Beyond ﬂowering time: diverse roles of an APETALA2-like transcription factor in shoot architecture and perennial traits

earlier than pep1 and carries a lesion in the A. alpina orthologue of the APETALA2 ( AP2 )-like gene, TARGET OF EAT2 ( AaTOE2 ). AaTOE2 is a ﬂoral repressor and acts upstream of SQUAMOSA PROMOTER-BINDING PROTEIN-LIKE 5 (AaSPL5). In the wild-type background, which requires cold treatment to ﬂower, AaTOE2 regulates the age-dependent response to vernalisation. In addition, AaTOE2 ensures the maintenance of vegetative growth by delaying axillary meristem initiation and repressing ﬂowering of axillary buds before and during cold exposure. (cid:1) We conclude that AaTOE2 is instrumental in ﬁne tuning different developmental traits in the perennial life cycle of A. alpina .


Introduction
Plants have evolved different life history strategies.Annuals are monocarpic and set seed and senesce after flowering.Most perennials are polycarpic, being able to maintain vegetative growth from axillary meristems (AMs) and restricting senescence only to the reproductive branches (Amasino, 2009;Albani & Coupland, 2010).The distinction between annuals and perennials is underlined by several traits.Typically, most perennials have a prolonged juvenile phase during which they are not competent to respond to flower inductive stimuli.The polycarpic behaviour of perennial plants is a result of the asynchronous and differential behaviour of AMs as some commit to reproductive development, while others develop into vegetative branches or arrest growth after producing a few leaves and forming a bud (Tan & Swain, 2006;Costes et al., 2014;Park et al., 2017;Vayssi eres et al., 2020).Thus, understanding the molecular mechanisms underlying the activity and fate of AMs can give insight into plant life history evolution.
Comparative studies between two Brassicaceae models, the annual Arabidopsis thaliana and the polycarpic perennial Arabis alpina have been successfully used as a tool to dissect at the molecular level the evolutionary changes driving the adoption of the annual or perennial life histories.These studies have demonstrated that several genes that regulate flowering time in Arabidopsis thaliana have similar and unique roles in A. alpina and, most importantly, that the A. alpina specific roles contribute to the perennial life cycle.One characteristic example is the MADS box transcription factor FLOWERING LOCUS C (FLC), a key floral repressor in the vernalisation pathway, in which prolonged cold exposure promotes flowering by stably silencing FLC mRNA levels in Arabidopsis thaliana (Michaels & Amasino, 1999;Sheldon et al., 2000).The FLC orthologue in A. alpina, PERPETUAL FLOWERING1 (PEP1) also regulates flowering in response to cold (Wang R. et al., 2009).pep1 mutants flower without cold exposure, whereas wild-type plants have an obligate vernalisation requirement to flower (Wang R. et al., 2009).The expression pattern of PEP1 differs from FLC in Arabidopsis thaliana so that PEP1 mRNA levels are upregulated in axillary branches after vernalisation to maintain vegetative development and sustain the polycarpic growth habit of A. alpina (Wang R. et al., 2009;L azaro et al., 2018).Other flowering time regulators that typically control flowering through the age pathway, such as the A. alpina orthologues of APETALA2 (PERPETUAL FLOWERING2, PEP2), SQUAMOSA PROMOTER-BINDING PROTEIN-LIKE15 (AaSPL15) and TERMINAL FLOWER1 (AaTFL1) also have similar and unique roles in A. alpina (Wang et al., 2011;Bergonzi et al., 2013;Hyun et al., 2019;L azaro et al., 2019).PEP2, AaTFL1 and AaSPL15 determine the age at which plants become competent to cold treatment, control the duration of cold exposure required for flowering and (similar to PEP1) ensure maintenance of vegetative development in some axillary branches (Wang R. et al., 2009(Wang R. et al., , 2011;;Bergonzi et al., 2013;Hyun et al., 2019;L azaro et al., 2019).The age and the vernalisation pathways are also integrated in different ways.PEP1 binds to the AaSPL15 locus, whereas PEP2 acts upstream of PEP1 to ensure the upregulation of PEP1 mRNA levels in axillary branches after vernalisation (Hyun et al., 2019;L azaro et al., 2019).
Similarly, microRNA156 (miR156) that targets most SPLs in the age pathway has similar and unique roles in A. alpina and in Arabidopsis thaliana.miR156 has similar expression patterns in both species and its accumulation is high in young seedlings and declines as plants get older (Wang J.W. et al., 2009a;Bergonzi et al., 2013).miR156 in A. alpina also regulates flowering in the age pathway, which is evident from A. alpina transgenics that constitutively express MIR156b and do not flower in response to vernalisation, whereas the ones with suppressed miR156 activity by target mimicry (35S:MIM156) respond to vernalisation at a younger age (Bergonzi et al., 2013).The miR156-SPL module also regulates other traits such as plastochron, branching and AM initiation (Wang et al., 2008;Wu et al., 2009;Tian et al., 2014;Jung et al., 2016;Gao et al., 2018).These additional roles of miR156 and some of its targets have been reported in Arabidopsis thaliana and other species but not in A. alpina (Bergonzi et al., 2013;Hyun et al., 2019).miR172 and its targets, AP2 and the AP2-like genes SCHLAFMUTZE (SMZ), SCHNARCHZAPFEN (SNZ), TARGET OF EAT1-3 (TOE1-3) are also placed in the age pathway (Aukerman & Sakai, 2003;Wu et al., 2009).In Arabidopsis thaliana, the accumulation of miR172 increases progressively as plants age and follows an opposite pattern to that of miR156 (Aukerman & Sakai, 2003;Wu et al., 2009).These complementary expression patterns of miR156 and miR172 are not observed in A. alpina, although one of the miR172 targets (A.alpina AP2/PEP2) regulates age-dependent flowering in response to vernalisation (Bergonzi et al., 2013;L azaro et al., 2019).
Our experiments have previously demonstrated that the pep1-1 mutant still responds to different durations of vernalisation, suggesting that there are other repressors regulating flowering in parallel with PEP1 (L azaro et al., 2018).Here we aimed at identifying such additional floral repressors and at characterising their contribution to the perennial life cycle.We performed an enhancer screen of pep1-1 and identified a mutant, enhancer of pep1 055 (eop055).We demonstrated that eop055 is a toe2 mutant in A. alpina.To characterise the role of AaTOE2 in the A. alpina life cycle and perennial traits, we introgressed the Aatoe2 mutation in the wild-type background that has an obligate vernalisation requirement to flower and an active PEP1 allele.We demonstrated that AaTOE2 regulates the age-dependent response to vernalisation and shoot architecture.We have previously shown that shoot architecture in A. alpina is organised in zones of differential bud activity and fate according to position on the plant named as V1 (zone of flowering axillary branches), V2 (zone of dormant buds) and V3 (zone of vegetative axillary branches) (L azaro et al., 2018;Vayssi eres et al., 2020).Here we demonstrate that AaTOE2 contributes to shoot architecture by repressing flowering in axillary branches and determining the number of metamers in each zone.Many of these roles have been tailored to contribute to the perennial growth habit and have not been described for TOE2 in Arabidopsis thaliana.

Plant material, growth conditions and phenotypic analysis
The pep1-1 mutant does not require vernalisation to flower and was previously obtained after the mutagenesis of the A. alpina accession Pajares (Wang R. et al., 2009b).Here, c. 6000 pep1-1 seeds were mutagenised with 0.35% ethyl methanesulphonate (Sigma) for 8-9 h.In total, 18 000 M2 seedlings (1500 M1 families) were screened in under long day (LD, 16 h : 8 h, light : dark) glasshouse and mutants that flowered earlier than pep1-1, named as enhancers of pep1 (eop), were selected.The phenotype of mutants was subsequently confirmed in the M3 generation.The eop055 mutant was crossed with wild-type Pajares, which requires vernalisation to flower, and an introgression line (Aatoe2_IL) was selected from the F2 segregating population that carried the single nucleotide polymorphisms (SNP) in AaTOE2 and lacked the SNP in PEP1 (responsible for the pep1-1 phenotype).Primers used for genotyping are listed in Supporting Information Table S1.The Arabidopsis thaliana toe2-1 mutant (SALK_065370) was obtained from the Nottingham Arabidopsis Stock Centre (NASC).
For most experiments, seeds were first stratified in darkness for 3-5 d at 4°C.Depending on the experiments, plants were grown in a long day (LD) or short day (SD) (8 h : 16 h, light : dark) glasshouse with light intensity varying from 200-500 µmol m À2 s À1 and temperature at c. 22°C.Vernalisation experiments were performed in a SD growth chamber at 4°C and light intensity 14 µmol m À2 s À1 .
Flowering time experiments with pep1-1, eop055 and Arabidopsis thaliana genotypes were performed by scoring total leaf number at flowering.Aatoe2_IL and Pajares plants were characterised for several traits: (1) Ability to flower without vernalisation, by scoring flowering in plants growing in a LD glasshouse for up to 22 wk.
(2) Juvenile phase, by growing plants for 3 wk in a LD glasshouse, cold treating them for 12 wk and scoring flowering after plants were returned to LD glasshouse conditions.
(3) Duration of vernalisation required for flowering, by exposing 8-wk-old plants grown in a LD glasshouse to 8, 12 and 18 wk of vernalisation and scoring flowering after they were returned to LD glasshouse conditions.
(4) Shoot architecture, by scoring bud activity and fate in every leaf node of flowering plants exposed to 12 wk of vernalisation and subsequently grown for 8 wk in a LD glasshouse.
(5) Leaf initiation rate, by monitoring total visible leaf number from 2-8 wk in a LD glasshouse.(6) AM initiation, by monitoring the presence or absence of a bud or an AM in each leaf node under the stereomicroscope.
All experiments were performed with at least 12 plants.Phenotypic data were analysed using Student's t-test.

Mapping-by-sequencing and fine mapping
A BC1F2 mapping population was created by backcrossing eop055 to pep1-1 and scored for flowering time in an LD glasshouse.From 450 eop055BC1F2 plants, 109 flowered with a similar number of leaves as eop055, suggesting that a single gene was segregating for flowering time.Genomic DNA was extracted from a pool of flower buds collected from 84 eop055BC1F2 early flowering plants.Flower buds from 48 pep1-1 plants were also harvested for DNA extraction.Genomic DNA samples were sequenced on an Illumina HiSeq2500 (Illumina, San Diego, CA, USA) at the Max Planck Genome Center Cologne (Germany) and yielded 203 042 704 reads for pep1-1 and 200 336 368 reads for eop055BC1F2.In total, 190 190 100 (93.67%) reads from pep1-1 and 194 532 678 (95.42%) reads from eop055BC1F2 were aligned to the A. alpina V5 reference genome using Bowtie 2 (Langmead & Salzberg, 2012;Jiao et al., 2017), resulting in an average coverage of 72.4 and 72.7 for the respective resequenced genome.We applied SAMTOOLS and BCFTOOLS to identify SNPs between samples and the A. alpina reference genome (Li et al., 2009).SHOREMAP v.3.6 was used to extract the SNPs and to visualise the allele frequency (AF) for EMS-induced mutations in eop055BC1F2 (Schneeberger et al., 2009;Sun & Schneeberger, 2015).SNPs with a quality score lower than 40 were filtered out from the eop055BC1F2 data.In addition, homozygous SNPs in the pep1-1 resequencing data were used for background correction and were also filtered out from the eop055BC1F2 data.
To fine map the EOP055 locus, 11 molecular markers were developed based on SNPs detected by the resequencing data to be polymorphic between eop055BC1F2 and pep1-1.For fine mapping, we screened 503 BC1F2 plants.Primers used for fine mapping are listed in Table S1.The raw and processed resequencing data have been submitted to GenBank under the accession no.GSE145718.

Plasmid construction and plant transformation
To generate the pAaTOE2:gAaTOE2 eop055 transgenic lines, 4128 bp upstream of the translational start of AaTOE2 and 3672 bp spanning AaTOE2 (from the ATG up to 776 bp downstream of its translational stop) were combined separately into the GATEWAY cloning vector pDONR207 (Invitrogen, Darmstadt, Germany) using the polymerase incomplete primer extension (PIPE) method (Klock et al., 2008).The generated 7.8 kb AaTOE2 DNA construct was then recombined into the pEarlyGate301 binary vector and transformed into the eop055 mutant.For construction of 35S:AaTOE2 and 35S:Aatoe2, AaTOE2 cDNA was amplified from pep1-1 and eop055, respectively.cDNAs were subsequently cloned into the destination vector pLEELA containing a double Cauliflower Mosaic Virus (CMV) 35S promoter and transformed into the Arabidopsis thaliana toe2-1 mutant.Primers used for plasmid constructions are listed in Table S1.Plant transformation in A. alpina and Arabidopsis thaliana was performed using the Agrobacterium-mediated floral dip method (Clough & Bent, 1998).

RNA extraction and expression studies
For main shoot samples, apices from 12-50 plants per sample were combined for all experiments.For V3 bud samples, buds from leaf axils 21-26 were harvested under a stereomicroscope from 24-48 plants per sample.Total RNA was isolated from plant tissues using the miRNeasy Plant Mini Kit (Qiagen, Hilden, Germany) and subjected to DNase treatment using DNA-free DNase (Ambion).For normal qRT-PCR reactions, 1 µg of total RNA was used as a template for the synthesis of cDNA by reverse transcription with SuperScript II Reverse Transcriptase (Invitrogen) and oligo(dT) (18) primer.For qRT-PCR of miRNA reactions, 200 ng of total RNA was used for reverse transcription using specific key primers.Quantitative PCRs to follow gene expression were carried out as described in L azaro et al. (2019) and for miRNAs as described in Bergonzi et al. (2013).For each sample, three technical replicates and three biological replicates were analysed.Transcript accumulation of selected genes in A. alpina samples was normalised using the A. alpina PROTEIN PHOSPHATASE 2A (AaPP2A) and RAN GTPASE 3 (AaRAN3) and in Arabidopsis thaliana samples using ACTIN.qPCRs for miR156 and miR172 were normalised using AasnoR101.Expression data were analysed using Student's t-test.Primers used in this analysis are listed in Table S1.

RNA-seq
For transcriptome analysis we dissected main shoot apices and V3 buds from Aatoe2_IL and wild-type plants as described above in three biological replicates per sample.RNA was extracted using the miRNeasy Plant Mini Kit (Qiagen, Hilden, Germany) and subjected to DNase treatment using DNA-free DNase (Ambion).PolyA enrichment, library preparation and sequencing were performed at the Max Planck Genome Center, Cologne (Germany).PolyA RNA was isolated from 1 µg of total RNA using NEBNext Poly(A) mRNA Magnetic Isolation Module (New England Biolabs, Frankfurt am Main, Germany) and used for library construction with NEBNext Ultra Directional RNA Library Prep Kit for Illumina (New England Biolabs).Sequencing was performed on an Illumina HiSeq 3000 system with 1 9 150-bp single-read lengths.
Sequencing reads were mapped and aligned to the reference genome (A.alpina V5) using STAR software (Dobin et al., 2013).Raw read counts per gene were quantified with HTSEQ v.0.5.4p1.The differentially expressed genes with more than a two-fold change and an adjusted P-value below 0.05 were Ó 2020 The Authors New Phytologist Ó 2020 New Phytologist Trust New Phytologist (2020) www.newphytologist.comobtained using DESeq2 (Anders & Huber, 2010) and selected for further analysis.The raw and processed data for RNA-seq analysis have been submitted to GenBank under the accession no.GSE145718.
Gene Ontology (GO) enrichment analysis was performed using the DAVID resource (Huang et al., 2008).DEGs with homologues in Arabidopsis thaliana were used as input data for GO analysis.GO terms with a Benjamini-Hochberg adjusted Pvalue lower than 0.05 were plotted using the R-based graphical tool BACA (Fortino et al., 2015).

Scanning electron microscopy (SEM) and light microscopy
For scanning electron microscopy, samples were fixed overnight at 4°C in 10 mM phosphate buffer containing 4% glutaraldehyde solution (pH 7.4), rinsed twice with 10 mM phosphate buffer and dehydrated in a graded ethanol series.Samples were critical point dried using a Leica CPD300 dryer (Leica Microsystems, Wetzlar, Germany) and mounted onto stubs with doublesided adhesive carbon tabs.Afterwards they were sputter-coated with platinum (Polaron SC 7640), images were taken with a Zeiss Supra 40VP SEM (Carl Zeiss, Jena, Germany).

Sequence analysis
For protein sequence alignments we used CLUSTAL OMEGA (Madeira et al., 2019).For synteny analysis, we compared the genomic regions containing AaTOE2 (LT669795.1)and TOE2 (AT5G60120) using the Graphic Alignment Tool for Comparative Sequence Analysis (GATA) plotter (Nix & Eisen, 2005).For phylogenetic analysis, the Maximum Likelihood method based on the JTT matrix-based model (Jones et al., 1992) was used to construct the phylogenetic tree with bootstrapping of 1000 iterations using MEGA7 software (Kumar et al., 2016).

Results
EOP055 is the orthologue of the AP2-like gene TOE2 and represses flowering in A. alpina and Arabidopsis thaliana To identify additional genes that regulate flowering and perennial traits in A. alpina, we performed an enhancer screen in the pep1-1 background.The enhancer of pep1-1 055 (eop055) mutant flowered earlier than pep1-1 in LD and SD glasshouse conditions (Fig. 1a,b).We combined mapping-by-sequencing and classical map-based cloning to identify the causal mutation in eop005 (Hartwig et al., 2012;Schneeberger, 2014;Andr es et al., 2015).SHOREMAP analysis indicated that this causal mutation in eop055 was located in a broad region spanning 40 Mb on the A. alpina chromosome 8 (Figs 1c, S1a; Table S2).We reduced the original mapping region (defined between markers M1 and M8) by screening 503 BC1F2 plants with markers designed based on SNPs from SHOREMAP.Fine mapping narrowed down the causal mutation to a 433 kb region (between markers M5 and M10), which contained a single SNP located in the CDS region of the A. alpina gene Aa_G234670 (Figs 1c,d, S1b).Aa_G234670 showed a higher degree of sequence similarity with the AP2-like transcription factor TOE2 in Arabidopsis thaliana (Fig. S2).This is one of six AP2-like genes in A. alpina, the others being (Fig. S2).Synteny of the flanking regions containing Aa_G234670 and TOE2 was conserved and Aa_G234670  S3a).Overall, these results suggested that Aa_G234670 is the orthologue of TOE2.
The SNP in AaTOE2 (Aa_G234670) causes a nonsynonymous amino acid substitution from glycine (G) to aspartic acid (D) at amino acid 171 within the first AP2 domain (AP2-R1) (Fig. S3b; Table S2).To verify whether the SNP in AaTOE2 was responsible for the early flowering phenotype in eop055, we complemented the eop055 mutant using the wild-type 7.8 kb AaTOE2 genomic region spanning 4128 bp upstream of its translational start and 776 bp downstream of its translational stop.Three independent T3 homozygous lines with a single copy insertion were selected and showed delayed flowering compared with eop055 (Fig. 2a,b).We also overexpressed the wild-type AaTOE2 and mutated Aatoe2 allele in Arabidopsis thaliana toe2-1.AaTOE2 complemented the early flowering phenotype of the toe2-1 mutant, whereas Aatoe2 did not (Figs 2c,d, S4).These results suggested that the role of AaTOE2 in flowering is conserved in Arabidopsis thaliana and that the G171D amino acid substitution affects its function.Taken together, these results demonstrate that the eop055 mutant phenotype was caused by the loss of function of AaTOE2, which represses flowering in A. alpina and Arabidopsis thaliana.
In Arabidopsis thaliana, TOE proteins regulate flowering through the photoperiod pathway (Yant et al., 2009;Zhang et al., 2015).In A. alpina, however, the role of the photoperiod pathway in flowering is not very clear and wild-type plants initiate floral buds during cold treatment under SDs (Wang R. et al., 2009).Flowering in response to prolonged cold treatment depends on plant age.Wild-type plants have a juvenile phase of 5 wk so that only plants older than 5 wk are competent to flower in response to prolonged cold treatment (Wang et al., 2011).We found that transcript accumulation of AaTOE2 in wild-type plants was high in the main shoot apex of vegetative plants and is transiently silenced during cold exposure.However, this is true only for plants that can initiate flowering during cold (adult, 8-wk-old) and not for cold-treated young seedlings (juvenile, 3-wk-old) which remain vegetative (Fig. 3).Transcript accumulation of AaTOE2 in AMs (within the V3 zone), which also remain vegetative, is similar during and after cold exposure (Fig. 3a).These results suggested AaTOE2 is a floral repressor that is highly expressed in vegetative apices and is differentially regulated spatiotemporally.

Aatoe2_IL responds to cold treatment at a young age
To understand the function of AaTOE2 in A. alpina we introgressed the Aatoe2 allele into the wild-type background by crossing eop055 with Pajares and selecting an Aatoe2_IL homozygous for Aatoe2 which contained a functional PEP1.We have previously shown that AP2 in A. alpina (PEP2) regulates flowering and perennial traits through the age pathway and plays a role in the vernalisation pathway by enhancing PEP1 mRNA levels (L azaro et al., 2019).The pep2-1 mutant does not require cold treatment to flower and has reduced PEP1 mRNA levels compared with the wild-type (Bergonzi et al., 2013).We checked whether AaTOE2 contributed to the obligate vernalisation requirement of A. alpina.We grew Aatoe2_IL plants in a LD glasshouse and tested whether they could flower without cold treatment.Similar to the wild-type, Aatoe2_IL plants did not flower in LDs (Fig. 4a).PEP1 mRNA levels also did not differ between 2-wk-old Aatoe2_IL and wild-type seedlings (Fig. 4b).This result suggested that AaTOE2 does not regulate flowering through PEP1.
AP2-like genes regulate the juvenile to adult phase change and TOE2 mRNA levels in Arabidopsis thaliana decrease in older plants (Jung et al., 2007;Wang, 2014).We tested AaTOE2 transcript accumulation in the main shoot apex of plants growing for 2-8 wk in a LD glasshouse without cold treatment.AaTOE2 transcript accumulation did not decrease as plants became older and resembled the expression patterns of PEP2 published previously (Fig. S5a; Bergonzi et al., 2013).Transcript accumulation of PEP1 in apices was similar between wild-type and Aatoe2_IL plants at all ages tested confirming again that AaTOE2 does not New Phytologist (2020) Ó 2020 The Authors New Phytologist Ó 2020 New Phytologist Trust www.newphytologist.comregulate flowering through PEP1 (Fig. S5i).Interestingly, PEP2 mRNA levels were slightly upregulated in Aatoe2_IL plants at most time points tested (Fig. S5j).AP2 proteins regulate flowering through several feedback mechanisms to upstream components in the age pathway (Yant et al., 2010).To test whether this is also the case in A. alpina, we also analysed the accumulation of miR156 and some of its target genes (AaSPL3/4/5, AaSPL9/15 and AaSPL10) (Figs 5a,b, S5b-f).miR156 accumulation was similar in both Aatoe2_IL and wild-type plants and reduced with plant age (Fig. 5a).Among all genes tested, only the accumulation of AaSPL5 transcripts differed in an age-dependent pattern between the Aatoe2_IL and wild-type (Figs 5b, S5b-f).AaSPL5 mRNA levels were higher in Aatoe2_IL and increased in the apices of older plants (Fig. 5b).In Arabidopsis thaliana, SPL3/4/5 transcript levels are upregulated in transgenic plants constitutively expressing 35S:MIR172b and are increased in double and triple mutants of AP2-like genes (Jung et al., 2011).To test whether miR172 accumulation was also coupled with the changes in AaSPL5, we compared miR172 levels between Aatoe2_IL and wild-type plants (Fig. 5c).The results we observed were opposite to what we would have expected based on studies in Arabidopsis thaliana and also in comparison with previous findings in A. alpina (Jung et al., 2011;Bergonzi et al., 2013).We found that miR172 accumulation increased with plant age and that in Aatoe2_IL it was lower compared with the wild-type (Fig. 5c).This result also correlated with the slight reduction in PEP2 mRNA levels in Aatoe2_IL apices (Fig. S5j).In Arabidopsis thaliana, SPL3/4/5 promote flowering by transcriptionally activating the floral meristem identity genes such as AP1, LFY and FUL in conjunction with the transcription factor FD (Yamaguchi et al., 2009;Wang J.W. et al., 2009a;Jung et al., 2016).In Aatoe2_IL plants growing in a LD glasshouse, the expression of meristem identity genes did not differ compared with the wild-type (Fig. S5g,h).These results were in agreement with the fact that Aatoe2_IL plants do not flower without cold treatment.
The effect of the age pathway in A. alpina is obvious after cold treatment (Wang et al., 2011;Bergonzi et al., 2013;Hyun et al., 2019;L azaro et al., 2019).To test whether AaTOE2 regulated the age-dependent response to vernalisation, we vernalised 3-wkold wild-type and Aatoe2_IL seedlings for 12 wk and tested whether they flowered (Fig. 5d).Aatoe2_IL seedlings flowered in response to cold treatment, whereas wild-type plants did not, suggesting that AaTOE2 regulated the response to vernalisation in an age-dependent manner (Fig. 5d).This result is also in agreement with the age-dependent differences observed in the silencing of AaTOE2 mRNA levels by cold.
Overall, these results suggested that AaTOE2 regulates the age-dependent response to vernalisation and influences miR172 and AaSPL5 mRNA levels at the shoot apical meristem (SAM).

Aatoe2_IL has more flowering axillary branches compared to the wild-type
In A. alpina, all genes identified so far that regulate the age-dependent response to vernalisation also contribute to the duration of vernalisation required for flowering and determine the fate of New Phytologist subapical axillary branches (Wang et al., 2011;Hyun et al., 2019;L azaro et al., 2019).To test whether this also stands for AaTOE2, we exposed 8-wk-old Aatoe2_IL plants to 8, 12 or 18 wk of cold treatment and tested flowering time after transfer to an LD glasshouse.Aatoe2_IL plants flowered earlier than wildtype after different durations of cold treatment (Fig. 6a).The biggest effect was observed in plants cold-treated for 8 wk when wild-type plants did not flower but Aatoe2_IL plants did (Fig. 6a,  b).This result suggested that AaTOE2 regulates the duration of vernalisation required for flowering in A. alpina.Interestingly, the shoot apical meristems of wild-type and Aatoe2_IL plants were both reproductive at the end of the 8 wk of cold treatment (Fig. S6), suggesting that AaTOE2 antagonises the commitment to flowering after vernalisation.
We also scored shoot architecture in plants exposed to 12 wk of vernalisation, as it was previously shown that this was sufficient to ensure the formation of zones with differential bud activity and fate (L azaro et al., 2018;Vayssi eres et al., 2020).As previously published, in wild-type plants the axillary branches at lower leaf nodes (V1) flowered in response to vernalisation and differed from inflorescence branches (I) because they only senesced partially after flowering (Fig. 7b; Wang R. et al., 2009b;L azaro et al., 2018;Vayssi eres et al., 2020).Wild-type A. alpina plants also maintained a zone of dormant axillary buds (V2) and vegetative axillary branches (V3), which are vital to maintain the perennial life cycle (L azaro et al., 2018;Vayssi eres et al., 2020).In Aatoe2_IL plants, the axillary branches in the leaf nodes corresponding to the V3 zone flowered (Fig. 7a,b).Flowering V3 branches in Aatoe2_IL behaved similarly to V1 branches and partially senesced (Figs 7d,e, S7).Interestingly, PEP1 mRNA levels in the V3 branches did not differ between wild-type and Aatoe2_IL plants, suggesting again that AaTOE2 regulates flowering in parallel to PEP1 (Fig. 7f).Furthermore, we observed a reduced number of V3 axillary branches and an increased number of V1 flowering axillary branches in Aatoe2_IL (Fig. 7b,c).This change in the number of metamers dedicated to V1, V2 and V3 zones was specific to the main shoot, as the architecture of the inflorescence was not influenced in Aatoe2_IL plants (Figs 7c, S8).
To understand how AaTOE2 regulates flowering in the main shoot and V3 axillary branches, we performed a transcriptome analysis.We exposed 8-wk-old Aatoe2_IL and wild-type plants to 4°C for 8 wk to induce flowering (Fig. S6), and dissected their SAM and V3 buds.Principal component analysis (PCA) analysis indicated that the variable explaining most of the variability (PC1) was tissue, which separated the SAM and V3 bud samples, whereas PC2 separated the samples according to their genotype (Fig. 8a).Most differentially expressed genes (DEGs) between genotypes were detected in V3 buds, with 422 genes being differentially expressed in V3 buds compared with 309 genes in the SAM (Fig. 8b; Table S3).In both tissues, GO analysis indicated enrichment in several biological processes, including response to various stresses and regulation of cellular, metabolic and developmental processes (Fig. S9).Several genes involved in stress responses were differentially expressed between Aatoe2_IL and the wild-type (Fig. 8c).In the SAM, many genes involved in the reproductive development were upregulated in the Aatoe2_IL (e.g.AP1, AGL5 and DYT1) (Fig. 8c; Table S3; Alejandra Mandel et al., 1992;Savidge et al., 1995;Zhang et al., 2006).This result suggested that Aatoe2_IL initiated flowering during cold treatment earlier than wild-type plants.Among the genes upregulated specifically in the V3 buds of Aatoe2_IL, we detected again SPLs (SPL3, 4, 5 and 10) and also several TCP genes (AaTCP4, 10, 24) known to be involved in leaf growth, flower development, flowering time and jasmonic acid (JA) metabolism (Fig. 8c; Table S3; Wu et al., 2009;Nag et al., 2009;Danisman et al., 2012;Xu et al., 2016;Jung et al., 2016;Lucero et al., 2017;Chang et al., 2018).In the V3 buds, miR172 accumulation was higher in Aatoe2_IL, whereas miR156 was lower compared with the wild-type (Fig. S10a).These results differed from the miR172 New Phytologist (2020) Ó 2020 The Authors New Phytologist Ó 2020 New Phytologist Trust www.newphytologist.comexpression patterns we obtained in the main shoot apex of vegetative plants in a LD glasshouse (Fig. 5c), although AaSPL5 mRNA levels followed a similar trend and were higher in both vegetative shoot apices and V3 buds of Aatoe2_IL plants (Figs 5, S10).V3 buds are formed during vernalisation (Ponraj & Theres, 2020) and therefore can commit to reproductive development if they acquire competence.To check whether V3 buds of Aatoe2_IL plants become reproductive during vernalisation, we cold-treated plants for 12 wk and looked at axillary buds in the V3 zone (Fig. S11).In wild-type plants, the axillary buds were vegetative whereas the ones from Aatoe2_IL plants had the characteristic dome shape of a flowering meristem (Fig. S11).These results suggested that flowering in V3 buds of Aatoe2_IL plants is initiated during vernalisation and that the increase of miR172 accumulation and AaSPL5 mRNA levels correlated with the initiation of flowering.
Taken together, these results suggested that AaTOE2 represses flowering in the SAM and V3 buds during vernalisation.

Initiation rate of AMs and leaves are affected in Aatoe2_IL
In Aatoe2_IL plants, the number of basal flowering axillary branches (corresponding to V1) was increased (Fig. 7b,c).As V1 axillary branches are formed before cold exposure, we compared  S12b).Components of the age pathway regulate other traits, including leaf initiation rate (plastochron) and branching (Wang & Wang, 2015).The toe2 mutants do not have a clear plastochron phenotype and in 35S: MIR172b plants the plastochron is not affected (Jung et al., 2011).Nevertheless, in the triple mutant spl3/4/5 plastochron is reduced (Jung et al., 2016).We followed leaf initiation in plants grown for up to 8 wk in LDs, and by contrast to the branching phenotype, we detected a statistically significant reduction in the leaf initiation rate in Aatoe2_IL plants compared with the wildtype (Fig. S12).
We also had a closer look at the branching phenotype in the Aatoe2_IL by scoring the presence or absence of an AM/bud in each leaf node at different developmental stages (Fig. 9b-d).In wild-type plants, it has recently been reported that during vegetative development the leaf axils close to the SAM do not contain an AM (Ponraj & Theres, 2020).These leaf axils fill up during vernalisation when the SAM transitions to reproductive development (Ponraj & Theres, 2020).Aatoe2_IL plants have a reduced number of empty leaf axils before and during vernalisation, suggesting that AM initiation is enhanced compared with the wild-type, which results in the increased number of basal axillary branches (Fig. 9e,f).Empty apical leaf axils in plants before vernalisation serve as sites where V3 buds will be formed during cold (Ponraj & Theres, 2020), explaining the reduced number of V3 axillary branches in Aatoe2_IL plants (Fig. 9e,g,h).
These results indicated that AaTOE2 regulates AM formation and shoot architecture in A. alpina.

Discussion
In perennials, regrowth from one year to the next is ensured by vegetative axillary or apical meristems.Therefore, perennial New Phytologist species usually show a complex shoot architecture that can consist of dormant buds and axillary vegetative and/or flowering branches (Costes et al., 2014;Vayssi eres et al., 2020).This strategy of splitting resources towards vegetative and reproductive development affects yield (Bazzaz et al., 1987).Thus, although perennial crops are considered to be the future in sustainable agriculture, they could eventually have a reduced yearly yield compared with established annual crops (Cox et al., 2006).Understanding shoot architecture in perennials can provide us with tools to improve crop yield.In Arabidopsis thaliana, the flowering phenotype of mutants such as toe2 is not very strong due to functional redundancy with other AP2-like genes (Aukerman & Sakai, 2003;Jung et al., 2007;Zhai et al., 2015).We describe that in A. alpina, although it contains six AP2-like genes, the toe2 mutant has a strong early flowering time phenotype.Thus, we concluded that functional redundancy between AP2like genes might be weaker in A. alpina.We also show that AaTOE2 contributed to the polycarpic growth habit by regulating the number and fate of axillary branches that will stay vegetative the following year.This role of AaTOE2 can affect the final yield as it modulates the balance between reproductive and vegetative development.AaTOE2 also controls the vernalisation response by regulating the age at which plants are sensitive to vernalisation as well as the duration of cold required for flowering.Hence, we concluded that AaTOE2 regulates different developmental traits that contribute to the perennial life cycle of A. alpina.
AaTOE2 regulates reproductive competence and the duration of cold treatment required for flowering in A. alpina Wild-type Pajares plants flower only if exposed to cold treatment at an age older than 5 wk (Wang et al., 2011;Bergonzi et al., 2013).Similar to other species, in A. alpina miR156 and members of the SPL and AP2-subfamilies also regulate the competence to flower and determine the age at which plants are able to respond to vernalisation (Bergonzi et al., 2013;Teotia & Tang, 2015;Hyun et al., 2019;L azaro et al., 2019).Transgenic A. alpina lines with reduced miR156 activity or overexpressing the miR156-resistant form of AaSPL15, flower when vernalised as young seedlings (grown for 2 or 3 wk before vernalisation) (Bergonzi et al., 2013;Hyun et al., 2019).We demonstrated that miR172 accumulation increased when A. alpina plants became older, resembling previous studies in Arabidopsis thaliana and other species, a mechanism regulated by SPL9, SPL10 and SPL3/ 4/5 (Aukerman & Sakai, 2003;Chuck et al., 2007;Wu et al., 2009;Xie et al., 2020).We also established a role for AaTOE2 in reproductive competence and placed AaSPL5 downstream of AaTOE2 (Fig. 10).In Arabidopsis thaliana SPL3/4/5 mRNA levels are upregulated in plants overexpressing miR172 and in toe1 toe2 double or toe1 toe2 smz triple mutants (Jung et al., 2011).Thus, it has been suggested that SPL3/4/5 are post-transcriptionally regulated by miR156 and transcriptionally by miR172 probably through the AP2-like genes (Jung et al., 2011).Our data indicated that in A. alpina there are parallel pathways to those described in Arabidopsis thaliana and feedback loops influence the expression of SPLs and miR172.However, by contrast with Arabidopsis thaliana, we detected that AaTOE2 in vegetative shoot apices positively regulates miR172 accumulation (Yant et al., 2010).SPL3/4/5 induced flowering by binding to the promoters of APETALA1, LEAFY, and FRUITFULL (Yamaguchi et al., 2009;Wang J.W. et al., 2009a;Jung et al., 2016) New Phytologist (2020) Ó 2020 The Authors New Phytologist Ó 2020 New Phytologist Trust www.newphytologist.comcommitment in cold, which is an important trait for alpine species adapted to environments with short growth seasons (Billings & Mooney, 1968;Aydelotte & Diggle, 1997;Wang R. et al., 2009b;L azaro et al., 2018).Interestingly, all genes that regulate competence to flower in A. alpina also influence the duration of vernalisation required for flowering (Fig. 10).We have previously published that A. alpina mutants in floral repressors (e.g.AaTFL1, PEP2) that regulate the age-dependent response to vernalisation also flower after a shorter duration of vernalisation, whereas mutants in floral promoters (e.g.AaSPL15) show strong floral reversion phenotypes after cold exposure (Wang et al., 2011;Bergonzi et al., 2013;Hyun et al., 2019).It is, therefore, very interesting to understand why and how these two traits are associated, especially because transcript accumulation of these genes does not differ between nonvernalised young and old seedlings (Wang et al., 2011;Bergonzi et al., 2013;Hyun et al., 2019).AP2 and AP2-like genes have been mainly studied for traits such as floral development, inflorescence and spikelet architecture (Greenwood et al., 2017;Franc ßois et al., 2018;Gattolin et al., 2018;Harrop et al., 2019;Debernardi et al., 2020).Similarly, TFL1 and SPL15 have also been reported to play a role in inflorescence architecture in other species (Jiao et al., 2010;Jiang et al., 2013).For AaTOE2, we did not observe a role in inflorescence architecture.However, the fact that the development of floral buds in A. alpina occurs during vernalisation may explain the role of AaTOE2 in the duration of vernalisation required for flowering.Our transcriptome analysis also indicated that AaTOE2 may influence how plants modify developmental transitions due to environmental stresses, as previously suggested for other flowering time genes (Riboni et al., 2014;Kazan & Lyons, 2016).
AaTOE2 regulates shoot architecture by inhibiting AM initiation and by repressing flowering in AMs formed during cold treatment Shoot architecture in A. alpina is organised in zones of consequent nodes that behave in a similar way and is determined by the differential response of AMs to cold treatment (L azaro et al., 2018;Vayssi eres et al., 2020).We have previously shown that the maintenance of vegetative growth is determined after vernalisation by PEP1, whose mRNA levels are upregulated in the subapical axillary branches (V3 branches) to repress flowering (Wang R. et al., 2009b;L azaro et al., 2018).AMs that will give rise to V3 axillary branches are formed during vernalisation in the axils of the leaf primordia close to the SAM (Ponraj & Theres, 2020).In the Aatoe2_IL plants, V3 AMs initiate flowering during cold treatment, suggesting that the fate of these meristems is regulated during vernalisation by AaTOE2.The initiation of flowering in V3 buds was also correlated with a strong increase of miR172 levels, which differed to our results on the vegetative shoot apices.Thus, the network between different components in the age pathway may vary between vegetative and flowering apices.AaSPL5, probably, does not participate in this differential AaTOE2/ miR172 feedback.The A. alpina AP2 (PEP2) regulates the fate of V3 branches by ensuring the upregulation of PEP1 mRNA levels after vernalisation (L azaro et al., 2019).In this respect, AaTOE2 differs from PEP2 as it does not influence PEP1 expression levels.Nevertheless, AaTOE2 has an effect on flowering only in vernalised plants suggesting that PEP1 plays a prominent role repressing flowering in parallel to AaTOE2.PEP2 regulates flowering through the age and the vernalisation pathways (L azaro et al., 2019).Thus, although AaTOE2 does not influence PEP1 mRNA levels, it could have overlapping functions with PEP2 through the age pathway.
AMs are initiated acropetally at a distance from the SAM and during vegetative development in Arabidopsis thaliana and A. alpina plants the leaf axils adjacent to the SAM lack an AM (Stirnberg et al., 1999;Reinhardt & Kuhlemeier, 2002;Greb et al., 2003;Ponraj & Theres, 2020).In Arabidopsis thaliana, the signal from the SAM that inhibits AM initiation may be auxin, as auxin resistant mutants (axr1-3 and axr1-12) have a reduced number of leaf nodes without an AM (Ponraj & Theres, 2020).Empty leaf axils close to the SAM are filled with an AM soon after A. alpina plants are exposed to cold (Ponraj & Theres, 2020).Cold treatment also causes a reduction of endogenous IAA levels in A. alpina stems (Vayssi eres et al., 2020).These results suggested that in A. alpina similar mechanisms might contribute to the inhibition and release of growth in the subapical leaf axils.In this study, we observed that Aatoe2_IL plants had an enhanced AM initiation rate and a reduced number of leaves without an AM.TOE2 or other AP2-like genes or even their regulator miR172 have not been reported in Arabidopsis thaliana to play a role in AMs initiation and/or to have a branching phenotype (Mathieu et al., 2009;Wu et al., 2009;Yant et al., 2010;Zhai et al., 2015).Nevertheless, miR156 and the SPLs contribute to branching and this role is also coupled with shortened plastochron lengths, which is again a trait that has not been reported for AP2-like genes and miR172 (Wang & Li, 2008;Schwarz et al., 2008;Jung et al., 2011;Zhou et al., 2013).OsSPL14, which is the orthologue of the Arabidopsis thaliana SPL15, regulates tiller number in rice (Jiao et al., 2010;Luo et al., 2012).The SPLs in Arabidopsis thaliana (specifically SPL9 and SPL15) suppress LATERAL SUPPRESSOR (LAS) expression, which is a central regulator of AM initiation (Greb et al., 2003;Tian et al., 2014).The spl9-4 spl15-1 double mutants have accessory buds which is an indicator of LAS overexpression (Tian et al., 2014).We observed differences in the number of leaf axils dedicated to V1 and V3 zones in Aatoe2_IL plants, which is a consequence of the enhanced AM initiation rate in vegetative plants before being exposed to cold treatment.Aatoe2_IL plants had more basal V1 axillary branches and fewer branches dedicated to the V3 zone.This phenotype has not been reported in the Aaspl15 mutant, suggesting that the role of AaTOE2 in AMs initiation is independent of AaSPL15 (Hyun et al., 2019).
Overall, we demonstrated that AaTOE2 regulates several developmental traits that contribute to the perennial life cycle of A. alpina.So far, the miR156/SPL module has been described as a regulatory hub due to its conservation in different species and the range of different phenotypes it influences (Wang, 2015; Ó 2020 The Authors New Phytologist Ó 2020 New Phytologist Trust New Phytologist (2020) www.newphytologist.comWang & Wang, 2015).We propose that AP2-like genes will be useful to breed for improved plant architecture and enhanced yield.Table S1 Primers used in this study.
Table S2 List of single nucleotide polymorphism within coding sequence regions in chromosome 8 for the Arabis alpina eop055 mutant.
Table S3 DEGs between Aatoe2_IL and wild-type in the main shoot apical meristem and V3 buds.
Please note: Wiley Blackwell are 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.

Fig. 1
Fig. 1 Arabis alpina TOE2 (AaTOE2) is the floral repressor EOP055.(a) Flowering phenotypes of pep1-1 and eop055 mutants grown in long days (LDs) for 51 d.(b) Total leaf number at flowering in LDs and short days (SDs).(c) Mapping of eop055.The eop055 mutation was originally mapped using SHOREMAP to the A. alpina chromosome 8 between 6 473 171 bp (M1) and 45 736 690 bp (M2).Using fine mapping, the candidate region was narrowed down to a 433 kb genomic region between markers M5 (41 140 691 bp) and M10 (41 574 114 bp).The detailed cloning approach is presented in Fig. S1.(d) Scheme of the AaTOE2 locus (Aa_G234670) showing the position of the mutation and the single nucleotide polymorphism (SNP) found in eop055.In (a) bar, 5 cm.In (b) n = 24-28.**Indicates significant differences at P < 0.01 according to Student's t-test.The boxes indicate the interquartile range (IQR), the horizontal line in the middle is the median, the vertical lines (whiskers) correspond to the maximum or minimum value within 1.5 9 IQR and the dots are the sample values.In (d) black boxes indicate the exons and white boxes indicate the untranslated regions.Red lines indicate the AP2 domains and the orange line indicates the miR172 binding site.G171D indicates the nonsynonymous amino acid substitution in the eop055 mutant from glycine (G) to aspartic acid (D) at amino acid 171.TOE2, TARGET OF EAT 2; EOP055, ENCHANCER OF PEP1; PEP1, PERPETUAL FLOWERING 1.

Fig. 2
Fig. 2 AaTOE2 complements the early flowering phenotypes of the Arabis alpina eop055 and Arabidopsis thaliana toe2-1 mutants.(a, b) Complementation of eop055.(a) Flowering phenotypes of pep1-1, eop055 and three independent pAaTOE2:gAaTOE2 eop055 transgenic lines grown in long days (LDs) for 57 d.(b) Total leaf number at flowering of pep1-1, eop055 and transgenic lines.(c, d) Complementation of Arabidopsis thaliana toe2-1 mutant.(c) Flowering phenotypes of Col, toe2-1 and transgenic lines constitutively expressing the wild-type allele (35S:AaTOE2) and mutated allele (35S:Aatoe2) of AaTOE2.The picture was taken 35 d after sowing in LDs.(d) Total leaf number at bolting of Col, toe2-1 and transgenic lines in LDs.AaTOE2 transcript accumulation in Arabidopsis thaliana transgenic lines is presented in Supporting Information Fig. S4.Bars: (a) 5 cm; (c) 2 cm.n = 12-15.ns, not significant.**Indicates significant differences compared with eop055 in (b) and toe2-1 in (d) at P < 0.01 according to Student's t-test.The boxes and the horizontal line in the middle represent the interquartile range (IQR) and the median, respectively.The whiskers correspond to the maximum or minimum value within 1.5 9 IQR, the dots indicate the sample values.TOE2, TARGET OF EAT 2.

Fig. 3 Fig. 4
Fig.3AaTOE2 transcript accumulation in wild-type Arabis alpina plants differs between meristems and is transiently silenced by cold treatment in a spatial and age-dependent manner.(a) AaTOE2 mRNA levels in the main shoot apex and subapical axillary buds (V3) of cold-treated 8-wk-old plants.(b) AaTOE2 mRNA levels in the main shoot apex of cold-treated 3wk-old plants.The 8-wk-old plants are adult and can initiate flowering in response to vernalisation, whereas 3-wk-old seedlings are juvenile and stay vegetative.Apices/axillary buds were harvested from wild-type plants grown in LD glasshouse conditions after 3 and 8 wk, after cold treatment for 8 and 12 wk and after the return to glasshouse conditions for 1 wk.Error bars represent the standard deviation of three biological replicates.TOE2, TARGET OF EAT 2.

Fig. 5
Fig. 5 AaTOE2 in Arabis alpina regulates the age-dependent response to cold treatment.(a-c) Expression patterns of (a) miR156, (b) AaSPL5 and (c) miR172 in apices of wild-type (WT) and Aatoe2_IL plants growing for up to 8 wk in a long day (LD) glasshouse.Transcript accumulation of additional genes using the same samples is presented in Supporting Information Fig. S5.(d) Flowering phenotypes of WT and Aatoe2_IL plants vernalised at the age of 3 wk.Seedlings were exposed to 12 wk of cold treatment followed by 4 wk in LDs.Error bars represent the standard deviation of three biological replicates.**Indicates significant differences between WT and Aatoe2_IL at each time point at P < 0.01 according to Student's t-test.Bar, 5 cm.TOE2, TARGET OF EAT 2; miR156, microRNA 156, SPL5, SQUAMOSA PROMOTER-BINDING PROTEIN-LIKE 5; miR172, microRNA 172.

Fig. 6
Fig. 6 AaTOE2 extends the duration of cold treatment required for flowering in adult Arabis alpina plants.(a) Days to flowering of wild-type (WT) and Aatoe2_IL plants exposed to different durations of cold treatment.Flowering time was measured in plants grown for 8 wk in long days (LDs), exposed to 8, 12 or 18 wk of cold treatment and subsequently returned to a LD glasshouse.Days to flowering were counted from the day that plants were transferred to LD glasshouse conditions after vernalisation.(b) Flowering phenotype of WT and Aatoe2_IL plants cold-treated for 8 wk, followed by 4 wk in LDs.The boxes indicate the interquartile range (IQR), the horizontal line in the middle is the median, the vertical lines (whiskers) correspond to the maximum or minimum value within 1.5 9 IQR and the dots are the outliers.**Indicates significant differences between WT and Aatoe2_IL at each cold treatment at P < 0.01 according to Student's t-test.Bar, 5 cm.n = 12.TOE2, TARGET OF EAT 2.

Fig. 7
Fig. 7 AaTOE2 regulates shoot architecture in Arabis alpina by influencing the number of flowering axillary branches.(a-e) Scoring of plant architecture in wild-type (WT) and Aatoe2_IL plants grown for 8 wk in long days (LDs), vernalised for 12 wk and transferred back to an LD glasshouse.(a) Flowering phenotype of vernalised WT and Aatoe2_IL plants after they were returned to an LD glasshouse for 4 wk, in which subapical axillary branches in Aatoe2_IL plants flowered (arrows).(b) Schematic representation of shoot architecture of flowering WT and Aatoe2_IL plants.Each column represents a single plant and each box within a column represents a single leaf axil numbered from the bottom to the top of the plant.Different zones are indicated on the right as described for a WT plant in Vayssi eres et al. (2020).V1, flowering axillary branches that partially senesce; V2, dormant buds; V3, vegetative axillary branches; I, inflorescence.Light orange denotes a flowering axillary branch and orange a flowering inflorescence branch.Green represents vegetative axillary branches and grey axillary buds or empty leaf axils.Boxes with circles indicate axillary branches in the V3 zone.(c) Number of leaf axils with or without an axillary flowering/vegetative branch in each zone from (b).(d) Flowering phenotype of vernalised WT and Aatoe2_IL plants after they were returned for 20 wk to a LD glasshouse, in which subapical axillary branches in Aatoe2_IL plants partially senesced (arrows).(e) V3 axillary branches of WT and Aatoe2_IL plants that are indicated in (d).V1 axillary branches on the same plants are shown in Supporting Information Fig. S7.Characterisation of inflorescence architecture on the same plants is presented in Fig. S8.(f) Relative expression of PEP1 in V3 axillary branches of WT and Aatoe2_IL plants during and after vernalisation.Bars, 5 cm (a-e).Error bars in (c) and (f) represent the standard deviation.In (c) ** indicates significant differences between WT and Aatoe2_IL plants at each zone at P < 0.01 according to Student's t-test.n = 11.TOE2, TARGET OF EAT 2; PEP1, PERPETUAL FLOWERING 1.

Fig. 10
Fig. 10 Schematic diagram showing the network of genes regulating flowering in the perennial Arabis alpina.The age and the vernalisation pathways in A. alpina are equally important for flowering.PERPETUAL FLOWERING 1 (PEP1), represses flowering through the vernalisation pathway.The A. alpina orthologues of APETALA2 (PERPETUAL FLOWERING2, PEP2), SQUAMOSA PROMOTER-BINDING PROTEIN-LIKE15 (AaSPL15), TERMINAL FLOWER1 (AaTFL1) and TARGET OF EAT2 (AaTOE2) regulate flowering through the age pathway.Most of these genes also regulate other traits related to flowering (blue boxes): (1) The duration of vernalisation required for flowering and the achievement of floral commitment during cold exposure and (2) The fate of axillary branches.In this study, we demonstrated that AaTOE2 also regulates axillary meristem initiation.Genes described in this study are shown in purple.The feedback loop between AaTOE2 and miR172 (microRNA 172) differs between vegetative (dotted line) and flowering (solid line) meristems.

Fig. S2
Fig. S2 Phylogenetic relationship of the AP2-like family in Arabis alpina and Arabidopsis thaliana.

Fig. S5
Fig. S5 Transcript accumulation of several flowering time regulators does not differ between Arabis alpina wild-type and Aatoe2_IL plants grown in continuous long day conditions.

Fig. S6
Fig. S6 Scanning electron microscopy images of the main shoot apex from wild-type and Aatoe2_IL plants vernalised for 8 wk.

Fig. S7
Fig.S7V1 axillary branches of vernalised Arabis alpina wild-type and Aatoe2_IL plants after they were returned for 20 wk to a long day glasshouse.

Fig. S8
Fig.S8Inflorescence architecture of Arabis alpina is not influenced in Aatoe2_IL plants.

Fig. S9
Fig.S9Gene ontology enrichment analysis of differentially expressed genes detected between wild-type and Aatoe2_IL plants in two different tissues: the main shoot apical meristem and V3 axillary buds.

Fig. S12
Fig. S12 Plastochron and number of axillary branches are increased in Arabis alpina Aatoe2_IL plants.