In perennial Arabis alpina, CONSTANS and FLOWERING LOCUS T have common and distinct effects on flowering and inflorescence architecture
Niharika Sashidhar, George Coupland
TL;DR
The study explores how CONSTANS and FLOWERING LOCUS T genes control flowering and plant structure in the perennial plant Arabis alpina.
Contribution
The paper reveals both shared and unique roles of CONSTANS and FLOWERING LOCUS T in regulating flowering and inflorescence architecture in a perennial plant.
Findings
AaCO and AaFT/TSFL mutations delay flowering under long days and affect inflorescence branching.
AaCO and AaFT/TSFL have distinct effects on axillary branch flowering and primary shoot development.
RNA-seq data shows gene responses to AaCO and AaFT/TSFL in different plant parts.
Abstract
Flowering of perennial Arabis alpina is differentially regulated on primary and axillary shoots. Although contributions of vernalization and ageing pathways have been analysed, those of photoperiodic flowering genes CONSTANS (CO), FLOWERING LOCUS T (FT), and TWIN‐SISTER OF FT (TSF) remain unexplored.CRISPR‐Cas9 mutations in AaCO and AaFT/TWIN SISTER OF FT‐LIKE (TSFL) were recovered. Aaco and Aaft/tsfl mutants in pep1 background were scored for flowering time, inflorescence branching, and floral phenotypes under long (LD) or short days (SD) and after vernalization. RNA‐seq data on primary and axillary branches were compared. AaCO activates AaFT/TSFL transcription in leaves, and Aaco and Aaft tsfl mutations delay flowering under LDs. Axillary branches flowered in Aaco mutants but not in Aaft tsfl mutants. Both lacked inflorescence branches and flowers on the primary shoot under LDs.…
Genes, proteins, chemicals, diseases, species, mutations and cell lines named across the full text — each resolved to its canonical identifier and authoritative record.
Click any figure to enlarge with its caption.
Fig. 1
Fig. 2
Fig. 3
Fig. 4
Fig. 5- —Deutsche Forschungsgemeinschaft (DFG) – Walter Benjamin Programme
- —DFG through Cluster of Excellence CEPLAS
- —DFG through ERA‐NET SusCrop2
- —BMBF grant Epi‐Brass
Peer Reviews
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
Taxonomy
TopicsPlant Molecular Biology Research · Plant Gene Expression Analysis · Plant Reproductive Biology
Introduction
Reproductive strategies in angiosperms are highly variable and their life history can be annual, biennial, or perennial (Friedman, 2020). Most Angiosperm species are perennial, but annuals have evolved repeatedly (Hjertaas et al., 2023). These differences in life cycle can have profound effects on the ecology and competitiveness of a plant and are largely dependent on variation in meristem fate. Annual plants such as Arabidopsis thaliana (L.) Heyn undergo rapid, single‐season growth, with the shoot apical (SAM) and axillary meristems (AMs) transitioning to reproductive growth. The plant then sets seeds and senesces, and all the meristems are consumed (Hensel et al., 1994). By contrast, perennials, such as Arabis alpina L., live for several years, alternating between vegetative and reproductive phases. Some AMs remain vegetative, while others undergo floral transition, with vegetative branches entering dormancy to regenerate new growth (R. Wang et al., 2009). The Brassicaceae family has emerged as a model to understand the genetics of diversification of life history, because sister annual and perennial species have evolved several times independently (Kiefer et al., 2017; Zhai et al., 2024).
The variation in meristem fate that underlies differences in the plant life cycle in the Brassicaceae is conferred by floral repressors in the MADS domain family, particularly in clades related to FLOWERING LOCUS C (FLC) (R. Wang et al., 2009; Madrid et al., 2021; Zhai et al., 2024). In A. thaliana, FLC confers a flowering response to winter cold (vernalization). Cold exposure silences FLC transcription, and this repression persists after returning to warm conditions through stable histone modifications at the FLC locus (Michaels & Amasino, 1999; Sheldon et al., 1999; Bastow et al., 2004; Sung & Amasino, 2004). The stable repression of FLC in the warm allows the SAM and all AMs to form flowers, leading to the production of seeds. In perennial A. alpina, the FLC orthologue PERPETUAL FLOWERING 1 (PEP1) is repressed during vernalization allowing floral transition to occur in all meristems (R. Wang et al., 2009; Lazaro et al., 2018). However, on return to warm, transcription of PEP1 is reactivated, maintaining meristems that have not undergone floral transition in the vegetative state (R. Wang et al., 2009). In A. alpina and several other Brassicaceae species, mutation of the FLC orthologue causes continuous flowering in which all AMs flower independently of vernalization (R. Wang et al., 2009; Zhou et al., 2013; Zhai et al., 2024). FLC‐related genes of the MADS AFFECTING FLOWERING (MAF) family and the related FLOWERING LOCUS M (FLM) or MAF‐RELATED (MAR) families also contribute to the repression of flowering in Brassicaceae perennials, and can be fully genetically redundant with the FLC orthologue (Madrid et al., 2021; Zhai et al., 2024).
In addition to vernalization, other environmental and developmental cues control floral transition in A. thaliana via a network of pathways that optimize the timing of reproduction (Kinoshita & Richter, 2020). Among these pathways is one involving the SQUAMOSA PROMOTER BINDING PROTEIN‐LIKE (SPL) transcription factors (J. W. Wang et al., 2009), which are negatively regulated by miR156 (Rhoades et al., 2002). The transcription of MIR156 genes is repressed in older plants, allowing expression of SPLs, and therefore, this pathway is considered an age‐related pathway to flowering. SPL15 is negatively regulated by miR156 and also transcriptionally repressed by FLC, making it a key point of convergence between the age‐related and vernalization pathways. In perennial species, this interaction enables an age‐dependent flowering response to vernalization (Bergonzi et al., 2013; Zhou et al., 2013; Hyun et al., 2019).
In contrast to vernalization and ageing, the role of the photoperiodic pathway in regulating floral transition in A. alpina and related perennials remains underexplored. This widely conserved pathway is important in synchronizing flowering with fluctuating day length and is likely to be key to understanding how A. alpina maximizes its reproductive success in its high‐altitude or high‐latitude habitats. A. thaliana and many other Brassicaceae species are facultative long‐day (LD) plants that flower much earlier when exposed to long photoperiods within each 24‐h cycle. Exposure of A. thaliana to long photoperiods leads to increased transcription of the CONSTANS (CO) gene and to stabilization of the B‐box transcription factor it encodes (Valverde et al., 2004; Sawa et al., 2007). CO then activates transcription of FT and its paralogue TWIN SISTER OF FT (TSF) in the companion cells of the phloem (Kardailsky et al., 1999; Kobayashi et al., 1999; An et al., 2004). The FT florigen protein is transported through the phloem to the SAM, where it initiates floral transition (Corbesier et al., 2007; Jaeger & Wigge, 2007; Mathieu et al., 2007). Under non‐inductive short days (SD), flowering occurs independently of the CO‐FT module through the activity of SPL transcription factors (J. W. Wang et al., 2009; Hyun et al., 2016). The roles of CO and FT orthologues have not been studied in detail in herbaceous perennials. However, FT genes were implicated in flowering control in Arabidopsis lyrata, a perennial relative of A. thaliana, in inflorescence development of woodland strawberry and in the regulation of bud dormancy in perennial trees (Böhlenius et al., 2006; Hsu et al., 2011; Kemi et al., 2019; André et al., 2022; Lembinen et al., 2023).
We studied the function of the CO‐FT regulatory module in flowering of the model perennial A. alpina. Using CRISPR‐Cas9, we generated mutations in the CO and FT/TSF genes of A. alpina and analysed their effects in the pep1 mutant background. We found that mutation of the FT or the CO genes strongly delayed flowering, but also had unexpectedly strong effects on inflorescence development on the primary shoot (PS), such that mutants failed to form inflorescence branches (inflorescence phase 1 (I1)) or flowers (inflorescence phase 2 (I2)). In addition, we found that on axillary branches (ABs) formed in the axils of leaves (V1) Aaco pep1 mutants flowered, whereas Aaft tsfls pep1 mutants did not form flowers on these ABs. We discuss the impact of these genes on the perennial life cycle of A. alpina.
Materials and Methods
Synteny and phylogenetic analyses
Synteny analysis was performed to investigate the genomic organization and conservation of genes across multiple Arabis species and Arabidopsis thaliana (L.) Heynh. Genomic sequence data for Arabis alpina L. accessions were generated using Pac‐Bio long‐read, pair‐end sequencing. All the obtained reads were mapped using the Pajares published reference genome (Willing et al., 2015). Genomic sequences for the target genes were retrieved from genome sequences of Arabis species and A. thaliana. The sequences were first reanalysed and mapped to their respective genomes using blast. Gene models were carefully validated and adjusted where necessary to ensure correct alignment. Synteny blocks were identified by comparing conserved regions between species using custom scripts in R in ggplot2 package. Additionally, to account for gene duplications commonly observed in Arabis, duplicate mappings were allowed during alignment to ensure all relevant paralogues were captured in the analysis. AaFT genes were previously referred to as AaFT1 to AaFT4 (Hyun et al., 2019); however, on the basis of the synteny analysis here, they were renamed AaFT, TSFL1, TSFL2, and TSFL3.
A phylogenetic tree was constructed using the protein sequences of AaCO, AaFT, and TSFLs. Amino‐acid sequences were aligned using the MUSCLE algorithm with default parameters. Phylogenetic trees were constructed using the Maximum Likelihood method implemented using MEGA11, applying the LG substitution model, which was selected based on its performance for protein sequence evolution. Statistical support for the tree topology was evaluated with 100 bootstrap replicates.
Gene expression analysis
Diurnal gene expression analysis was performed on a fully developed fifth leaf from plants grown under LD and SD conditions. For each genotype, three biological replicates were collected every 2 h in LDs and every 3 h in SDs. Total RNA was extracted using the RNeasy kit (Qiagen, Hilden, Germany), and 1 μg total RNA was treated with DNAse using the Turbo DNAse kit (Thermo Fisher Scientific, Waltham, MA, USA). cDNA synthesis was performed using oligo(dT)18 primer and Superscript II reverse transcriptase (Roche) according to the manufacturer's instructions. Quantitative real‐time RT‐PCR using a BioRad iQ5 apparatus and SYBR Green I detection was performed using 2 μl of a 1 : 10 dilution of cDNA:H_2_O. Arabis alpina PROTEIN PHOSPHATASE 2A (AaPP2A) was used as a reference gene to normalize expression data. Relative gene expression was determined using the 2^−ΔΔCt^ method (Livak & Schmittgen, 2001).
The tissue‐specificity of AaFT, TSFL, and AaCO gene expression was analysed by qRT‐PCR on RNA from leaves, flowers, and siliques of plants grown either in LD or SD conditions. Shift assays were performed 5 wk after germination in the respective growth conditions. A time–course series was performed and leaf samples were harvested each week at ZT16 and fortnightly at ZT8, in LD and SD, respectively. A new fully formed leaf at weekly intervals was harvested to measure the mRNA expression. Flowers and siliques were harvested at ZT6 from plants that were vernalized. The first two completely opened flowers and young siliques at 6 d after pollination were harvested for RNA isolation.
Plant material and growth conditions
The perennial Spanish Arabis alpina accession ‘Pajares’ was used to analyse expression of photoperiodic pathway genes. To generate CRISPR‐Cas9 mutant lines, pep1 mutant plants were used because they flower perpetually from all meristems, leading to the formation of more flowers and thereby increasing the efficiency of transformation. Furthermore, we used the pep1 background to study the effect of photoperiod on A. alpina plants that produce flowers on all ABs independently of vernalization. Seeds were stratified at 4°C in the dark for 3 d and plants were grown in controlled glasshouse conditions or growth cabinets either in SDs (8 h : 16 h light : dark) or LDs (16 h : 8 h, light : dark) at a light intensity of 200 to 500 μmol m^−2^ s^−1^ at 22°C. Plants were vernalized in a growth chamber at 4°C and at a light intensity of 14 μmol m^−2^ s^−1^ either in LDs or SDs. Days to first open flower (DTFO) were scored as the days from germination. Between 5 and 15 plants were used for phenotypic analyses.
Confocal imaging of shoot apical meristems in pep1 background
SAMs were harvested from plants grown under various conditions. Three meristems per time point for each genotype were analysed. Apices were harvested on ice and fixed with 4% (w/v) paraformaldehyde in PBS by vacuum infiltration three times for 10 min. Samples were then incubated overnight in the dark and washed with PBS buffer twice for 10 min. The tissue was cleared using ClearSee (Kurihara et al., 2015) until it was transparent. The SCRI Renaissance 2200 cell wall‐specific marker was added to the ClearSee solution and samples were kept in the dark at room temperature until imaging. Confocal images with z‐stack intervals were acquired using a Leica SP8 confocal microscope (Leica Microsystems, Wetzlar, Germany).
Cloning and generation of CRISPR‐Cas9 lines
Two target sites were used to knockout AaFT and TSFL1, TSFL2 and TSFL3 genes in A. alpina. Target 1 (5′‐ACGGATATTCCTGCCACAAC‐3′) is specific to exon 3 of AaFT, and target 2 (5′‐AGTCCAAGCAACCCTTACC‐3′) is conserved in exon 2 of TSFL1, TSFL2 and TSFL3. For AaCO, two target sites in exon 1 were used (5′‐CGAGTCATGTGAGCGTGCCC‐3′ and 5′‐CGAGACAACAATGACCAATC‐3′). Cloning of the guide RNAs was performed as described in (Xing et al., 2014). The final binary plasmid was cloned into Agrobacterium tumefaciens strain GV3101 (pSOUP). Following transformation by floral dipping of pep1 plants (Clough & Bent, 1998; J. W. Wang et al., 2009), T_0_ seeds were screened on solid agar plates containing MS medium without sucrose and 100 mg ml^−1^ hygromycin (Roche). Positive transformants were selected on the basis of longer hypocotyls and roots, and were transferred to soil. Transformants were verified by PCR for the presence of the transgene. To identify gene editing, genes were amplified with gene‐specific primers and the resulting PCR amplicons were cloned into the pGEM‐T vector and sequenced by Sanger sequencing (Supporting Information Table S3).
Propagation of CRISPR‐Cas9‐edited plants
Transformation with the AaFT‐TSFL‐sgRNA‐Cas9‐Hyg cassette resulted in 28 T_1_ transformants in pep1 background, and sequencing of the AaFT and TSFL genes in each transformant revealed various combinations of Aaft and tsfl mutations (Fig. S5; Table S2). Self‐pollination of the T_1_ plants led to various mutant genotypes in the T_2_ generation, either through segregation of edited alleles identified in the T_1_ or the identification of newly edited alleles (Table S2). Selected mutants were propagated through subsequent generations until the mutations of interest were homozygous. Similarly, 10 T_1_ transformants were obtained when transformed using the AaCO‐sgRNA‐Cas9‐Hyg cassette. Two mutants with frameshift deletions were selected for further analysis (Table S2). These plants were self‐pollinated to produce T_2_ and T_3_ generations, which were then used for phenotyping analysis.
Transcriptome analysis of Aaco and Aaft tsfl mutants
RNA‐seq analysis was performed along a time–course on material from plants grown in LDs and tissues were harvested at ZT16. SAMs and AMs were harvested at each time point for RNA sequencing analysis. Apical tissue enriched for SAM was excised under a stereo microscope and visible leaf primordia were removed. A total of c. 20 million reads per sample were obtained as 100 bp paired‐end reads from BGI, Poland. Raw sequencing data were processed using Galaxy (Blankenberg et al., 2014). Reads were aligned to the A. alpina reference genome (Willing et al., 2015) using STAR (Dobin et al., 2013), and gene‐level counts were generated with FeatureCounts (Liao et al., 2014). Normalization was performed using the DESeq2 median of ratios method, which corrects for differences in sequencing depth and RNA composition across samples. Differential gene expression analysis was conducted using the DESeq2 Wald test to calculate log_2_ FC and associated P‐values for each gene. P‐values were adjusted for multiple testing using the Benjamini–Hochberg false discovery rate correction to control for type I errors (Love et al., 2014).
Gene Ontology (GO) term enrichment analysis was performed using clusterprofiler (Yu et al., 2012), focusing on biological process terms with A. thaliana annotations. To reduce redundancy among enriched GO‐terms, semantic similarity‐based clustering was performed using rrvgo (Yu et al., 2010; Sayols, 2023).
Statistical analysis
To minimize phenotypic plasticity, plants were randomly assigned to the glasshouse cabinets. The data are presented as means ± SD from three to five independent biological replicates, depending on the type of experiment. Statistical significance between the means of different groups was assessed using a one‐way ANOVA followed either by the Tukey post hoc test or the Wilcoxon test. Differences were considered significant at *, P < 0.05, highly significant at **, P < 0.01, and very highly significant at ***, P < 0.001. Figures were generated using R Studio, for analysing and making graphs (R Core Team, 2021; Wickham, 2016; Xu et al., 2021).
Results
Flowering response to daylength in perennial Arabis alpina correlates with transcription of
CO ‐ FT genes
The well‐characterized Spanish accession Pajares was used to investigate the CO‐FT module in the herbaceous perennial A. alpina. A CO homologue was identified on A. alpina chromosome 8, flanked by orthologues of the genes neighbouring CO in A. thaliana (At5G15830 and AT5G15850 (AtCOL1)) (Fig. 1a). Phylogenetic analysis confirmed it is more closely related to AtCO than to other AtCOL proteins, supporting its identity as the AtCO orthologue (Fig. S1A). In addition, four homologues of FT were identified. One of these genes, AaFT, showed conservation of synteny with FT and FAS1 of A. thaliana, and was previously named AaFT1 (Adrian et al., 2010) (Fig. 1b). However, the other three genes showed no conservation of synteny with FT or its paralogue TSF (Fig. S2). Phylogenetic analysis was performed on the protein sequences of FT and TSF of A. thaliana, and the FT homologues identified in A. alpina accessions and related species (Fig. S1B; Table S1). This analysis indicated that AaFT is present in the same clade as FT, consistent with the conservation of synteny. However, the remaining three proteins were more closely related to TSF of A. thaliana. Therefore, we named the corresponding genes TWIN SISTER OF FT‐LIKE (TSFL1), TSFL2, and TSFL3. The FT and TSFL2 subclades contained proteins from all Arabis species. However, TSFL1 was only found in the A. alpina accessions suggesting that it might have arisen relatively recently by duplication in the A. alpina lineage.
Genome organization and expression of photoperiodic flowering genes, and architectural plots of Arabis alpina Pajares and pep1. (a, b) Conservation of synteny of AaCO and AaFT among Arabis alpina accessions, other Arabis species and Arabidopsis thaliana. In (b) the AaFT promoter elements (Block a–e) are described in Zicola et al. (2019). Relative distances of the genes on the respective chromosome from the first flanking gene are shown on x‐axis. (c, d) Boxplots and representative architecture plots showing days to visible buds and days to first open flower (DTFO) of the shoot structure of Pajares and pep1 plants grown under 10 wk SD → 12 wk SD‐V → SD and 5 wk LD → 12 wk SD‐V → LD, respectively. Letters below boxplots indicate statistically significant differences in flowering time (Tukey HSD, P < 0.05). Refer to Supporting Information Fig. S4 for a detailed illustration of node/branch types. (e) A line plot showing the profiles of AaCO, AaFT, and TSFL mRNAs in leaves of A. alpina Pajares and pep1 plants grown under 5 wk LD → 12 wk SD‐V → LD and the x‐axis shows the indicated number of weeks (w). Leaf number (L) corresponds to the sampled node position. Grey shading marks the duration of vernalization in short days (SDV). Expression values represent the mean ± SD from three biological replicates, normalized to reference genes PP2A.
The photoperiodic flowering responses of A. alpina accession Pajares, which requires vernalization to flower, and its mutant pep1, which flowers independently of vernalization, were then determined. Both genotypes were grown for 5 wk under LD at 22°C followed by 12 wk‐SD vernalization at 4°C, then returned to LD, 22°C. Pajares formed visible floral buds 2 wk after return to LD at 22°C, whereas pep1 already had floral buds upon return to the same conditions (Fig. S3A). Ten‐wk‐old plants of both genotypes were then grown under continuous SDs and exposed to vernalization for 12 wk. Pajares formed visible floral buds on the PS, but these buds failed to progress further 10 wk after return to warm SDs. Furthermore, Pajares did not form any floral buds on the AB (V1) even 10 wk after return to warm SDs (Fig. S3B). By contrast, pep1 plants flowered on the PS within 5 wk of return to warm SDs, but then reverted to forming leaves before forming more flowers, and ultimately the shoot terminated in a floral bud (Figs 1c, S3B). To decouple vernalization and photoperiod effects, pep1 plants were grown under continuous LD and SD conditions. Plants flowered earlier in LDs than in SDs, and under SDs, the architecture of the PS showed dormant buds, vegetative branches, and terminated with a few flowers (Figs 1d, S3C, S4). Under LDs, the pep1 plants formed V1 branches that eventually flowered, followed by I1 branches and then solitary flowers on the PS (Figs 1D, S3C, S4), as expected (R. Wang et al., 2009). However, in SDs, pep1 mutants produced more V1 branches. These were followed by several nodes containing dormant buds, then flowers subtended by bracts and a few solitary flowers. The dormant buds were present at positions where axillary I1 branches were expected, and no I1 branches developed under SDs (Figs 1d, S3C, S4). In Pajares plants grown under continuous SDs before and after vernalization, floral transition was incomplete and plants showed aborted floral buds (Figs 1c, S3B). These findings indicate that A. alpina is a facultative LD plant and that photoperiod has a strong influence on both flowering time and inflorescence development.
To further dissect photoperiodic control of flowering, AaCO and AaFT/TSFL mRNA levels were measured in leaves at ZT16 (LDs) and ZT8 (SDs) at the end of the light period, and plants were grown under simulated natural conditions (5 wk LD 22°C → 12 wk SD 4°C → LD 22°C). RT‐qPCR analysis showed that AaFT mRNA levels were relatively high under LDs, whereas the TSFL genes were expressed at lower levels under the same conditions (Fig. 1e). By contrast, the abundance of AaFT/TSFL2 mRNA was extremely low in younger (L17) and older (L1) leaves under SDs 4°C. Upon return to LD, 22°C, the expression of AaCO, AaFT, and TSFL2 increased in the newly formed (L20) and already formed (L1) leaves after vernalization, correlating with floral transition and inflorescence development (Fig. S5A). To further verify the decrease of AaCO/AaFT/TSFLs expression under SD 4°C, shift assays were conducted, maintaining the constant developmental growth period. Plants grown under SDs at 22°C for 5 wk before shifting to 4°C either in SDs or LDs for vernalization. Expression of AaCO and AaFT increased only in the leaves of the plants that were exposed to LD conditions (Fig. S5B,C). This result was consistent with the previously detected low levels of expression under SDs. In addition to these experiments, pep1 plants were used to measure the expression profiles under continuous LD conditions because we used this genotype for the generation of CRISPR mutants. Under continuous LDs, the expression profiles were similar to those observed under LD cold experiments (Fig. S5D). Consistent with this result, pep1 flowered earlier in LDs than in SDs (Fig. 1d).
AaCO
and AaFT TSFLs all promote photoperiodic flowering but have different effects on inflorescence development in the primary and axillary shoots in the pep1 background
To genetically characterize the AaCO‐AaFT/TSFL module, CRISPR‐Cas9‐induced mutations of AaCO, AaFT and TSFLs were created in the pep1 background. Two Aaco mutants, containing either an 8‐bp deletion or a 1‐bp insertion, referred to as Aaco‐1 pep1 and Aaco‐2 pep1, respectively, were recovered (Fig. S6A; Table S2). Phenotypic analysis showed that under LDs, both Aaco pep1 mutants flowered later than pep1, and flowered only from the V1 ABs (Figs 2a,c, S6B). On the PS, the Aaco pep1 mutants never produced typical I1 branches or the solitary flowers characteristic of the pep1 inflorescence. This phenotype suggests that the flowering of the PS is more dependent on a regulatory pathway involving AaCO than the flowering of V1 ABs. To study whether the differential effect of AaCO on the PS and V1 ABs is mediated by AaFT and TSFL, mutations were recovered in these genes. Three quintuple mutants (Aaft tsfl1 tsfl2 tsfl3 pep1 (hereafter Aaft tsfls‐1 pep1 to Aaft tsfls‐3 pep1)) were obtained among the T_1_ transformants (Figs 2b,c, S7A–C). These mutants did not form flowers, but formed dormant buds at the positions where I1 nodes were expected, as observed in Aaco pep1 mutant under LDs and in pep1 under SDs. Confocal imaging of the SAM of pep1 and Aaft tsfls‐1 pep1 mutants at 6 wk showed that the pep1 mutant had already formed floral primordia, but the Aaft tsfls‐1 pep1 SAM was still vegetative. By 28 wk, many of the pep1 mutant inflorescences had senesced, but the Aaft tsfls‐1 pep1 quintuple mutants continued to form vegetative leaves until the end of the experiment at 41 wk (Figs 2b, S7D). These experiments indicate that AaFT and TSFL genes are essential for flowering of the pep1 mutant under LDs.
Architectural phenotypes of Arabis alpina Aaco pep1 and Aaft tsfls pep1 mutants under inductive and non‐inductive conditions. (a, b) Days to first open flower (DTFO) and shoot architecture of Aaco pep1, and Aaft tsfl pep1 mutants grown under long‐day (LD) conditions. Statistical significance was performed to compare the branch type of each mutant to that of pep1 plants. Fisher's exact test was used when one or more genotypes contained only non‐flowering plants; otherwise, one‐way ANOVA followed by Tukey's HSD test was applied. Letters above boxplots indicate statistically significant differences between genotypes (P < 0.05). (c) Representative plant images of T3 homozygous Aaco pep1 and Aaft tsfls pep1 mutants grown continuously under LDs. The primary shoot of Aaco pep1 mutants did not form flowers or typical I1 inflorescence branches (inset shows the PS apex of Aaco‐1 pep1). At 16 wk, Aaco pep1 and Aaft tsfls pep1 mutants did not flower from the primary shoot and continued to grow vegetatively until the end of the experiment. Blue triangle shows the SAM. By contrast, Aaft tsfls pep1 mutants failed to form flowers on any shoot, including axillary V1 branches, under inductive LDs. Bar, 9 cm. (d) Diurnal expression profiles of AaCO, AaFT and TSFLs in leaves of pep1 under LDs and SDs. Leaves were sampled every 2–3 h over a 24‐h period. Expression values represent mean ± SD from three biological replicates. Grey shading indicates the dark period. (e) Shoot architecture and days to first open flower (DTFO) of Aaco pep1, and Aaft tsfls pep1 mutants grown under SD conditions. Letters above boxplots indicate statistical groups. Red letters indicate groupings for the primary shoot (pink boxes), while blue letters indicate groupings for axillary shoots (blue boxes). Different letters represent significant differences (P < 0.05) among genotypes when compared to pep1. Fisher's exact test was used when one or more genotypes contained only non‐flowering plants; otherwise, ANOVA followed by Tukey's HSD test was applied.
The abundance of the mRNAs of the AaFT and TSFL genes was then measured in leaves under LDs and SDs at 22°C. Consistent with results from previous studies in A. thaliana, the levels of AaFT and AaCO mRNA peaked near the end of the light at ZT14 and decreased during the night in pep1 plants (Suárez‐López et al., 2001; Sawa et al., 2007). A minor peak in TSFL1 mRNA abundance was observed at ZT4, whereas TSFL2 mRNA showed peaks at both ZT4 and ZT14 (Fig. S8A). TSFL3 mRNA was not detected. AaFT and TSFL expression levels were below detectable levels under SDs in pep1 plants (Fig. 2d). Moreover, the levels of AaFT and the TSFL mRNAs were reduced to undetectable levels in the leaves of both Aaco pep1 mutants compared with their levels in pep1 plants (Fig. S8B).
Annual A. thaliana plants eventually flower and form a normal inflorescence under SDs, demonstrating that photoperiod‐independent pathways can override its LD requirement. We tested whether Aaco pep1 and Aaft tsfls pep1 mutations enhanced the late‐flowering and impaired inflorescence development phenotypes of pep1 observed under SDs. Under SDs, Aaco pep1 mutants flowered at a similar time to pep1 mutants, but notably, under these conditions the Aaco pep1 mutants formed some flowers on the PS and also on the ABs (Figs 1d, 2e, Fig. S9A–C), whereas they did not under LDs on the PS (Fig. 2a). Aaft tsfls‐1 pep1 showed a similar growth pattern under LDs and SDs, and failed to transition to flowering from all nodes when grown under SDs (Figs 2b,e, S9D–F). In conclusion, plants in which all AaFT and TSFL genes are inactive, fail to form I1 branches or flowers under LDs, emphasizing their essential roles in floral induction and inflorescence development, whereas those lacking AaCO fail to flower only on the PS. Also, under SDs, the Aaft Aatsfls pep1 mutants did not form flowers, emphasizing the importance of AaFT and TSFL genes in inflorescence development under non‐inductive conditions.
To resolve the individual contributions of each FT/TSFL gene, all double and triple mutants for AaFT, TSFL1, TSFL2, and TSFL3 were generated in pep1 background and phenotyped alongside the previously described higher‐order mutants. Each genotype was scored for flowering time, solitary flower formation on the PS, I1 branching, and AB (V1) flowering. Under LDs, all double mutants flowered significantly later than pep1 (Figs S10A, S11A). The pep1 mutant formed c. 13 V1 nodes, followed by c. 3 I1 branches and then 10–20 solitary flowers (I2). The Aaft‐1 pep1 and Aaft‐2 pep1 mutants eventually formed c. 30 distal partially vegetative V1 nodes. These were followed by the formation of 20–30 additional proximal nodes that produce simpler V1 branches similar to those observed in pep1 under SDs (Fig. S3C). Some plants then formed flower‐like structures subtended by a bract before reverting to producing leaves. The Aaft‐1 pep1 and Aaft‐2 pep1 mutants never formed solitary flowers. Each of the tsfl pep1 mutants developed significantly more V1 nodes than pep1 mutants, with only tsfl1 pep1 producing I1 branches and eventually formed some nodes containing solitary flowers. Combinations of mutant alleles of the four AaFT and TSFL genes resulted in more extreme late‐flowering and impaired inflorescence development phenotypes. Several triple mutant combinations exhibited vegetative characteristics on the PS inflorescence, such as the formation of bracts and reversion to forming leaves, indicating that maintenance of inflorescence meristem identity was impaired (Figs S10B–D, S11B, S12A). The quadruple mutants did not form solitary flowers, but after producing c. 30 vegetative nodes in the V1 zone, they produced c. 30 nodes that contained a leaf subtending a dormant bud in its axil. Quadruple mutants containing Aaft and combinations of tsfl mutations in the pep1 background did not flower on the primary SAM and continued growing vegetatively until the end of the experiment (200 d) (Figs S10E–H, S11C, S12B).
Aaco and Aaft tsfls mutants in pep1 background can be induced to flower by prolonged vernalization
Many A. alpina accessions show an obligate vernalization requirement with dominant PEP1 expression to flower (R. Wang et al., 2009; Wunder et al., 2023). Notably, pep1 knockout mutants still responded to vernalization, resulting in accelerated flowering compared to non‐vernalized plants. Therefore, the Aaft tsfls pep1 mutants were vernalized to assess whether this would overcome their inability to flower. Five‐week‐old plants grown under LDs at 22°C were exposed to either 12 or 24 wk of vernalization at 4°C under SDs and then returned to LDs at 22°C (Fig. 3a–e). After exposure to 12 wk of SD vernalization, visible flower buds were present on pep1 plants 1 wk after return to LDs (Fig. 3a), whereas on the quintuple mutants, the presence of floral buds was severely delayed and only occurred 20 wk after return to LDs (Fig. 3b,d). Moreover, the mutants only formed a few malformed flowers (Fig. 3d), and then the PS reverted to vegetative development by first forming flowers subtended by a bract, and then leaves (Fig. 3e,f). Therefore, flowering of Aaft tsfls pep1 quintuple mutants did occur after 12 wk of SD vernalization, but was extremely delayed compared with that of control pep1 plants, and only a few flowers were produced. Consistent with a role for AaFT and TSFL2 in flowering of pep1 mutants after 12 wk vernalization, both genes, as well as AaCO, were expressed after vernalization in young and mature leaves of pep1 and Pajares (Fig. S5A).
Extended cold exposure promotes floral transition in Aaco pep1 and Aaft tsfls pep1 quintuple mutants. Plants were grown for 3 to 5 wk in LDs at 22°C (warm), exposed to 4°C (cold) for up to 24 wk in SDs and were then returned to 22°C to LDs for 21 wk. The respective growth conditions are shown for each panel. (a) After 12 w vernalization, pep1 plants flowered, whereas Aaft tsfls pep1 quintuple mutants showed floral reversion. (b–e) Images of Aaft tsfls pep1 quintuple mutants. (b) Dormant buds formed after the V1 zone, and no branch elongation was observed (red bracket). (c) Close‐up image of the PS in B with a V2‐like zone. (d) Floral buds formed after the dormant bud zone and were mostly infertile, but occasionally siliques containing fewer than five seeds were produced. Yellow asterisks show floral buds in (b) and (d). (e) Floral reversion led to a leaf‐like structure subtending a single aborted flower bud, which was considered to be a bract (red asterisks). Bars for (a–e), 9 cm. (f) Boxplots and representative architecture plots showing days to first open flower (DTFO) and shoot structure of plants grown under 5 wk LD → 12 wk vernalization conditions. One‐way ANOVA was performed to compare the respective branch type flowering of respective mutant to that of pep1 plants. Letters above boxplots indicate statistically significant differences (Tukey HSD, P < 0.05). (g) Left to right: Shoot apical meristems of pep1 and Aaft tsfls pep1 produced leaf primordia after 1 wk cold, and floral primordia after 4 wk cold only in pep1. After 23 wk cold, visible floral buds were present on the PSs of Aaft tsfls pep1, whereas pep1 showed elongation of the internodes bearing buds. Following return to LDs at 22°C, pep1 produced flowers with normal floral organs, whereas Aaft tsfls pep1 produced aberrant flowers. Bar for confocal images, 50 μm and for plant apices, 0.5 cm. (h) Boxplots and representative architecture plots of pep1 and Aaft tsfls pep1 grown under 5 wk LD → 24 w vernalization conditions. (i) Boxplots and representative architecture plots showing days to first open flower (DTFO) and shoot structure of Aaco pep1 mutants grown under LDs at 22°C for 5 wk and then exposed to 24 wk of cold at 4°C. One‐way ANOVA was performed to compare the respective branch type flowering of respective mutant to that of pep1 plants. Letters above boxplots indicate statistically significant differences (Tukey HSD, P < 0.05). Bar, 9 cm. Red letters indicate groupings for the primary shoot (pink boxes), while blue letters indicate groupings for axillary shoots (blue boxes). Different letters represent significant differences (P < 0.05) among genotypes when compared to pep1. (j) pep1 plants showed no juvenility requirement to flower when exposed to 24 wk of cold after 3 wk of growth at 21°C in LDs, whereas Aaft tsfls pep1 did show a juvenility response to cold. Even after returning to LD at 22°C, Aaft tsfls pep1 did not flower and remained vegetative until the end of the experiment. (k) DTFO and architecture plots for pep1 and Aaft tsfls pep1 grown under 3 wk LD → 24 wk SD vernalization.
Aaft tsfls pep1 mutants were then subjected to a longer vernalization period of 24 wk under SDs. After this treatment, the mutants transitioned to flowering and continued to form flowers on the PS without reverting to the vegetative state (Figs 3g,h, S13A). The morphology of the Aaft tsfls pep1 mutant SAM was examined using confocal microscopy. After 1 wk of vernalization, the SAM of the PS of pep1 and Aaft tsfls pep1 mutants was still in the vegetative phase, surrounded by leaf primordia (Fig. 3g). After an additional 3 wk of cold exposure, floral transition was visible in pep1 plants, but not in Aaft tsfls pep1 quintuple mutants. However, after 23 wk in the cold, the Aaft tsfls pep1 mutant had formed visible floral buds, and 3 wk after return to warm LDs, produced visible petals. After 24 wk vernalization, the Aaft tsfls pep1 mutants formed flowers or floral‐like structures at each node on the PS, but did not produce fully developed I1 branches (Figs 3h, S4, S13B,C). Similarly, when Aaco pep1 mutants were exposed to cold for 24 wk, they flowered later than pep1 mutants but formed flowers on the PS, and showed similar aborted I1 branches as in Aaft tsfls pep1 mutants (Figs 3i, S13D–H).
In Pajares, sensitivity to vernalization is age‐dependent and plants flower when exposed to cold at 5 wk old, but do not flower if exposed at 3 wk old (Bergonzi et al., 2013; Hyun et al., 2019). To assess the age dependence of the vernalization response of Aaft tsfls pep1 mutants, they were exposed to 24 wk of cold after 3 wk of growth in LDs. Although pep1 plants transitioned to flowering and formed open flowers 1 wk after return to warm LDs, the Aaft tsfls pep1 mutants remained vegetative 10 wk after transfer to LDs, but did form flower buds after 14 wk (Fig. 3j,k). However, after the formation of two to three flowers subtended by bracts, these plants reverted to forming leaves on the PS. This result suggests that development of flowers on Aaft tsfls pep1 plants after extended exposure to vernalization requires an age‐dependent flowering pathway that is active 5 wk but not 3 wk after germination and therefore probably involves SPL transcription factors (Bergonzi et al., 2013; Hyun et al., 2019). In parallel, the Aaft tsfls pep1 mutants were vernalized in LDs after 3 wk and 5 wk of growth in LDs. In both cases, the Aaft tsfls pep1 mutants flowered significantly later than pep1 plants but produced several flowers and terminated with a floral bud (Fig. S14).
The AaCO‐AaFT module regulates hormone, floral identity, and developmental gene networks during floral transition
To investigate the molecular basis of the phenotypic differences between pep1, Aaco pep1 and Aaft tsfls pep1 mutants, RNA‐seq analysis was performed from AMs (V1) and SAMs across different developmental stages. Initially, differentially expressed genes (DEGs) in the SAM and AMs of pep1 compared to Aaco pep1 or Aaft tsfls pep1 mutants at three developmental stages (5, 7, and 13 wk) were identified (Fig. 4a; Table S3). In both tissues, only a small number of DEGs were detected at 5 and 7 wk, indicating small effects of the AaFT TSFLs and AaCO on the transcriptome of these meristems before floral transition. However, by 13 wk, the number of DEGs increased substantially, correlating with the transition to flowering and floral bud formation in pep1 mutants. In the SAM of pep1 compared to Aaco pep1, 98, 8, and 1765 genes were upregulated, whereas 10, 9, and 187 genes were downregulated at 5, 7, and 13 wk, respectively. In comparison of the SAM of pep1 with that of Aaft tsfls pep1, 80, 11, and 2246 genes were upregulated, whereas 30, 23, and 255 genes were downregulated at 5, 7, and 13 wk, respectively. The high overlap in the DEGs of the SAM in Aaco pep1 and Aaft tsfls pep1 at 13 wk (1632 upregulated and 65 downregulated) indicates that most of the effect of AaCO on the SAM is through AaFT TSFL activation (Fig. 4a). Following the comparison of the numbers of DEGs, GO analysis was performed to identify the biological processes affected. Several GO‐terms related to floral development and floral organs were identified in the SAM DEGs of the comparisons between pep1 and Aaco pep1 or Aaft tsfls pep1 at 13 wk (Fig. 4b,c).
Spatial and temporal gene expression profiles in Aaco‐1 pep1 and Aaft tsfls‐3 pep1 mutants. (a) Venn diagrams showing the numbers of shared and independent up‐ and downregulated differentially expressed genes (DEGs) across different timepoints and tissues for the indicated mutant comparisons. (b, c) Gene Ontology (GO) biological process enrichment analysis of DEGs in 13 wk SAMs of (b) pep1 vs Aaco‐1 pep1 and (c) pep1 vs Aaft tsfls‐3 pep1 mutants. (d) Pie‐donut plots to visualize temporal changes in expression of key regulatory genes across developmental stages. Each pie‐donut represents one gene, with the outer ring showing absolute TPM expression values at different time points (weeks), and the inner circle indicating the proportional contribution of each genotype. (e) Line plots showing log2 (TPM + 1) expression values for selected genes in the axillary meristem (AM) and shoot apical meristem (SAM) in the indicated genotypes at three developmental stages for SAMs (5, 7, and 13 wk) and two stages for AMs (7 and 13 wk). Letters indicate statistically significant differences between genotypes at each time point (ANOVA with Tukey's HSD, P < 0.05).
In the AMs, the number of common genes in the comparisons of pep1 with Aaco pep1 or Aaft tsfls pep1 was greatly reduced. Notably, in the pep1 and Aaft tsfls pep1 comparison, 2073 genes were upregulated, whereas in the pep1 and Aaco pep1 comparison, 343 were upregulated, and 327 were in common. This analysis supports the conclusion that Aaco pep1 has a less severe effect on the AMs than Aaft tsfls pep1, but also indicates that the effect AaCO does have on the AM transcriptome is mainly through AaFT/TSFL activation. Direct comparison of the transcriptomes of the AMs of Aaft tsfls pep1 and Aaco pep1 showed that regulators of flowering, floral organ development, and meristem development are expressed at lower levels in Aaft tsfls pep1 than Aaco pep1 (Figs S15, S16A,B).
Analysis of the specific DEGs associated with developmental processes showed that in the SAM of Aaco pep1 and Aaft tsfls mutants at 13 wk, floral integrators, including AaSOC1, AaAGL42, and AaSPL15, were significantly upregulated compared to pep1 plants (Figs 4d,e, S15). Floral repressors such as AaTFL1 and AaSVP were also more highly expressed in Aaco pep1 and Aaft tsfls pep1 SAMs at 13 wk, consistent with their later flowering phenotype (Figs 4e, S15). Furthermore, the floral repressors AaTEM1 and AaTEM2 were downregulated in Aaft tsfls pep1 mutants at 7 and 13 wk in both the AM and SAM (Figs 4d, S15), indicating that AaFT and TSFLs are required to promote the transcription of these FT repressors, suggesting a possible negative feedback loop. Consistent with their flowering phenotype, genes that act early in floral development, such as AaAP1, AaCAL, AaAGL15, AaCOL2, AaSPL4, and AaSPL5, were upregulated in pep1 mutants compared to Aaco pep1 and Aaft tsfl pep1 mutants (Fig. 4d,e) (Fernandez et al., 2000; Ledger et al., 2001; Lazaro et al., 2018). Moreover, floral organ development genes such as AaPI, AaAG, AaAP3, and AaSEP2‐4 (Fig. 4d) exhibited significantly higher expression at 13 wk in pep1 SAMs than those of Aaco pep1 and Aaft tsfls pep1.
In the SAM, differential expression of the HECATE (HEC) genes, which encode basic helix–loop–helix transcription factors, was detected. Notably, AaHEC1 was upregulated at 7 wk in pep1 plants compared to Aaco pep1 and Aaft tsfls pep1, coinciding with the downregulation of WUSCHEL (WUS), a key regulator of meristem maintenance (Figs 4d, S15). These changes are consistent with differences in meristem morphology and regulation during floral transition. Additionally, AaHEC2 and AaHEC3 were significantly upregulated at 13 wk in pep1 (Figs 4d, S15), likely indicating the initiation of floral primordia development (Schuster et al., 2014; Gaillochet et al., 2017). Furthermore, the FANTASTIC FOUR (FAF) genes were upregulated in the A. thaliana SAM during floral transition (Wahl et al., 2010), and AaFAF1 was upregulated across developmental stages, AaFAF2 was specifically upregulated at 13 wk, whereas AaFAF3 and AaFAF4 were downregulated in pep1 plants (Fig. 4d). AaSWEET10, AaSWEET11, and AaSWEET1 were also upregulated in pep1 plants compared to Aaco pep1 and Aaft tsfl pep1, suggesting the occurrence of FT‐dependent regulation of sugar transport and consistent with SWEET10 activation by FT in A. thaliana (Figs 4d, S15) (Andrés et al., 2020; Yuan et al., 2024).
The delay in the floral transition at the SAM of Aaco pep1 and Aaft tsfls pep1 mutants is associated with an increase in the formation of dormant buds and leaves at the shoot apex. This phenotype may involve the reduction in expression of genes encoding strigolactone biosynthetic enzymes, AaCCD7 and AaLBO, that is observed in the mutants (Fig. 4d), because lower strigolactone levels enhance axillary branching (Brewer et al., 2016; Yoneyama & Brewer, 2021). Interestingly, BRC1 expression was elevated at 13 wk in Aaco pep1 and Aaft tsfls pep1 mutants, indicative of suppressed axillary bud outgrowth (Aguilar‐Martínez et al., 2007), which correlates with the observed mutant phenotypes (Figs 2a,c, 4d, S15). Another class of AM regulators is the R2R3‐type MYB transcription factors RAX1, RAX2, and RAX3, which function during early vegetative stages to promote AM initiation (Keller et al., 2006; Müller et al., 2006). AaRAX1 was reduced at 5 wk in Aaft tsfls pep1 mutants, whereas AaRAX2 and AaRAX3 were upregulated at 13 and 7 wk, respectively, in pep1 plants compared to the mutants. Furthermore, other genes playing roles in diverse developmental and regulatory pathways were differentially regulated between the pep1 and Aaco pep1, Aaft tsfls pep1 mutants (Fig. S16A,B). Collectively, these results suggest roles for AaFT and TSFL in coordinating gene regulatory networks involved in floral transition and branching (Figs 4d, S15), and that AaCO mainly acts through AaFT TSFL but has a less severe effect on the AMs formed in V1.
AaCO
, AaFT , and TSFL are expressed in flowers and promote proper floral development after vernalization
The Aaco pep1 and Aaft tsfls pep1 mutants were defective in inflorescence and floral development after vernalization; therefore, to determine whether the wild‐type genes act directly in these tissues, their expression was measured in mature flowers and siliques by RT‐qPCR. AaFT mRNA was relatively highly expressed in these tissues, and TSFL2 mRNA was also detected but at much lower levels (Figs 5a, S17A). Notably, the expression of AaCO, AaFT, and TSFL2 was detected in flowers of Pajares and pep1 formed after vernalization (Figs 5a, S17A). This expression pattern was reflected in the floral phenotypes of Aaft tsfls pep1 and Aaft pep1 and Aatsfl pep1 mutants, which developed aborted petals and stamens and reduced pistils after 12 wk of vernalization or formed supernumerary petals following prolonged vernalization for 24 wk (Fig. 5b,c). A Pearson correlation analysis was performed to examine relationships among flowering traits across various growth conditions in pep1 plants. Flower and silique formation were consistently positively correlated across all conditions, whereas the correlation between silique number and I2 stem length was weak to moderate (Fig. 5d). The formation of fertilized siliques is highly dependent on LDs and FT TSFLs, as the Aaft tsfl mutants produced no or few seeds (Fig. S17B,C).
*Schematic summary of flowering outcomes in Arabis alpina mutants lacking FT and TSFL genes under different developmental stages and cold treatments. (a) Left, normalized expression values for each gene in flowers and siliques of pep1 mutants grown under LDs. ND: not detectable. Right, box plots showing the total number of flowers and siliques produced on the primary shoot by the indicated single mutants grown under long‐day (LD) conditions. Letters indicate statistically significant differences between genotypes at each time point (Kruskal‐Wallis followed by Dunn test, P < 0.05). The normalized expression values are shown above each bar plot. ND: not detectable (b) Representative images showing malformed flowers and flowers with extra petals observed in Aaft tsfls pep1 mutants following the indicated cold exposure. Bar, 100 μm. (c) Box plots showing the total number of flowers and siliques formed on the primary inflorescence in Aaft tsfls pep1 quintuple mutants and pep1 mutants grown under different growth conditions. Asterisks above the boxes indicate statistically significant differences between genotypes or conditions based on ANOVA followed by Tukey's HSD test: *, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001. (d) Pearson correlation analysis of traits of Aaft tsfls pep1 mutants under different growth conditions. Ellipses indicate the strength and direction of correlations between the measured traits across different growth conditions (specified in each panel). The colour intensity and shape of ellipses represent the magnitude of correlation (r value), ranging from white (no to weak correlation) to dark blue (strong correlation). Correlation coefficients are shown numerically within each cell. Asterisks denote the statistical significance of the correlations: **, P < 0.01; **, P < 0.001. (e) Lollipop plot showing the log2 fold changes (log2FC) of the topmost DEGs from the RNA‐seq analysis at 13 wk in AM. Each horizontal line represents the confidence interval for each gene, marking the estimated log2FC. The size of each bubble reflects the statistical significance of differential expression (e.g., larger bubble size = smaller adjusted P‐value). Genes are coloured according to their functional categories. Genes with positive log2FC are upregulated in pep1 compared with Aaco pep1 mutants and with negative log2FC are downregulated in pep1 compared with the mutants. (f) A model for regulation of flowering and bud fate in A. alpina. Under long days (LD) before vernalization, FT and TSFLs mRNAs are expressed in leaves and the proteins may be transported to the meristem (solid arrow). At this stage, miR156/7 is reduced in abundance as the plant ages. In vernalization under SDs PEP1 is downregulated, SPL15 is expressed and initiates floral transition. After vernalization under LDs under warm conditions, FT TSFLs are reactivated and might move to the axillary branches (dotted arrows) and in flowers transcription of FT and TSFLs occurs, thereby facilitating commitment to reproduction and the formation of siliques. Symbols that are not transparent represent the current growth condition at the corresponding developmental stage and yellow leaves and cotyledons represent senescence. [Correction added on 27 February 2026, after first online publication: the labels in panel (b) have been updated.]
Additionally, RNA‐seq analysis of non‐vernalized plants revealed that many of the top DEGs were associated with floral and pollen development, and showed significantly higher expression in pep1 plants (Fig. 5e). A lollipop plot was used to compare AMs of Aaco pep1 mutants that can form fertile flowers with those of vegetative AMs of Aaft tsfls pep1 mutants (Fig. S17D) at 13 wk. Although the significant upregulation of many genes in pep1 compared with Aaco pep1 mutants probably reflects the delayed floral transition in Aaco pep1 mutants, the log_2_‐FC was substantially higher in this comparison and in Aaft tsfls pep1 mutants, even higher, which might be an indirect effect of loss of flower formation (Fig. S17D). Together, these data suggest that the CO‐FT module affects flower and pollen development as well as floral transition in A. alpina.
Discussion
Arabidopsis thaliana is an intensively studied model system, but as a short‐lived annual species, it represents a relatively small part of the diversity present in the Brassicaceae family. In A. thaliana, CO‐FT/TSF primarily controls flowering time in response to exposure to LDs, and mutations in these genes cause late flowering under these conditions (Putterill et al., 1995; Kardailsky et al., 1999; Kobayashi et al., 1999). Here, we used CRISPR‐Cas9‐based reverse genetics to characterize the role of the AaCO‐AaFT/TSFL module in the arctic alpine perennial plant A. alpina, a member of the Arabideae super tribe of the Brassicaceae. We used the A. alpina pep1 mutant, which flowers without a requirement for vernalization, to enhance the transformation frequency and accelerate the genetic analysis. We demonstrated that this mutant shows a strong flowering response to day length and that the roles of the AaCO‐AaFT/TSFL genes in promoting flowering in response to LDs are conserved with those of the orthologous genes in A. thaliana. During this process, AaCO activates AaFT and TSFL2 transcription in leaves under LDs to promote earlier flowering. However, Aaco pep1 and Aaft tsfls pep1 mutations also have broader effects on reproductive development. Notably, they strongly impair I1 branching and solitary flower development on the PS, having a much more severe effect on these structures than the corresponding mutations in A. thaliana. Flowering of ABs in the axils of vegetative leaves is strongly impaired by Aaft tsfls pep1, but only weakly affected by Aaco pep1. Wild‐type A. alpina Pajares only flowers during vernalization, and ABs formed towards the end of or after vernalization remain vegetative until exposed to vernalization a second time (R. Wang et al., 2009). In the future, it will be interesting to assess how the Aaco and Aaft tsfl mutations affect this seasonal flowering response of A. alpina Pajares.
Photoperiodic regulation of flowering in A. alpina by the CO‐FT/TSF module
Many Brassicaceae species are facultative LD plants, flowering more rapidly under LDs than SDs. In A. thaliana, this response is driven by the CO‐FT/TSF module (Putterill et al., 1995; Kardailsky et al., 1999; Kobayashi et al., 1999; Suárez‐López et al., 2001), whereas under SDs, flowering occurs independently of CO‐FT/TSF through a pathway involving the phytohormone gibberellin (GA) and SPL transcription factors (Wilson et al., 1992; J. W. Wang et al., 2009; Hyun et al., 2016).
In A. alpina, we found that the canonical CO‐FT regulatory module is conserved and promotes early flowering of the non‐vernalization requiring pep1 mutant under LDs. The Aaft pep1, tsfl1 pep1, tsfl2 pep1 and tsfl3 pep1 mutants all flowered later than the pep1 progenitor, suggesting that all four AaFT/TSFL genes contribute to early flowering. However, only AaFT and AaTSFL2 mRNAs were reproducibly detected in leaves under LDs. We assume that TSFL3 is expressed at specific stages in development or in precise spatial patterns that we have not yet detected, allowing it to promote flowering. Moreover, under SDs, when AaFT and TSFL are not transcribed in leaves, pep1 plants flower later, and on the PS form only a few solitary flowers and no I1 branches. Under these conditions the pep1 mutants exhibit strong vegetative characteristics on the PS, including dormant buds, bract development, and reduced fertility. Aaco pep1 and Aaft/tsfls pep1 mutants under LDs resembled pep1 mutants under SDs, supporting the idea that the AaCO‐AaFT module confers the LD flowering response. However, Aaft tsfls pep1 showed stronger effects, particularly on flowering of V1 AB, suggesting that AaFT TSFL have AaCO‐independent functions on some aspects of reproductive development.
Role of
AaCO ‐ AaFT / TSFL genes in inflorescence development of A. alpina
The effects of Aaco pep1 and Aaft tsfls pep1 mutations on inflorescence and flower development of A. alpina are more severe than the corresponding mutations in A. thaliana. Under SDs, the ft tsf mutant flowers with a similar number of rosette leaves to wild‐type A. thaliana, but show increased numbers of cauline leaves and branches, consistent with the mutations impairing floral meristem identity independently of their effects on flowering time (Müller‐Xing et al., 2014; Romera‐Branchat et al., 2025). Moreover, under LDs, the flowers of ft tsf mutants show variable numbers of sepals and greatly reduced numbers of petals (Romera‐Branchat et al., 2025). FT also prevents reversion to the vegetative state in the inflorescence of A. thaliana (Liu et al., 2014). Although distinctive, these inflorescence phenotypes of ft tsf are less severe than those observed in A. alpina Aaft tsfls pep1 mutants. In A. thaliana, FT interacts with FD to activate flowering‐time genes such as SOC1 and FUL in the SAM, and later is proposed to activate transcription of the floral meristem identity gene AP1 early in floral development (Schmid et al., 2003; Abe et al., 2005; Wigge et al., 2005; Melzer et al., 2008; Torti et al., 2012). Later in the floral meristem, FT‐FD activates the SEP2 and SEP3 genes (Romera‐Branchat et al., 2025). In the shoot apex of A. alpina, flowering‐time genes such as AaSPL15 and AaSOC1 were expressed at 13 wk at higher levels in Aaco pep1 and Aaft tsfls pep1 mutants than in pep1, and other flowering genes such as AaSPL5 were increased in expression compared to earlier time points, suggesting part of the regulatory process involved in floral induction is activated in the mutants. However, the mRNAs of genes involved in floral development, including AaAP1 and floral organ identity genes, are markedly reduced at 13 wk in Aaft tsfls pep1. Moreover, AaFT mRNA was detected in developing flowers by RT‐qPCR (Fig. 5a), suggesting that this gene acts directly in the flower to influence floral development, as recently proposed for FT of A. thaliana and FTL1 of rice (Giaume et al., 2023; Gao et al., 2025; Romera‐Branchat et al., 2025). Also, as discussed below, vernalization of Aaft tsfls pep1 mutants promotes the initiation of flower development, but the resulting flowers lack floral organs, particularly petals and stamens. This phenotype is related to that of the ft tsf mutant, which showed strongly reduced petal development (Romera‐Branchat et al., 2025), but again is more severe in A. alpina. Taken together, the data suggest a major function for AaFT in developing flowers of A. alpina to promote floral identity and floral organ development.
In Aaft tsfls pep1 plants, dormant buds are consistently observed at the nodes expected to form inflorescence branches (I1). Therefore, in A. alpina, AaFT TSFL have a critical role in activating I1 axillary bud outgrowth. In A. thaliana, FT moves into the AM to promote flowering of inflorescence branches, and this role of FT is repressed in the AM by the TCP transcription factor BRANCHED 1 (BRC1) (Aguilar‐Martínez et al., 2007; Niwa et al., 2013). However, FT transcription was also recently detected in the axil of the cauline leaf adjacent to the AM, which might contribute to the floral transition of the AB (Gao et al., 2025). In A. thaliana, FT is not reported to promote outgrowth of inflorescence branches, but to accelerate their transition to flowering (Niwa et al., 2013), as described in a more extreme form here for AMs in the vegetative nodes (V1) where flowering is prevented in Aaft tsfls pep1 plants but outgrowth occurs. Nevertheless, BRC1 does inhibit outgrowth of AB in A. thaliana, and one possibility is that in A. alpina, antagonism between BRC1 and AaFT TSFL extends to activation of axillary buds in the inflorescence as well as flowering.
Interactions between CO‐FT TSFL and vernalization
Exposure to cold (vernalization) for 12 wk partially alleviated the defects of Aaft tsfls pep1 mutants, so that some fertile flowers were formed on the PS. After forming these flowers, the vernalized plants reverted to vegetative development. In A. alpina (Pajares), flowering is repressed by PEP1 until the plant is exposed to vernalization (R. Wang et al., 2009; Albani et al., 2012), but a response to vernalization was observed in the Aaft tsfls pep1 background lacking an active PEP1 gene. Therefore, neither the AaFT TSFL nor PEP1 is essential for a vernalization response. We hypothesize that in Aaft tsfls pep1, additional cold‐responsive pathways contribute to floral induction in the absence of PEP1. In A. thaliana, MAF genes act redundantly with FLC to delay flowering (Ratcliffe et al., 2003; Gu et al., 2013), and maf mutants of other Brassicaceae species showed early flowering (Zhai et al., 2024). In A. alpina, orthologues of MAF genes may also function in parallel to PEP1 to regulate flowering (Madrid et al., 2021). The observation that prolonged vernalization (24 wk) can restore flowering in Aaft tsfls pep1 mutants suggests that extended cold might relieve floral repression by such MAFs, or activate alternative flowering pathways that bypass the canonical FT/TSFL pathway. However, Aaco pep1 and Aaft tsfls pep1 mutants still showed impaired inflorescence architecture even after vernalization, including reduced formation of I1 branches (Fig. S13), indicating that the requirement for the CO‐FT module on these phenotypes cannot be fully overcome by vernalization.
Arabis alpina Pajares plants show an age‐dependent response to vernalization because the SPL15 gene promotes flowering of A. alpina and is directly repressed both by PEP1 and the microRNAs miR156 and miR157 (Bergonzi et al., 2013; Hyun et al., 2019; Roggen et al., 2025). Young plants contain high levels of these microRNAs, and, therefore, cannot express AaSPL15, even when PEP1 is repressed by vernalization. However, older plants express AaSPL15 in response to cold, because, in addition to PEP1 repression, miR156/7 levels are reduced in older meristems. We found that the flowering response of Aaft tsfl plants to vernalization is age‐dependent, suggesting that the bypass mechanism that occurs in vernalization requires SPL transcription factors that are targets for miR156/7, such as SPL15. In A. thaliana, SPL15 and FT TSF show genetic redundancy in the control of flowering time (Hyun et al., 2019), and therefore activation of expression of SPL transcription factors during vernalization might contribute to overcoming the effects of Aaft tsfls pep1 mutations. Nevertheless, in the RNA‐seq analysis, we detected relatively high levels of AaSPL15 and AaSPL5 mRNA in apices of Aaco pep1 and Aaft tsfls pep1 mutants at 13 wk without exposure to vernalization. Further information on the mechanism by which the non‐flowering phenotype of the Aaft tsfls pep1 mutations is overcome by vernalization could be obtained by performing RNA‐seq on vernalized plants and by studying the effect of incorporating the Aaspl15 mutation into the Aaft tsfls pep1 background.
Previously, AaSPL15 was shown to be critical for flowering of A. alpina Pajares in SD vernalization (Hyun et al., 2019), and the role of FT TSF in flowering of perennial A. alpina under these conditions was proposed to be less important compared to annual A. thaliana. In support of that idea, we found that AaFT mRNA level is strongly reduced in SD vernalization, the stage when A. alpina Pajares flowers through SPL15 activity. Furthermore, our experiments conducted exclusively under SDs revealed that, despite reduced PEP1 levels and SPL15 expression, plants failed to develop floral meristems on axillary V1 branches, and floral meristems on the PS aborted prematurely when plants were returned to SDs after vernalization. These findings suggest that low miR156 and PEP1 levels allow SPL15 to become fully activated during the SD vernalization phase, establishing floral commitment, but that transcription of FT and TSFL genes is required after vernalization to promote inflorescence emergence and fertile flower development (Fig. 5f). Therefore, in addition to the redundant functions of AaFT TSFL and AaSPL15 in controlling floral induction (Hyun et al., 2019; Roggen et al., 2025), AaFT TSFL have non‐overlapping functions with AaSPL15 in inflorescence development. Moreover, we cannot exclude that AaFT might contribute to the acquisition of competence along with miR156, because AaFT mRNA was detected under LDs before vernalization.
Contribution of the AaCO‐AaFT/TSFL module to the perennial growth habit
One of the key features of the growth habit of perennial species is the ability to maintain different developmental fates in AMs enabling the plant to alternate between cycles of flowering and vegetative growth. In A. alpina, this effect is controlled by age‐related and cold‐responsive genes such as MIR156/7, PEP1 and TFL1 (R. Wang et al., 2009; Wang et al., 2011; Bergonzi et al., 2013) in addition to loss of LATERAL SUPPRESSOR (LAS) function, which results in an annual growth habit by loss of AM initiation (Ponraj & Theres, 2020). Our findings reveal that the influence of AaCO on flowering is spatially restricted to the PS and is not required for flowering of AB formed in the axils of vegetative leaves (V1), thereby linking photoperiodic regulation to perennial growth architecture. In A. alpina, the reduced dependence on AaCO for flowering of AMs suggests the presence of AaCO‐independent pathways that activate AaFT TSFL to induce flowering of these lateral shoots. FT activity is sufficient to activate flowering in other contexts, such as the promotion of flowering independently of vernalization by gain‐of‐function FTa1 alleles in Medicago truncatula (Jaudal et al., 2013). A CO‐independent pathway is probably not active in A. alpina leaves because AaFT TSFL expression was extremely low in leaves of the Aaco pep1 mutant, but flowering of AB still occurred; therefore, AaFT TSFL transcription might occur directly in AMs independently of AaCO. Although we could not reliably detect AaFT mRNA in the transcriptome of pep1 AMs, it might be expressed transiently or in a spatially restricted pattern, as recently observed in A. thaliana (Gao et al., 2025), so that it is present in the transcriptome at low levels. By contrast, photoperiodic activation of AaFT by AaCO controls flowering time in the PS, axillary branch formation in the inflorescence and solitary flower formation on the PS, and may be relevant for other metabolic and developmental pathways controlled by CO (Romero et al., 2024). This modular diversification of the CO‐FT regulatory network might contribute to the distinct flowering strategies that define annual vs perennial growth habits within the Brassicaceae. This situation parallels bourse shoots in apple or spurs in cherry, which initiate flowers even when the primary axis remains vegetative (Costes et al., 2014). Our findings demonstrate that the CO‐FT module can be spatially decoupled within a single plant, enabling perennials to sustain vegetative growth along a central axis while redeploying florigenic signals laterally.
Competing interests
None declared.
Author contributions
NS and GC initiated the project, contributed to the funding and wrote the article. NS and GC designed and interpreted the experiments and NS conducted the experiments.
Disclaimer
The New Phytologist Foundation remains neutral with regard to jurisdictional claims in maps and in any institutional affiliations.
Supporting information
Fig. S1 Phylogenetic relationships of CO, FT and TSF‐LIKE proteins in Arabis alpina and related species. Fig. S2 Conserved synteny of TSFL gene loci across Arabis alpina accessions and related Arabis species. Fig. S3 Phenotypic characterization of Pajares and pep1 plants grown under defined conditions. Fig. S4 Phenotypic characterization of node identities scored in this study. Fig. S5 Temporal expression profiles of AaCO, AaFT, AaTSFL, PEP1, and AaSPL15 genes in leaves of Arabis alpina Pajares and pep1. Fig. S6 Genotypes and phenotypes of the CRISPR‐Cas9‐generated homozygous T_3_ lines of Aaco pep1. Fig. S7 Genotypes and phenotypes of the CRISPR‐Cas9‐generated homozygous T_3_ lines of Aaft tsfls mutants in pep1 background. Fig. S8 Diurnal expression of photoperiodic genes. Fig. S9 Phenotypes of Aaco and Aaft tsfls mutants grown under short days in pep1 background. Fig. 10 Flowering phenotypes of double and higher‐order mutants of Aaft and tsfl genes (A) Aaft and tsfl single mutations in pep1 background after growth in long days. Fig. S11 Phenotypes of single and higher‐order mutants of Aaft and tsfl genes in pep1 background. Fig. S12 Genotypes of different Aaft and tsfl mutants in pep1 background. Fig. S13 Scoring of I1 and I2 transition when exposed to longer vernalization in short days. Fig. S14 Phenotypes of quintuple Aaft tsfls pep1 mutants exposed to vernalization in long days. Fig. S15 Bubble plot showing log_2_‐fold changes in gene expression for selected genes across genotype comparisons: pep1 vs Aaco‐1 pep1 and pep1 vs Aaft tsfls‐3 pep1 in axillary meristem (AM), primary meristem (SAM) at 13 wk. Fig. S16 Temporal expression of genes involved in hormone and other pathways. Fig. S17 Gene expression profiles and flower and silique formation.
Table S1 Gene annotation of AaCO, AaFT, and AaTSFL genes. Table S2 Mutant alleles generated in this study and their predicted protein disruption. Table S3 Total number of upregulated and downregulated DEGs identified for each genotype at each developmental stage. Table S4 Primers used in this study.Please note: Wiley is not responsible for the content or functionality of any Supporting Information supplied by the authors. Any queries (other than missing material) should be directed to the New Phytologist Central Office.
The reference list from the paper itself. Each links out to its DOI / PubMed record.
- 1Abe M , Kobayashi Y , Yamamoto S , Daimon Y , Yamaguchi A , Ikeda Y , Ichinoki H , Notaguchi M , Goto K , Araki T . 2005. FD, a b ZIP protein mediating signals from the floral pathway integrator FT at the shoot apex. Science 309: 1052–1056.16099979 10.1126/science.1115983 · doi ↗ · pubmed ↗
- 2Adrian J , Farrona S , Reimer JJ , Albani MC , Coupland G , Turck F . 2010. cis‐Regulatory elements and chromatin state coordinately control temporal and spatial expression of FLOWERING LOCUS T in Arabidopsis. Plant Cell 22: 1425–1440.20472817 10.1105/tpc.110.074682 PMC 2899882 · doi ↗ · pubmed ↗
- 3Aguilar‐Martínez JA , Poza‐Carrión C , Cubas P . 2007. Arabidopsis BRANCHED 1 acts as an integrator of branching signals within axillary buds. Plant Cell 19: 458–472.17307924 10.1105/tpc.106.048934 PMC 1867329 · doi ↗ · pubmed ↗
- 4Albani MC , Castaings L , Wötzel S , Mateos JL , Wunder J , Wang R , Reymond M , Coupland G . 2012. PEP 1 of Arabis alpina is encoded by two overlapping genes that contribute to natural genetic variation in perennial flowering. P Lo S Genetics 8: e 1003130.23284298 10.1371/journal.pgen.1003130 PMC 3527215 · doi ↗ · pubmed ↗
- 5An H , Roussot C , Suarez‐Lopez P , Corbesier L , Vincent C , Pineiro M , Hepworth S , Mouradov A , Justin S , Turnbull C et al. 2004. CONSTANS acts in the phloem to regulate a systemic signal that induces photoperiodic flowering of Arabidopsis. Development 131: 3615–3626.15229176 10.1242/dev.01231 · doi ↗ · pubmed ↗
- 6André D , Marcon A , Lee KC , Goretti D , Zhang B , Delhomme N , Schmid M , Nilsson O . 2022. FLOWERING LOCUS T paralogs control the annual growth cycle in Populus trees. Current Biology 32: 2988–2996.35660141 10.1016/j.cub.2022.05.023 · doi ↗ · pubmed ↗
- 7Andrés F , Kinoshita A , Kalluri N , Fernández V , Falavigna VS , Cruz TMD , Jang SA‐O , Chiba Y , Seo M , Mettler‐Altmann T et al. 2020. The sugar transporter SWEET 10 acts downstream of FLOWERING LOCUS T during floral transition of Arabidopsis thaliana . BMC Plant Biology 20: 53.32013867 10.1186/s 12870-020-2266-0PMC 6998834 · doi ↗ · pubmed ↗
- 8Bastow R , Mylne JS , Lister C , Lippman Z , Martienssen RA , Dean C . 2004. Vernalization requires epigenetic silencing of FLC by histone methylation. Nature 427: 164–167.14712277 10.1038/nature 02269 · doi ↗ · pubmed ↗