FAD2 RNAi transgenic B. napus L. was obtained through RNAi technology (Zhou et al., 2021b), and multiple generations of GM plants were obtained through stable inheritance. The phenotype of transgenic B. napus L. analyzed in this study and the specific planting and seed treatment procedures can be found in a previous study (Zhou et al., 2021b). T3, T7, and WT plants were used in this study (Zhou et al., 2021b). WT is wild-type B. napus L., and T3 and T7 are different generations of transgenic B. napus L., corresponding to the third generation and seventh generation, respectively. The two genotypes were provided by Jiangsu Academy of Agricultural Sciences (Zhou et al., 2021). Rapeseed plants were planted in a paddy field at Hunan Agricultural University, and the latitude and longitude were 28.18°CN and 113.08°CE, respectively. The sampling time was the summer season in 2018. We used a five-point sampling method. The soils for the experiments were collected at five points in each plot from the rapeseed field (1–30 cm depth) and mixed into one sample. The loose soil of the root system around the surface was shaken off the roots, and the soil that adhered strongly to the roots was carefully brushed from the roots and kept as rhizosphere soil. Stones and debris were removed manually and combined into one sample. Four composite soil samples were collected for each treatment, totalling 12 soil samples. Three treatments were designed, and each experimental treatment was replicated four times and repeated four times. Samples were stored at −80°C until pretreatment. Among the samples, Samples 1 to 4 were wild-type rapeseed (WT), Samples 5 to 8 were the third generation of FAD2 RNAi transgenic B. napus L. (T3), and Samples 9 to 12 were the seventh generation of FAD2 RNAi transgenic B. napus L. (T7).

Microbial community genomic DNA was extracted from the collected samples using the E.Z.N.A.^® soil DNA Kit (Omega Bio-Tek, Norcross, GA, USA) according to the manufacturer’s instructions. The hypervariable region ITS of the fungal gene was amplified with the primer pairs ITS3F (5’-GCATCGATGAAGAACGCAGC-3’) and ITS4R (5’-TCCTCCGCTTATTGATATGC-3’), and the hypervariable region V3–V4 of the bacterial 16S rRNA gene was amplified with the primer pairs 338F (5’-ACTCCTACGGGAGGCAGCAG-3’) and 806R (5’-GGACTACHVGGGTWTCTAAT-3’) (Toju et al., 2012; Liu et al., 2016). PCR reactions of the 16S rRNA gene and ITS gene, containing 25 μl of 2× Premix Taq (Takara Biotechnology, Dalian Co. Ltd., China), 1 μl of each primer (10 μM), and 3 μl of DNA (20 ng/μl) template in a volume of 50 µl, were amplified by thermocycling: 5 min at 94°C for initialization; 30 cycles of 30 s denaturation at 94°C, 30 s annealing at 52°C, and 30 s extension at 72°C; followed by 10 min final elongation at 72°C. The PCR instrument was Bio-Rad S1000 (Bio-Rad Laboratory, CA, USA). The PCR product was extracted from a 1% agarose gel, purified using the E.Z.N.A. Gel Extraction Kit (Omega, USA), according to the manufacturer’s instructions and quantified using a Quantus™ Fluorometer (Promega, USA) (Zuo et al., 2018). Sequencing libraries were generated using NEBNext^® Ultra™ II DNA Library Prep Kit for Illumina^® (New England Biolabs, MA, USA) and index codes were added. The library quality was assessed on the Qubit@ 2.0 Fluorometer (Thermo Fisher Scientific, MA, USA). Lastly, purified amplicons were pooled in equimolar amounts and paired-end sequenced on an Illumina Nova6000 platform and 250-bp paired-end reads were generated (Guangdong Magigene Biotechnology Co., Ltd. Guangzhou, China).

Fifty milligrams of soil sample was accurately weighed, and the metabolites were extracted using a 400-µl methanol:water (4:1, v/v) solution. The mixture was allowed to settle at −20°C and treated by a high-throughput tissue crusher (Wonbio-96c, Shanghai Wanbo Biotechnology Co., Ltd) at 50 Hz for 6 min, followed by vortexing for 30 s and ultrasound at 40 kHz for 30 min at 5°C. The samples were placed at −20°C for 30 min to precipitate proteins after centrifugation at 4°C for 15 min. The supernatant was transferred to sample vials for LC−MS/MS analysis (Wang and Xie, 2020).

Multivariate statistical analysis was performed by using the ropls R package (Version 1.6.2, http://bioconductor.org/packages/release/bioc/html/ropls.html). Principal component analysis (PCA) using an unsupervised method was applied to obtain an overview of the metabolic data. All metabolite variables were scaled to unit variances prior to conducting the PCA. Partial least squares discriminate analysis (PLS-DA) was used for statistical analysis. All metabolite variables were scaled to Pareto scaling prior to conducting the PLS-DA. The model validity was evaluated from model parameters R ² and Q ², which provide information for interpretability and predictability, respectively, and avoid the risk of overfitting (Wang et al., 2021). Variable importance in the projection (VIP) was calculated in the PLS-DA model, and p-values were estimated with paired Student’s t-test on single-dimensional statistical analysis. The variables with significant differences of p < 0.05 and VIP values > 1 were defined as key metabolites.

The raw 16S rRNA gene-sequencing reads and raw ITS gene-sequencing reads were demultiplexed, quality-filtered by fastp Version 0.20.0, and merged by FLASH Version 1.2.7. Operational taxonomic units (OTUs) with a 97% similarity cutoff were clustered using UPARSE Version 7.1 (Edgar, 2013), and chimeric sequences were identified and removed. QIIME software and R language tools were used to analyze the community structure of samples at different classification levels. The taxonomy of each OTU representative sequence was analyzed by RDP Classifier Version 2.2 against the 16S rRNA database (silva132/16s_bacteria) and ITS database (unite7.2/its_fungi) using a confidence threshold of 0.7 (Wang et al., 2007).

Statistical analyses were performed by SPSS 19.0 and summarized as the means ± standard errors (SE). The level of significance of various treatments was set at p < 0.05 using one-way analysis of variance (ANOVA).

The number of common and specific OTUs of groups is displayed in Figure 1 to show the overlaps of OTUs among the rhizosphere soil from WT, T3, and T7 plants. The OTUs of bacteria and fungi totalled 4,526 and 1,453 in the rhizosphere soil of WT, respectively. The OTUs of bacteria and fungi totalled 4,631 and 1,624 in the rhizosphere soil of T3, respectively, and the OTUs of bacteria and fungi totalled 4,540 and 1,335 in the rhizosphere soil of T7, respectively. Compared with that of the WT, the number of genera in the root soil of T7 decreased relatively, while the number of genera in T3 increased relatively. The number of common OTUs of bacteria and fungi in root soil was 4,004 among WT, T3, and T7, representing 66.97%, 64.01%, and 68.15% of the total number of OTUs in WT, T3, and T7, respectively. The composition of the microbial communities of WT, T3, and T7 is shown in Figure 2 . There was rarely any difference in rhizosphere microbial abundance between wild-type and transgenic plants. Bacteroidia, Gammaproteobacteria, Ascomycota, and Sordariomymycetes were found in high proportions in each sample. Moreover, Ascomycota accounted for a very high proportion of the fungal communities. In addition, the abundance of Bacteroidia in sample No. 7 of T3 (T3_7) was as high as 44%, and the abundance of Sordariomymycetes in this sample was also high. There was no difference in the composition of microbial species at the class level in the samples, but the abundance of microbial species was slightly different. Zhou et al. (2021a) found that Bt corn residues had no direct effect on soil bacterial communities compared with the decomposition time and environment. The functional diversity and metabolic activity of the microbial community indicated that there was no significant difference between Bt maize IE09S034 and the control group. A study found differences in microbial community structures between Bt and non-Bt maize fields, and there were also variations between the chemical and biochemical properties of rhizosphere soils under Bt and non-Bt maize cultivation; they concluded that these differences could be related to agricultural practices and varieties (van Wyk et al., 2017).

In Figure 3 , the dominant species with high abundance at the genus level of rhizosphere bacteria in wild-type and transgenic plant were Proteobacteria and Bacteroidia. At the genus level of rhizosphere fungi in wild-type and transgenic plant, the dominant species was Ascomycota. They had a high percentage in all samples. All other soil samples of T3, T7, and WT at the genus level showed very small differences in microbial abundance. As shown in Figure 3 , the abundance of Proteobacteria, Bacteroidetes, Acidobacteria, Verrucomicrobia, Chloroflexi, and Ascomycota was relatively high at the genus level. Some studies have used Proteobacteria and Acidobacteria as indicators of land use change and soil pollution (Schneider et al., 2017; Kim et al., 2021). Members of the class Bacteroidia are Gram-negative, non-spore-forming, mostly anaerobic bacteria that decompose organic substances, and they thrive in diverse environments, such as soils and subsurface sediments (Garcia-Lopez et al., 2019; Podosokorskaya et al., 2020). Ascomycota is by far the largest group in the fungal kingdom. Ecologically important mutualistic associations, such as mycorrhizae and lichens, have evolved in this species and are regarded as pivotal innovations that supported the evolution of land plants (Beimforde et al., 2014). These results showed that the soil microbial community structure was stable, and the general microbial community structure of rhizosphere soil was not changed by the transgenic plants. Zou et al. found that 5-year-old transgenic Poplar 741 could not cause ecological risk and could not affect the microbial community structure or functional diversity (Zou et al., 2018). Fan et al. (2020) analyzed the effects of transgenic poplar on related bacterial communities from various aspects, including bacterial OTU composition, alpha and beta diversity, gene function, and differences among groups. Overall, transgenic poplar did not significantly affect endophytic or soil bacteria. Our results on the structural composition and diversity of the microbial community are consistent with these studies. Therefore, we preliminarily concluded that transgenic B. napus L. had no harmful effects on microbial communities in the environment.

Alpha diversity refers to the diversity in a specific region or ecosystem. The commonly used indicators to characterize alpha diversity are Chao, ACE, and Shannon. The ACE index, Chao index, and Shannon index were obtained at a 97% similarity level using Mothur software. The Shannon index measures the average degree of uncertainty in predicting to what species a randomly selected individual from a set of S species and N individuals will belong to, and the value increases as the number of species increases and individuals become evenly distributed among species (Pylro et al., 2014). The ACE index and Chao1 index reflect the number of species in the community. Figure 4 shows the alpha diversity analysis of the microbial community and significance difference test. The results of some samples showed fluctuations but few changes within a certain range, and the diversity indicators of WT, T3, and T7 were close to each other. Compared with transgenic plants and wild-type plants, the ACE index value and Chao index showed little difference between transgenic plants and wild-type plants, indicating that the transgenic plants did not cause obvious changes in the richness of the bacterial community. The Chao index of the fungal community from T7 was slightly smaller than that of WT and T3, indicating that the fungal community had higher richness in WT and T3. By comparing the two indicators, no obvious difference in bacterial community diversity between transgenic plants and wild-type plants was found, while the fungal community diversity of T7 was slightly smaller than that of the WT and T3. Overall, there was no significant difference in the alpha diversity of the fungal community and bacterial community.

Zhang et al. (2015) detected no OsrHSA protein in the roots of OsrHSA transgenic rice and concluded that OsrHSA transgenic rice and the rHSA protein produced by transgenic rice did not change the functional diversity of rhizosphere microbial communities. Another study also confirmed that microbial community diversity was not significantly influenced by the presence of GM plants (Riglietti et al., 2006). Our results were similar to these studies. In summary, through RNAi technology, B. napus L. had no obvious effect on the richness and diversity of the rhizosphere bacterial community and fungal community.

As shown in Figure 5 , the heatmap graph color represents the size and proportion of data in a two-dimensional matrix or table, thus showing the composition and structure of the microbiome and reflecting the similarity and differences of the community composition of different groups or different samples at each classification level. In the heatmap of bacterial and fungal abundance, the abundance of each microbial species in WT, T3, and T7 was similar. Flavobacterium in T3 was significantly more than that in WT and T7, and Ascomycota was abundant in WT, T3, and T7, which was consistent with the above analysis. The dominant genus and proportion between transgenic rapeseed plants and wild plants were consistent with the structural analysis at the genus level. A study found that compared with the parental line, the transgenic strain had no obvious effect, and for all evaluated microbial communities, soil type and the year of planting maize were the dominant factors affecting their structures, and no difference in the total number of bacteria between the rhizospheres of GM and parental plant lines was observed (Saxena and Stotzky, 2001). Chun et al. (2012) found no difference between transgenic rice and its parental nontransgenic counterpart (cultivar Dongjin) in the diversity index and community composition of the microbial community. In contrast, this change in the microbial community depended strongly on the growth stage and year. Therefore, they did not observe the adverse effects of these transgenic rice crops on the microbial community in paddy soils.

PLS-DA can maximize the intergroup distinction and is conducive to finding differential metabolites, and it belongs to beta diversity approaches. PLS-DA obtained two principal components: the contribution rate of Principal Component 1 was 56.40%, the contribution rate of Principal Component 2 was 13.70%, R ²X = 0, R ²Y = 0.9721, and Q ² = 0.6466. PLS-DA was verified for alignment (n = 200, i.e., 200 alignment experiments). In the model verification ( Figure 6B ), the horizontal line corresponds to R ²Y and Q ² of the original model, and the red dot and blue dot represent R ²Y’ and Q ²’ of the model after replacement of y, respectively. R ²Y’ and Q ²’ were both smaller than R ²Y and Q ² of the original model; i.e., the corresponding points did not exceed the corresponding lines, which indicated that the model was meaningful and that its differential metabolites could be screened according to VIP value analysis. According to the PLS-DA model ( Figure 6A ), 726 metabolome data points were analyzed. The samples of RNAi-T3 were distributed on the left side of the confidence interval, and the samples of WT were distributed on the right side of the confidence interval. The distinguishing effects of the two samples were abundant and obvious.

The root exudates of rapeseed mainly included alcohols, hydrocarbons, acids, and esters, and the contents of alcohols and hydrocarbons were relatively high. Hydrocarbons, esters, and acids or secondary metabolites have certain chemical influences, including the inhibition or stimulation of seed germination in other crops and an influence on microorganisms in the soil, in keeping with the results of Yang (2006).

Comparing the multiple changes in quantitative information for metabolic components in WT and RNAi samples, log2 treatment was carried out on the difference multiple. The top 20 differentially expressed metabolic components with changes are shown in Figure 7 . Compared with WT plants, five types of amino acids and their derivatives [proline, L-(+)-arginine, N-(3-indolylacetyl)-L-alanine, Nα-acetyl-L-glutamine, and L-phenylalanine], one aldehyde and its derivatives (4-hydroxy-3,5-diisopropylbenzaldehyde), one amide and its derivatives (N,N-dimethylformamide), one chalcone (4,2’,4’,6’-tetrahydroxychalcone), one alcohol and its derivatives (caffeyl alcohol and sclareol), one ketone and its derivatives (α-lonone), one glycoside and its derivatives (deoxyguanosine), one pyrimidine and its derivatives (thymidine), and two others [acid orange 20 and MAG(18:1) isomer1] increased in relative content in RNAi-T plants. However, one guanidine and its derivatives (1-dimethylbiguanide), one polyether and its derivatives (salicin), one alcohol and its derivatives (chrysoeriol), two glycosides and their derivatives [O-glucuronic acid-O-hexoside and delphinidin 3-O-rutinoside (Tulipanin)], and one aldehyde and its derivatives (1 p-coumaricaldehyde) were reduced in relative content in RNAi-T plants.

Based on the PLS-DA results, the VIP of the PLS-DA model was analyzed from the obtained multivariate analysis. A total of 88 upregulated differential metabolites were detected, namely, 19 amino acids and their derivatives; 15 organic acids and their derivatives; 14 lipids; 6 nucleotides and their derivatives; 5 flavanones; 4 alkaloids; 3 phenylpropanoids, flavonoids, and flavonols; 2 alcohols, polyphenols, and procyanidins; 1 phenolamine, terpenoid, and vitamin and their derivatives; 1 indole and its derivatives; and 6 other metabolites. A total of 24 downregulated differential metabolites were detected, namely, 10 phenylpropanoids, 3 lipids, 3 flavonols, 2 polyphenols, 2 organic acids and their derivatives, 1 amino acid and its derivatives, 1 phenolamine, 1 nucleotide and its derivatives, and 1 vitamin. A volcano map was used to visualize the differential data for observation and analysis. It was found that there were both upregulated and downregulated substances with significant differences. Metabolites with upregulated expression were more abundant than metabolites with downregulated expression. However, there was no significant change in the expression of most metabolites.

All metabolites were analyzed by t-test and two-tailed test, and the differences in metabolites were plotted ( Figure 8 ). A point represented a metabolite, and a horizontal coordinate represented a pair of values. As shown in Figure 8 , only a few metabolites showed significant differences. The expression levels of more differential metabolites increased in GM rapeseed than in wild-type rapeseed, and the expression levels of some metabolites were reduced. Specifically, there were six metabolites with significantly upregulated and downregulated changes. The main metabolites significantly upregulated were quinic acid O-glucuronic acid, N-(3-indolylacetyl)-L-alanine, L-tyramine, and phenylacetyl-L-glutamine. There were two significantly downregulated differential metabolites, homovanilloyl quinic acid and N-caffeoyl agmatine. It was necessary to demonstrate that the presence of these metabolites and expressed changes were not harmful to the environment based on relevant studies. Alanine is an amino acid with a neutral charge that can be sorbed into soil particles by (1) cation exchange, (2) anion exchange, or (3) ligand exchange (Apostel et al., 2017). Alanine is also the richest amino acid in soluble organic substances (Fischer et al., 2007). Therefore, this kind of metabolite is common in the environment. In addition, a study found that feruloyl tyramine, feruloyl tryptamine, and feruloyl dopamine were effective against S. aureus 209 and S. pyogenes, and the MIC was between 190 and 372 μM (Georgiev et al., 2012). Glutamine affected the level of other amino acids utilized by microorganisms, and the addition of glutamine reduced (P<0.05) the use of alanine and glycine by E. coli while stimulating (P<0.05) the use of threonine in the bacteria (Dai et al., 2013). Aromatic compounds are produced by microorganisms in the synthesis of the aromatic components of their own cellular materials. In addition to these anabolic syntheses, some microorganisms produce aromatic metabolites in the process of oxidative assimilation of nonaromatic materials. Quinic acid (1,3,4,5-tetrahydroxycyclohexanecarboxylic acid) was attacked by Micrococcus chinicus, resulting in protocatechuic acid (Rogoff, 1958). In Neurospora crassa and Aspergillus nidulans, the exogenous supply of quinic acid as a growth carbon source led to the production of three enzymes that were essential for the catabolism of quinic acid into protocatechuic acid (Watts et al., 2002). Such acids can be used by microorganisms and do not harm the environment. Polyamines play a variety of roles in every living organism, including physiological responses to pathogens in plants (Jiménez-Bremont et al., 2014). In fungi, they are involved in metabolism, regulation (Valdés-Santiago et al., 2012), and stress-coping functions (Valdés-Santiago and Ruiz-Herrera, 2014), and they also do not cause harm to the environment. In conclusion, the metabolites with variable expression levels were not harmful to the environment.