Rhizosphere microbiota can delay the onset of flowering of Wt Arabidopsis
Multiple generations of experimental adaptation/acclimation could be used to observe microbially mediated mechanisms of plant growth and reproduction [24,25,26]. To test whether the multi-generations of rhizosphere microbiota can induce earlier or delay flowering time, we measured the phenotypic parameters of Arabidopsis plants growing in soil for three generations (G1, G2, or G3) inoculated with different soil microbiomes (Fig. 1a). The phenotypic parameters of Wt Arabidopsis did not change significantly in plants treated with microbiota isolated from the roots of wild-type plants (Wt-M treatment) compared to the corresponding control (growing in sterilized soil without addition of soil microbiomes) during the first two generations (Additional file 1: Table S1). This indicates that plants in sterilized soil could grow as well as those grown in sterilized soil inoculated with a live microbial slurry during the first two generations. Wt-M rhizosphere microbiota significantly affected flowering and reproduction by the third generation. Flowering time in the G3-Wt-M group was significantly delayed, by approximately 3 days, and silique number increased significantly.
The richness and diversity of the rhizosphere microorganisms tended to decrease in parallel with the changes in Arabidopsis physiology over the three generations. Species richness and diversity based on four indices (Chao1, abundance-based coverage (ACE), number of species, and the Shannon index) significantly decreased after the two and three generations (Fig. 1b and Additional file 1: Figure S1). The relative abundances of Proteobacteria and Acidobacteria, the dominant bacterial phyla in the rhizosphere, decreased by generations 2 and 3 relative to generation 1. In contrast, Bacteroidetes and cyanobacterial abundances increased by generation 3 (Fig. 1c and Additional file 2). It was demonstrated that some early (or late)-flowering plant could be associated to specific soil microorganism families [25] and that plant microbiota could be associated with changes in different plant growth phases [27]. The results suggest that the selected enriched microbes during the three generations play crucial roles in modulating plant flowering time.
Rhizosphere microbiota advanced the flowering time of the pgr5 mutant
To better understand the relationships between microbiota, exudates, and flowering time, we used an Arabidopsis mutant (for the PGR5 gene that encodes a novel thylakoid membrane protein) that grows as well as Wt plants in the vegetative phase [28, 29]. The pgr5 mutant is deficient in antimycin A-sensitive cyclic flow from ferredoxin to plastoquinone, which is one of the most crucial physiological processes for efficient photosynthesis [28]. Because of defects in photosynthesis, the pgr5 mutant produces different exudates from the Wt Arabidopsis. The phenotype of the pgr5 mutant was unchanged in generations 1 and 2 in the group treated with pgr5 microbiota (pgr5-M) compared to the corresponding control (without the addition of soil microbiomes), as for the Wt Arabidopsis. The flowering time of the pgr5-M-treated group, however, was nearly 4 days earlier by generation 3, in contrast to Wt, and the silique number was significantly lower (Additional file 1: Table S1). The changes in plant reproduction induced by the microbial manipulations were probably not driven by rhizosphere diversity, because the Shannon and richness indices of the pgr5-M- and Wt-M-treated groups did not generally differ over the three generations (Additional file 1: Figure S1). The ACE data, though, differed after one generation (Fig. 1b). The relative abundances of the phyla differed only marginally between the two Arabidopsis lines in each generation (Fig. 1c).
Rare rhizosphere microbes potentially affected flowering time
The abundances of microbial phyla were relatively constant (see above), but abundances between the WM- (microbiota from the third-generation Wt cultures) and the PM (microbiota from the third-generation pgr5 cultures)-treated groups differed more at lower taxonomic levels. The relative abundances of 77 rhizosphere genera differed significantly between the WM and PM treatment by generation 3, by at least a factor of two. A total of 41 genera were enriched in the WM treatment relative to the PM treatment, and 36 genera were enriched in the PM treatment (Additional file 1: Table S2 and Additional file 3).
Most of the enriched rhizosphere microorganisms were initially rare (relative abundance < 1%), such as Emticicia, Methylobacterium, and Filimonas, suggesting that rare rhizosphere microbes might play a role in modulating Arabidopsis flowering. Rare microbes can be involved in soil biochemical processes and as active modulators of plant growth and resistance to pathogens [30]. The enriched microbes in the WM treatment mostly have key roles in rhizosphere N regeneration or in maintaining plant growth (Additional file 1: Table S3) [16, 31,32,33,34]. Indeed, Bacillus, enriched in the WM treatment, can potentially contribute to soil N fixation [35]. Potential denitrifying organisms such as Stenotrophomonas and Emticicia, though, were enriched in the PM treatment [31, 36]. We hypothesize that an increase in N fixation and cycling in Wt Arabidopsis associated with the microbiota, and the potential increase in denitrification in the pgr5 mutant, could help increase the duration of N bioavailability in Wt Arabidopsis relative to the pgr5 mutant. Plant pathogenic genera enriched in PM, such as Panacagrimonas and Filimonas, may also contribute to the earlier flowering time, because infected hosts preferentially allocate resources toward reproduction [37, 38]. This hypothesis linking N availability to flowering time is discussed further below within the newly proposed molecular network that modulates flowering time.
Verification of microbial function
We unambiguously demonstrated that flowering time was directly associated with the rhizosphere microbiota. Microbiota from the third-generation pgr5 (PM) or Wt cultures (WM) was used to inoculate cultures of three Arabidopsis lines (Wt and two mutants of the photosynthetic apparatus, pgr5 and pnsB4, deficient in cyclic electron flow from NADPH to plastoquinone) for one generation. The microbiota of the third-generation Wt culture delayed flowering time and increased shoot growth in all cases compared to the treatment using microbiota from the pgr5 cultures. The addition of Wt microbiota delayed flowering by 3.3, 5.5, and 5.7 days in Wt, the pnsB4 mutant, and the pgr5 mutant, respectively (Fig. 2a). The shoot fresh weight of the plants treated with Wt microbiota also increased significantly in the three lines compared to the treatments with the addition of pgr5 microbiota (Fig. 2b).
These results clearly indicate that flowering time can be affected by the rhizosphere microbiota. This effect disappeared when the soil slurry was sterilized before inoculation, indicating that heat-stable exudates alone were not modulating Arabidopsis flowering time (Fig. 2c, d). It is interesting that the difference between WM- and WM-S (sterilized soil slurry)-treated plants was not very obvious (Additional file 1: Figure S4). The reason may be that sterilized soil slurry included more N, and other nutrients to influence flowering time, which was in accordance with the results in Fig. 3. Therefore, we speculated that the metabolites in the sterilized slurry also played a role in influencing the flowering time and plant stature.
The addition of the two soil slurries did not change bulk-soil pH, available soil K or P contents (Additional file 1: Table S4), the abundance of key genes involved in the C cycle, or the activities of β-glucosidase or chitinase in the rhizosphere (Additional file 1: Figure S2 and S3) but did affect the abundance of genes (normalized to the abundance of 16S rRNA gene) involved in N cycling (Fig. 2e–h) and the amounts of NH4+ (Fig. 3A) and NO3− (Fig. 3C) in soil. The concentration of bioavailable N species was generally lower after the addition of the pgr5 microbiota compared to the treatment with the addition of Wt microbiota (Fig. 3A, C). This decrease in N availability in the cultures treated with pgr5 microbiota was accompanied by an increase in the abundance of genes involved in denitrification (nirK and nosZ) and a decrease in the abundance of genes involved in nitrogen fixation (nifH) and nitrification (amoA) compared to the Wt-treated groups (Fig. 2e–h). In the Wt-treated group, the activity of urease was higher (Fig. 3B) and nitrate reductase was lower (Fig. 3D) than those in the PM group, which probably resulted in higher NH4+ and NO3−.
Several lines of evidence support the hypothesis that the rhizosphere microbiota modulated N cycling and bioavailability, leading to N deficiency earlier in the pgr5-treated groups and thus earlier flowering. Flowering can be triggered by low nitrate levels [39], but plants maximize growth before flowering under the conditions of N sufficiency [25].
Different root exudates in the two Arabidopsis lines
Root exudates can act as key substrates or signaling molecules that affect microbial composition [27], so we tested the hypothesis that exudate concentrations and compositions differed between the Arabidopsis lines. A metabolomic analysis found that 34 exudates involved in 10 metabolic pathways were differentially released in the two lines (Wt and pgr5) (see Additional file 1: Table S5, the principal component analysis in Fig. 4a, and ≥ 2- or ≤ 0.5-fold changes and p values < 0.05 in Fig. 4b). Four of the 10 biochemical pathways were upregulated in the Wt cultures relative to the pgr5 cultures (Fig. 4b). Thymine was the most differentially released exudate (Additional file 1: Table S5). Thymine can be degraded by bacteria, perhaps accounting for the increase in NH4+ content in the WM groups (Fig. 3A).
Trp and its derivatives, phenols, and some carboxylic acids were preferentially exuded in the Wt cultures (Additional file 1: Table S5). The concentrations of amino acids were generally higher in the Wt cultures, consistent with a higher abundance of Bacillus subtilis in the Wt cultures, for which amino acids are chemoattractants. These metabolomic results are consistent with the hypothesis that differences in exudates affect rhizosphere microbiota [40].
IAA delayed flowering time by downregulating genes involved in flowering
Auxins regulate plant growth in many ways [13, 41, 42]. One of the important auxins for plants is IAA, which is soluble in aqueous solutions and, when protonated, diffuses passively across cell membranes without the need of a specific transporter [43]. IAA has also been hypothesized to have a floral-inductive signaling role by regulating multiple aspects of embryonic and postembryonic development [44, 45]. Microorganisms can produce IAA from Trp [46,47,48,49]. One of the enriched rhizosphere microorganisms, Arthrobacter (Additional file 1: Table S3), has been reported to be beneficial for plant growth by having the ability to produce IAA [16]. Trp and its derivatives were enriched in the Wt exudates, so the generation of IAA by microorganisms may control flowering time by a novel molecular network. Trp content in the soil of generation 3 of the Wt cultures decreased, and the IAA content increased 3.03-fold, suggesting that the Wt microbiota rapidly converted Trp into IAA (Fig. 4c, d).
We explored the possibility that microbially generated IAA delayed flowering time by adding Trp, 5-hydroxytryptophan (5-HTP), and IAA to the hydroponic cultures and monitoring the flowering time of each line. Adding 5 and 25 nM IAA decreased the bolting proportion of Wt Arabidopsis by 20–30% (Fig. 5a) when 80% of the control plants had floral buds, implying that IAA was the direct driver that delayed flowering time. Changes in the expression of genes involved in flowering further supported the role of IAA in regulating flowering time (Fig. 5b). IAA treatment induced changes in the relative transcription rates of genes associated with flowering. The rates for some of the genes comprising the autonomous, GA, and vernalization pathways changed significantly, and these changes were observable at the early (when 10% of control plants had floral buds), intermediate (when > 50% of control plants had floral buds), and late (when > 80% of control plants had floral buds) stages of flowering. The downregulation of gene activity was most consistent in the photoperiod pathway, where multiple genes were downregulated at all flowering stages.
Shimada et al. [14] reported that IAA relieved the inhibitory effects of aspterric acid on pollen growth and thus speculated that IAA accelerated reproductive growth in Arabidopsis. To our knowledge, our microcosm study is the first to demonstrate that IAA is one critical signal that delays flowering in Arabidopsis, although Wagner et al. [50] also found a similar phenomenon, while not proposing any related mechanism.
Combined with earlier reports, the weight of evidence suggests that IAA stimulates the development of floral organs but delays the flowering time of Arabidopsis. Our data strongly suggest that IAA delays flowering and acts as a signal of optimal growth conditions in the absence of N limitation for growth.