Reproductive competence is regulated independently of vegetative phase change in Arabidopsis thaliana

SUMMARY

A long-standing question in plant biology is how the acquisition of reproductive competence is related to the juvenile-to-adult vegetative transition. We addressed this question by examining the expression pattern and mutant phenotypes of two families of miRNAs—miR156/miR157 and miR172—that operate in the same pathway and play important roles in these processes. The phenotype of mutants deficient for miR156/miR157, miR172, and all three miRNAs, demonstrated that miR156/miR157 regulate the timing of vegetative phase change but have only a minor effect on reproductive competence, whereas miR172 has a minor role in vegetative phase change but has a major effect on reproductive competence. MIR172B is directly downstream of the miR156/SPL module, but temporal variation in the level of miR156 in the shoot apex and leaf-to-leaf variation in miR156 expression in young primordia was not associated with a change in the level of miR172 in these tissues. Additionally, although miR172 levels increase from leaf-to-leaf later in leaf development, this variation is largely insensitive to changes in the abundance of miR156. Our results indicate that the acquisition of reproductive competence in Arabidopsis is regulated by miR172 by a mechanism that is independent of the vegetative phase change pathway.

eTOC Blurb

Zhao et al show that the acquisition of reproductive competence in Arabidopsis thaliana is regulated independently of vegetative phase change, and that miR156 does not play a major role in the regulation of miR172 in the shoot apex. These results challenge the existing model for the age-dependent flowering time pathway in Arabidopsis.

INTRODUCTION

Shoot development in plants can be divided into distinct phases based on successive phenotypic changes, such as changes in leaf shape and size, branching pattern, and other species-specific traits^1–3. For example, in Arabidopsis, the cotyledons and first few rosette leaves are round, have a smooth leaf margin, and lack abaxial trichomes, whereas later formed leaves are more elongated, have a serrated leaf margin and produce abaxial trichomes. The transition between these vegetative phases is referred to as the juvenile-to-adult transition or vegetative phase change. In addition to vegetative phase change, plants undergo a transition between a vegetative phase of shoot growth and a reproductive phase, in which they produce structures involved in sexual reproduction. This vegetative-to-reproductive transition typically occurs after vegetative phase change. However, the interval between vegetative phase change and flowering varies considerably in different accessions of Arabidopsis⁴, and varies from days to years in different plant species; furthermore, species that flower in a juvenile vegetative phase exist in nature^5–7 and mutations or transgenes that result in the constitutive expression of genes that promote the juvenile vegetative phase do not have a major effect on flowering time in Arabidopsis or maize under conditions that promote flowering^5,6,8–14. These observations suggest that vegetative phase change is not necessarily associated with an increase in reproductive competence and raise the possibility that these transitions are regulated independently in many species.

In flowering plants, vegetative phase change and floral induction are regulated by two families of evolutionary-conserved microRNAs, miR156/157 and miR172^13,15,16. In Arabidopsis, miR156/157 repress 10 members of the Squamosa Promoter Binding Protein/SBP-Like (SBP/SPL) transcription factor family¹⁷, whereas miR172 represses 6 members of the Apetala2-like (AP2-like) transcription factor family^18–20, including TARGET OF EAT 1 and 2 (TOE1 and TOE2), SCHLAFMUTZE (SMZ) and SCHNARCHZAPFEN (SNZ). The phenotypes of plants with elevated or reduced levels of miR156/157 or their SPL targets demonstrate that SPL transcription factors promote most, if not all, adult phase-specific vegetative traits, and affect flowering time to a lesser or greater extent depending on environmental conditions^{12,13,21–24}. AP2-like transcription factors inhibit floral induction by directly repressing the transcription of key floral integrator genes, such as FLOWERING LOCUS T (FT), SUPRESSOR OF OVEREXPRESSOR OF CONSTANS 1 (SOC1), FRUITFULL (FUL) and APETALA1 (AP1) ^18,25–28, and by repressing the activity of the circadian-regulated floral promoter, CONSTANS (CO) ²⁸. They repress abaxial trichome production (an adult vegetative trait) by associating with the abaxial regulator, KANADI1, to repress the transcription of GLABRA1^29,30.

Under floral inductive conditions, most species only produce flowers after a period of vegetative growth, implying that they are reproductively incompetent early in their development. In Arabidopsis, we proposed that the acquisition of reproductive competence is regulated by the following genetic pathway, which has come to be known as the “age-dependent flowering time pathway”: miR156/157 --| SPL → miR172 --| AP2-like --| floral induction¹³. According to this model, reproductive competence is acquired when a decline in miR156/157 produces an increase in the level of their direct targets, SPL transcription factors, which promote the transcription of genes encoding miR172, leading to the downregulation of the floral repressors, AP2-like transcription factors. An obvious implication of this model is that the acquisition of reproductive competence is linked to vegetative phase change.

Support for this model comes from the expression pattern of miR156/157 and miR172, and from the phenotype of plants with reduced or elevated levels of these miRNAs. miR156/157 are expressed at a very high level in the first few leaves, and at a much lower and slowly declining level in leaves produced later in the shoot development^13,23,31. Over-expression of miR156 under a strong constitutive promoter prolongs the expression of all known juvenile vegetative traits, and delays flowering to a small extent in long days (LD) ^12,13,21,23, and a much larger extent in short days (SD) ²⁴, whereas loss of miR156/157 accelerates the appearance of adult vegetative traits and slightly accelerates flowering^12,13,23,31. miR172 has the opposite expression pattern and the opposite effect on vegetative phase change and flowering time. In LD, miR172 is present at a low level early in shoot development and increases in abundance as the shoot develops^13,18,25. Over-expression of miR172 accelerates flowering^18,25,26,32 and the appearance of some adult vegetative traits¹³, whereas loss of miR172 accelerates the appearance of abaxial trichomes¹³ and delays flowering in LD and SD^{13,18,32–34}. Expression of miR156 under a strong constitutive promoter, or loss-of-function mutations in the SPL genes repressed by miR156, decreases the abundance of miR172^{12,13,24,35–40}. These observations, and the evidence that both SPL9 and SPL15 bind the promoter of MIR172B^13,24, support the hypothesis that the miR156/SPL module regulates reproductive competence and phase-specific epidermal traits through their effect on the expression of miR172¹³.

Developmental changes in reproductive competence are typically defined by the ability of a plant to flower in response to environmental conditions that are conducive to flowering. Recent studies have shown that SPL15 and miR172 play important roles in flowering under environmental conditions that repress flowering^24,33,34,41, but the extent to which the miR156/miR157 -| SPL -> miR172 -| AP2-like pathway controls reproductive competence under conditions that are conducive to flowering remains unclear. In particular, the extent to which vegetative phase change leads to a change in reproductive competence has never been carefully examined. To answer this question, we compared the expression patterns of miR156 and miR172 during leaf and shoot development and characterized the vegetative and reproductive phenotypes of plants with reduced or elevated levels of these miRNAs. We focused on the effect of these genotypes on the acquisition of the ability to respond to a photoinductive long-day stimulus because photoperiod has a major effect on flowering time and the molecular mechanism of the photoperiod response is well understood. We found that under both a non-inductive and inductive photoperiod, the expression of miR172 is not strongly correlated with the expression of miR156. We also found that variation in the abundance of miR156 has little to no effect on the sensitivity of the shoot to a LD stimulus, whereas variation in the abundance of miR172 has a major effect on this response. Together, these results indicate that the age-dependent increase in the photoperiodic competence of Arabidopsis is controlled primarily by miR172. Although miR156 can influence the acquisition of reproductive competence through its effect on miR172, it is not a significant factor in this process. We conclude that the age-dependent increase in reproductive competence is regulated by miR172 independently of vegetative phase change.

RESULTS

The expression of miR172 is poorly correlated with the expression of miR156

We first examined the relationship between the expression patterns of miR156 and miR172 in the shoot apices of wild-type Col-0 growing under short-day (SD) conditions because miR172 expression is induced by long days (LD)¹⁹. RT-qPCR revealed that miR156 decreased dramatically during the first 3 weeks after germination and then more slowly after that, whereas miR172 declined very slightly and with much greater variation from week 1 to week 7 (Figure 1A). We then examined the abundance of miR156 and miR172 in successive 2 mm leaf primordia (LP) grown in SD. miR156 was expressed at a very high level in LP1&2, and at much lower levels in all other LP (Figure 1B). In contrast, miR172 increased only slightly from LP1&2 to LP17 (Figure 1B). To confirm these results, we examined a next-generation sequencing small RNA dataset from FRIGIDA (FRI) FLOWERING LOCUS C (FLC) and FRI flc-3 plants growing in LD³¹ (Figure 2A). Consistent with our RT-qPCR results, miR156 was expressed at much higher levels than miR172 and declined rapidly with time in the shoot apices of both genotypes, while the abundance of miR172 remained essentially unchanged over the same period.

Figure 1. miR156 regulates the amount but not the temporal expression of miR172.

RT-qPCR analysis of the amount of miR156 and miR172 in (A) shoot apices and (B) successive 2 mm leaf primordia of wild-type plants (SD). (C) The abundance of miR156 and miR172 in successive leaf primordia of natural accessions of Arabidopsis (SD). p values are shown for a comparison of each sample with the first sample. (D) RT-qPCR analysis of the amount of miR172 in 2 mm leaf primordia and 18–20 mm leaves of wild-type and miR156ac miR157ac quadruple mutant plants (SD). (E) RT-qPCR analysis of the amount of miR172 in the entire shoots of wild type, the miR156ac miR157ac quadruple mutant, and plants transformed with 35S::MIR156A (LD). n = 3 biological replicates for all analyses. Data are relative to the first sample in each graph (e.g. 1 week, LP1/2) and are shown as the mean ± standard error of the mean. Means were compared using an unpaired two-tailed t-test with Welch’s correction. *p<0.05; **p<0.005; ***, p<0.0005. n.s., not significant. SD = short days; LD = long days. See also Figure S1.

Figure 2. Developmental variation in the absolute amount of miR156 and miR172.

(A) The abundance all miR156 and miR172 transcripts in the shoot apex and in fully expanded cotyledons and leaves of FRI FLC and fri-3 FLC plants grown in LD. (B) The abundance of miR172 transcripts from different MIR172 genes in fully expanded cotyledons and leaves of FRI FLC and fri-3 FLC plants grown in LD. Data are the number of reads obtained by next-generation sequencing normalized to 3 million total reads, and are the average of two biological replicates.

To determine the generality of these expression patterns, we compared the expression of miR156 and miR172 in leaf primordia of Col-0 and 4 natural accessions (Aitba-2, Dra-3, Sij-4 and Voeran-1) grown in SD (Figure 1 C). In three of these accessions, the amount and expression pattern of miR156 and miR172 was similar, if not identical to Col. The only difference was that in two accessions (Sij-4 and Dra-3) miR172 increased slightly from LP1&2 to LP3. Voeran-1 had much higher levels of miR156 and miR172, and in this accession miR172 increased significantly from LP1&2 to LP4, in association with a decline in the abundance of miR156. These results are consistent with a previous study⁴ in indicating that while the expression pattern of miR156 is conserved within Arabidopsis, the abundance of miR156 can vary considerably between accessions. The situation is more complicated in the case of miR172, as both the expression pattern and the abundance of this miRNA vary between accessions.

These results were unexpected because previous studies of RNA extracted from whole shoots have shown that miR172 increases in a complementary fashion to the decline in miR156^13,18,25. To explore the basis for this discrepancy, we compared the level of miR172 in 2 mm LP and 18–22 mm leaves in both wild-type plants and in the mir156a-2 mir156c-1 mir157a-1 mir157c-1 (mir156ac mir157ac) quadruple mutant, which has approximately 20% of the normal level of miR156 and almost no miR157³¹ (Figure 1D, see also Figure S1). In SD, there was no significant difference in the amount of miR172 in successive LP of wild-type plants. In contrast, 18–22 mm leaves had significantly more miR172 than 2 mm LP, and the amount of miR172 increased with leaf position. The miR156ac miR157ac mutant had significantly more miR172 than wild-type plants, but the pattern of miR172 expression in successive 2 mm LP and 18–22 mm leaves was otherwise similar to wild-type plants (Figure 1D). These results were confirmed by the abundance of miR156 and miR172 in next-generation sequencing data from fully expanded leaves of FRI FLC and FRI flc-3 plants grown in LD (Figure 2A). In both genotypes, miR156 was most abundant in cotyledons, and decreased dramatically in subsequent leaves. miR172 was more abundant in fully expanded leaves than in LP and increased in abundance from leaf-to-leaf in both genotypes.

We then examined the expression of miR172 in whole shoots of wild-type plants, mir156ac mir157ac mutants, and transgenic plants expressing miR156a under the regulation of the constitutive 35S promoter (35S::MIR156A) growing in LD (Figure 1E). As previously reported^13,18,25, miR172 levels increased from 3 to 12 days after planting (DAP) in wild-type plants. Although mir156ac mir157ac had significantly more miR172 than wild-type, this mutant displayed a similar temporal increase in miR172. 35S::MIR156A plants had reduced levels of miR172, but also exhibited a temporal increase in miR172 expression.

These results suggest that the age-dependent increase in miR172 in whole shoots is attributable both to an increase in miR172 during leaf expansion and to a leaf-to-leaf increase in miR172 that is only evident late in leaf development. They also suggest that the increase in the abundance of miR172 in whole shoots and in successive leaves is only partly attributable to a decline in miR156.

The temporal expression of miR172 is attributable to transcriptional regulation of MIR172A and MIR172B.

miR172 is encoded by 5 genes in Arabidopsis thaliana. Next-generation sequencing of small RNAs in the cotyledons and fully expanded leaves of FRI FLC and FRI flc-3 grown in LD revealed that MIR172A and B collectively contribute more than 99%, MIR172C and D contribute less than 1%, and MIR172E makes an insignificant contribution to the mature miR172 pool (Figure 2B). Consistent with our RT-qPCR results (Figure 1D), all of these species increased in abundance from cotyledons to leaf 11 in both non-flowering (FRI FLC) and florally-induced (FRI flc-3) plants (Figure 2B).

To study the function of miR172, we generated loss-of-function mutations in MIR172A and MIR172B using CRISPR-Cas9 (Figure 2B, see also Figure S2). The miR172a-4 allele contains a 204bp deletion that removes the 5’ end of the miR172 stem-loop and 8 nucleotides of the star strand, whereas the miR172b-4 allele contains a 64bp deletion that removes the entire star strand and most of the loop region. RT-qPCR demonstrated that in SD each mutation reduces the amount of miR172 by slightly less than 50%, and that the miR172ab double mutant has less than 2% of the amount of miR172 present in wild-type plants. These results are consistent with our small RNA sequencing data and indicate that MIR172A and MIR172B are the major sources of miR172 in leaves.

Under SD conditions, the amount of miR172 in whole shoots increased approximately 2-fold from 1 to 3 weeks after planting (Figure 3A). The amount of miR172 in mir172a-4 and mir172b-4 also increased by approximately two-fold, demonstrating that MIR172A and MIR172B contribute roughly equally to this increase. Under LD conditions, miR172 increased 8–10 fold in whole shoots during the first two weeks after planting (Figure 3A), or more than 4 times as much as in plants grown in SD conditions. In contrast, LD had little effect on the expression pattern of miR172 in the shoot apex, where it was present at a nearly constant level from 6 to 12 days after planting (Figure 3B, see also Figure 2A).

Figure 3. MIR172A and MIR173B produce most of the miR172 in leaves and are regulated at a transcriptional level.

(A) RT-qPCR analysis of the amount of miR172 in whole shoots of wild-type and mir172a-4, mir172b-4, and mir172a-4 mir172b-4 mutant plants growing in SD and LD. Values are relative to the amount of miR172 in wild-type plants at the first time point. (B) RT-qPCR analysis of the amount of mature miR172 in the shoot apex of wild-type and mir172a-4 mir172b-4 double mutant plants at 6, 8 and 12 DAP (LD). (C) RT-qPCR analysis of the amount of the unprocessed primary transcripts of MIR172A and MIR172B in whole shoots of wild-type plants and mir172a-4 mir172b-4 double mutants at 3 and 7 DAP (LD). n = 3 biological replicates for all analyses, except (A). Data are relative to the first sample in each graph and are shown as the mean ± standard error of the mean. Means were compared using an unpaired two-tailed t-test with Welch’s correction. *p<0.05; **p<0.005; ***, p<0.0005. n.s., not significant. SD = short days; LD = long days. See also Figure S2.

The deletions in mir172a-4 and mir172b-4 remove, respectively, some or all of the star strand and are therefore expected to block the processing of the primary miR172a and miR172b transcripts. Consistent with this expectation, the mutant forms of pri-miR172a and pri-miR172b were significantly more abundant than the WT transcripts (Figure 3C). Both mutant transcripts increased in abundance from day 3 to day 7 in whole shoots grown in LD. These results indicate that the increase in the amount of miR172 during this period is due to the transcriptional activation of MIR172A and MIR172B, not to an increase in their processing efficiency.

MIR156 regulates vegetative phase change independently of miR172

miR156 and miR172 are major regulators of vegetative phase change and flowering time in Arabidopsis thaliana and other species^42–44. However, it has been difficult to determine their specific roles in these transitions because loss-of-function mutations in the genes that encode miR172 have only recently become available^33,34. We took advantage of our loss-of-function alleles of MIR172A and MIR172B as well as previously characterized loss-of-function alleles of genes encoding MIR156 and MIR157³¹ to characterize the developmental phenotypes of plants deficient for miR172, miR156/157, and plants lacking both miRNAs.

In LD, mir172a-4 produced significantly more leaves without abaxial trichomes and more rosette and cauline leaves than wild-type plants (Figure 4A). mir172b-4 did not have a significant effect on abaxial trichome production or rosette leaf number. This observation indicates that although these genes make similar amounts of miR172 (Figure 3A), MIR172A is more important for floral induction than MIR172B. This could be because these genes are expressed in different tissues, or because MIR172A is expressed at a slightly higher level than MIR172B and the AP2-like targets of miR172 are hypersensitive to this small difference. Although the expression patterns of reporter genes for MIR172A and MIR172B have been described in the shoot apex and young leaf primordia^33,34, there is still no information about the expression of these genes in older leaves, where miR172 is present at much higher levels (Figure 1D, Figure 2A).

Figure 4. miR156 regulates vegetative traits independently of miR172.

(A) The effect of miR172a-4, miR172b-4 and miR172a-4 miR172b-4 on abaxial trichome production and leaf number (LD). n=24 for each genotype. Data were analyzed using a one-way ANOVA followed by a Tukey HSD test; significant differences (p=<0.05) are indicated by different letters. (B) The angle of the leaf base in fully expanded leaves of wild-type plants, and plants mutant for genes encoding miR156, miR157 and miR172 (LD). Mutant leaf angles were compared to the wild-type leaf angle using an unpaired two-tailed t-test with Welch’s correction. n=24 (C) The first leaf with abaxial trichomes in wild-type plants and plants mutant for genes encoding miR156, miR157 and miR172 (LD). Dots represent individual samples; The mean and S.E.M are indicated, n=24. (D) The morphology of fully expanded cotyledons and rosette leaves of wild-type plants and plants mutant for genes encoding miR156, miR157 and miR172 (LD). (E) The first leaf with abaxial trichomes in wild-type plants and the miR156-insensitive mutant, spl15-D. n=24 (F, G) The relative expression of miR172 in wild type plants and mutants in genes miR156, miR157 and miR172 in 18-day-old shoot apices (F) and 1-week-old rosettes (G). n = 3. Data are shown as the mean ± standard error of the mean. Means were compared using an unpaired two-tailed t-test with Welch’s correction. *p<0.05; **p<0.005; ***, p<0.0005. n.s., not significant. See also Figure S1 and S2.

The mir172ab double mutant produced abaxial trichomes significantly later than miR172–4 and miR172b-4 and had significantly more rosette leaves than these single mutants (Figure 4A). mir172ab leaves were not significantly different in shape from wild-type leaves in SD (Figure 4B) but were slightly rounder than wild-type leaves in LD (Figure 4D). This latter difference may be due to the difference in the flowering time of these genotypes, as late flowering plants tend to have rounder rosette leaves than early flowering plants⁴⁵. These results indicate that MIR172A and MIR172B have overlapping functions in the regulation of vegetative phase change and flowering time, and that MIR172A is a functionally more important source of miR172 than MIR172B.

The mir156ac mir157ac quadruple mutant had a reduced number of leaves lacking abaxial trichomes in SD (Figure 4C) and produced fewer rosette leaves in LD; furthermore, all its rosette leaves had the narrow shape typical of adult leaves (Figure 4D). The mir156abcd mir157ac hextuple mutant had an even stronger phenotype. All of its rosette leaves had abaxial trichomes (Figure 4C) and were similar to wild-type adult leaves in both shape and size (Figure 4D). To determine if this phenotype is dependent on miR172, we generated mir156ac mir157ac mir172ab hextuple and mir156abcd mir157ac mir172ab octuple mutant lines. In SD, the timing of abaxial trichome production in these lines was intermediate between that of the mir156ac mir157ac or mir156abcd mir157ac line and the mir172ab line (Figure 4C), but was closer to the mir156ac mir157ac or mir156abcd mir157ac line than to the mir172ab line. The rosette leaves of miR156ac miR157ac miR172ab and mir156abcd mir157ac mir172ab were similar in shape to miR156ac miR157ac and mir156abcd mir157ac, but were slightly rounder than these mir156/mir157 mutant lines (Figures 4B and 4D). These results suggest that in addition to regulating abaxial trichome production through their effect on miR172, miR156-targeted SPL genes regulate this trait independently of miR172. They also show that rosette leaf shape is regulated primarily by the miR156/miR157-SPL module, with only a small input from the miR172-AP2-like module.

At least two of the targets of miR156 — SPL9 and SPL15 — directly promote the transcription of MIR172B^13,23,24. To examine the relationship between miR156 and miR172 more directly, we took advantage of spl15-D, a point mutation in the miR156 target site that renders SPL15 insensitive to miR156⁴⁶. We found that spl15-D accelerates abaxial trichome production and that the phenotype of the spl15-D mir172ab triple mutant was intermediate between that of spl15-D and miR172ab (Figure 4E). This result demonstrates that the abaxial trichome phenotype of spl15-D is only partly dependent on MIR172A and MIR172B and implies that SPL15 regulates abaxial trichome production independently of miR172.

An alternative explanation for these results is that SPL15 (and possibly other SPL genes) promote the expression of other MIR172 genes, in which case the phenotype of miR156/imR157 mutants could be due to an increase in the expression of these genes. To address this possibility, we asked whether mir156abcd mir157ac —which has elevated levels of all miR156-regulated SPL genes³¹—affects the abundance of miR172 in combination with mir172ab. We found that mir156abcd mir157ac had elevated levels of miR172 in both the shoot apex and in whole shoots and that mir156abcd mir157ac mir172ab had the same low amount of miR172 as mir172ab (Figures 4F and 4G). This result demonstrates that the effect of mir156abcd mir157ac on miR172 levels is mediated entirely through MIR172A and/or MIR172B, and supports the conclusion that the targets of these miRNAs regulate leaf shape and abaxial trichome production independently of miR172.

The acquisition of photoperiodic competence is largely independent of vegetative phase change

Plants that over-express miR156 flower slightly later than normal in LD^12,23,47,48. This effect is thought to be mediated by miR172 because plants that over-express miR172 flower early^18,25,32, and loss of miR156 leads to an increase in miR172¹³. However, conclusive evidence for this hypothesis is lacking. We took advantage of mutants deficient for miR156/miR157 and miR172 to examine the role of these miRNAs in determining the competence to respond to a LD floral inductive stimulus.

Under LD conditions, the date of the appearance of floral buds in the rosette and the date of the appearance of the first open flower were not significantly different from wild type in both the miR156ac miR157ac mutant and in a transgenic line constitutively over-expressing miR156 under the regulation of the 35S CaMV promoter (Figures 5A and 5B). In contrast, plants overexpressing miR172, as well as plants mutant for the miR172 targets TOE1, TOE2, SMZ, and SNZ, flowered early in LDs (Figures 5A and 5B). Conversely, plants deficient for miR172 were late flowering (Figures 5A and 5B). The flowering time of the miR156ac miR157ac miR172ab mutant was not significantly different from that of miR172ab. These results are consistent with previous studies indicating that miR172 promotes flowering in LD^18,25,27,32, and demonstrate that miR172 is more important for floral induction under these conditions than miR156/miR157.

Figure 5. Effect of the miR156/SPL and miR172/AP2-like modules on reproductive competence.

(A) Days to first visible floral bud and (B) days to first open flower in LD- grown wild-type plants and plants transformed with 35S::miR156a (miR156ox), 35S::miR172b (miR172ox), or mutant for miR156, miR157, miR172, or AP2-like genes. (C) Scheme of the double shift assay for floral induction. Plants were grown in SD for 7 days, then moved to LD for different lengths of time, and then moved back to SD until bolting. The floral induction efficiency was calculated as described in the text. (D to G) The induction efficiency of a transient LD exposure on wild-type and mutant plants at one week (D and E) and two weeks (F and G) after planting. The asterisk indicates a statistically significant difference between wild-type and (D and F) miR172a/miR172ab or (E and G) toe1 toe2/toe1 toe2 smz snz. (H) Scheme of the single shift floral induction assay. Seeds were transferred from 4°C to LD (Day 0), grown for the indicated number of days and then transferred to SD until bolting. Floral induction efficiency was calculated as described in the text. (I) The number of LDs required to induce flowering in wild-type plants and in plants over-expressing miR156 or miR172, or mutant for these miRNAs. n = 24 for all genotypes. Data are the mean ± S.E.M.; Samples were compared using an unpaired two-tailed t-test with Welch’s correction *p<0.05; **p<0.005; ***, p<0.0005. See also Figure S3.

Floral induction depends on the ability of the shoot to perceive a floral stimulus as well as factors that transduce this initial signal into the production of an inflorescence. Although a great deal is known about the mechanism of floral induction, the basis for the developmental increase in the ability of the shoot to respond to floral stimuli under environmental conditions that promote flowering (i.e., the acquisition of reproductive competence) is unknown. We explored the role of miR156/SPL and miR172/TOE play a role in this process by examining the effect of mutations in MIR156/157 and MIR172 genes on the ability of plants to respond to transient exposure to LD.

In an initial experiment, plants were grown under SD conditions for 1 week and then transferred to LDs for different periods of time. After the LD induction period, plants were transferred back to SD until flowering (Figure 5C). Inspired by the method used by Torti et al⁴⁹, the effect of these treatments was measured by comparing the difference between the leaf number of plants exposed to a LD stimulus (experimental plants) and plants grown continuously in SD, relative to the difference between the leaf number (LN) of plants continuously in LD and SD, i.e., (LN_SD – LN_exp)/(LN_SD - LN_LD). In the case of wild-type plants, a 3-day exposure to LD was 8.4% as effective in inducing flowering as continuous exposure to LD, whereas 5, 7, and 9 LD treatments were, respectively, 53.4%, 74.7%, and 89.5% as effective as continuous LD conditions (Figures 5D and 5E, see also Figure S3A). The response of mir172b-4 was not significantly different from the wild type, but mir172a-4 was less sensitive to LD than the wild type, and the mir172ab double mutant was even less sensitive (Figure 5D, see also Figure S3A). Loss-of-function mutations in the targets of miR172 had the opposite phenotype. In response to a 3 LD treatment, smz snz exhibited less than 10% of the efficiency of continuous LD, but the toe1 toe2 double mutant exhibited 30%, and the toe1 toe2 smz snz quadruple mutant displayed nearly 70% of the floral induction efficiency of continuous LD (Figure 5E, see also Figure S3B). We used this same transfer assay to examine the photoperiodic sensitivity of 2-week-old plants and observed a similar pattern of induction efficiencies (Figures 5F and 5G, see also Figures S3C and S3D), although all genotypes were more sensitive to LD than 1-week-old plants (Figures 5D and 5E).

To determine if the miR156/SPL and miR172/TOE modules regulate the acquisition of photoperiodic competence, we germinated seeds under LDs and then transferred seedlings to SDs at successively later times (Figure 5H). As in the experiment described above, the effect of this treatment was assessed by comparing the difference between the leaf number of experimental plants and plants grown continuously in SD, to the difference between plants grown continuously in LD and SD (Figure 5I, see also Figure S3E). This approach revealed that wild-type Col-0 begins to become committed to flowering starting 6 DAP, and is fully committed by 10 DAP. Plants deficient for miR156 and miR157 (mir156ac mir157ac) were not significantly different from wild-type, and plants overexpressing miR156 were delayed 1–2 days compared to wild-type plants. In contrast, plants overexpressing miR172 began to become florally committed by 2 DAP and were fully committed by 5 DAP, whereas the mir172ab mutant began to become committed by 9–10 DAP and did not become completely committed until 15 DAP. The mir156ac mir157ac mir172ab line was essentially identical to mir172ab. These results demonstrate that the acquisition of photoperiodic competence in Arabidopsis is regulated primarily by the miR172/AP2-like module, with only minor input from the miR156/SPL module.

DISCUSSION

In Arabidopsis, floral induction is controlled by a variety of external and internal factors, including photoperiod, prolonged cold, ambient temperature, gibberellin, and the age of the shoot. The effect of shoot age is illustrated by the fact that even when all other conditions are optimal, plants make several rosette leaves before flowering. This transition between a reproductively incompetent and a reproductively competent state of shoot development is widespread in the plant kingdom and is typically assumed to reflect a global process of shoot maturation that includes changes in vegetative morphology. However, there is little evidence for the existence of a single “shoot maturation program”. In the perennial Arabis alpina⁴¹, reproductive competence requires both vernalization and low levels of miR156, but in both maize¹¹ and Arabidopsis^12,13 juvenilized mutants or transgenics with high levels of miR156 flower at nearly the same time as wild-type plants. These observations suggest that the acquisition of reproductive competence may be largely independent of vegetative phase change, at least in some species.

We investigated the relationship between vegetative phase change and photoperiodic competence in Arabidopsis by characterizing the genetic interaction between miR156/157 and miR172, two families of miRNAs that have important roles in these transitions. We found that miR156/miR157 regulate vegetative traits associated with vegetative phase change—specifically, leaf shape and size and abaxial trichome production—but have no significant effect on the age at which seedlings become competent to respond to a floral-inductive LD photoperiod. In contrast, miR172 regulates abaxial trichome production and has a small effect on leaf shape, but has a major effect on the sensitivity of plants to LD. miR172 also has a significant effect on flowering under SD conditions^33,34, when Arabidopsis flowers weeks after vegetative phase change and miR156 is at a very low level¹². These results are consistent with a comparison of the timing of vegetative phase change and flowering in natural accessions of Arabidopsis, which showed that flowering time was poorly correlated with the level of miR156 or the timing of vegetative phase change⁴. Our evidence that the expression pattern of miR172 varies more in different accessions of Arabidopsis than the expression pattern of miR156 may provide an explanation for this observation.

The relationship between the vegetative phase of a plant and its reproductive competence may differ depending on the ecology of a species or an ecotype. For example, plants that overwinter before flowering may use a combination of high levels of FLC (or another regulator of vernalization) and miR156/157 to ensure that they do not flower prematurely⁵⁰. The vegetative phase of the shoot may be less important in species that have less stringent environmental requirements for flowering. The relationship between the vegetative phase and reproductive competence may also vary depending on the specific level of miR156. In Arabidopsis³¹, Vachellia collinsii⁵¹ and poplar⁵², miR156 is present at very high levels in the first few juvenile leaves and at much lower and gradually declining levels in subsequent juvenile leaves. The high level of miR156/157 very early in shoot development may repress flowering, whereas the lower level of miR156/157 later in the juvenile phase may be sufficient to promote juvenile vegetative traits but not to repress flowering. This may explain why some trees flower in a vegetatively juvenile state. Finally, miR156/157 and miR172 expression may be regulated independently by the same factors, in which case vegetative phase change and reproductive competence would change in a coordinated fashion without being linked. This latter possibility is suggested by the fact that many plants undergo vegetative phase change a few weeks or months after germination, but flower years later.

To determine how miR172 regulates photoperiodic competence, it is necessary to determine where in the plant miR172 operates. In LD, wild-type Col-0 becomes fully committed to flowering between 6 and 10 DAP. Several observations suggest that this process is the result of an increase in the level of miR172 during the expansion of cotyledons and leaves 1–3, rather than by an increase in the level of miR172 in the shoot apex. Under LD conditions, miR172 promotes flowering primarily by promoting the expression of the floral inducer, FT²⁵. FT is expressed in cotyledons, expanding leaves, but not the shoot apex, leaf primordia, or roots^53,54. Under our LD conditions, 10-day-old seedlings have fully expanded cotyledons and 3 leaves that are larger than 2 mm, suggesting that the effect of miR172 on photoperiodic competence is mediated by these organs. Consistent with this prediction, the abundance of miR172 increases during the expansion of leaves 1–3 and from leaves 1&2 to leaf 3. In contrast, miR172 is expressed at a relatively low and constant level in 2 mm leaf primordia during the period when plants become florally committed. These observations suggest that the shoot apical meristem and young leaf primordia play little or no role in the acquisition of photoperiodic competence, and also support the conclusion that this process is regulated independently of vegetative phase change because the only organs that are capable of expressing FT from day 6 to 10 are juvenile organs. We suspect that, in Arabidopsis, the developmental increase in reproductive competence (i.e., the “age-dependent pathway”) does not result from a change in gene expression in successive leaves, but rather from a change in gene expression during the expansion of cotyledons and leaves 1–3.

Our observation that the increase in the expression of miR172 in successive leaves was largely unaffected by 35S::MIR156A, or by a significant reduction in miR156 in the mir156ac mir157ac mutant, is surprising given that the overall level of miR172 was, respectively, lower or higher than normal in these genotypes. This suggests that the increase in miR172 expression in successive leaves is mediated by factors other than miR156. One possibility is that miR156 sets the upper limit of miR172 expression, while leaf-to-leaf variation in miR172 expression is regulated by other factors. Candidates include cytokinin⁵⁵ and the transcription factors SHORT VEGETATIVE PHASE^56,57 and TEMPERANILLO1/2^58,59, all of which regulate the level of miR172 independently of miR156.

In conclusion, the evidence presented here demonstrates that, in contrast to our original proposal¹³, miR156 plays a relatively minor role in the regulation of miR172 and does not have a major effect on the ability of plants to respond to a floral inductive stimulus. Our original model was based primarily on the phenotype of transgenic plants with abnormally elevated levels of miR156 or miR172. Future research on the biological functions of miR156/157 and miR172 will need to examine this question by manipulating the abundance of these miRNAs in specific organs, within their normal range of expression.

STAR★Methods

RESOURCE AVAILABILITY

Lead contact

Further information and requests should be directed to and will be fulfilled by the lead contact, Dr. R. Scott Poethig (spoethig@sas.upenn.edu)

Materials availability

The new materials generated in this study are available from the lead contact, R. S. Poethig, and from the Arabidopsis Biological Resource Center (https://abrc.osu.edu).

Data and code availability

• This study did not generate any unique code.

• Any additional data required to reanalyze the data reported in this paper are available from the lead contact upon request.

EXPERIMENTAL MODEL AND SUBJECT DETAILS

Arabidopsis thaliana, accession Columbia-O (Col-0) was used as wild-type and all mutants were in a Col-0 background. Seeds were sown on Farfard #2 soil in 96-well 1020 flats and placed at 4°C for 3 days before moving to long-day (16 hrs light/8 hrs dark; 110 μmol m⁻² s⁻¹) or short-day conditions (8 hrs light/10 hrs dark; 110 μmol m⁻² s⁻¹) with illumination provided by a 3:1 ratio of broad spectrum (Philips TL950) and red light-enriched (Interelectric Gro-lite) T8 fluorescent lights.

METHOD DETAILS

Plant material and growth conditions

Generation of mir172a and mir172b loss-of-function mutations

The miR172A and miR172B loss-of-function mutations were generated by CRISPR-Cas9 mutagenesis in a ku70–1 (SALK_123114c) background using the psgR-Cas9-At plasmid as described⁶⁰. Oligonucleotides for three small guiding RNAs (sgRNA) (sgRNA-A: 5’-ATG CCG ATT TGT CTT GTT GA-3’, sgRNA-B: 5’-CAT CCA TCA ACA AGA CAA AT-3’ and sgRNA-C: 5’-CAA CAA GAC AAA TCG GCA TC-3’) were synthesized by Sigma Corp, and phosphorylated and annealed on a BioRad T100 Thermo cycler using T4 polynucleotide kinase (New England Biolabs) in T4 ligation buffer (New England Biolabs) using the program: 30°C 30 minutes, 95°C 5 minutes, ramp down to 25°C at 5°C/minute. The plasmid psgR-Cas9-At was digested with BbsI, gel-purified, and ligated with the annealed double-stranded oligonucleotides. The sequence of the primers used for this construction is provided in Table S1.

Plant transformation

The pNapin:eGFP-Cas9-MIRNA172A/B-sgRNA plasmids were transformed into Agrobacterium tumefaciens strain GV3101, which was then used to transform the ku70–1 mutant by floral dipping⁶¹. T1 seeds were screened using a Leica MZ FLIII stereomicroscope and long-bandpass GFP filter (ET480/40x/ET510 LP) to identify transgenic seeds displaying strong green fluorescence. T2 seeds were collected separately from individual T1 plants. Genomic DNA was extracted from approximately 300 T2 seeds, PCR-amplified using the primer pairs (5’-TGG AAA CTC TTC CTC TGT TTT TG-3’ and 5’-CCA CAG GAG TTT TTG AAT GAA C-3’), and the resulting products were run on a 3% agarose gel to identify families segregating deletions in MIR172A or MIR172B. Non-fluorescent T2 seeds were planted from lines containing deletions, and individual plants were assayed by PCR to identify individuals homozygous for the deletion. Amplicons were subsequently sequenced to determine the nature of the mutation. The miR172a-4 and miR172b-4 mutants were then backcrossed to Col-0, the F1 plants were selfed, and the resulting F2 plants were then genotyped to identify miR172 mutants lacking ku70–1. The sequence of the primers used to genotype these mutants is provided in Table S1.

RNA extraction and quantification

Samples were collected from plants grown under SD or LD conditions. Shoot apex samples consisted of the shoot apical meristem and leaf primordia less that 0.5 mm in length, leaf primordia were harvested at a length of 2mm or 18–22mm as indicated in the text, and whole shoot samples consisted of all above-ground tissue. Total RNA was extracted using Trizol (Invitrogen) according to the manufacturer’s instructions. 1ug of total RNA was treated with RNase-free DNaseI (Ambion), and 1.5ug of the product was reverse transcribed using Superscript III (Invitrogen). MicroRNA-specific reverse transcription primers were used to amplify miR156, miR172, snoR101 and Oligo(dT)₂₁. Three-step real-time PCR protocols were used and run in triplicate. The average was calculated and normalized to snoR101 (for miRNAs) and for eIF4a (for protein-coding genes). The sequence of the primers used for these experiments is provided in Table S1.

Small RNA Sequencing

Fully expanded leaves of FRI FLC and FRI flc-3 plants grown in LD conditions⁴⁵ were used to generate sequencing libraries. A lab-assembled small RNA library sample preparation tool was used to prepare the libraries, followed by high-throughput sequencing with Illumina’s Genome Analyzer II platform³¹. Small RNA sequence data are available in the NCBI Gene Expression Omnibus database under series accession number GSE72303.

Floral induction assay

Plants of the indicated genotypes were sown on Farfard #2 soil and kept at 4°C for 3 days before being transferred to a growth chamber programmed for SD (8-hour light/16-hour dark, 22 °C) or LD (16-hour light/8-hour dark, 22 °C), depending on the experiment. Seedlings were thinned to individuals of the same size shortly after germination. Transient SD or LD treatments were performed by transferring plants to chambers programmed for these photoperiods and then back to the original chamber. Plants were watered with dilute fertilizer (Peters Professional^® Water Soluble Fertilizer 20-10-20) every week while growing in SD.

QUANTIFICATION AND STATISTICAL ANALYSIS

All statistical tests and analyses were conducted in GraphPad Prism 9 (GraphPad Software, La Jolla, CA, USA) or R version 3.6.1. The number of samples analyzed, the statistical test used for analysis, and the precision measures, are provided in the figure legends.

KEY RESOURCES TABLE

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Critical commercial assays
SuperScript™ III First-Strand Synthesis System	Invitrogen	18080051
Turbo DNA-free™ Kit	Invitrogen	AM1907
2x SYBR Green qPCR Master Mix	Bimake.com	B21203
Experimental models: Organisms/strains
A. thaliana: Columbia-0
A. thaliana: 35S::miR156a (miR156ox)	Xu et al.12, Wu et al13, Wu et al21, Schwarz et al22, Wang et al23 and Hyun et al24	N/A
A. thaliana: 35S::miR172b (miR172ox)	Wu et al13	N/A
A. thaliana: toe1–2	Arabidopsis Biological Resource Center	SALK_069677
A. thaliana: toe2–1	Arabidopsis Biological Resource Center	SALK_065370
A. thaliana: smz	Arabidopsis Biological Resource Center	SALK_110654
A. thaliana: snz	Arabidopsis Biological Resource Center	SALK_030031
A. thaliana: toe1–2 toe2–1	Yant et al26	N/A
A. thaliana: smz snz	Yant et al26	N/A
A. thaliana: toe1–2 toe2–1 smz snz	Yant et al26	N/A
A. thaliana: miR156a-2 miR156c-1 miR157a-1 miR157c-1	He et al31	N/A
A. thaliana: miR156b-1	He et al31	N/A
A. thaliana: miR156d-1	Arabidopsis Biological Resource Center	SALK_40772
A. thaliana: miR156a-2 miR156b-1 miR156c-1 miR156d-1 miR157a-1 miR157c-1	This study	N/A
A. thaliana: miR172a-4	This study	N/A
A. thaliana: miR172b-4	This study	N/A
A. thaliana: miR172a-4 miR172b-4	This study	N/A
A. thaliana: miR172b-4 spl15–1D	This study	N/A
A. thaliana: miR172a-4 miR172b-4 spl15–1D	This study	N/A
A. thaliana: miR156a-2 miR156b-1 miR156c-1 miR156d-1 miR157a-1 miR157c-1 miR172a-4 miR172b-4	This study	N/A
Software and algorithms
GraphPad Prism 9.0	https://www.graphpad.com
R Version 3.6.1	www.r-project.org

Highlights.

• Reproductive competence is independent of vegetative phase change in Arabidopsis.

• miR172 regulates the age-dependent increase in reproductive competence.

• miR156 plays a minor role in the regulation of miR172.