Consortia of anti-nematode fungi and bacteria in the rhizosphere of soybean plants attacked by root-knot nematodes

Cyst and root-knot nematodes are major risk factors of agroecosystem management, often causing devastating impacts on crop production. The use of microbes that parasitize or prey on nematodes has been considered as a promising approach for suppressing phytopathogenic nematode populations. However, effects and persistence of those biological control agents often vary substantially depending on regions, soil characteristics and agricultural practices: more insights into microbial community processes are required to develop reproducible control of nematode populations. By performing high-throughput sequencing profiling of bacteria and fungi, we examined how root and soil microbiomes differ between benign and nematode-infected plant individuals in a soybean field in Japan. Results indicated that various taxonomic groups of bacteria and fungi occurred preferentially on the soybean individuals infected by root-knot nematodes or those uninfected by nematodes. Based on a network analysis of potential microbe–microbe associations, we further found that several fungal taxa potentially preying on nematodes (Dactylellina (Orbiliales), Rhizophydium (Rhizophydiales), Clonostachys (Hypocreales), Pochonia (Hypocreales) and Purpureocillium (Hypocreales)) co-occurred in the soybean rhizosphere at a small spatial scale. This study suggests how ‘consortia’ of anti-nematode microbes can derive from indigenous (resident) microbiomes, providing basic information for managing anti-nematode microbial communities in agroecosystems.


43
Plant pathogenic nematodes, such as cyst and root-knot nematodes, are major threats to crop 44 production worldwide (Barker & Koenning 1998; Abad et al. 2008). Soybean fields, in 45 particular, are often damaged by such phytopathogenic nematodes, resulting in substantial 46 yield loss (Wrather et al. 1997;Wrather & Koenning 2006). A number of chemical 47 nematicides and biological control agents (e.g., nematophagous fungi in the genera 48 Purpureocillium and Clonostachys) have been used to suppress nematode populations in 49 farmlands (Schmitt et al. 1983;Li et al. 2015). However, once cyst and root-knot nematodes 50 appear in a farmland, they often persist in the soil for a long time (Meyer & Roberts 2002), 51 causing high financial costs in agricultural management. Therefore, finding ways to suppress 52 pathogenic nematode populations in agroecosystems is a key to reducing risk and 53 management costs in production of soybean and other crop plants. By an Illumina sequencing analysis of bacteria and fungi in a soybean (Glycine max) 89 field, we examined how root and rhizosphere microbiome structures varied among host plant 90 individuals that differed in damage by root-knot nematodes (Meloidogyne sp.). Based on the 91 data of microbiomes at a small spatial scale, we statistically explored microbial species/taxa 92 that occurred preferentially in the roots or rhizosphere soil of nematode-infected soybean 93 individuals. We further investigated the structure of networks depicting co-abundance 94 patterns of microbial species/taxa within the soybean field, thereby examining whether 95 multiple anti-nematode bacteria and fungi form consortia (assemblages) on/around the plant 96 individuals infected by root-knot nematodes. Overall, this study suggests that various 97 taxonomic groups of anti-nematode bacteria and fungi are present within indigenous 98 microbiomes. Our results also suggest that microbiome assembly at fine spatial scales is a key 99 to manage populations and communities of such functional microbes. 100

Sampling 103
Fieldwork was conducted at the soybean field on the Hokubu Campus of Kyoto University, 104 Japan (35.033 ºN, 135.784 ºE). In the field, the soybean strain "Sachiyutaka" was sown at 15 105 cm intervals in two lines ( Supplementary Fig. 1) on July 4, 2016 [basal fertilizer, N:P 2 O 5 :K 2 O 106 = 3:10:10 g/m 2 ]. In the lines, 69 and 62 individuals ("set 1" and "set 2", respectively), 107 respectively, were sampled every other positions (i.e., 30 cm intervals) ( Fig. 1) on October 7, 108 2016. The sampled soybean individuals were classified into three categories: normal 109 individuals with green leaves ("green"), individuals with yellow leaves ("yellow"), and those 110 with no leaves ("no leaf") ( Fig. 1A-C). Among them, "green" individuals exhibited normal 111 growth, while "no leaf" individuals were heavily infected by root-knot nematodes: "yellow" 112 individuals showed intermediate characters. In total, 97 "green", 19 "yellow", and 15 "no leaf" 113 individuals were sampled (Fig. 1D). 114 For each individual, two segments of 5-cm terminal roots and rhizosphere soil were 115 collected from ca. 10-cm below the soil surface. The root and soil samples were transferred 116 into a cool box in the field and then stored at -80ºC until DNA extraction in the laboratory. 117 The whole bodies of the individuals were placed in drying ovens at 80 ºC for 72 hours to 118 measure dry mass. The dry mass data indicated that "green", "yellow", and "no leaf" soybean 119 individuals significantly differed in their biomass (Fig. 1C). 120

DNA Extraction, PCR, and Sequencing 122
The root segments of each individual were transferred to a 15 mL tube and washed in 70% 123 ethanol by vortexing for 10 s. The samples were then transferred to a new 15 mL tube and 124 then washed again in 70% ethanol by sonication (42 Hz) for 5 min. After an additional 125 sonication wash in a new tube, one of the two root segments were dried and placed in a 1.2 126 mL tube for each soybean individual. DNA extraction was then performed with a 127 cetyltrimethylammonium bromide (CTAB) method (Sato & Murakami 2008) after 128 pulverizing the roots with 4 mm zirconium balls at 25 Hz for 3 min using a TissueLyser II 129 (Qiagen). 130 For DNA extraction from the rhizosphere soil, the ISOIL for Beads Beating kit (Nippon 131 Gene) was used as instructed by the manufacturer. For each sample, 0.5 g of soil was placed 132 into a 2 mL microtubes of the ISOIL kit. To increase the yield of DNA, 10 mg of skim milk Relationship between the number of sequencing reads and that of detected OTUs was 216 examined for each dataset (root prokaryote, root fungal, soil prokaryote, or soil fungal 217 dataset) with the "rarecurve" function of the R vegan package. Likewise, relationship between 218 the number of samples and that of OTUs was examined with the vegan "specaccum" function. 219 For each dataset, difference in OTU compositions among "green", "yellow", and "no leaf" 220 To control effects of sampling positions (lines) on the community structure, the information 223 of sampling sets (set 1 or set 2) was included as an explanatory variable in the 224 PERMANOVA. The variation in OTU compositions was visualized with nonmetric 225 multidimensional scaling (NMDS) using the vegan "metaMDS" function. To examine 226 potential relationship between root/soil microbial community structure and plant biomass, an 227 additional PERMANOVA was performed for each dataset. The information of sampling sets 228 was included in the models. To explore signs of spatial autocorrelation in the community data, 229 a Mantel's correlogram analysis was performed with the vegan "mantel.correlog" function. 230 The "Bray-Curtis" metric of β-diversity was used in the PERMANOVA, NMDS, and 231 Mantel's correlogram analyses. 232 233

Screening of Host-state-specific OTUs 234
To explore prokaryote/fungal OTUs that preferentially occurred on/around "green", "yellow", 235 or "no leaf" soybean individuals, a randomization test was performed by shuffling the plant 236 state labels in each of the root prokaryote, root fungal, soil prokaryote, and soil fungal data 237 matrices (100,000 permutations). We then evaluated preference of a prokaryote/fungal OTU (i) for a plant state (j) ("green", "yellow", or "no leaf") as follows: 239 where N observed (i, j) denoted the mean number of the sequencing reads of OTU i among state j 241 soybean samples in the original data, and the Mean (N ranodomized (i, j)) and SD (N ranodomized (i, j)) 242 were the mean and standard deviation of the number of sequencing reads for the focal OTU- OTUs per sample were observed, respectively, from the root prokaryote, root fungal, soil 264 prokaryote, and soil fungal dataset after filtering and rarefaction steps ( Supplementary Fig. 2). 265 The total number of OTUs observed was 1387, 346, 1191, and 769 for the root prokaryote, 266 root fungal, soil prokaryote, and soil fungal datasets, respectively ( Supplementary Fig. 3). 267 In the soybean field, the prokaryote community on roots was dominated by the bacterial 268 classes Proteobacteria, Actinobacteria, Chloroflexi, and Bacteroidetes, while that of 269 rhizosphere soil consisted mainly of Proteobacteria, Actinobacteria, and Acidobacteria, and 270 the archaeal lineage Thaumarchaeota (Fig. 2A). The fungal community of roots was 271 dominated by the fungal orders Hypocreales, Sordariales, Plesporales, while that of soil 272 consisted mainly of Hypocreales, Agaricales, Eurotiales, Mortierellales, and Filobasidiales 273 (Fig. 2B). Regarding the order level compositions of fungi in the rhizosphere soil, the 274 proportion of Orbiliales reads was much higher in "yellow" (3.62 %) and "no leaf" (4.82 %) 275 soybean individuals than in "green" ones (0.89 %) (Fig. 2). 276 In each dataset (i.e., root prokaryote, root fungal, soil prokaryote, or soil fungal data), 277 microbial community structure varied among "green", "yellow", or "no leaf" soybean 278 individuals, although the effects of sampling sets on the community structure were much 279 stronger (Fig. 3). Even within each sampling set, spatial autocorrelations of bacterial/fungal 280 community structure were observed ( Supplementary Fig. 4). Significant relationships between 281 microbial community structure and soybean biomass were observed in the soil prokaryote and 282 soil fungal datasets but not in the root prokaryote and root fungal datasets (Table 1). 283 284

Screening of Host-state-specific OTUs 285
In the root microbiome, only an unidentified fungal OTU showed a strong preference for 286 "green" soybean individuals, while 18 bacterial and 4 fungal OTUs occurred preferentially on 287 "no leaf" host individuals ( Table 2). The list of the bacteria showing preferences for "no leaf" 288 soybean individuals included OTUs whose 16S rRNA sequences were allied to those of 289 Dyella, Herbaspirillum, Labrys, Phenylobacterium, Gemmata, Chitinophaga, Pedobacter, 290 Niastella, and Streptomyces (Table 2). The four fungal OTUs showing preferences for "no 291 leaf" hosts were unidentified basidiomycetes (Table 2).

Microbe-microbe Networks 311
The structure of microbe-microbe networks (Fig. 4) were more complicated in the soil 312 microbiome data (Fig. 4C-D) than in the root microbiome data (Fig. 4A-B). Within the 313 network representing co-abundance of microbes across root samples, the Clonostachys OTU 314 (F_0257) had a significant link with a Streptomyces OTU, while Dactylellina was absent from 315 the root microbiome network data (Fig. 4A). Within the positive co-abundance network of the 316 rhizosphere soil microbiome (Fig. 4C), the Clonostachys (F_0257) and Dactylellina (F_0163) 317 nematophagous fungal OTUs were connected with each other (Table 4). In addition, the 318 Clonostachys OTU was linked with two bacterial OTUs (Ralstonia and Rhizobiales) and 319 fungal OTUs in the genera Calonectria and Purpureocillium (Table 4). Likewise, the Pochonia, Purpureocillium (Table 4). Within the root microbiome analyzed, various taxonomic groups of bacteria 340 preferentially occurred on "no leaf" soybean samples (Table 2). Among them, the genus 341 Streptomyces is known to involve some species that suppress nematode populations, Within the soybean rhizosphere soil microbiome, diverse taxonomic groups of not only 351 bacteria but also fungi preferentially occurred around "no leaf" soybean individuals (Table 3). preferentially found in the rhizosphere of "no leaf" soybean individuals. Meanwhile, the list 357 of the fungal OTUs frequently observed in the rhizosphere of "no leaf" soybeans included 358 some fungi whose ability to suppressing nematode populations had been well documented 359 (Table 3). Clonostachys rosea, for example, has been known as a prospective biological and fungi that frequently co-occur with the nematophagous fungi (Fig. 4). In the root 373 microbiome, Clonostachys and a Streptomyces OTU showed positively correlated 374 distributions across soybean samples (Table 4). In the rhizosphere microbiome, Clonostachys 375 and Dactylellina showed significant co-abundance patterns (Table 4)  For each dataset (i.e., root prokaryote, root fungal, soil prokaryote, or soil fungal data), a 745 PEMANOVA model of community structure was constructed. The information of the 746 sampling set ("set 1" or "set 2") and the dry mass of host soybean individuals were included 747 as explanatory variables. 748 1 The prokaryote/fungal OTUs that showed strong preferences for "green" or "no leaf" soybean individuals (preference value ≥ 3) are shown. The 3 taxonomic assignment results based on the QCauto-LCA pipeline are shown with the top-hit results of NCBI BLAST searches. The OTU code 4 starting with P (P_xxxx) and F (F_xxxx) are prokaryotes and fungi, respectively.