2.1 Surface sterilization efficiency
The results showed that no colony was observed in NA medium after a certain period of cultivation, it reflected that the method of surface sterilization was effective, and the surface-sterilized samples can be used to subsequent tests.
2.2 Analysis of sequencing data and alpha diversity
A total of 634,016 and 781,749 effective tags were obtained for 16S sequencing of the P. sibirica Linnaeus and P. crotalarioides Buchanan-Hamilton ex Candolle tissues samples, respectively, after filtering and removing chimeric particles, with the library coverage of the samples being higher than 0.994, which indicated that the sequencing data can fully reflect the structure of endophytic bacteria community of the samples (Table 1). With the increasement of sequencing effort, the rarefaction curves of the samples tended to be stable based on the number of observed species, which indicated that the amount of sequencing data gradually tended to be reasonable (Fig. 3S).
In all libraries, 130 and 222 OTUs were shared in different tissue samples of the two medicinal plants, respectively. The numbers of OTUs that occurred only in root (A1), stem (A2), and leaf (A3) samples of P. sibirica Linnaeus were 320, 113 and 107, respectively, while the numbers of OTUs that occurred only in root, stem and leaf samples P. crotalarioides Buchanan-Hamilton ex Candolle were 402 (A4), 196 (A5), and 63 (A6), respectively (Fig. 1a, b). Fig. 1c showed that 318 OTUs were exclusively recoverd from P. sibirica Linnaeus, 629 OTUs were exclusively recoverd from P. crotalarioides Buchanan-Hamilton ex Candolle, and 516 OTUs were shared both. In addtion, flower plot showed that the specific OTUs of tissue samples were different in the two medicinal plants (Fig. 1d). The OTUs of root (A4) and stem samples (A5) of P. crotalarioides Buchanan-Hamilton ex Candolle were 1.7 times and 3.7 times more than that of P. sibirica Linnaeus. But in leaf samples, OTUs of P. sibirica Linnaeus was 1.6 times than that of P. crotalarioides Buchanan-Hamilton ex Candolle.
Information on the richness, homogeneity and diversity of species in tissue samples can be obtained through alpha diversity analysis, results showed that, according to the diversity index of Chao1, the relative abundance of endophytic bacteria in different tissues of the two medicinal plants, in descending order, was root>stem>leaf. By shannon's diversity index, endophytic bacteria diversity in different tissues of P. sibirica Linnaeus, in descending order, was stem (A2)>root(A1)>leaf(A3), while that in P. crotalarioides Buchanan-Hamilton ex Candoll was leaf (A6) >stem (A5) >root (A4) (Table 1). The data were subjected to Kruskal-Wallis rank sum test which showed significant differences (P < 0.05).
Table 1 Community diversity of endophytic bacteria of different samples
Medicinal Plants
|
Sample
|
Effective tags
|
Shannon
|
Chao1
|
Goods_coverage
|
P. sibirica Linnaeus
|
A1 (Root)
|
217,667
|
4.00
|
331.06
|
0.99
|
A2 (Stem)
|
210,099
|
5.74
|
194.27
|
0.96
|
A3 (Leaf)
|
206,250
|
2.76
|
153.87
|
0.98
|
P. crotalarioides Buchanan-Hamilton ex Candolle
|
A4 (Root)
|
266,050
|
5.30
|
431.27
|
1.00
|
A5 (Stem)
|
257,527
|
5.71
|
353.00
|
1.00
|
A6 (Leaf)
|
258,172
|
6.07
|
224.13
|
0.98
|
2.3 Beta diversity analysis
To understand the degree of dissimilarity among tissue samples, beta diversity was employed to determine the extent of microbial ecosystem differentiation. PCoA clustering analysis based on OTUs features indicated that different tissue samples formed distinguishable clusters in roots (A1), stem (A2) and leaf (A3) samples of P. sibirica Linnaeus (Fig. 2a), and that clusters in root (A3), stem (A4) and leaf (A5) samples of of P. crotalarioides Buchanan-Hamilton ex Candolle (Fig. 2b).
The phylogenetic tree was constructed by selecting the top20 sequences of endophytic bacteria in total relative abundance with the unweighted pair group method with arithmetic mean (UPGMA) to investigate the correlations between different tissues (Fig. 3). Different clusters were found in the UPGMA tree at OTU level. Fig. 3 clarified that the bacterial compositions in roots and stems of P. sibirica Linnaeus were more similar versus leaves (a), and that of stems and leaves of P. crotalarioides Buchanan-Hamilton ex Candolle were more similar versus roots (b).
2.4 Composition and difference analysis
OTUs were assigned into 23 phyla and 226 genera of endophytic bacteria in P. sibirica Linnaeus, and 23 phyla and 275 genera of that in P. crotalarioides Buchanan-Hamilton ex Candolle (Data 1S). The endocytic bacteria of top10 relative abundance at a phylum level in P. sibirica Linnaeus were revealed in Fig. 4a. The dominant phylum across all samples was Proteobacteria, with relative abundances ranging from 73.57% to 96.38%, then was Acinetobacteria. At a genus level, Burkholderia-Caballeronia-Paraburkholderia was dominant in root samples (47.52%, A1), Acinetobacter was dominant in stem samples except for the unassigned (11.89%, A2), and Candidatus_Portiera was dominant in leaf samples (82.15%, A3) (Fig. 4b).
The endocytic bacteria of top10 relative abundance at a phylum level in P. crotalarioides Buchanan-Hamilton ex Candolle was revealed in Fig. 4c. The dominant phylum across all samples was also Proteobacteria, with relative abundances ranging from 48.70% to 74.19%, then were Acinetobacteria and Bacteroidetes. At a genus level, Amycolatopsis was dominant in root samples except for the unassigned (9.61%, A4), Methylobacterium was dominant in stem samples (21.90%, A5) and leaf (7.33%, A6) samples (Fig. 4d).
LDA Effect Size(LEfSe) analyse was used to find the biomarker of each group based on homogeneous OTU table, as visualized in Fig. 5. Firstly, the significant difference in abundance included nine commities in roots (A1) of P. sibirica Linnaeus, and five in nine communities contributed the differentiation, including orders Betaproteobacteriales, Xanthomonadales, KF_JG30_C25 and Diplorickettsiales in class Gammaproteobacteria, and family Micropepsaceae in order Micropepsales (Fig. 5a, b; Fig. 4S, b). In stem samples, there was only one commity significantly different in abundance, which was family Intrasporangiaceae in order Micrococcales (Fig. 5a, b; Fig. 4S, c), and there were three communities contributed the differentiation in abundance in leaves (A3), including genus Candidatus_Portiera in family Halomonadaceae, family AKYH767 in order Sphingobacteriales and family Peptostreptococcaceae in order Clostridiales (Fig. 5a, b; Fig. 4S a).
Secondly, in P. crotalarioides Buchanan-Hamilton ex Candolle, the significant difference in abundance included six commities in roots (A4). Two communities contributed the differentiation in abundance, including order Xanthomonadales in class Gammaproteobacteria and Pseudonocardiales in class Actinobacteria (Fig. 5c, d; Fig. 5S, a). In stem sampls, family Beijerinckiaceae (especially genus Beijerinckiaceae) in order Rhizobiales was significant different in abundance (Fig. 5c, d; Fig. 5S, b). In leaves (A6), order Oceanospirillales in class Gammaproteobacteria, genus Aquabacterium in family Burkholderiaceae and family Geodermatophilaceae (especially genus Geodermatophilus) in order Frankiales were significantly different in abundance (Fig. 5c, d; Fig. 5S, c).
2.5 Analysis of endophytic bacteria correlations
Alpha diversity analysis showed that the relative abundance and diversity of endophytic bacteria in different tissues were different. Whether the endophytic bacteria had correlations with each other? Some of them may play important roles in the endophytic microecology? We analyzed the correlations of endophytic bacteria of top 20 in relative abundance in P. sibirica Linnaeus and P. crotalarioides Buchanan-Hamilton ex Candolle, respectively.
In P. sibirica Linnaeus, results showed that there were three phyla had strong correlations with others (Fig. 6a). Phylum Dependentiae had positive correlations with phyla Chlamydiae, Patescibacteria and Spirochaetes, the correlation coefficients were 0.93, 0.80 and 0.83, respectively. Phylum Chlamydiae had positive correlations with phyla Dependentiae, Patescibacteria and Spirochaetes, the correlation coefficients were 0.93, 0.94 and 0.86, respectively. While phylum Proteobacteria had negative correlations with phyla Deinococcus-Thermus, Bacteroidetes and Actinobacteria, the correlation coefficients were negative of 0.81, 0.90 and 0.90, respectively. At a genus level, all correlations were positive, mainly occurred between phyla Proteobacteria and Actinobacteria (Fig. 6b). Although phylum Proteobacteria had negative correlations with phylum Actinobacteria (Fig. 6a), genus Microbacterium and Cutibacterium in phylum Actinobacteria had positive correlations with genus Novosphingobium in phylum Proteobacteria, where the coefficients were 0.95 and 0.96, and two of them also had correlations with genus Cupriavidus in phylum Proteobacteria, where coefficients were 0.82 and 0.90. The two genera also had a positive correlation with each other, where the coefficient was 0.94. Additionally, Genus Novosphingobium had correlations with other three genera in phylum Proteobacteria (Fig. 6b). In all, genera Microbacterium, Cutibacterium and Novosphingobium played important roles in endophytic microecology of P. sibirica Linnaeus.
In P. crotalarioides Buchanan-Hamilton ex Candolle, phylum Proteobacteria had four negative correlations with phyla Actinobacteria, Patescibacteria, Planctomycetes and Verrucomicrobia, where the coefficients were 0.95, 0.80, 0.81 and 0.82. Each of the four phyla of Patescibacteria, Planctomycetes, Chloroflexi and Armatimonadetes had correlations with each other (Fig. 6c). At genus level, Curtobacterium in phylum Actinobacteria had strong correlations with three bacterial commities in phylum Proteobacteria. And genera Acinetobacter, Candidatus Hamiltonella and Candidatus Portiera in phylum Proteobacteria had correlations with each other (Fig. 6d). In all, relative to P. sibirica Linnaeus, the bacterial commities in P. crotalarioides Buchanan-Hamilton ex Candolle were less correlated at genus level.
2.6 Functional prediction
PICRUSt2 was used to predict the function of the endophytic bacteria based on KEGG (Kyoto Encyclopedia of Genes and Genomes). Results showed that six functions were predicted, and metabolism was the most abundance function of endophytic bacteria in different tissues of the two medicinal plants, where the relative abundances were about 78.4%~84.2%. The metabolism abundance of endophytic bacteria in roots (A4) of P. crotalarioides Buchanan-Hamilton ex Candolle was the most, and that in leaves (A3) of P. sibirica Linnaeus was the lowest (Fig. 7).
To find out what pathways were mainly included in metabolism of endophytic bacteria and which tisssue was the most abundance, clustering analysis of pathway with top30 abundance was performed. Results showed that the most abundant metabolic pathway in roots (A1) included three pathways of terpenoids and polyketides, biosynthesis of other secondary metabolites, glycan biosynthesis and metabolism, and that five pathways in stems (A2) were amino acids metabolism, xenobiotics biodegradation and metabolism, metabolism of other amino acids, lipid metabolism, carbohydrate metabolism, and then that two pathways in leaves (A3) included energy metabolism, the metabolism of cofactors and vitamins (Fig. 8a).
In P. crotalarioides Buchanan-Hamilton ex Candolle, the most abundant metabolic pathway in roots (A4) included seven pathways of metabolism of terpenoids and polyketides, lipid metabolism, glycan biosynthesis and metabolism, biosynthesis of other secondary metabolites, amino acid metabolism, carbohydrate metabolism, xenobiotics biodegradation and metabolism, and that two pathways in stems (A5) were metabolism of other amino acids, energy metabolism, and that two pathways in leaves (A6) were metabolism of cofactors and vitamins, nucleotide metabolism (Fig. 8b). KEGG pathways with relative abundance≥1% were selected for variance analysis by using PICRUSt, and result showed that ten metabolic pathways of endophytic bacteria with the most significant differences in abundance in the two medicinal plants. Carbohydrate metabolism, amino acid metabolism, metabolism of cofactors and vitamins were the three most abundant metabolic pathways of endophytic bacteria in the two medicinal plants (Fig. 6S).