Time-resolved small RNA sequencing reveals different dynamics of miR156 and miR157 within RISCs in apices

To provide a quantitative analysis of the dynamics of miR156 and miR157 levels in shoot apices during shoot development and floral transition, time-resolved small RNA (sRNA) sequencing was performed on sRNAs purified using Trans-kingdom, rapid, affordable Purification of RISCs (TraPR) [40]. This approach enriches for sRNAs loaded into RNA-induced silencing complexes (RISCs). The importance of the miR156/miR157–SPL module on flowering differs between SD and LD conditions [12,15]; therefore, miRNA levels were compared in wild-type plants grown under both daylengths (Fig 1A, B).

Download:

• PNG
  larger image
• TIFF
  original image

Fig 1. Time-resolved small RNA-sequencing reveals different expression dynamics of miR156 and miR157 families.

The abundance of miR156 and miR157 isoforms in apices of wild-type (Col-0) plants in (A) long-day (LD) and (B) short-day (SD) time courses. Material was harvested at the indicated time points. Based on microscopic analysis of plants under the same growth conditions, the 7LD and 11LD samples are vegetative apices, 14 LD is during the I1 to I2 transition, and 18LD samples are I2 [33]. The 1wSD to 5wSD samples represent the extended vegetative phase under SDs, whereas the 6wSD samples are I1 apices (Fig 4A). Lines represent fitted generalised additive models (see Methods). Shaded areas represent approximate 95% confidence intervals. The expression of miR156 isoforms (C) and miR157 isoforms (D) in Col-0 (dark blue), mir156a-4 mir156c-4 (purple), mir157c-3 mir157d-2 (light blue), and mir157a-3 mir157b-3 mir157c-3 (green) mutants. Data points represent means of three independent biological replicates. Error bars represent standard deviation. Asterisks in (C) and (D) represent significance differences in miR156 or miR157 levels between mutant and Col-0 at a given time point as determined by pairwise Wilcoxon t-tests at p < 0.05.

https://doi.org/10.1371/journal.pgen.1011799.g001

Shoot apices from late-flowering SD-grown plants were harvested weekly between 1 and 6 weeks after germination, whereas samples from early-flowering LD-grown plants were collected 7, 11, 14 and 18 days after germination. After TraPR purification and sRNA sequencing, the sRNA reads were mapped to the genome and counted (see Material and Methods). Most MIR156 paralogues encode an identical miR156 isoform, and MIR157A to MIR157C encode the same miR157 isoform [41]; therefore, miRNA abundance was analysed for miR156 and miR157 separately, but not for each individual paralogue.

In the LD and SD time courses, the levels of miR156 in RISCs in shoot apices were highest at the first time point and decreased steeply afterwards (Fig 1A, B). At 7LDs, the level of miR156 was on average twice as high as that of miR157 (Fig 1A and S1 Table). However, at the 11LD and 14LD time points, both miRNAs were approximately equally abundant. After 18LDs, Col-0 plants have undergone floral transition and formed floral primordia [42], and at this time point the miR156 level decreased to 60% of that of miR157 (S1 Table). Thus, in apices of LD-grown plants, miR156 was more abundant in RISCs than miR157 during early vegetative growth but decreased to a similar and then to a lower level than that of miR157 in mature inflorescence apices.

In apices of SD-grown plants, miR156 was initially four times more abundant in RISCs than miR157 (1wSD, S1 Table), but miR156 levels decreased steeply between 1wSD and 2wSD, whereas the abundance of miR157 showed a more moderate decrease (Fig 1B). Consequently, miR156 and miR157 were approximately equally abundant in apices of 2-week-old SD-grown plants, and subsequently the levels of miR157 were consistently higher than those of miR156 (S1 Table). At 5wSD and 6wSD, when Col-0 plants initiate floral transition, mature miR156 levels were approximately one-third of those of miR157. We conclude that although miR156 and miR157 levels in RISCs in apices decrease as the plants age, they are still present at the shoot apex around the time of floral transition (14LDs or 6wSDs) and in the early inflorescence meristem, and that particularly under SDs, miR157 is more abundant in RISCs at floral transition than miR156.

To determine the contributions of miR156 and miR157 to flowering time and inflorescence development, CRISPR-Cas9 reverse genetics was used to recover mutant alleles of MIR156 and MIR157 genes (see Material and Methods). First, the levels of all MIR156 and MIR157 precursor RNAs in apices during vegetative development and early inflorescence development were assessed under LDs using RNAseq data [42] (S1A Fig). As previously described, MIR156A and MIR156C genes are the most highly expressed MIR156 genes [9] and are reduced in expression in apices as plants age [13]; therefore, we introduced new mutations into these genes to increase the allelic variation at these loci [17] and combined the mutations in a double mutant. MIR156D is also relatively highly expressed in apices, but no mutant allele could be recovered for this gene. Novel deletion alleles were generated in all MIR157 genes (MIR157A, MIR157B, MIR157C and MIR157D; S1B Fig). The mir157a-3 mir157b-3 mir157c-3 triple mutant was then constructed to assess the contribution of MIR157D alone, which appeared to be only expressed during inflorescence development (16LD; S1A Fig). Similarly, the mir157c-3 mir157d-2 double mutant was constructed to assess the effect of inactivating the two MIR157 genes mainly expressed during inflorescence development. To assess the reduction in the abundance of miR156 and miR157 caused by these combinations of mutations, sRNA sequencing was performed after carrying out TraPR purification on apical samples of mir156a-4 mir156c-4, mir157c-3 mir157d-2 and mir157a-3 mir157b-3 mir157c-3 mutants grown from vegetative development (1 and 3 weeks under SDs) to floral transition (6 weeks under SDs) (Fig 1C, D). The levels of the miRNAs in RISCs at each time point were then compared with those of Col-0 and the other mutants (Fig 1C, D). At 1wSD, when its expression was highest in all genotypes, miR156 levels in mir156a-4 mir156c-4 were approximately one-third of those in Col-0 and were extremely low at later time points (Fig 1C and S1 Table). By contrast, the abundance of miR156 in mir157c-3 mir157d-2 and mir157a-3 mir157b-3 mir157c-3 was comparable to that of Col-0 at all three time points (Fig 1C). The miR157 levels in mir157c-3 mir157d-2 were reduced to approximately 25% of Col-0 levels early in development (1wSD) and to 3% of Col-0 at floral transition (6wSD), consistent with these two genes contributing the vast majority of miR157 at shoot apices during floral transition and inflorescence development (Fig 1D and S1 Table). In mir157a-3 mir157b-3 mir157c-3, miR157 levels were very low during vegetative development, but clearly increased at floral transition, consistent with MIR157D being increased in expression at this stage (Fig 1C, D and S1 Table). Therefore, the new CRISPR-Cas9-induced deletions in MIR156 and MIR157 genes effectively reduce miR156 or miR157 abundance in RISCs in a family-specific manner, and support the idea that MIR157C and MIR157D are mainly expressed later in development during floral transition and early inflorescence development.

miR156 and mir157 mutations differentially affect vegetative and cauline leaf development

The morphology of rosette leaves changes acropetally along the main shoot axis as a consequence of vegetative phase change, and these differences are regulated by miR156/miR157 [9,11,43]. Therefore, whether mir157 mutations have a stronger effect on later formed leaves, consistent with the miRNA sequencing results, was tested. The effects of the mir156 and mir157 deletion alleles on leaves were determined by calculating the length-to-width ratio (LWR) of all rosette leaves that are formed during the vegetative phase of Col-0 and mutant plants at 4wSD (Fig 2A, B and S2A, B Fig). For Col-0, the LWR of rosette leaves increased progressively from the second to the sixth leaf, where it reached its maximum (Fig 2A). The leaves formed later on the shoot had not completely expanded at 4wSDs and therefore showed a lower LWR. The LWR of the cotyledons and the first two leaves was greater in rSPL15 plants than in Col-0 (Fig 2A and S2A, B Fig), and the fifth leaf of rSPL15 plants had already reached a maximum LWR, consistent with an accelerated transition to the adult vegetative phase compared with Col-0. The rosette leaf phenotypes of mir156a-4 mir156c-4 double mutants were most similar to those of rSPL15 plants, and the LWR of the first two leaves was much higher than that of Col-0 and close to the maximum value. Therefore, MIR156AC play a major role in conferring juvenile vegetative phase represented by leaf morphology, as previously described [9,44,45]. A more irregular phyllotaxy compared with Col-0 was also observed for mir156a-4 mir156c-4 (Fig 2A and S2C Fig). The variation in divergence angle between successive rosette leaves was similarly high in rSPL15 plants (S2C Fig). In contrast to mir156a-4 mir156c-4, the increase in the LWR of successive leaves of mir157c-3 mir157d-2 was largely comparable to that of Col-0 plants, but higher maximum LWR values were achieved for leaves 6, 7 and 8 (Fig 2A). Moreover, mir157c-3 mir157d-2 and mir157a-3 mir157b-3 mir157c-3 mutants showed no obvious defect in phyllotaxis (S2A, C Fig). Increased miR156 expression and reduced SPL activity also cause a reduction in plastochron length [46]. Therefore, we tested for differences in the rate of leaf initiation during vegetative growth under SDs. The mir156a-4 mir156c-4 plants showed a lower leaf initiation rate (longer plastochron), consistent with increased SPL activity, whereas mir157c-3 mir157d-2 plants unexpectedly appeared to produce leaves faster than that of Col-0 plants (Fig 2C). Collectively, these results indicate that MIR156AC are important early during shoot development to confer juvenile vegetative phase, but that MIR157 genes have a more marked function in later rosette leaves, consistent with miR157 being more abundant than miR156 in RISCs in older apices.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 2. Analysis of mir156 and mir157 mutants reveals different roles for mir156 and mir157 in the regulation of leaf number and shape.

(A) The length/width ratio of cotyledons and rosette leaves after growth for 4 weeks in short days (4wSD). Dashed lines show the minimum and maximum mean values for Col-0. (B) Rosette phenotype of Col-0, mir156a-4 mir156c-4, mir157c-3, mir157d-2, mir157c-3 mir157d-2 and rSPL15 plants grown for 4wSD. (C) Leaf initiation rate of each genotype defined as number of leaves formed per day and measured at 4-weeks short days. (D–E) The number of cauline leaves formed by each genotype under short days. (F) Cauline leaf phenotype of Col-0, mir156a-4 mir156c-4, mir157c-3, mir157d-2, mir157c-3 mir157d-2 and rSPL15 plants grown under long days for five weeks. (G) Length/width ratio of cauline leaves compared with Col-0, leaves are shown in F. Letters indicate significant differences calculated with ANOVA and Tukey’s Honest Significant Difference test; p < 0.05. Error bars indicate standard deviation in all panels.

https://doi.org/10.1371/journal.pgen.1011799.g002

We therefore assessed the effects of miR156 and miR157 on the shape and number of cauline leaves formed in the inflorescence during the I1 phase. The mir157c-3 and mir157d-2 mutants both formed significantly more cauline leaves than Col-0, particularly under SDs (Fig 2D and S2 Table), and the double mutant showed an additive phenotype (Fig 2D and S2 Table). The triple mutant mir157a-3 mir157b-3 mir157c-3 formed fewer cauline leaves than mir157c-3 mir157d-2 (S2 Table). Moreover, mir156a-4 mir156c-4 mutants produced the same number of cauline leaves as Col-0 (Fig 2E), consistent with MIR156AC having a smaller effect than MIR157 CD genes on inflorescence development. The morphology of the cauline leaves of mir157c-3, mir157d-2 and mir157c-3 mir157d-2 mutants also differed from Col-0 (Fig 2F, G). The cauline leaves of these mutants were more serrated in terms of the number and the depth of the serrations than those of Col-0 and mir156a-4 mir156c-4. Moreover, the LWR of the cauline leaves was lower in mir157d-2 than Col-0, and unexpectedly this effect was suppressed in the mir157c-3 mir157d-2 double mutant (Fig 2G). Taken together, this analysis demonstrates that MIR157 CD genes are more important in regulating cauline leaf number and morphology than MIR156AC.

SPL15 is one of the miR156/miR157 targets that regulates floral transition under SDs [15], and therefore the effect of SPL15 on the enhanced cauline leaf number of mir157c-3 mir157d-2 was tested. Under SDs, spl15–1 (11.33 cauline leaves; Fig 2D and S2 Table) was epistatic to mir157c-3 mir157d-2 (14.94 cauline leaves) with respect to cauline leaf number in the spl15–1 mir157c-3 mir157d-2 triple mutant (11.28 cauline leaves), suggesting that increased SPL15 expression might confer this phenotype. Overall, consistent with the sRNA sequencing data, analysis of leaf morphology suggests that MIR156AC are important for regulating juvenile vegetative leaf morphology and phyllotaxis, whereas MIR157 CD are more important in controlling cauline leaf number and shape in the inflorescence.

mir156a-4 mir156c-4 and mir157c-3 mir157d-2 mutants differ in their flowering responses to daylength

The SPL TFs and miR156/miR157 regulate floral transition [10,12,15]; therefore, the flowering time of mir156a-4 mir156c-4 and mir157c-3 mir157d-2 mutants was scored under LDs and SDs, and compared with that of Col-0 and rSPL15 plants. Overexpression of MIR156 genes or mutation of SPL genes caused a shorter plastochron than in Col-0 [46,47], therefore, we concentrated on measuring flowering time by assessing the number of days until the inflorescence stem started to elongate above the rosette (bolting time, see Material and Methods). However, rosette leaf number, cauline leaf number, days to first open flower and bolting time are provided for all genotypes in S2 Table. Under LDs and SDs, rSPL15 plants bolted significantly earlier than Col-0, but the difference was more extreme under SDs in which bolting of Col-0 was delayed (Fig 3A, B and S2 Table). The mir157c-3 mir157d-2 double mutant also consistently bolted earlier than Col-0 under SDs, whereas the mir156a-4 mir156c-4 plants did not (Fig 3B). Therefore, MIR157 CD play a more important role than MIR156AC in repressing bolting time under SDs. By contrast, under LDs, mir157c-3 mir157d-2 mutants bolted at the same time as Col-0 (Fig 3A), whereas the bolting time of mir156a-4 mir156c-4 relative to Col-0 varied among experiments. The mir156a-4 mir156c-4 mutant bolted significantly earlier than Col-0 under LDs in some experiments (Fig 3A), whereas in others the bolting time of these genotypes did not differ (S2 Table). These data demonstrate that under SDs, when plants transition to flowering later, MIR157 CD have a more important role than MIR156AC in repressing bolting, and that under LDs MIR156AC have a stronger effect, although this varied among experiments presumably due to redundancy among MIR156/MIR157 genes [17].

Download:

• PNG
  larger image
• TIFF
  original image

Fig 3. Analysis of mir156 and mir157 mutants reveal different roles in the regulation of flowering.

(A–B) Bolting time of wild-type (Col-0) plants compared with mir156a-4 mir156c-4, mir157c-3 mir157d-2 and rSPL15 in (A) long-day (LD) and (B) short-day (SD) conditions. (C, D) AP1 and FUL expression levels quantified by RT-qPCR in apices of different genotypes grown in (C) LD and (D) SD conditions. Values are means ± standard deviation. (E–G) Flowering time of the indicated genotypes under SDs expressed as the number of days between sowing and bolting. Letters indicate significant differences calculated with ANOVA and Tukey’s Honest Significant Difference test; p < 0.05.

https://doi.org/10.1371/journal.pgen.1011799.g003

FUL and SPL15 contribute to the early-bolting phenotype of mir157c-3 mir157d-2 mutants

SPL TFs activate transcription of FUL, which promotes the expression of LFY and AP1 in floral primordia, and AP1 mRNA acts as a molecular marker for floral development [12,28]. Therefore, the abundance of FUL and AP1 mRNAs was quantified in apices of plants of different genotypes grown under LDs and SDs as markers for SPL activity and floral primordia development. Under SDs, the level of FUL mRNA in rSPL15 and mir157c-3 mir157d-2 plants was higher than in Col-0 and mir156a-4 mir156c-4 at 5w, 6w and 7w (Fig 3D), consistent with the bolting phenotypes of these genotypes (Fig 3B). Similarly, AP1 mRNA was more abundant in mir157c-3 mir157d-2 and rSPL15 than in Col-0 and mir156a-4 mir156c-4 at 7wSD (Fig 3D). Under LDs, the differences were less pronounced, but the level of FUL mRNA was slightly higher in mir156a-4 mir156c-4 compared with other genotypes at 11LDs, and AP1 mRNA was higher in the same genotype at 14LDs (Fig 3C). Overall, the gene expression analysis supports the hypothesis that mir157c-3 mir157d-2 and rSPL15 cause early floral transition under SDs, whereas mir156a-4 mir156c-4 does not.

FUL mRNA abundance was increased in mir157c-3 mir157d-2, and ful mutants show delayed flowering under LDs and SDs [48,49]; therefore, whether FUL contributes to the early bolting of mir157c-3 mir157d-2 under SDs was tested by constructing the triple mutant ful-2 mir157c-3 mir157d-2 (Fig 3G). The ful-2 mutation was fully epistatic to mir157c-3 mir157d-2 and mir157c-3 with respect to bolting time, indicating that FUL is essential for the early bolting caused by mir157c-3 mir157d-2. SPL15 activates FUL and loss-of-function mutations in SPL15 delay bolting and flowering under SDs [15]; therefore, to determine whether SPL15 is required for the early-bolting phenotype of mir157c-3 mir157d-2, the triple mutant spl15–1 mir157c-3 mir157d-2 was constructed. The spl15–1 mir157c-3 mir157d-2 plants bolted later than mir157c-3 mir157d-2 mutants and Col-0 (Fig 3E), indicating that SPL15 contributes to the early-bolting phenotype of mir157c-3 mir157d-2. However, spl15–1 mir157c-3 mir157d-2 plants bolted much earlier than spl15–1 mutants, suggesting that increased expression of other SPL TFs must also contribute to the mir157c-3 mir157d-2 phenotype. The bolting times of mir157c-3 and spl15–1 mir157c-3 were also compared. The mir157c-3 mutant bolted at a similar time to mir157c-3 mir157d-2, suggesting that the early-bolting phenotype of the double mutant is mainly caused by the mir157c-3 mutation (Fig 3E, G and S2 Table). Moreover, the spl15–1 mir157c-3 double mutant showed a similar bolting time to spl15–1 mir157c-3 mir157d-2, and bolted later than mir157c-3 but earlier than spl15–1 (Fig 3E). To account for the possibility that additional SPL genes might be involved, the spl4 spl15–1 mir157c-3 triple mutant was generated. The spl4 spl15–1 mir157c-3 plants bolted later than mir157c-3 and spl4 mutants (Fig 3F and S2 Table). However, the triple mutant plants still bolted earlier than the spl15–1 and spl4 spl15–1 mutants, suggesting that SPL4 and SPL15 are major contributors to the early bolting of mir157c mutants, but that other SPL genes also contribute (Fig 3F and S2 Table). Mutations in MIR156A and MIR156C did not accelerate bolting under SDs (Fig 3B and S2 Table). Nevertheless, whether mir156a-4 mir156c-4 affected the extreme late-bolting phenotype of spl15–1 was also tested. The triple mutant spl15–1 mir156a-4 mir156c-4 bolted slightly earlier than spl15–1, but much later than Col-0 and spl15–1 mir157c-3 mir157d-2 (S2 Table). Taken together, these results support the conclusion that MIR157 CD are more important in delaying floral transition under SDs than MIR156AC and that they do so through negatively regulating FUL and partially through downregulating SPL15.

Spatio-temporal expression of SPL15 and FUL in mir157 cd mutants

The spl15–1 mutation partially suppressed the early bolting of mir157c-3 mir157d-2 plants; therefore, we examined whether SPL15 expression is increased in mir157c-3 mir157d-2. Apices were harvested from plants grown under SDs (2wSD to 6wSD; Fig 4 or under LDs (9, 11, 14 and 16LD) and analysed by confocal microscopy (see Material and Methods). First, apices of Col-0 plants carrying the pSPL15::V9A:rSPL15 (rSPL15) or the pSPL15:: V9A:SPL15 (pSPL15) reporter [15] were analysed for comparison under SD (Fig 4A) where the mir157c-3 mir157d-2 early bolting phenotype was detected (Fig 3B). Consistent with previous analyses, the spatial expression patterns of pSPL15 and rSPL15 in the Col-0 SAM and stem cortex were similar, but rSPL15 was expressed earlier and more strongly in both tissues (Fig 4A; 21). Then, pSPL15 introduced into the mir157c-3 mir157d-2 double mutant was analysed, and the fluorescence signal of VENUS9A:SPL15 was stronger in the apex of mir157c-3 mir157d-2 than in Col-0 from at least 4wSD onwards, and was stronger in the cortex of the stem below the SAM at 5wSD and 6wSD (Fig 4A). Moreover, floral transition and elongation of the apex occurred faster in pSPL15 mir157c-3 mir157d-2 than in pSPL15 at 5wSD and 6wSD (Fig 4A). The pattern of VENUS9A-SPL15 was also compared in pSPL15 mir157c-3 mir157d-2 and pSPL15 plants grown from 9LDs to 16LDs (Fig 4B). VENUS9A-SPL15 was detected in the SAM and stem cortex in both genotypes at similar levels, consistent with mir157c-3 mir157d-2 plants bolting at similar times to Col-0 under LDs (Fig 3A). Taken together, these data indicate that MIR157 CD are important in the repression of SPL15 expression in the SAM and in the stem cortex of inflorescence stem internodes under SDs. The higher SPL15 activity in mir157c-3 mir157d-2 was also supported by increased expression of the FUL:VENUS protein in the SAM at 6wSD (Fig 4C).

Download:

• PNG
  larger image
• TIFF
  original image

Fig 4. Floral transition of mir157 mutants in short days is accelerated by the premature accumulation of SPL15 and its target FUL at the shoot apical meristem.

(A) Expression of pSPL15::V9A-SPL15 in wild type (Col-0), mir157c-3 mir157d-2 transgenic plants, and pSPL15::V9A-rSPL15 plants grown under SD. (B) Expression of pSPL15::V9A-SPL15 in wild type (Col-0) and mir157c-3 mir157d-2 transgenic plants grown under LD. (C) Confocal images of shoot apical meristems of ful-2 and ful-2 mir157c-3 mir157d-2 mutants expressing pFUL::FUL: 9AV grown under SDs. Each image is representative of five analysed samples. Fluorescence from the Venus protein is artificially colored in yellow. Asterisks mark images acquired with a lower laser intensity (see Methods). Scale bar = 50 µm. V9A, Venus-9Alanina; 9AV, 9Alanina-Venus; CZ, Central zone; RB, Rib meristem; PZ, Primordia zone.

https://doi.org/10.1371/journal.pgen.1011799.g004

Spatial and temporal patterns of transcription of MIR157 and MIR156 genes at the shoot apex

The sRNA-sequencing experiment showed a progressive reduction in miR156 abundance in Col-0 apices over time, but a more constant abundance of miR157. To analyse the temporal expression patterns of individual MIR genes at the shoot apex, the RNA levels of the MIR156AC and MIR157 CD genes were first quantified by RT-qPCR (Fig 5A–C and S3 Fig). The levels of MIR156A and MIR156C RNA in apices of SD-grown Col-0 decreased more or less exponentially, from 1wSD until basal levels were reached after 4wSD (Fig 5A). A comparable expression pattern was also observed in LD-grown plants, where the levels of MIR156A and MIR156C decreased rapidly from 4LD to 11LD (S3A Fig), similar to changes in the level of the mature miR156 (Fig 1B) and as previously reported in seedlings [12] and apices [13]. By contrast, the levels of MIR157A and MIR157C mRNAs decreased more moderately from 1wSD and stabilised after 3wSD (S3 Fig). The MIR157C and MIR157D transcripts were therefore studied from 3wSD to 7wSD. MIR157C transcripts were detected at all time points and their abundance decreased by half between 4wSD and 5wSD and was subsequently maintained at a constant level (Fig 5B), whereas MIR157D expression was barely detectable in vegetative apices (until 6wSD), but increased when plants underwent floral transition (7wSD, Fig 5B). These results are consistent with the extremely low abundance of miR157 detected between 1wSD and 6wSD in mir157a-3 mir157b-3 mir157c-3 mutants, in which MIR157D was the only functional gene encoding miR157 (Fig 1D). MIR157C and MIR157D RNAs were also detected in cauline leaves of 36LD plants (Fig 5C), demonstrating that miR157 is still present in leaves formed late in shoot development on the inflorescence and consistent with the effects of mir157 cd mutations on cauline leaf number and shape (Fig 2).

Download:

• PNG
  larger image
• TIFF
  original image

Fig 5. Characterization of MIR156 and MIR157 temporal and spatial expression patterns.

(A) Expression levels of endogenous MIR156A and MIR156C genes in apices of plants grown in short days. (B) Endogenous expression of MIR157C and MIR157D genes was quantified in apices of wild-type seedlings under short days. (C) MIR157C and MIR157D expression in leaves of Col-0 plants after 36 long days. RL = youngest rosette leaf, C1 = first (lowest) cauline leaf, C2 and C3 = second and third cauline leaf respectively. Data points in panels A–C indicate the mean ± SD of three biological replicates. (D–F) Confocal images of dissected apices of (D) MIR157C::VENUS:GUS in short days (SD) and long days (LDs), (E) MIR157D::VENUS:GUS under LDs and SDs. Arrows indicate expression. (F) MIR156A::GFP and MIR156C::VENUS:GUS in LDs. Plants were harvested at the indicated time points after sowing. Each image is representative of five analysed samples. Fluorescence from the Venus protein is artificially coloured in yellow, fluorescence from the GFP protein is artificially coloured in green. All reporter constructs contain a nuclear localization signal. Scale bars = 50 µM.

https://doi.org/10.1371/journal.pgen.1011799.g005

Transcriptional reporter lines were then constructed to characterize the spatial expression patterns of MIR156B, MIR156C, MIR157A, MIR157C and MIR157D at the shoot apex and the previously constructed MIR156A:GFP [17] was included for comparison. The MIR156A::GFP and MIR156C::VENUS:GUS reporters were broadly expressed throughout apices of young plants, and these spatial patterns of expression were similar in LDs and SDs (Fig 5F and S4A, B Fig). Consistent with the temporal decrease in mature miR156 levels (Fig 1A, C), MIR156A and MIR156C reporter expression decreased in apices with increasing plant age, as observed previously for MIR156C [13]. This decrease occurred more rapidly under LDs, in which the signals were barely detectable at 11LD and 14LD (Fig 5F), than under SDs, when the signals were clearly detectable at 14SDs but strongly attenuated at 21SDs (S4 Fig). The spatial expression pattern of MIR156A::GFP differed from that of MIR156C::VENUS:GUS at the SAM. In LDs and SDs, MIR156C::VENUS:GUS was detected in the SAM one week after germination, before being restricted to the more distal rib zone at later time points. By contrast, MIR156A::GFP was only present in the rib region at all time points under both growth conditions. Although the expression of both reporters decreased in the shoot apex and in newly formed primordia, signal persisted lower in the shoot for longer, for example at 11LD (Fig 5F). In contrast to MIR156A::GFP and MIR156C::VENUS:GUS, MIR156B::VENUS:GUS was expressed exclusively in the epidermis, but was absent from the epidermis of the central zone of the SAM (S5 Fig). Expression of MIR156B::VENUS:GUS did not appear to reduce with the age of the plant.

Consistent with the sRNA-sequencing and the RT-qPCR data, signal of MIR157C::VENUS:GUS expression was detected in LDs and SDs below the SAM in the stem (Figs 5D and S7A, B Fig). Expression persisted in the stem until at least 5wSD, and in LDs was still present after floral induction at 14LD (Fig 5D). At 14LD, MIR157C::VENUS:GUS expression was not detected in the elongating internodes representing the bolting inflorescence stem, but was present in the lower internodes (Fig 5D). This pattern together with the early-bolting phenotype of mir157c mutants under SD suggests that MIR157C acts in the stem to inhibit internode elongation and bolting under these conditions. Expression of MIR157C::VENUS:GUS was also observed in cauline leaves (S7B Fig), as well as in the hypocotyl epidermis (Fig 6G and S7B Fig).

Download:

• PNG
  larger image
• TIFF
  original image

Fig 6. AamiR157c exhibits similar behaviour to miR157c in regulating flowering in Arabis alpina.

(A) Phylogenetic tree of miR156/7 sequences from Arabidopsis (black) and Arabis alpina (green). The tree was constructed using the Maximum Likelihood method and bootstrapped with 1000 replicates. The scale bar indicates the branch length that corresponds to the number of substitutions per nucleotide. (B, C) High-resolution analysis of conservation of synteny in genomic regions from Arabidopsis (top) and A. alpina (bottom) containing (B) MIR157A/B and (C) MIR157C genomic regions. The black boxes indicate MIR157A/B/C genes in A. thaliana, and the green boxes indicate homologous regions in A. alpina. Each panel represents a genomic region, with the dashed line separating top and bottom strands of DNA. Coloured arrows in green/yellow represent protein coding sequences, blue represents transcribed regions, and the full extent of the gene is shown in grey. For each pairwise comparison, light pink boxes represent regions of sequence similarity, with putatively homologous sequences being connected by red shading. (D) Flowering time and total leaf number of the indicated genotypes under long days (LDs). Letters indicate significant differences calculated with ANOVA and Tukey’s Honest Significant Difference test; p < 0.05. (E) Flowering phenotypes of the indicated genotypes of plants grown for 60LD. Scale bar = 1 cm. (F) Confocal images of A. alpina AaMIR157C::mSc at 4–5 weeks and Arabidopsis MIR157C::VENUS:GUS plants grown for 9LDs. Scale bar = 100 µm.

https://doi.org/10.1371/journal.pgen.1011799.g006

The expression of the MIR157D::VENUS:GUS reporter was not detected in vegetative apices (Fig 5E and S7C Fig). At floral transition, MIR157D::VENUS:GUS expression was observed at the adaxial base of cauline leaves, in the vicinity of the axillary meristem (Fig 5E). Moreover, MIR157A::VENUS:GUS expression was only detected in the vasculature of young seedlings and in cell clusters below young leaf primordia, and expression was absent from the SAM (S6 Fig).

Therefore, although MIR156AC expression is progressively repressed at the apex during vegetative growth in SDs, MIR157C expression persists in the stem until floral transition, after which it is expressed in the rosette stem and cauline leaves. By contrast, MIR157D is not expressed prior to floral induction, although it might be expressed specifically in stomatal cells [50], but later shows a characteristic expression pattern on the adaxial side and axil of cauline leaves in the inflorescence.

Conserved function of MIR157C in Arabis alpina accelerates floral transition

In A. alpina, a perennial relative of Arabidopsis, the miR156–SPL module regulates the age-dependent response to vernalization [37,39]. The mir157c-3 and mir157d-2 alleles described above indicate that MIR157C has a major role in regulating bolting and flowering in Arabidopsis; therefore, we tested whether this function is conserved in A. alpina. First, we identified the AaMIR156 and AaMIR157 genes present in the A. alpina genome. The MIR157 family exhibits lower conservation compared with MIR156, so that only two out of the four genes present in Arabidopsis (MIR157A and MIR157C) are conserved in A. alpina (Fig 6A). Conversely, both species retain all eight genes encoding the miR156 family. Therefore, the origin of the MIR156 family paralogues preceded the divergence of A. alpina and Arabidopsis, since which time frame duplications or losses of MIR157 family members have occurred so that A. alpina contains two fewer copies than Arabidopsis.

To further investigate the evolutionary relationship between AaMIR157A and AaMIR157C genes in Arabidopsis and A. alpina, we performed a microsynteny analysis of their respective genomic regions (Fig 6B, C). Consistent with the phylogeny (Fig 6A), a high conservation of synteny was found between the MIR157C regions of Arabidopsis and A. alpina, and the tandem duplication of MIR157A and MIR157B in Arabidopsis was present as a single copy in A. alpina. At the position corresponding to MIR157D in Arabidopsis no homologous gene was observed in the A. alpina genome.

To address the function of AaMIR157A and AaMIR157C genes during floral transition, CRISPR-Cas9 reverse genetics was used to recover mutant alleles (see Material and Methods; S8 Fig). Analysis of the resulting mutants showed that Aamir157a did not significantly influence flowering, but all of the genotypes carrying Aamir157c mutations exhibited early flowering from both the main shoot and lateral branches (Fig 6D, E). Apart from accelerated flowering, Aamir157c mutations caused reductions in total leaf number, implicating AaMIR157C in the regulation of shoot architecture and reproductive timing. Transcriptional reporter lines were constructed to characterize the spatial expression pattern of AaMIR157C. Expression of AaMIR157C::mScarlet reporter was not detected in the A. alpina SAM (Fig 6F), similar to what was described for the MIR157C::VENUS:GUS reporter in Arabidopsis (Figs 5D and 6G); however, it was broadly expressed in the epidermis of hypocotyls (Fig 6F) similar to in Arabidopsis (Fig 6G). The genetic and expression data suggest a conserved role for AaMIR157C in repressing flowering, illustrating the functional conservation of this paralogue with Arabidopsis MIR157C.

Redundancy between SPL15 and FT TSF in promoting the vegetative to I1 transition

The miR157–SPL15 pathway is most important for flowering under SDs (Fig 3A, B, E and S2 Table; 21). These results suggest that the major effectors of the photoperiodic pathway, FT TSF, overcome the requirement for SPL15 under LDs by promoting extremely early flowering. In support of this, spl15–1 mutations strongly enhanced the late-flowering phenotype of ft-10 tsf-1 double mutants under LDs [39] (Fig 7A). Therefore, we investigated the interaction between SPL15 and FT TSF in more depth by phenotypically scoring flowering time and inflorescence development of spl15–1 ft-10 tsf-1 under LDs and SDs.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 7. Genetic interactions between SPL15 and the FT TSF flowering pathway.

(A, B) Bolting time, (C, D) number of rosette leaves and (E, F) number of cauline leaves of the indicated genotypes under (A, C, E) long-day (LD) and (B, D, F) short-day (SD) conditions. Letters indicate significant differences calculated with ANOVA, Tukey’s Honest Significant Difference test; p < 0.05.

https://doi.org/10.1371/journal.pgen.1011799.g007

Under LDs, spl15–1 did not affect bolting time or rosette leaf number of Col-0, but greatly enhanced the increased bolting time and rosette leaf number of ft-10 tsf-1, confirming that SPL15 and FT TSF redundantly regulate floral transition under these conditions (Fig 7A, C and S2 Table). Moreover, ft-10 tsf-1 formed about four times more cauline leaves than Col-0, whereas spl15–1 formed the same number as Col-0, and spl15–1 ft-10 tsf-1 formed a similar number to ft-10 tsf-1 (4.42 for Col-0; 16.2 for ft-10 tsf-1; 15.96 for spl15–1 ft-10 tsf-1; Fig 7E and S2 Table). Therefore, under LDs the effect of SPL15 on the transition from vegetative phase to I1, in which cauline leaves and axillary inflorescence branches are formed, is redundant with FT TSF, but in the transition from I1 to I2 phase, in which flowers are formed, only FT TSF plays a major role and SPL15 does not affect this transition, even if FT TSF are mutated.

Under SDs, bolting time and rosette leaf number are strongly increased in spl15–1 mutants compared with Col-0 (Fig 7B, D). FT TSF are not transcribed in leaves of Col-0 under SDs, and the effect of spl15–1 on bolting time is only weakly enhanced by ft-10 tsf-1. This result supports the proposal that SPL15 performs the role of initiating the transition from vegetative to I1 phase under SDs, and that FT TSF has little effect even in spl15–1. Moreover, both FT TSF and SPL15 relatively weakly promote the I1 to I2 transition, so that ft-10 tsf-1 and spl15–1 form, respectively, about 50% and 25% more cauline leaves than Col-0, and the triple mutant has an additive effect, forming approximately double the number of cauline leves (8,91 for Col-0; 12,38 for ft-10 tsf-1; 10,86 for spl15–1; 16,32 for spl15–1 ft-10 tsf-1; Fig 7E, F and S2 Table). These data suggest that FT TSF and SPL15 both promote vegetative to I1 transition, but their relative importance depends on photoperiod, and that particularly under LDs, FT TSF promotes I1 to I2 transition, whereas SPL15 has a weaker or no effect on this transition.

Spatial and temporal patterns of expression of MIR156 and MIR157 genes

The dominant concept in the regulation of miR156 and miR157 is that their levels progressively fall as the plant ages [2,10–12]. This progressive downregulation is linked to cell division and therefore developmental age of the tissue, and involves accumulation of trimethylation at lysine 27 of histone 3 (H3K27me3) [13,51]. We used TraPR small RNA sequencing to quantify the levels of miR156 and miR157 within RISCs in shoot apices. In this analysis, levels of miR156 fell rapidly in apices of older plants, whereas levels of miR157 fell more slowly at the same stage and were significantly more abundant than levels of miR156. The relatively high level of miR157 within RISCs in apices of older plants is supported by the higher levels of mature miR157 than miR156 that were found in sRNAs extracted from inflorescences [17]. He et al [9] found a higher level of miR157 than miR156 in 11-day-old seedlings, which contrasts with our description of higher levels of miR156 than miR157 in RISCs in apices of younger plants. This discrepancy might be explained by our use of TraPR for purification of sRNAs, because this method only detects miRNA within RISCs, and miR157 was previously proposed to load less efficiently than miR156 into AGO1 [9].

MIRNA paralogues can be expressed in different spatial or temporal patterns. In shoot apices, MIR156C transcription was previously detected in the rib zone but not the central or peripheral zones of the SAM from 3 days after germination [17]. We defined the patterns of reporters for MIR156A, MIR156B and MIR156C as well as MIR157A, MIR157C and MIR157D at apices by confocal microscopy. MIR156B::VENUS:GUS was expressed throughout the epidermis at the apex, except at the central zone of the SAM from which it was excluded. This pattern did not appear to alter with the age of the plant, and MIR156B might therefore modulate SPL activity in the epidermis throughout the life of the plant. In young plants, MIR156C was initially detected in the SAM and then later was restricted to the rib region, as previously described [17], before being fully repressed even in apices of older plants. The pattern of expression of MIR156A was similar to that of MIR156C, but MIR156A was not detected in the SAM even in very young plants. Expression of both MIR156A and MIR156C is almost undetectable throughout the apex by the time flowers are formed under LDs and SDs. However, under LDs, MIR156C was still detected in the lower rib region at 11LD and may act in these tissues around that time to weakly delay internode elongation and bolting. These results suggest that MIR156AC expression is undetectable at apices during flowering under SDs, when SPL TFs are most important for floral induction.

The spatial patterns of expression of MIR157 genes have not previously been described at shoot apices. We found that MIR157C expression was similar to that of MIR156AC in the rib region or upper stem of younger plants, but it then persisted for longer until floral transition at 5wSD or 14LD, at least in the stem. By contrast, MIR157D expression was undetectable prior to floral transition and appeared in the axils of cauline leaves and in flowers in the inflorescence. This developmental regulation of MIR157D does not follow the age-related reduction shown by MIR156AC. MIR157A expression was not detected in the apex, but was present in developing leaf vasculature. Overall, therefore, MIR156B and MIR157D are developmentally regulated in apices and do not show age-dependent repression, whereas MIR157C is slowly reduced in expression with age but persists in the upper stem until flowering. Therefore, among those genes characterized by confocal microscopy, only MIR156AC showed the characteristic high-level of expression in juvenile apices and reduction in adult apices that is most associated with these miRNA families.

The eight members of the Arabidopsis MIR156 family were conserved in genomic location and sequence in A. alpina, whereas the MIR157 family was more variable and only two (MIR157A and MIR157C) of the four members present in Arabidopsis were also present in A. alpina. By contrast, all twelve Arabidopsis genes were conserved in the more closely related A. lyrata and Capsella rubellum [52]. The presence of mature miR156 was reported in all vascular plants tested [53,54] and in moss [55], but these studies did not distinguish between miR156 and miR157. In addition to miR156, SPL mRNAs are targeted by miR529 in mosses and other members of the Embryophyta as well as monocotyledonous species, but this miRNA is absent in core eudicots [53,54]. In grasses, MIR529 genes are more strongly expressed in the panicle whereas MIR156 genes are more highly expressed in juvenile vegetative tissues [56–58]. These observations together with our expression analysis of MIR157 genes in Arabidopsis, suggest that they might have evolved more recently than the core MIR156 genes, be more variable in the Brassicaceae than MIR156, and have roles in regulating SPLs in the inflorescence as observed for MIR529 in panicles of grasses.

Roles of miR156 and miR157 in stem elongation, flowering and inflorescence development

Under LDs, flowering occurs rapidly through the florigen pathway and is initiated early in plant development when levels of miR156 in RiSCs are still higher than those of miR157. Accordingly, under these conditions, strong reductions in miR156 in the mir156ac mutant caused earlier bolting and flowering in some experiments, but mir157 cd mutations had no effect on these phenotypes under LDs. A similar result was obtained with vegetative phase change, because leaves formed early during vegetative development showed morphological characteristics of adult leaves in mir156ac mutants, but in mir157 cd mutants were similar to those of Col-0. This result supports the idea that miR156 is a stronger repressor of vegetative phase change than miR157 [9]. Similarly, mir156ac mutants showed the longer plastochron length during vegetative development that is expected for increased SPL expression, whereas mir157 cd mutants showed a shorter plastochron at this stage. The basis of the shorter plastochron of the latter mutants is not clear, but might be related to the different spatial patterns of expression of the MIR157 and MIR156 genes, because expression of MIR156 genes from heterologous promoters was previously shown to have different effects on plastochron length [46]. Complete loss of miR156 in the hextuple mir156 mutant and complete loss of miR156 and miR157 in the duodecuple mutant caused extreme early flowering under LDs [17]. These data also suggest that miR156 is more important in flowering of Col-0 plants early in development under LDs than miR157, but that miR157 contributes in mir156 mutants. In addition, miR157 does affect inflorescence development under LDs, because in the mir157 cd mutant the number of cauline leaves was increased under these conditions but was unaffected by mir156ac.

Under SDs, the FT florigen pathway is not active and the SPLs are more critical for floral transition [12,15]. Col-0 plants flower later under these conditions, and as the life-cycle is extended, miR157 levels within RiSCs persist at a higher level than those of miR156. Accordingly, we found that the MIR157 CD genes were more important than MIR156 genes in repressing bolting and flowering under SDs, and that they had effects on cauline leaf number and shape that MIR156 genes did not. In addition, in the smaller MIR157 family in A. alpina, the role of the MIR157C orthologue in delaying flowering was conserved.

Contribution of the SPL15 miR157 module in flowering and inflorescence development

Five SPL genes (SPL3, SPL4, SPL5, SPL9 and SPL15) that are negatively regulated by miR156/miR157 have been reported to promote flowering [12,15,27–30]. Mutations in SPL15 are sufficient to strongly delay flowering under SDs, whereas transgenes expressing miR156/miR157-resistant versions of SPL15 mRNA cause early bolting and flowering [15], and genetic epistasis experiments showed that spl15 mutations delayed bolting of mir157c and mir157 cd mutants. Moreover, the fluorescence signal of VENUS9A:SPL15 was stronger in the apex of pSPL15 mir157c-3 mir157d-2 double mutant than in Col-0. These experiments support the idea that miR157 directly targets SPL15 mRNA, and interestingly, miR157c and mir157d are one nucleotide longer than miR156a or miR156c, and in the case of miR157c this extends the complementarity with SPL15 mRNA [9]. Furthermore, spl15 mir157c mir157d triple mutants bolted much earlier than spl15 mutants, suggesting that under SDs MIR157C delays bolting by repressing other SPL genes in addition to SPL15. A major contributor to this effect is SPL4, because spl4 spl15 mir157c bolted significantly later than spl15 mir157c. SPL9, the paralogue of SPL15, and SPL3 and SPL5, paralogues of SPL4, might also contribute. FUL is bound and activated by several of these SPL family members [12,15,28,59] and ful mutations delay flowering under SDs [32,49]. We found that FUL is more highly expressed at the SAM of mir157 cd mutants from around 5wSDs, and that ful mutations were fully epistatic to mir157c and mir157 cd in terms of bolting time, so that the higher-order mutants bolted at the same time as ful and later than Col-0. Therefore, FUL is required to promote bolting downstream of MIR157C, and increased expression of SPLs in mir157c does not overcome the requirement for FUL. In mir157 cd mutants, the abundance of SPL15 at the SAM and in the cortex of the stem below the SAM increased earlier in development. Nevertheless, at flowering MIR157C expression was only detected in the stem and not in the SAM, whereas MIR157D expression was only detected in the axils of cauline leaves. Therefore, MIR157 CD reduce expression of SPL15 in the SAM and cortex where they are not expressed. This non-autonomy might be explained by movement of the miR157 expressed from MIR157C in the lower stem into the upper internodes and SAM, or by MIR157C indirectly affecting expression of SPL15 in the SAM via an intermediate mobile signal that is repressed in the lower internodes. The higher level of rSPL15 accumulation in these regions suggest that SPL is directly regulated by movement of miR157; moreover, movement of miR156 through vasculature was previously proposed in potato [60], miR156 and miR157 were detected in the phloem sap of Brassica napus [61] and in Arabidopsis movement of miR156 between the SAM and leaf primordia was proposed to influence leaf development [2,62].

Cauline leaf formation occurs during the I1 phase of inflorescence development, and stops on the transition to I2 when flowers are formed. The cauline leaves of mir157d-2 plants were shorter and wider, consistent with increased SPL activity. This effect was suppressed in the mir157c-3 mir157d-2 double mutant, perhaps due to miR157 repressing SPLs in a different spatial pattern in the cauline leaf leading to suppression of the phenotype. Further explanation of this interaction will require understanding the roles of different SPL transcription factors in controlling cauline leaf shape. In principle, more cauline leaves can be formed if plants transition earlier from vegetative development to I1 or by delayed transition from I1 to I2. Moreover, an increase in cauline leaf number can be caused by an increased rate of organ initiation in I1, for example due to enlargement of the meristem during floral transition [63,64]. Notably, mir157c and mir157d both caused an increase in cauline leaf number, particularly under SDs and this was enhanced in the double mutant. The different expression patterns of MIR157C and MIR157D and the increase in cauline leaf number in the double mutant suggest that they might influence cauline leaf number by different mechanisms. The mir157c mutant also bolts earlier, as observed for rSPL15, suggesting it might cause an increase in cauline leaf number due to premature transition from vegetative to I1. By contrast, mir157d does not cause early bolting and is expressed in the axils of cauline leaves, suggesting that MIR157D might reduce cauline leaf number by accelerating acquisition of floral meristem identity and transition from I1 to I2. Therefore, MIR157D might repress SPLs in the axils of cauline leaves that would otherwise repress acquisition of floral meristem identity. Notably, SPL9 and SPL15 were previously proposed to contribute to the repression of LATERAL SUPPRESSOR transcription and axillary meristem formation in the axils of cauline leaves [65]. Furthermore, under SDs the ful mutation strongly increases cauline leaf number compared with Col-0 [32,49,66], probably by impairing floral meristem identity, and cauline leaf number is further strongly enhanced in the mir157 cd ful triple mutant compared with ful or mir157 cd. These results indicate how different modulators of cauline leaf and axillary branch development can shape inflorescence development and mutually enhance their effects.

In A. alpina, the Aamir157c mutation caused early flowering. AaSPL15 has an important role in the promotion of flowering in A. alpina, and rAaSPL15 caused early flowering. Therefore, the early flowering of the Aamir157c mutant might at least partly be due to increased activity of AaSPL15 [39]. Our analysis of the Aamir157c mutant was performed in the A. alpina pep1 mutant background, which does not require vernalization to flower [67]. In the A. alpina Pajares genotype, which is wild type for PEP1 and shows an obligate vernalization requirement, the AamiR156/AaMIR157 families delay responsiveness to vernalization by reducing the activity of SPL15 after vernalization of juvenile plants [37,39]. Our observation that Aamir157c pep1 mutants are early flowering suggests that AaMIR157C is a major contributor to the juvenile flowering response in the A. alpina Pajares background.

SPL15 and FT TSF show genetic redundancy in flowering time that is dependent on the daylength to which the plants are exposed. Under LDs, FT TSF strongly promote vegetative to I1 transition and the I1 to I2 transition. Any effect of SPL15 on these characters is almost undetectable under LDs in the presence of FT TSF. However, in the triple spl15 ft tsf mutant the redundant role of SPL15 in vegetative to I1 transition is revealed, but this also showed that only FT TSF regulate I1 to I2 transition under LDs. Under SDs, the relative contribution of FT TSF and SPL15 to the vegetative to I1 transition are reversed, so that spl15 mutants show a greatly increased rosette leaf number, days to bolting and days to flowering, whereas ft tsf are similar to Col-0. In transition from I1 to I2, both spl15 and ft tsf cause slight increases in cauline leaf number that are enhanced in the triple mutant. Therefore, the redundancy between SPL15 and FT TSF mainly occurs in the transition from vegetative to I1, and which of these genes are most important depends on daylength. In the transition from I1 to I2, SPL15 generally has a minor role, although this is slightly greater in SDs, whereas FT TSF influences this transition under both conditions but is more important under LDs. In the I1 to I2 transition, FT TSF have been proposed to activate AP1 in floral primordia and this may not be mediated by SPL15 which is not expressed in primordia [15,68], whereas both pathways activate FUL in the SAM [15,35,36]. Other SPLs are also likely to play roles in both pathways as the level of SPL3, SPL4 and SPL5 are activated in apices by the FT–FD module [29,35,69] and SPL9 is expressed in primordia and is involved in AP1 activation [12,27]. Additionally, it has been proposed that SPL3, SPL4 or SPL5 forms a transcriptional co-activator complex with FD to activate floral meristem identity genes during the induction flowering under LDs [27,28]. Further elaboration of the common and distinct target genes of FT TSF, SPL15 and related SPLs, as well as their spatio-temporal expression patterns and the contribution of MIR157 CD in regulating these will allow more detailed elucidation of the daylength-dependent gene regulatory networks required for floral transition, stem bolting and floral development.

Plant material and growth conditions

Plants were grown on soil in long-day (16 h light/8 h dark) or short-day (8 h light/16 h dark) conditions at 20–22°C with 150–180 μmol m⁻² s⁻¹ light intensity. All genotypes were in the Arabidopsis thaliana (L.) ecotype Columbia-0 (Col-0) background. All the Arabis alpina genotypes were in a pep1 background. The spl15–1 [47], spl4 [72], pSPL15::V9A-SPL15 [15], pSPL15:: V9A:rSPL15 [15], ful-2 [48], pFUL::FUL:9AV ful-2 [42], MIR156A::GFP [17], ft-10 tsf-1 [39] and spl15–1 ft-10 tsf-1 [39] genotypes were published previously. The following genotypes were generated in this study by CRISPR-Cas9 mutagenesis: miR156a-4, mir156c-4, mir157a-3, mir157b-3, miR157c-3, mir157d-2, Aamir157a-1, Aamir157c-1, Aamir157c-2 and Aamir157a-1 Aamir157c-1. To generate mir157c-3 mir157d-2 pFUL::9AV:FUL ful-2 and mir157c-3 mir157d-2 pSPL15::9AV:SPL15, mir157c-3 mir157d-2 was crossed with the previously published pFUL::FUL:9AV ful-2 [42] and pSPL15::V9A-SPL15 [15] and genotyped with the primers listed in S3 Table. The mir156a-4 mir156c-4 spl15–1, mir157c-3 mir157d-2 spl15–1, mir157c-3 spl4 spl15–1, spl4 spl15–1, mir157c-3 ful-2, mir157d-2 ful-2, mir157c-3 mir157d-2 ful-2, mir156a-4 mir156c-4, mir157c-3 mir157d-2 and mir157a-3 mir157b-3 mir157c-3 genotypes were obtained by crossing and genotyping with the primers listed in S3 Table. The reporters MIR156B::VNG, MIR156C::VNG, MIR157A::VNG MIR157C::VNG, MIR157D::VNG and AaMIR157C::2HA-mScarlet were generated in this study and are described below.

Cloning of MIR156/MIR157 and AaMIR157C reporter constructs

The design of MIR156B, MIR156C, MIR157A, MIR157C, MIR157D and AaMIR157C reporter lines followed a previously published strategy [32]. Full intergenic regions of MIR156B, MIR156C, MIR157A, MIR157C and MIR157D were amplified by PCR and cloned into a pDONR entry vector via BP reaction (Invitrogen). Subsequently, the miRNA-encoding stem–loop region was replaced by the coding region of Venus-GUS or 2HA-mScarlet using Gibson Assembly [70]. The entry clones were subcloned via LR reaction (Invitrogen) into the binary vector pEarlyGate301 for MIR157C::VENUS::GUS and AaMIR157C::mSc, and pBGW.0 for MIR156B::VENUS::GUS, MIR156C::VENUS:GUS, MIR157A::VENUS:GUS and MIR157D::VENUS:GUS. Next, the plasmids were introduced into Agrobacterium tumefaciens strain GV3101 and Col-0 Arabidopsis and pep1 A. alpina plants were transformed via floral dipping [71]. Seeds transformed with the bar selection cassette were selected on soil by spraying with 0.001% glufosinate-ammonium. Surviving plants were genotyped via PCR to confirm the presence of the transgene. Several independent lines were generated and analysed, two representative lines were further imaged and presented in the manuscript. Primers used for plasmid construction and genotyping are listed in S3 Table.

Generation of MIR156/MIR157 CRISPR-Cas9 alleles

New mutations in Arabidopsis MIR156 and MIR157 genes were generated by CRISPR-Cas9. In brief, sgRNAs were designed that targeted the stem–loop region. We took advantage of the high sequence similarity between the precursor genes to design multi-target sgRNAs (S3 Table). We designed one sgRNA to target either MIR156A, MIR156C or MIR156D individually, or, MIR157A and MIR157B, or MIR157C and MIR157D. The U6-26 promoter-sgRNA cassette was cloned into the pGreen0229 vector as described in [72]. The plasmids were introduced into A. tumefaciens strain GV3101 and floral dipping was used to transform wild-type Arabidopsis plants [71]. T1 transgenic plants harbouring the sgRNA-Cas9 construct were selected on soil for the presence of the bar gene. CRISPR-Cas9-free T2 and T3 plants containing mutations were selected via PCR. The mutations were identified with a combination of PCR and Sanger sequencing (S1B Fig). The mir157a-3 mir157b-3 and mir157c-3 mir157d-2 mutants each arose from double editing events within the same plant. The mir157c-3 and mir157d-2 mutants were obtained via backcrossing.

The A. alpina CRISPR-Cas9 mutants were generated using a similar strategy: the U6-26 promoter-sgRNA cassette was cloned into the pHEE401, which allows the incorporation of two specific-target sgRNAs. The resulting construct, which targeted both AaMIR157A and AaMIR157C, was transformed into pep1 mutant plants by the Agrobacterium-mediated floral dip method [39]. Primers used for plasmid construction and genotyping are listed in S3 Table.

DNA extraction and genotyping

For genotyping, genomic DNA (gDNA) was extracted using the BioSprint 96 DNA Plant Kit (Qiagen) in combination with the Biosprint automation platform according to the manufacturer’s instructions. The extracted gDNA was eluted to a final volume of 100 µL with ddH₂O. GoTaq DNA Polymerase and GoGreen Buffer (Promega GmbH) were used for all genotyping assays. All primers used for genotyping are listed in S3 Table.

Confocal imaging

Shoot apices at different developmental stages were dissected under a stereo microscope and fixed as described in [63]. Tissue was cleared in ClearSee [73] for 2–5 days, depending on the developmental stage. Cell walls were stained with SCRI 2200 Renaissance [74] (0.1% in ClearSee) overnight at room temperature. A. alpina samples were imaged as live tissue. Confocal images were acquired with a TPS SP8 confocal microscope (Leica Microsystems GmbH). The excitation wavelength was 405 nm for Renaissance, 488 nm for GFP, 514 nm for VENUS, 561 nm for mScarlet. Image collection was performed at 430–470 nm for Renaissance, 500–520 nm for GFP, 520–534 nm for VENUS and 580–620 nm for mScarlet. Images were processed in ImageJ, which involved applying consistent brightness settings and generating maximum intensity projections of 10-µm optical sections at the meristem centre. Imaging and processing parameters for fluorescent proteins were determined for the first time point and remained constant throughout the experiment, unless stated otherwise.

RNA extraction and RT-qPCR

Plant tissue was harvested at the indicated times and immediately snap frozen in liquid nitrogen. Total RNA was isolated using the miRNeasy Tissue/Cells Advanced Mini Kit (Qiagen) according to the manufacturer’s instructions. The RNA was treated with Ambion DNase I (Thermo Fisher Scientific Inc) and 1 μg total RNA was used for cDNA synthesis with the Superscript IV first-strand synthesis system (Thermo Fischer Scientific Inc). Transcript levels were quantified by quantitative real-time PCR with iQ SYBR green supermix (Bio-Rad) in a CFX384 Touch Real-Time PCR Detection system (Bio-Rad). Three technical replicates were performed for three independent biological triplicates, and expression of PEROXIN4 (PEX4) and PP2A (PROTEIN PHOSPHATASE 2A) genes was used to normalize the data. Primers used for expression analyses are listed in S3 Table.

Small RNA extraction and data processing

Apices were harvested in three biological replicates for all the indicated time points. Small RNAs were extracted with the TraPR Small RNA Isolation Kit (Lexogen). Sample extraction and library preparation were carried out at the Max Planck Genome Centre Cologne. Circa 10 million single-end 150-bp reads were sequenced per sample using the Illumina HiSeq 3500 system. Adapter sequences were trimmed from raw reads using Cutadapt version 3.4. The parameters were selected for a minimum read length of 17 nucleotides and a minimum Illumina quality score of 20. Trimmed sequence reads were analysed with DeconSeq version 0.4.3 to filter out non-plant contaminations [75]. Mapping of the sRNA reads to the Arabidopsis genome was performed with the Manatee tool, version 1.2 [76] and used annotations from the Araport11 release [77] supplemented with the miRbase 22 release [78]. Annotations were limited to miRNAs, short-interfering RNAs and other non-coding RNAs. Other annotations, including protein-coding genes, were excluded from the analysis.

Phylogenetic analysis

Sequences similar to Arabidopsis miR156 and miR157 were obtained from A. alpina genome database (https://ucscbrowser.mpipz.mpg.de) through BLAT search using default parameters. Sequence alignments were performed by ClustalW [79], and Gblocks [80] was used to remove less conservative regions. A phylogenetic tree was elaborated using Maximum Likelihood method (Bootstrapped with 1000 replicates). Alignment and phylogeny were carried out in MEGA version X.

Microsynteny analysis

Microsynteny analysis was performed using the Genome Evolution Analysis (GEvo) tools of the Comparative Genomics Platform (CoGe) [81]. Briefly, similar regions between the Arabidopsis and A. alpina genomes were searched using CoGeBlast and then further examined using GEvo. We used the default setting to define the minimum number of collinear genes for two regions to be called syntenic.

Statistical analysis

Data were analysed using R version 4.0.3 and associated statistical packages [82] and GraphPad Prism 9. Statistical differences were determined with one- or two-factor ANOVA comparisons. For clearly non-parametric data, a Kruskal-Wallis (KW) omnibus test was applied. If significant differences were identified, post-hoc multiple comparisons were performed. Tukey’s HSD multiple comparison test was used to determine differences identified by ANOVA (both from base R) and a Nemenyi post-hoc test (PMCMRplus package) was used to analyse differences after the Kruskal-Wallis test (base R). Levene’s test of equal variance was employed to determine differences in sample variation. For the box plots, the boxes represent the 25% and 75% quartiles. The horizontal line inside the box indicates the median, and each point represents an individual plant.