In this study, the composition and diversity of the bacterial microbiota inhabiting perennial sorghum N778 roots at the maturing stage under cold stress conditions were examined.
Coamplification of chloroplast DNA and amplification of bacterial rDNA from sorghum root samples via the primer pair 338F and 806R
In the winter of 2019, root samples, bulk soil and rhizosphere soil of the perennial sweet sorghum cultivar N778 and a control sorghum line P607 were collected. After total metagenomic DNA was extracted, amplicon deep sequencing of the 16S rDNA V3-V4 region was carried out via Illumina MiSeq. A total of 818,955 clean tags were generated, with an average number of 45,497 clean tags per sample (Table S1), and the average number of effective taxonomic tags at 97% similarity, including or excluding chloroplast DNA (CpDNA) and mitochondrial DNA (MtDNA), was 35,484 or 28,422 per sample, respectively (Table S1). Compared to bulk soil or rhizosphere soil samples, the average number of effective tags for bacterial rDNA (V3-V4 region) per root sample of N778 and P607 was only 15,783.3 and 14,481.7, respectively (Table 1).
Table 1
Amplification of bacterial rDNA and coamplification of nontarget DNA from bulk soil, rhizosphere soil and roots of N778 and P607 via primer pair 338F-806R of the 16S rRNA gene.
Effective tags
|
N778BS (mean ± SD)
|
N778RS (mean ± SD)
|
N778RT (mean ± SD)
|
P607BS (mean ± SD)
|
P607RS
(mean ± SD)
|
P607RT
(mean ± SD)
|
Total
Effective tags
|
33076.3 ± 1875.20
|
35873.7 ± 3373.66
|
37023.7 ± 2263.21
|
33082.7 ± 1022.21
|
38566.0 ± 176.77
|
35283.0 ± 4150.47
|
CpDNA #
|
54.3 ± 45.98 a
|
18.3 ± 8.08 a
|
20887.3 ± 11928.94 b
|
156.3 ± 89.23 a
|
88.0 ± 23.26 a
|
20375.0 ± 6852.12 b
|
MtDNA
|
ND
|
ND
|
353.0 ± 159.15
|
5.7 ± 4.04
|
6.7 ± 2.89
|
426.3 ± 350.28
|
Bacterial
rDNA #
|
33022.0 ± 1870.14 a
|
35855.3 ± 3380.22 a
|
15783.3 ± 11610.61 b
|
32920.7 ± 1112.74 a
|
38471.3 ± 162.97 a
|
14481.7 ± 6340.25 b
|
Note: SD indicates the standard deviation (n = 3). N778 and P607 indicate the perennial sweet sorghum cultivar NaPSB778 and the control sorghum line P607, respectively. BS, RS, and RT indicate the bulk soil, rhizosphere soil and root samples, respectively. CpDNA, MtDNA and bacterial rDNA refer to effective tags of chloroplast DNA, mitochondrial DNA and bacterial 16S rRNA genes, respectively. One-way ANOVA followed by Tukey’s HSD post hoc test was the statistical method used here. # indicates that critical values were not monotonic for these data, and substitutions were made to ensure monotonicity; therefore, the type I error was smaller. The values in bold letters refer to a significant difference (P < 0.05) among the different groups at the maturation stage. ND refers to “not detected”. |
In contrast, the average number of effective tags for CpDNA per root sample of N778 and P607 was 20,887.3 and 20,375, respectively, which was noticeably higher than that of bulk soil or rhizosphere soil samples (Table 1). Additionally, the average number of effective tags for bacterial rDNA (V3-V4 region) per root sample of N778 or P607 was less than those for CpDNA (Table 1).
Even though the total effective tags at 97% similarity per sample were greater than 30,984 (Table S1 and S3), only 7,371 tags of each sample were retained for alpha and beta diversity analyses, and the normalized OTU table was obtained (Table S2).
Specific amplification of bacterial rDNA from sorghum root samples via the primer pair 799F and 1193R
Because the primer pair 338F and 806R coamplified significantly more chloroplast DNA from sorghum root samples, the primer pairs 799F-1392R and 799F-1193R were chosen for nested PCR amplification of the 16S rDNA V5-V7 region of bacterial communities in root samples of sorghum N778 and TP60 grown in the experimental field at the HNU campus.
A total of 1,189,466 clean tags were generated with an average number of 66,081 tags per root sample (Table S4), and the average number of effective tags at 97% similarity, including or excluding MtDNA and CpDNA, was 42,609 or 42,170 per root sample, respectively (Table S4). Coamplification of CpDNA was completely eliminated in the root samples (Table 2), although a few tags of MtDNA were coamplified per root sample by the primer pairs 799F-1392R and 799F-1193R (Table 2). Additionally, the average length of clean tags was 377 nt, and the average Q30 value was 96.17% (Table S4). Eighteen root samples of N778 and TP60 yielded a total of 2015 OTUs, which belonged to 991 species, 554 genera and 34 phyla (Table S5, S6).
Table 2
Amplification of bacterial rDNA and coamplification of nontarget DNA from root samples of N778 and the control line TP60 via primer pairs 799F-1193R of the 16S rRNA gene.
Effective tags
|
NaFARt (mean ± SD)
|
NaFPRt (mean ± SD)
|
NaMARt (mean ± SD)
|
NaMPRt (mean ± SD)
|
TPMARt
(mean ± SD)
|
TPMPRt (mean ± SD)
|
Total Effective tags
|
42732.7 ± 3035.51
|
47152.0 ± 5903.35
|
40927.7 ± 3531.01
|
46132.0 ± 393.78
|
36216.3 ± 12976.92
|
42495.7 ± 11839.02
|
CpDNA
|
ND
|
ND
|
ND
|
ND
|
ND
|
ND
|
MtDNA
|
364.0 ± 100.86 ab
|
980.7 ± 540.73 b
|
511.0 ± 282.09 ab
|
382.0 ± 81.51 ab
|
148.7 ± 53.59 a
|
250.0±
8.54 a
|
Bacterial
rDNA
|
42368.7 ± 2944.20
|
46171.3 ± 5632.94
|
40416.7 ± 3265.59
|
45750.0 ± 443.2
|
36067.7 ± 12923.50
|
42245.7 ± 11847.29
|
Note: SD indicates the standard deviation (n = 3). Na and TP indicate the perennial sweet sorghum cultivar NaPSB778 (N778) and control sorghum line TP60, respectively. F and M indicate the flowering and maturing stages, respectively. ARt and PRt refer to the latest aerial root and the primary root (taproot) samples, respectively. CpDNA, MtDNA and bacterial rDNA refer to effective tags of chloroplast DNA, mitochondrial DNA and bacterial 16S rRNA genes, respectively. One-way ANOVA followed by Tukey’s HSD post hoc test was the statistical method used here. The values in bold letters indicate a significant difference (P < 0.05) among the different groups at the maturation stage. ND refers to “not detected”. |
Alpha diversity analysis of bacterial microbiota between N778 and TP60
After excluding tags of MtDNA, effective tags of bacterial rDNA recovered from all root samples were normalized according to the minimal number of 21,145 tags from TP_1MARt (Table S4 and S7). Six alpha diversity indices, including Observed Species (Sobs), Chao1, Shannon's diversity, phylogenetic diversity, Simpsoneven, and Good's coverage at the OTU level, were chosen for pairwise group comparisons, and Welch’s t- test was used to determine whether there were significant differences between pairwise groups. (Fig. 1). The Good's coverage index of all samples exceeded 98.5% (Fig. 1f), and rarefaction curves of Good's coverage and the other five indices nearly reached the saturation plateau (Fig. S1), indicating that the sequencing depth contained enough detectable species and captured the diversity of bacterial endophyte communities.
Among the different genotypes, the alpha diversity indices of N778 during the maturing stage did not differ significantly from those of TP60, with the exception of the Sobs and phylogenetic diversity indices of the N778 aerial root sample (NaMARt), which were significantly lower than those of TP60 (TPMARt) (Fig. 1a, d). At different growth stages, the Simpsoneven index of N778 aerial root samples at the flowering stage (NaFARt) was noticeably lower than that of N778 aerial root samples (NaMARt) and N778 primary root samples (NaPARt) at the maturing stage (Fig. 1e). Additionally, the aerial root samples of N778 at the maturing stage (NaMARt) had a considerably lower phylogenetic diversity index than the primary root samples of N778 during the flowering stage (NaFPRt) (Fig. 1d). The remaining four indices of samples of N778 aerial roots or primary roots at the flowering stage did not substantially differ from those of N778 at the maturation stage (Fig. 1a-c, f).
Beta diversity analysis of bacterial microbiota between N778 and TP60
The variations in root samples between N778 and TP60 were initially investigated at the OTU level using principal coordinate analysis (PCoA). Adonis and ANOSIM statistical analysis of PCoA results based on Bray‒Curtis or weighted UniFrac (WUF) distance metrics (Table S8, S9) showed that there were no significant differences in the beta diversity of bacterial microbiota among various root samples between N778 and TP60 (Fig. 2a, 2b). In addition, there were no significant alterations in the beta diversity of the N778 root microbiota at different growth stages, although these groups appeared to be separate from the other groups (Fig. 2a, 2b). However, the statistical analysis results from Adonis and ANOSIM based on Bray‒Curtis distance metrics or WUF distance metrics showed that the difference between sample groups was more significant than that within the group (Fig. 2a, 2b).
Then, similarity of the bacterial microbiota composition of root samples between N778 and TP60 was carried out by nonmetric multidimensional scaling (NMDS) analysis. The NMDS results, followed by Adonis and ANOSIM statistical analysis, revealed no noticeable differences in the beta diversity of bacterial microbiota between N778 and TP60 root samples (Table S9), although the phylogenetic beta diversity of bacterial endophyte communities in N778 root samples at the flowering stage appeared to be distinct from those at the maturing stage (Fig. 2d). Additionally, both stress values (Fig. 2c, 2d) were less than 0.2, indicating the validity of the NMDS analysis results.
Dominant bacterial OTUs and genera in root samples of the perennial sweet sorghum N778
A total of 594 of the 2015 OTUs (Tables S5 and S6) were shared by all six groups, whereas 28 and 32 OTUs, respectively, were unique to the aerial and primary roots of N778 at the maturing stage (Fig. S2a). Additionally, the relative abundances of the top nine OTUs exceeded 50% (Fig. S2b). Moreover, 670 OTUs were shared between the aerial and primary roots of N778 at the flowering and maturing stages (Fig. 3a), whereas 688 OTUs were shared between the aerial and primary roots of N778 and TP60 during the maturing stage (Fig. 3b). At the maturation stage of N778, the top 5 OTUs in the aerial roots were OTU1902, OTU989, OTU1207, OTU1726, and OTU414 (Fig. 3c), which accounted for 14.43%, 12.29%, 8.69%, 8.22% and 4.88% of the relative abundances (Fig. S2c), respectively, while the top 5 OTUs in the primary roots were OTU1726, OTU1902, OTU989, OTU901, and OTU1262 (Fig. 3c), which accounted for 17.29%, 15.06%, 4.41%, 4.23%, and 3.74% of the relative abundances (Fig. S2d), respectively. The three most prevalent OTUs, OTU1902, OTU1726, and OTU988, were shared by both the aerial and primary roots of N778 at the maturing stage (Fig. S2c, d), which belonged to the genera Streptomyces, Pseudomonas and an unidentified genus of the family Comamonadaceae, respectively (Table S7).
Among 554 genera, the top ten genera were Streptomyces, unclassified_f__Comamonadaceae, Pseudomonas, Lechevalieria, Allorhizobium (Neorhizobium-Pararhizobium-Rhizobium), unclassified_f__Oxalobacteraceae, unclassified_f__Enterobacteriaceae, Steroidobacter, Klebsiella, and Pantoea, which accounted for 19.788%, 18.872%, 9.162%, 4.317%, 2.744%, 2.143%, 2.046%, 1.860%, 1.556% and 1.496% of all bacterial taxa with absolute abundances in all 18 root samples, respectively (Table S6, sheet of genus).
Based on the normalized OTU taxonomic table (Table S7), the top ten genera with high relative abundances were similar to the top ten genera with absolute abundances, with the exception of Klebsiella being replaced by Flavobacterium (Fig. 3d). Furthermore, at the maturing stage of N778, the top ten abundant genera in the aerial roots were Streptomyces (25.98%), unclassified_f__Comamonadaceae (14.23%), Pseudomonas (9.99%), Pantoea (8.69%), Lechevalieria (7.34%), unclassified_f__Oxalobacteraceae (3.32%), Massilia (1.49%), Steroidobacter (1.39%), Pseudarthrobacter (1.06%) and Ideonella (0.99%) (Fig. 3d and S3c), while the top ten abundant genera in the primary roots were Streptomyces (26.27%), Pseudomonas (18.14%), unclassified_f__Comamonadaceae (6.03%), unclassified_f__Oxalobacteraceae (4.06%), Pseudarthrobacter (3.74%), Flavobacterium (3.50%), Lechevalieria (2.91%), Myceligenerans (2.23%), Variovorax (1.51%), and Rhizobacter (1.51%) (Fig. 3d and S3d).
Bacterial taxa enriched in root samples of the perennial sweet sorghum N778 at the maturing stage
At the maturing stage, pairwise comparison using Student's t- test revealed that the proportions of OTU434, OTU220, OTU1589, OTU1304 and OTU80, among the top 18 distinct OTUs, were noticeably higher in the aerial root sample of N778 (NaMPRt) than in TP60 (TPMPRt) (Fig. S4a). Conversely, the proportions of OTU901, OTU1589, OTU1304, OTU499, OTU475, OTU1101, and OTU1175, among the other top 18 different OTUs, were significantly higher in the primary root sample of N778 at the maturing stage (NaMPRt) than in TPMPRt (Fig. S4b). Among these OTUs, only the relative abundance of OTU1304, which belonged to an unclassified species of Streptomyces, was significantly higher in both aerial root and primary root samples of N778 compared with TP60 at the maturing stage (Fig. S4).
The bacterial composition of aerial root samples from N778 and TP60 was then compared using ternary plot analysis at both the flowering and maturation stages (Fig. 4a). With a relative abundance of 2.91%, OTU1207 from the genus Pantoea was enriched in NaMARt compared to the other two groups (Fig. 4a). The proportions of OTU803, OTU836 and OTU1170 in NaMARt were 86.9%, 72.3% and 62.5%, respectively, which were significantly greater than those in NaFARt and TPMARt (Fig. 4a). OTU803, OTU1170 and OTU836 belong to Massilia, Pseudomonas, and an unclassified genus of the family Oxalobacteraceae, respectively, and accounted for 0.505%, 0.705%, and 1.31%, on average, of the relative abundance per group in the 3 aerial root sample groups.
When comparing the bacterial composition in primary root samples, it was found that OTU1262 and OTU836 were more prevalent in NaMPRt than in NaFPRt and TPMPRt, with proportions of 83.0% and 87.0%, respectively (Fig. 4b), and are members of the genus Pseudarthrobacter and an unclassified genus of the family Oxalobacteraceae, respectively, which accounted for 1.50% and 0.885% of the average relative abundance in the three primary root sample groups, respectively.
Then, three group comparisons via Kruskal–Wallis H test were performed, and the relative abundance of OTU836 was found to be significantly higher in the primary root sample of N778 at the maturing stage (NaMPRt) than that in NaFPRt and TPMPRt (Fig. S5b). However, OTU803, OTU1170, OTU1207 and OTU1262 were found to be not significantly different among the three group comparisons (Fig. S5). In addition, the proportions of OTU434 and OTU1304 were noticeably higher in the aerial root sample of N778 at the maturing stage (NaMARt) than those in NaFARt and TPMARt (Fig. S5a).
Finally, six group comparisons via Kruskal–Wallis H test were carried out. The proportions of OTU1262, OTU836 and OTU1304 were noticeably higher in the primary root sample of N778 at the maturing stage (NaMPRt) than those in the other five groups (Fig. 5). Additionally, compared to the other four groups, the proportion of OTU434 was significantly higher in the aerial and primary root samples of N778 at the maturing stage (NaMARt, NaMPRt) (Fig. 5).
Overall, the root sample of N778 was significantly enriched with OTU1262, OTU836, OTU434 and OTU1304 at the maturing stage.
Phylogenetic analysis of OTU1262 particularly enriched in the primary root sample of the perennial sweet sorghum N778 at the maturing stage
According to a normalized OTU table, OTU1262 was an unidentified species of the genus Pseudarthrobacter with a relative abundance of approximately 1.11% in all root samples (Table S7). To predict the potential function of OTU1262, FastTree was used for phylogenetic analysis, and the phylogenetic tree was built in R (v3.3.1) using the maximum likelihood (ML) approach. OTU1262 was found to be strongly associated with Pseudarthrobacter psychrotolerans YJ56, a known psychrotolerant bacterial species (Fig. 6).