Table S1).

3.2 FWD affects protistan intra-kingdom network

Intra-kingdom co-occurrence network analysis was conducted to explore how FWD impacted the intra-kingdom relationships within protistan ZOTUs. The number of nodes and edges, clustering coefficient, average degree, and network density, which collectively convey a network’s complexity, were all higher in the diseased network than in the healthy one (Figure 2a and Table S2). A higher proportion of nodes belonging to Cercozoa were recorded in the diseased (31%) than the healthy (26%) network (Figure 2a). Also, the closeness centrality of nodes and the number of hub nodes (degree > 50) in the diseased network were greater than those in the healthy network (1 hub node in healthy network vs. 24 hub nodes in diseased network) (Figure 2e). Importantly, 57% of the hub nodes in the diseased network were Cercozoa ( Figure 2e), and the degree of nodes belonging to Cercozoa (P = 7.7e−05) and Ciliophora (P = 0.045) were higher in the diseased network than in the healthy network (Figure 2f, g). In addition, healthy and diseased networks present 33 overlapped nodes of Cercozoa and 24 overlapped nodes of Ciliophora; all the former and most of the latter nodes featured an average degree that was higher in the diseased than healthy network (Figure S6a, b). Compared with the healthy network, a higher proportion of FWD-enriched ZOTUs belonging to phagotrophic protists (healthy/diseased: 1.9%/7.7%) were recorded in the diseased network (Figure 2b). Furthermore, 5.06% and 3.16% of nodes in the healthy network belonged to core ZOTUs and FWD-depleted ZOTUs (i.e., healthy enriched), while 2.98% and 11.31% nodes in the diseased network belonged to core ZOTUs and FWD-enriched ZOTUs, respectively (Figure 2c, d).

3.3 Potential correlations between bacteria, fungi, and phagotrophic protists under FWD

After studying the patterns and processes that influenced the diversity and composition of protist microbiomes in plants, we studied the correlations among protists, bacteria and fungi in the healthy and FWD impacted pepper systems; interkingdom network analysis has been performed to integrate the current protist microbiome data, together with the bacteria and fungi microbiome data published in 2021 (Gao et al., 2021).
The diseased network contained a higher number of fungal and protistan nodes but a lower number of bacterial nodes, compared with healthy network (Figure 3a). 43.3% of hub nodes (degree > 50) in the diseased network were fungal taxa while all hub nodes in the healthy network belonged to bacterial taxa, no fungal and protistan taxa were inferred (Figure 3b). Also, the proportion of nodes belonging to FWD-enriched ZOTUs was higher in the diseased network (9.5%) than FWD-depleted ZOTUs in the healthy network (4.7%) (Figure S7). On the other hand, more negative edges representing bacterial–fungal (BF) and bacterial–protistan (BP) interkingdom correlations occurred in the diseased network than in the healthy one; the diseased network had more positive edges denoting fungal–fungal (FF), fungal–protistan (FP), and protistan–protistan (PP) correlations (Figure 3c). When comparing the interkingdom networks in the organs, higher number of protistan ZOTUs and BP correlations were observed in roots, bottom stems, and fruit networks in the diseased networks, compared to the healthy networks (Figure S8 and Figure S9). Moreover, the number of nodes belonging to Cercozoa and Ciliophora was higher in the diseased than the healthy network (Figure 3d and Figure S6c, d); the degree of nodes belonging to Cercozoa was also significantly higher in the diseased than the healthy network (P = 0.042, Figure S6c). For those Cercozoa and Ciliophora nodes shared by healthy and diseased networks, most of their connectivity was also higher in diseased network than in healthy networks (Figure S6e, f).
The nodes belonging to the phagotrophic protists Cercozoa and Ciliophora, as well as related nodes that interacted with these two protistan taxa were selected for networks’ visualization. Focusing on Cercozoa in the networks, the correlations of Cercozoa–Actinobacteria, Cercozoa–Alphaproteobacteria, and Cercozoa–Gammaproteobacteria were primarily negative, whereas those of Cercozoa–Sordariomycetes, Cercozoa–Cercozoa, Cercozoa–Ciliophora, and Cercozoa–Discoba were predominantly positive, in both healthy and diseased networks (Figure 3d, e). Importantly, numbers of these correlations were increased in the diseased network when compared with the healthy. Similar patterns were also observed in the networks focused upon Ciliophora (Figure 3d, e). Among the correlations of Cercozoa/Ciliophora–Alphaproteobacteria, the genera Sphingomonasand Methylobacterium were the main two linking members of the Alphaproteobacteria (Figure S10). Although the correlations between Cercozoa/Ciliophora and Actinobacteria/Alphaproteobacteria/Gammaproteobacteria were generally negative, we did detect several positive correlations among them (Figure 3f). Specifically, some of these implied a linkage between Cercozoa/Ciliophora and potential beneficial bacteria, such as that of Cercozoa–Bacillus in the healthy network, as well as that for Cercozoa/Ciliophora–Lysobacter , Cercozoa–Streptomyces , Cercozoa–Bacillus , and Cercozoa-Flavisolibacter in the diseased network (Figure 3f; Figure S10; Table S3). Several positive correlations were found linking Cercozoa/Ciliophora to potential pathogenic fungi in the diseased network, such as cases of Cercozoa/Ciliophora–Diaporthe , and Cercozoa–Gibellulopsis (Figure S10 b, d).
Since the proportion and the links associated with phagotrophic protists ZOTUs had increased from healthy to FWD networks, structural equation modeling (SEM) was utilized to gain deeper insights on how phagotrophic protists influenced the communities of bacteria, fungi, and protists under FWD (Figure S11 and Table S4). The modeling results revealed that phagotrophic protists had a significant, direct negative effect on the bacterial community, but a significant, direct positive effect on the fungal community, in both healthy and diseased SEM (Figure S11a, b). FWD increased strongly the path coefficients of phagotrophic protist–bacteria (healthy/diseased: 0.15/0.36) and phagotrophic protist–fungi (0.18/0.38) relationships (Figure S11a, b; Table S5), which indicated that the effects of phagotrophic protists on bacterial and fungal communities strongly enhanced by FWD.
As FWD strongly enhanced the microbial correlations between bacteria, fungi, and protists, we further hypothesized that the increased microbial interaction would increase the relative contribution of the deterministic processes in microbiome assembly of the diseased plants. Consistent with this hypothesis, null model analysis demonstrated a higher relative contribution of deterministic processes (|βNTI|≥2), in protistan (healthy/diseased: 41.1%/44.6%), bacterial (determinism, healthy/diseased: 66.9%/69.5%), and fungal communities (healthy/diseased: 62.2%/80.2%) of the diseased plant, compared with the healthy plants (Figure 4a). Remarkably, FWD increased heterogeneous selection in the assembly of the three microbial communities and increased homogeneous selection in fungal community assembly (Figure 4b).

3.4 FWD enhanced functional traits related to prey defense in root endosphere microbiomes

We further performed metagenomic analysis to characterize the functional genes in the pepper root-associated microbiomes. The data revealed that the relative abundance of several functional genes associated with bacterial prey defense, such as those associated with hydrogen cyanide (increased by 106%–1842%), cyclic lipopeptides (increased by 295%), and type III secretion systems (increased by 13.6%–138%) were higher in the microbiome of diseased vis-à-vis healthy pepper plants (Figure 4c).
Metagenomic binning successfully generated 15 and 41 bacterial MAGs of sufficient quality from the healthy and diseased pepper samples, respectively. These MAGs could be assigned to four phyla: Proteobacteria (31 MAGs), Bacteroidota (11), Actinobacteriota (10), and Patescibacteria (4) (Figure 5a). Among these MAGs, 8 MAGs harbored genes associated with bacterial prey defense (1 MAG from healthy samples, 7 MAGs from diseased samples) (Figure 5b). Intriguingly, MAGs that contained bacterial prey defense genes all belonged to Actinobacteriota and Gammaproteobacteria (Figure 5a), which was negatively correlated with phagotrophic protists Cercozoa and Ciliophora in the interkingdom networks (Figure 3d, e). Furthermore, 7 of these 8 MAGs were more abundant on the microbiome of the diseased plants compared to the healthy plants (Figure 5c).

4 DISCUSSION

While considering the assembly and the interkingdom interactions of soil and plant associated microbiomes under pathogen invasion, previous studies have predominantly focused on the bacterial or fungal communities. The responses of these communities to pathogen invasion have been extensively studied in various agricultural crops (Berendsen et al., 2018; Carrión et al., 2019; Gao et al., 2021; Kwak et al., 2018; Liu et al., 2020). The purpose of our research was to broaden this perspective by delving into the diversity and structure of protistan communities and exploring their responses to pathogen invasion. Additionally, our work sought to unravel the intricate interactions among protists, bacteria, and fungi, thereby providing a more comprehensive understanding of soil and plant microbiome assembly.
Our data showed that the protistan communities were influenced by sampling locations and plant compartments. A significant portion of protists display predatory behavior and selectivity in their feeding preferences, targeting specific bacterial and fungal groups. The distinct taxonomic and functional diversity harbored within plant compartments can influence the community structure and functions of bacteria and fungi, leading to variation in the protist communities (Nguyen et al., 2023). Also, FWD increased the relative abundance of phagotrophic protists in pepper root. Previous studies showed that protists constitute an important microbial component that is highly sensitive to fertilization (Guo et al., 2021; Zhao et al., 2019), aridity (Chen et al., 2022), and seasonal variation (Walden et al., 2021). Expanding upon these observations, our results present compelling evidence that soil and plant associated protistan communities could be drastically altered by disease incurred by pathogen invasion. Members of Cercozoa and Ciliophora are widely acknowledged as dominant and integral component of the protistan communities in soils and plants (Bates et al., 2013; Oliverio et al., 2020; Sapp et al., 2017). Both groups also prominently featured in the intra-kingdom networks of protists in our study. FWD-mediated enrichment of Cercozoa and Ciliophora seemed to exert a significant influence on the diseased intra- and interkingdom networks, implying that these two groups are crucial in microbe–microbe interactions.
By unraveling the intricate ecological relationships among different microbial groups, including predation, competition, and cooperative correlations, we can gain a more comprehensive understanding of microbe-microbe interactions within soil and plant ecosystems. The complexity of these ecological relationships can be represented through microbial networks, which is fundamental for characterizing the correlations of microbial taxa and could reflect dynamics and functions of ecosystems (Delgado-Baquerizo et al., 2020; Wagg et al., 2019; Yuan et al., 2021). Our network analysis revealed that the intra-kingdom correlations between protists and fungi were predominantly positive, while both groups demonstrated negative correlations with bacteria. This preponderance of negative correlations between protists and bacteria suggests a mutual exclusivity in species abundance between the two. Specifically, members of Cercozoa and Ciliophora were largely negatively associated with members of the Actinobacteria, Alphaproteobacteria, and Gammaproteobacteria. According to a microbiota reconstitution experiment, members of Alphaproteobacteria and Gammaproteobacteria demonstrated lower resistance to grazing by the leaf-associated protist Cercomonads (Rhizaria: Cercozoa) (Flues et al., 2017). This concurrence suggests that the observed negative correlations between protists and bacteria in our networks may reflect potential predatory relationships. Such predation could augment microbial loops, as the consumption of bacteria or fungi by protists releases nutrients that can then be utilized by other microbes or assimilated by host plants (Bonkowski, 2004; Clarholm, 1985; Gao et al., 2019; Geisen et al., 2018). Our analysis also identified several positive correlations between Cercozoa/Ciliophora and potentially beneficial bacteria, as well as certain Sordariomycetes fungi. Collectively, these findings imply that protistan predation may play a key role in modulating microbial community structures, significantly impacting the distribution and abundance of specific taxa.
Our metagenomic data revealed that that the relative abundance of functional genes associated with bacterial prey defense mechanisms, including hydrogen cyanide, cyclic lipopeptides, and type III secretion systems, increased in the microbiome of diseased pepper plants when compared with healthy ones. Bacteria have evolved a range of adaptive strategies, such as morphological adaptation, secondary metabolites and secretions, in response to the selective pressure exerted by predatory protists (Gao et al., 2019; Jousset, 2012; Jousset et al., 2009). These adaptive mechanisms can, in turn, impact the abundance of predatory protists (Becks et al., 2012; Hiltunen et al., 2014). Given that the growth of predator population was affected more by prey evolution than by prey abundance (Becks et al., 2012), we propose that the elevated abundance of prey defense traits and phagotrophic protists in the diseased plants may represent the main components of an eco-evolutionary feedback loop in the predator–prey system. This feedback loop, mediated by predator-prey interactions could further drive the functions of the microbiome (Becks et al., 2012); particularly, the stimulation of the prey defense traits could influence the capacity of the microbiome to suppress the disease invasion of host plant. For example, the toxic secondary metabolite hydrogen cyanide produced by bacteria can be induced by protists to reduce predation pressure (A. Jousset et al., 2010), and this metabolite has been a potent deterrent of various plant fungal pathogens (Rijavec et al., 2016). Similarly, antifungal metabolites cyclic lipopeptides are known to confer resistance to predation by protists (Mazzola et al., 2009; Nielsen et al., 2003). The overlapping suite of the traits linked to predation defense and pathogen suppression could be considered an adaptation which beneficial functions are promoted throughout the microbiome (Gao et al., 2019). Another persuasive piece of evidence is that the presence of protists can stimulate the antifungal activity of beneficial bacteria; therefore, co-inoculating beneficial bacteria and protists (Rosculus terrestris , Bodomorpha sp., and Cercomonas lenta ) significantly enhanced faba bean to suppress against soil-borne pathogen Fusarium solani (Bahroun et al., 2021). Further, organic fertilization can stimulate predator-prey interactions that catalyze keystone interactions between protistan predators and plant-beneficial bacteria, thus decreasing bananaFusarium wilt disease incidence and ultimately enhancing plant health and yield (Guo et al., 2022). Drawing on these findings, our results propose that, similar to the influence of organic fertilizer inputs, fungal pathogen invasions could also stimulate predator-prey correlations.
Beyond catalyzing associations between protists and bacteria, FWD also amplified the associations between various other microbes within the networks. This augment may be related to the disruption of the stable systems maintained by host plants. Lots of evidences support that the composition of the microbial communities could be stabilized during plant growth and development (Edwards et al., 2018; Xiong et al., 2021; Zhang et al., 2018). When mature plants are infected by the pathogens, this would spur a series of changes in the physiochemical conditions, such as water losses, withering, and litter fall (Michielse et al., 2009), potentially altering the nutrients and space available to microbes in the soil and plants directly or indirectly. Moreover, the plant’s immune system response can trigger alterations to root exudates or other volatile substances to recruit protective microbial members (Berendsen et al., 2018; H. Liu et al., 2019; Liu et al., 2020; Liu et al., 2019), or provide legacy for the next generation under pathogen stress (Bakker et al., 2018). Changes of these complex substances and resources may alter the selection pressure of host plant to microbes, thereby inducing competition or cooperation among them. The increased deterministic process, particularly heterogeneous selection observed in the assembly process of the diseased plant microbiome, lends further support to this perspective. Still, However, selective predation by phagotrophic protists might play a crucial role in maintaining microbial balance within such unstable communities instigated by pathogen infection. Existing research has demonstrated that phagotrophic protists are main determinants of crop health and yield by interacting with plant-beneficial microorganisms (Bahroun et al., 2021; Guo et al., 2021), and influencing the pathogenicity of pathogens (Chakraborty et al., 1983; Xiong et al., 2020). Given that plant health, particularly for future generations, requires a favourable microenvironment, predation by phagotrophic protists may assist in shaping a microbiome more conducive to the fitness of host plants.