Learning from Seed Microbes: Trichoderma Coating Intervenes in Rhizosphere Microbiome Assembly

ABSTRACT Seed-associated microbiomes can impact the later colonization of a plant rhizosphere microbiome. However, there remains little insight into the underlying mechanisms concerning how alterations in the composition of the seed microbiome may intervene in the assembly of a rhizosphere microbiome. In this study, the fungus Trichoderma guizhouense NJAU4742 was introduced to both maize and watermelon seed microbiomes by seed coating. Application was found to significantly promote seed germination and improve plant growth and rhizosphere soil quality. The activities of acid phosphatase, cellulase, peroxidase, sucrase, and α-glucosidase increased significantly in two crops. The introduction of Trichoderma guizhouense NJAU4742 also led to a decrease in the occurrence of disease. Coating with T. guizhouense NJAU4742 did not alter the alpha diversities of the bacterial and fungal communities but formed a key network module that contained both Trichoderma and Mortierella. This key network module comprised of these potentially beneficial microorganisms was positively linked with the belowground biomass and activities of rhizosphere soil enzymes but negatively correlated with disease incidence. Overall, this study provides insights into plant growth promotion and plant health maintenance via seed coating in order to influence the rhizosphere microbiome. IMPORTANCE Seed-associated microbiomes can impact the rhizosphere microbiome assembly and function display. However, there remains little insight into the underlying mechanisms concerning how alterations in the composition of the seed microbiome with the beneficial microbes may intervene in the assembly of a rhizosphere microbiome. Here, we introduced T. guizhouense NJAU4742 to the seed microbiome by seed coating. This introduction led to a decrease in the occurrence of disease and an increase in plant growth; furthermore, it formed a key network module that contained both Trichoderma and Mortierella. Our study provides insights into plant growth promotion and plant health maintenance via seed coating in order to influence the rhizosphere microbiome.

rhizosphere microbial community. In this study, we coated seeds of maize as well as seeds of watermelon with Trichoderma guizhouense NJAU4742 to evaluate its plantgrowth-promoting effect and to assess its influence on the rhizosphere microbial community assembly. We hypothesized that (i) seeds coated with T. guizhouense NJAU4742 would improve agronomic traits of crops and (ii) the impact of coating T. guizhouense NJAU4742 on crop performance may be partially explained by a shift in the rhizosphere microbial community.

RESULTS AND DISCUSSION
Seed coating with Trichoderma sp. improves plant growth and rhizosphere soil enzyme activities. At the early stage of a plant's life span, seedlings are exposed to various biotic and abiotic stresses, such as pathogens, herbivores, and drought (24). Thus, manipulating the plant rhizosphere microbiome at an early stage via seed coating with T. guizhouense NJAU4742 would be expected to promote plant growth and reduce disease incidence. Here, we tested the promotional effect of T. guizhouense NJAU4742 coating on watermelon and maize. Results showed that the germination rate of coated seeds were significantly increased by 25% for maize after 3 days postplanting ( Fig. 1A and C) and 35% for watermelon after 8 days postplanting, compared with the control (Fig. 1B and D). Seed coating was also found to benefit plant growth, as plant height, stem diameter, and biomass production were significantly higher than in the control groups (see Table S2 in the supplemental material). Compared with the control, the final plant height of maize and watermelon was also improved by 20% and 50% ( Fig. 1E and F) with a 24% increase in dry shoot weight for watermelon as well as a 39% and 57% increase in dry shoot weight and stem diameter for maize, respectively ( Fig. 1G and H; Table S2). Since the plant root system plays a vital role in nutrient uptake and physical support, increasing root biomass and improving root architecture are some of the principal goals of plant breeding (25). Here, seed-coated maize and watermelon exhibited a higher number of primary and lateral roots with the main root lengths significantly elongated by 58% and 72%, respectively, than the control plants ( Fig. 1G and H; see Fig. S1 in the supplemental material). This finding was accompanied by an increase of 22% (maize) and 39% (watermelon) in dry root weight ( Fig. 1G and H).
Previous research has shown that the inoculation of Trichoderma sp. can promote plant growth and reduce the disease occurrence of Fusarium wilt (28). To this end, we explored whether the method of coating with T. guizhouense NJAU4742 can achieve the same results. We found that the incidence of maize stalk rot in the coated treatment was significantly reduced by 28% compared with that in the control (Fig. 1G), and the incidence of watermelon with coating was reduced by 37% (Fig. 1H) at 60 days after pathogen application. Trichoderma spp. have been used widely as biocontrol agents in commercial crop production with previous methods of application principally using combinations of bioorganic fertilizer and microbial inoculum (29). Beneficial effects depend primarily on the metabolic activity of Trichoderma sp. as well as beneficial interactions with host plants (30). For example, this strain of Trichoderma was reported to efficiently stimulate lateral roots by The germination rate of maize and watermelon after seeds were coated with or without T. guizhouense NJAU4742. (E and F) The plant height of maize and watermelon after seeds were coated with or without T. guizhouense NJAU4742. (G) Plant root height, root dry weight, shoot dry weight, and disease incidence of maize after seeds were coated with or without T. guizhouense NJAU4742. (H) Plant root height, root dry weight, shoot dry weight, and Fusarium wilt incidence of watermelon after seeds were coated with or without T. guizhouense NJAU4742. Error bars indicate the standard error based on six replicates. **, 0.001 , P , 0.01; ***, P , 0.001. TM, maize seeds coated with T. guizhouense NJAU4742; CM, maize seeds coated without T. guizhouense NJAU4742; TW, watermelon seeds coated with T. guizhouense NJAU4742; CW, watermelon seeds coated without T. guizhouense NJAU4742.

Engineering Rhizosphere Microbiome via Seed Coating
Microbiology Spectrum the secretion of swollen protein (31). In summary, the seed coating strategy implemented here appears to be a high-throughput, lower cost method to achieve similar results to rhizosphere application. Seed coating with Trichoderma sp. intervenes in rhizosphere microbiome assembly. Microbial community composition and diversity are critical for maintaining agroecosystem sustainability and productivity (32). Since it was apparent that seed coating was resulting in beneficial effects on plant growth and health, it was necessary to further explore the effect of seed coating on the rhizosphere microbiome. Several studies have reported the reduction of rhizosphere microbial diversity after application of Trichoderma sp. as a biofertilizer, resulting in lower complexity and stability of the rhizosphere microbial community (33,34). Our results showed that coating with T. guizhouense NJAU4742 did not affect rhizosphere bacterial and fungal diversity (see Fig. S3A and B in the supplemental material), indicating the "safety" of T. guizhouense NJAU4742 coating on the rhizosphere microbiome. At the phylum level, the relative abundance of Proteobacteria was the highest, followed by Acidobacteria, and Gemmatimonadetes having a relatively high abundance. Moreover, in coated samples with both maize and watermelon, there was no significant difference in the relative abundance of bacteria compared with the control, except for Bacteroidetes, which were higher than in the control (Fig. S3C). Accordingly, the subgroups Chryseolinea, Crocinitomix, and Niabella were significantly enriched in the seedcoating treatment (see Table S4 in the supplemental material). These groups may potentially act antagonistically against plant pathogens, as it has been reported that Chryseolinea was found to be the key taxa contributing to the microbiome in disease suppressive soils against banana Fusarium wilt (35). Other studies have reported that growth-promoting bacteria, such as Pseudoxanthomonas (36), Herbaspirillum (37), and Flavobacterium (38), were enriched in the rhizosphere of coated watermelon and maize seeds and that they may promote plant growth by altering root architecture and producing various phytohormones (39). In terms of fungal composition, the dominant phylum present was Ascomycota which represented 45% of the relative abundance at the phylum level (Fig. S3C). Previously, we found lower abundances of Ascomycota and Basidiomycota in non-Fusarium wilt diseased soils (40), indicating that these two phyla were not beneficial for plant health in the context of Fusarium wilt disease. Here, the relative abundances of Ascomycota and Basidiomycota were lower in the coated treatment (Fig. S3C). As expected, the relative abundance of Trichoderma sp. in the coated treatment was increased within both the maize and watermelon rhizosphere fungal communities, indicating the effective rhizosphere colonization of Trichoderma sp. via seed coating (see Table S6 in the supplemental material).
In complex ecosystems, co-occurrence networks have been used to explore interactions between microbes. Here, bacterial and fungal co-occurrence networks were constructed based on Spearman correlations among operational taxonomic units (OTUs) to investigate how T. guizhouense NJAU4742 coating impacted microbial interactions in the rhizosphere soil. Based on network topological comparisons, we found that the proportion of negative edges in both bacterial and fungal networks increased in the seed-coating treatment ( Fig. 2B and H; see Table S5 in the supplemental material). This finding indicates that coating with T. guizhouense NJAU4742 may improve the stability of microbial communities by increasing microbial competition. To further evaluate community stability, natural connectivity was tested by the random removal of nodes. By removing the same proportion (from 1% to 60%) of nodes, results showed that the natural connectivity of fungal networks in coated maize and watermelon was higher than that in the control which indicated that coating with T. guizhouense NJAU4742 could enhance fungal community stability ( Fig. 2C and I). Traditionally, the resilience and stability of beneficial microbial communities in response to a disturbance and their ability to moderate or suppress invasion and disease caused by a pathogen are some of the most important characteristics of ecosystem and organismal health (41). Here, the rhizosphere microbiome stability was shown to be associated with plant health, especially against soilborne diseases.
To evaluate the potential relationship between plant biomass and rhizosphere bacterial/fungal OTUs, we calculated correlations between network module-based eigengenes and plant biomass. In bacterial networks, there was no significant correlation between network modules and plant physiological traits (see Fig. S4 in the supplemental material). For fungal networks, only the coated samples exhibited a significant correlation between network modules and plant physiological traits ( Fig. 2D and J). In the fungal network of coated maize, submodule 7 was positively correlated with root length and stem diameter (P , 0.01) while negatively related to disease incidence (P , 0.05) (Fig. 2D), and in the fungal network of coated watermelon, submodule 3 was positively correlated with root length and root dry weight (P , 0.01) while negatively related to disease incidence (P , 0.05) (Fig. 2J). Moreover, in coated maize samples, the submodule 7 was positively correlated with urease (r = 0.75), acid phosphatase (r = 0.78), alkaline phosphatase (r = 0.76), and cellulase (r = 0.8) enzyme activities. In coated watermelon samples, the submodule 3 was positively correlated with a-glucosidase (r = 0.78), polyphenol oxidase (r = 0.6), and cellulase (r = 0.85) enzyme activities (see Fig. S5 in the supplemental material). The partial least-squares path model (PLS-PM) analysis also showed that the microbial communities of fungal submodule 7 (maize) and submodule 3 (watermelon) in the coating treatment might affect plant growth directly or indirectly by improving enzyme activities in the rhizosphere soil (Fig. 3). These results suggested that the enhancement of rhizosphere soil enzyme activities was principally driven by the microorganisms within fungal submodule 7 (maize) and submodule 3 (watermelon). Previous findings have demonstrated clear linkages between rhizosphere fungal communities and rhizosphere enzyme activities (8). One proposed explanation was that soil fungi were the most vital contributor to soil enzymes which improved soil quality, further promoting the increase in plant biomass (42). Further analysis showed a variety of plant mutualists, including Trichoderma and Mortierella (43,44), that were enriched in submodule 7 and submodule 3 ( Fig. 2E and K; Table S6), and they were identified as key hubs with high betweenness centrality and degree ( Fig. 2F and L; see Table S7 in the supplemental material). These essential microorganisms, such as Trichoderma sp., possess high potency to produce enzymes (45).
To further validate whether the formation of submodule 3 and submodule 7 was predominantly driven by the introduction of Trichoderma sp., we recalculated the microbiome co-occurrence network of maize seed coating treated with spores (TM) and watermelon seed coating treated with spores (TW) after removing the Trichoderma genus from the OTU table or reducing the abundance of Trichoderma sp. in the coated samples to that of the control. None of the modules in the reconstructed coated fungal networks, after the removal of the Trichoderma genus, were found to be similar to the previous submodule 7 (maize) and submodule 3 (watermelon) by comparing the proportion of shared nodes in the two targeted modules (see Table S8 in the supplemental material). Identical results were found after reducing the abundance of Trichoderma sp. in the coated maize. One of the modules in the coated watermelon (by reducing the abundance of Trichoderma) was similar to the previous submodule 3 (watermelon), but five shared OTUs were not the key hubs in submodule 3 (watermelon) ( Table S8). Network hubs have been proposed as keystone taxa and ecological generalists critical for the maintenance of community structure, function, and overall ecosystem stability (46). In this study, we found that seed coating Conclusion. Our study evaluated the effect of seeds coated with T. guizhouense NJAU4742 on plant growth and the associated rhizosphere microbial community. The results revealed that seed coating enhanced seed germination and improved the root architecture of the plants, including increasing the main root length and the number of primary and lateral roots. In addition, it also increased the above/belowground biomass of the plant and reduced the occurrence of disease. Furthermore, some growthpromoting microbes like Trichoderma and Mortierella in the coated microbiome were found to principally contribute to the positive correlations between the microbiome composition and function (plant growth promotion and disease resistance). Our results demonstrated that the manipulation of the rhizosphere microbiome by seed coating with T. guizhouense NJAU4742 was an efficient strategy to achieve plant growth promotion as well as effective disease control.

MATERIALS AND METHODS
Preparation of Trichoderma-containing seed coating agent. The coating agent used in this experiment was prepared by adding 1% starch and 10% iron oxide red (coloring agent) to a prepared T. guizhouense NJAU4742 spore solution (1 Â 10 8 spore/mL). The Trichoderma spore solution was prepared as follows: T. guizhouense NJAU4742 (GenBank accession number LVVK01000017; stored in our laboratory, and the spores of this strain were preserved in 30% glycerin at 280°C to prevent spawn degeneration) was used in this experiment and stored at 280°C as a spore solution, and 1 mL of spores was placed in 30 mL potato dextrose broth (PDB) and incubated for 7 days at 26°C and 180 rpm in a thermostat incubator chamber. Spores were then harvested by filtering three times with sterile gauze to Microbiology Spectrum remove mycelium. The spore concentration was adjusted to 1 Â 10 8 spore/mL, based on the measurement of the hemocytometer. Pot experiment. We used watermelon ("Lenong No.3") and maize ("Zhongnuo 309") to test the biological effect of seed coating with T. guizhouense NJAU4742. For seed coating with T. guizhouense NJAU4742, 2 mL of coating agent was used for 100 maize seeds and 1 mL was used for 100 watermelon seeds, resulting in a concentration of T. guizhouense NJAU4742 spores at 1 Â 10 6 spore per seed. For the control treatments, identical numbers of maize and watermelon seeds were coated with a coating agent comprised of inactivated spores. The soil for the pot experiment was collected from the Baima Teaching and Research Base of Nanjing Agricultural University (31°369N latitude and 119°109E longitude). The physicochemical properties of the soil were presented in Table S1 in the supplemental material. One coated seed was planted in each cell (6.8 cm Â 7.5 cm Â 10cm, containing 100 g soil) of seeding trays (21 cells per tray) that were placed in a greenhouse with 16 h photoperiod (120 mmol photons m 22 Á s 21 ) at 23/20°C day/night temperatures. Each crop type was comprised of a coated treatment (10 6 spores per seed) and control, as follows: maize seed coating treated with spores (TM), control maize seed coating with inactivated spores (CM), watermelon seed coating treated with spores (TW), and control watermelon seed coating with inactivated spores (CW). The 2 treatments of maize seeds contained 168 seedlings divided into 6 replicates with 28 seedlings per replicate. The 2 treatments of watermelon seeds contained 144 seedlings divided into 6 replicates with 24 seedlings per replicate.
The seedling germination rate of maize was recorded on the 3rd day after planting. After planting for 8 days, the watermelon seedling germination rate was recorded. The plant height was measured every 3 days until irrigating pathogen solution at the 35th day after planting. At the same time, six plants from each replicate were harvested to measure stem diameter, root parameter (measured by root scanner, GXY-A; Zhejiang Top Cloud Agri Technology Co., Ltd.), and biomass, and rhizosphere soil was collected for subsequent determination.
To test the biocontrol effect of the seed coating with T. guizhouense NJAU4742, we followed an inoculation method used in previous publications (47,48). Briefly, watermelon was irrigated with a conidial suspension (Fusarium oxysporum f. sp. niveum was isolated in our lab [49]) of 1 Â 10 5 spores per mL. A pathogen for maize (Fusarium graminearum) was kindly provided by Xiao (Heilongjiang Academy of Agricultural Sciences) (48), and the inoculation method was the same as mentioned above. After 25 days of inoculation with the pathogen, the disease incidence did not increase anymore, and we recorded the number of watermelon and maize with obvious disease symptoms (the watermelon seedlings exhibited stunted growth, with yellowing and wilting of the leaves starting from the base of the plant; the maize seedlings exhibited slow growth and stem discoloration). The whole experimental period lasted 60 days.
Rhizosphere soil genome DNA extraction and sequencing. Before DNA extraction, 0.5 g of soil adhered tightly with roots was placed into a 2 mL centrifuge tube containing 1 mL of phosphate-buffered saline solution and several sterilized glass beads, after which the mixture was vortexed at maximum speed for 15 min. The suspension (without root materials) was then transferred to a new 2-mL centrifuge tube and centrifuged for 30 min at 15,000 rpm. The supernatant was discarded, and the precipitate was used for DNA extraction. In total, 20 samples were collected and total DNA was extracted from the precipitate using a FastBeat SoilPure PowerSoil DNA Isolation Kit (BOLAZ) according to the manufacturer's protocol. The DNA quality and quantity were measured via a 1.2% agarose gel and a NanoDrop 1000 spectrophotometer (Thermo Scientific, USA).
Both the ITS1-2 and V4 regions of 16S rRNA gene libraries were sequenced using the Illumina MiSeq platform at Guangdong Magigene Biotechnology, Co., Ltd., China. The primer pair 515F (GTGYCAGCMGCCGCGG TAA) and 806R (GGACTACNVGGGTWTCTAAT) for bacteria and ITS1F (CTTGGTCATTTAGAGGAAGTAA) and ITS2R (GCTGCGTTCTTCATCGATGC) for fungi were used. The barcoded PCR amplicon library was built separately for bacterial and fungal community analysis. Bacterial and fungal amplicon library preparation and high-throughput sequencing were performed as described previously (40). The DNA marker was provided by Sangon Biotech, Shanghai (B500351) with a range of 100-5000 bp.
For sequencing data analyses, usearch (V. 10.1) and vsearch (V. 0.6.3) were implemented. First, the "vsearch -fastq_mergepairs" script was used for merging paired-end sequences, "vsearch -fastx_filter" script was used to cut primers, "vsearch -derep_fulllength" script was used for finding unique read sequences, and "usearch -unoise3" script was used for cluster OTUs. The "usearch -otutab" script was used to create an OTU table. The "vsearch -sintax" script, Silva tax database, and unite tax database were used for bacterial and fungal OTU annotation.
Statistical analysis. Statistical analysis in the study was performed using the statistical programming language R 4.1 (63). Differences between treatments of plant physiological parameters were analyzed by Wilcoxon signed-rank test. The significance threshold was set as adjusted P values of ,0.05.
The P values were corrected by the Benjamini-Hochberg false discovery rate (FDR) procedure for multiple comparisons. Visualization analysis was done by R package "ggplot2" (64).
A normalized number of sequences were randomly extracted from each sample to calculate alpha diversity indices with the R package "vegan" (65). Wilcoxon nonparametric tests were used for the detection of significance for Shannon diversity, Pielou evenness, and the Chao1 indexes using the "EasyStat" package. Before the calculation of beta diversity, relative abundances were used to standardize the OTU profiles. Bray-Curtis similarity matrices were prepared using the "vegan" R packages. Permutational multivariate analysis of variance (PERMANOVA; Adonis, transformed data by Bray-Curtis, permutation = 999) was used to test if the beta diversity differed among treatments. Nonmetric multidimensional scaling (NMDS) plots were generated according to Bray-Curtis similarity matrices created using the R package "ggplot2" (66). R package "edgeR" were used to identify differentially abundant microbes between treatments (67). Network analysis was performed using R package "ggClusterNet" (68). In the eigengene analysis, each network module is represented by principal-component analysis (PCA) of the abundance profile called module eigengene "WGCNA" (69). The R package "WGCNA" was used to calculate the correlations between module-based eigengenes and plant physiological enzyme activities. The partial least-squares path model (PLS-PM) was performed to further infer potential direct and indirect effects of the bacterial community, fungal community, network module, and soil enzyme activities on plant physiological indices. All PLS-PM analyses were conducted using R package "plspm" (70). The module paired was calculated according to Yuan (46).
Data availability. Raw sequence data obtained in this study have been deposited in the Genome Sequence Archive in the BIG Data Center, Chinese Academy of Sciences under accession code CRA007067.

SUPPLEMENTAL MATERIAL
Supplemental material is available online only. SUPPLEMENTAL FILE 1, PDF file, 1.5 MB.