Auxin Response Factors promote organogenesis by chromatin-mediated repression of the pluripotency gene SHOOTMERISTEMLESS

Specification of new organs from transit amplifying cells is critical for higher eukaryote development. In plants, a central stem cell pool maintained by the pluripotency factor SHOOTMERISTEMLESS (STM), is surrounded by transit amplifying cells competent to respond to auxin hormone maxima by giving rise to new organs. Auxin triggers flower initiation through Auxin Response Factor (ARF) MONOPTEROS (MP) and recruitment of chromatin remodelers to activate genes promoting floral fate. The contribution of gene repression to reproductive primordium initiation is poorly understood. Here we show that downregulation of the STM pluripotency gene promotes initiation of flowers and uncover the mechanism for STM silencing. The ARFs ETTIN (ETT) and ARF4 promote organogenesis at the reproductive shoot apex in parallel with MP via histone-deacetylation mediated transcriptional silencing of STM. ETT and ARF4 directly repress STM, while MP acts indirectly, through its target FILAMENTOUS FLOWER (FIL). Our data suggest that – as in animals- downregulation of the pluripotency program is important for organogenesis in plants.

P lants give rise to new organs continuously throughout their life. During reproduction, primordia that give rise to flowers and cryptic bracts (henceforth termed reproductive primordia), arise from a population of transit amplifying cells at the flanks of the inflorescence shoot apex 1,2 . Flowers are plant organs important for reproductive success and yield 3 . Perception of a local auxin maximum triggers transcriptional responses that enable specification of reproductive primordium founder cells 4,5 . When auxin levels are low, small nuclear Aux/IAA proteins bind to MONOPTEROS/Auxin Response Factor 5 (MP; ARF5) and recruit co-repressor complexes to prevent activation of auxin responsive genes 6,7 . Increased auxin levels lead to ubiquitinmediated degradation of Aux/IAA proteins 8 , releasing the corepressors and allowing MP to recruit the SWI/SNF family chromatin remodelers BRAHMA and SPLAYED (SYD) 9 . In this fashion, MP directly upregulates genes important for flower development, such as those encoding the transcription factors FILAMENTOUS FLOWER (FIL) and LEAFY (LFY) 9,10 . MP also directly modulates expression of genes that control hormone accumulation and response [11][12][13] .
Auxin Response Factors are divided into three evolutionarily conserved classes, A, B and C 14 . Class A ARFs are classified as transcriptional activators, while class B and C ARFs are classified as transcriptional repressors 15,16 . MP is a class A ARF with an unstructured glutamine-rich central domain that can interact with SWI/SNF family chromatin remodelers 9 . Besides MP, only two other ARFs are strongly expressed in founder cells of the reproductive primordia: ETTIN (ETT; also called ARF3) and ARF4 17 . Both ETT and ARF4 are 'repressive' class B ARFs 15,16 . ETT and ARF4 have partly redundant roles in several developmental processes in Arabidopsis, including lateral root initiation and leaf polarity [18][19][20] .
A key pluripotency gene in plants is the class I KNOX homeobox transcription factor SHOOTMERISTEMLESS (STM) 21 . STM promotes meristematic fate in part by upregulating biosynthesis of the hormone cytokinin, which acts in a positive feedback loop to promote stem cell fate 22,23 . STM frequently acts in concert with other class I KNOX genes such as BREVIPEDI-CELLUS (BP) and the activity of the class I KNOX proteins is modulated by a different class of homeodomain proteins, the BELlike proteins (see 24 for review). STM expression is high throughout the shoot apical meristem, but is downregulated at sites of auxin maxima in transit amplifying cells 25,26 . Neither the functional significance nor the mechanism for STM downregulation in primordium initials are currently understood.
We show here that STM repression promotes initiation of reproductive primordia. We further demonstrate that the class B ARFs ETT and ARF4, which are expressed in incipient reproductive primordia, act in parallel with MP to downregulate STM as well as BP. We implicate the MP target FIL in direct repression of the class I KNOX genes together with ETT and ARF4. Finally, we reveal that silencing of the pluripotency gene STM and BP by FIL and ETT/ARF4 is mediated by histone deacetylation.

Results
ETT and ARF4 promote flower initiation with MP. To probe whether ETT and ARF4 contribute to flower initiation, we performed a genetic enhancer test using the hypomorphic mp allele mp-S319 27 . mp-S319 has a T-DNA insertion in the second to last exon of MP 28 and displays weaker phenotypes than mp null mutants, including partially compromised flower initiation (Fig. 1a, b) 10 . Loss of ETT and ARF4 activity in this background caused formation of naked inflorescence 'pins' that lack flowers (Fig. 1a, b). We also generated triple mutants between the MP interacting chromatin remodeler SYD and ETT/ARF4. Like mp-S319 ett arf4, syd-5 ett arf4 mutants formed naked inflorescence pins ( Supplementary Fig. 1a, b). These data suggest that the class B ARFs ETT and ARF4 promote flower initiation.
Stipules formed on syd-5 ett arf4 inflorescence pins, pointing to increased meristematic activity 29,30 . Indeed, molecular characterization of the flower initiation defects of syd-5 ett arf4 and mp-S319 ett arf4 revealed a striking expansion of the STM expression into incipient reproductive primordia ( Supplementary  Fig. 1c, Fig. 1c). The expression domain of the shoot apical  meristem stem cell marker CLAVATA3 (CLV3) was not dramatically increased, suggesting that the expansion of the STM expression domain is not accompanied by an increase in the size of the CLV3-expressing stem cell pool. (Fig. 1c, Supplementary Fig. 1c). As in the wild type (WT) 25,26 , STM was downregulated at the flanks of the shoot apex of mp-S319 and ett arf4 mutants (Fig. 1c, Supplementary Fig. 2). Slightly elevated expression of STM was also detected by quantitative reverse transcriptase PCR (qRT-PCR) in entire mp-S319 ett arf4 or syd-5 ett arf4 inflorescences (Fig. 1d, Supplementary Fig. 1d). The observed increase in STM levels in mp-S319 ett arf4 is probably an underestimate, because mp-S319 ett arf4 mutants lack STM expressing flower meristems, in contrast to the other genotypes tested. We propose that the overexpression of STM in mp-S319 ett arf4 is likely to be functionally important, because it triggered a marked increase in the expression of the STM target IPT7, a cytokinin hormone biosynthesis gene 23 (Fig. 1d).
To probe whether ETT and ARF4 act in parallel with MP, we generated triple mutants between the mp-12 null mutant 31 and ett arf4. While most of the plant body is elaborated post embryogenesis, the root, hypocotyl and embryonic leaves (cotyledons) are formed during embryo development 32,33 . mp-12 ett arf4 mutants, like mp-12 mutants, did not form roots (Fig. 2a, b). In addition, mp-12 ett arf4 seedlings exhibited severe defects in cotyledon initiation and had enlarged shoot apical meristems ( Fig. 2a-d). In situ hybridization in mp-12 ett arf4 seedlings demonstrated an expansion of the STM expression domain into the region from where primordia initiate (Fig. 2e). qRT-PCR confirmed strongly increased expression levels of STM and IPT7 in the triple mutant relative to the parental lines (Fig. 2f). An expanded STM expression domain was also apparent in mature embryos segregating in mp-12/+ ett/+ arf4/-plants that expressed a pSTM:GUS reporter 34 (Fig. 2g). Based on these results, we conclude that ETT and ARF4 act in parallel with MP to promote organogenesis by repressing STM expression.
STM downregulation is important for flower initiation. To directly test for a biological role of STM in flower initiation, we used a steroid inducible version of STM, p35S:STM-GR 22 (Supplementary Fig. 3a) to conditionally elevate STM levels in mp-S319 mutant inflorescence meristems. Increased STM accumulation significantly enhanced the mp-S319 flower initiation defect, causing formation of naked inflorescence pins (Fig. 3a, b). In agreement with the known positive feedback loop between STM and cytokinin 22,23 , treatment with cytokinin (zeatin) likewise enhanced the flower initiation defects of the hypomorph mp-S319 mutant ( Supplementary Fig. 4). On the other hand, introduction of a previously characterized artificial microRNA that specifically targets STM 35 into the mp-S319 ett arf4 triple mutant significantly rescued the floral initiation defect of the triple mutant (Fig. 3c, d) despite displaying only a partial decrease in STM accumulation ( Supplementary Fig. 3b). The combined data suggest that downregulation of the STM pluripotency gene promotes reproductive primordium initiation.
ETT directly and MP indirectly repress STM. The effect of ETT/ ARF4 or MP on STM expression could be direct or indirect. To distinguish between these possibilities, we employed plant lines expressing biologically active genomic constructs for ETT-GFP 36 or MP-6xHA 9 ( Supplementary Fig. 3c) to conduct chromatin immunoprecipitation (ChIP) followed by qPCR. We assayed binding to five evolutionarily conserved regions of the STM locus (Fig. 4a). ETT-GFP bound strongly and specifically to STM, with the strongest binding observed in the second intron, which has many putative ETT binding sites (Fig. 4a, b). MP did not associate with any of the regions of the STM locus tested (Fig. 4a, c), but did bind the previously identified direct target FIL 10 . Although our genetic analyses suggest that MP acts in parallel with ETT and ARF4 to repress STM expression, the chromatin immunoprecipitation data point to a possible indirect role of MP in this process. This conclusion was supported by examination of a second class I KNOX gene, BREVIPEDICELLUS (BP), whose expression is also elevated in mp-S319 ett arf5 and mp-12 ett arf4  Supplementary Fig. 5a). When we tested their association with this locus by ChIP-qPCR, we found that ETT, but not MP bound to the BP locus (Supplementary Fig. 5e-g).
The MP target FIL represses STM together with ETT/ARF4. We next asked whether one or several MP targets might act in concert with ETT and ARF4 to directly repress class I KNOX gene expression. One obvious candidate is the YABBY transcription factor FIL, a direct MP target with a role in flower initiation 9 . FIL has been shown to suppress class I KNOX gene expression in leaves 37 . In addition to its expression in the abaxial domain of developing leaves and flowers, FIL is expressed at the flanks of the shoot apex in the primordium founder cells from where primordia initiate [38][39][40] . We compared the accumulation of FIL mRNA with that of ETT, ARF4 and MP in the reproductive shoot apex by situ hybridization and confirmed that all four genes are expressed in the primordium founder cells ( Supplementary  Fig. 6), the region where STM is depleted 26 . In agreement with a role for FIL in flower initiation, fil ett arf4 triple mutant inflorescences displayed defects that ranged from initiation of filamentous structures in lieu of flowers to pins in older fil ett arf4 inflorescences ( Fig. 5a, b). Treatment with low doses of an auxin transport inhibitor caused formation of naked inflorescence pins in fil ett arf4 plants, but not in any of the control genotypes ( Fig. 5a, b, Supplementary Fig. 7). Furthermore, fil ett arf4 mutant inflorescences also displayed increased expression of STM, IPT7 and BP relative to the parental lines ( Fig. 5c, Supplementary  Fig. 5c). To test whether FIL directly regulates STM expression, we generated a biologically active, 6xHA-tagged genomic construct for FIL ( Supplementary Fig. 3d). ChIP qPCR revealed that FIL bound to the STM locus, with the strongest binding observed at the proximal STM promoter (Fig. 5d), a conserved region with a FIL binding motif 41 . FIL also bound to the BP class I KNOX gene locus ( Supplementary Fig. 5i). These data suggest that FIL directly represses class I KNOX genes including STM downstream of MP. A role for FIL and ETT/ARF4 in STM repression is further supported by the finding that fil ett arf4 mutants also displayed leaf phenotypes characteristic of class I KNOX overexpressing plants 30 . These phenotypes were already apparent in fil/+ ett arf4 mutants. Instead of the simple leaf typical of wild-type Arabidopsis plants, fil/+ ett arf4 leaves were lobed and divided (Fig. 6a, b). The leaf phenotype of the fil ett arf4 triple mutants was even more severe. fil ett arf4 mutants formed deeply divided leaves or leaves bearing leaflets with ectopic meristems (Fig. 6a,  b). Moreover, leaves of fil/+ ett arf4 and of fil ett arf4 mutants had increased STM, IPT7 and BP expression ( Fig. 6c; Supplementary  Fig. 5c). The haploinsufficiency of the fil/+ ett arf4 mutant suggests that FIL and ETT or ARF4 may form a complex. Indeed, FIL was identified as an ETT interacting protein in a recent yeasttwo-hybrid screen 36 . We confirmed that FIL interacts with ETT and ARF4 by yeast-two-hybrid tests, bifluorescence molecular complementation, and co-immunoprecipitation ( Supplementary  Fig. 8). The combined data implicate a YABBY/ARF complex in direct repression of STM during reproductive primordium initiation and formation of simple leaves in Arabidopsis.
We therefore tested for a biological role of HDA19 in initiation of reproductive primordia. mp-S319 hda19 double mutants had significantly enhanced flower initiation defects relative to the parental lines (Fig. 7a, b). The naked pin inflorescence phenotype of mp-S319 hda19 was accompanied by increased expression of STM, IPT7 and BP (Fig. 7c, Supplementary Fig. 5e). Treatment with histone deacetylase inhibitors also significantly enhanced the flower initiation defect of mp-S319 ( Supplementary Fig. 9). To further probe the biological role of HDA19 in initiation of reproductive primordia, we generated double mutants between HDA19 and the SWI/SNF chromatin remodeler SYD, which acts together with MP to upregulate genes that promote flower initiation 9 . Like mp-S319 hda19, syd-5 hda19 inflorescences formed inflorescence pins and initiated significantly fewer flowers than the parental lines ( Supplementary Fig. 1). In addition, syd-5 hda19 inflorescences displayed elevated STM and ITP7 expression ( Supplementary Fig. 1).
To assess whether the histone deacetylase HDA19 binds to class I KNOX loci, we performed ChIP-qPCR using pHDA19: HDA19-GFP 9 plants. HDA19 associated strongly with multiple regions of the STM and BP loci (Fig. 7d, Supplementary Fig. 5j). If FIL and ETT/ARF4 repression of STM is mediated by histone deacetylation, the triple mutant should display increased histone acetylation at the class I KNOX gene loci. We therefore tested for an increase in H3K27 acetylation in fil ett arf4 mutants compared to parental lines. H3K27ac accumulation relative to that of H3 was elevated at both the STM and BP loci in fil ett arf4 mutants (Fig. 7e, Supplementary Fig. 5). H3K27ac/H3 accumulation correlates with STM expression levels in different plant tissues ( Supplementary Fig. 10). The combined data point to a role of FIL, ETT and ARF4 in STM and BP silencing by generating a repressive chromatin state via histone deacetylation. They further reveal that STM downregulation is important for primordium founder fate and that the 'repressive' class B ARFs ETT and ARF4 cooperate with the 'activating' class A ARF MP to silence STM together with HDA19 through a YABBY/ARF co-repressor complex (Fig. 7f).

Discussion
Organogenesis is critical to establish the body plan of higher eukaryotes. Here we investigate the initiation of reproductive primordia that give rise to flowers, plant organs critical for reproductive success. Prior studies uncovered the gene expression programs activated by MP and the mechanism for this activation in response to auxin accumulation in the founder cells of reproductive primordia 9,10 . Downregulation of expression of the pluripotency gene STM has been observed in the primordium initials, transit amplifying cells that have perceived an auxin maximum 25,26 . Using complementary gain-and loss-of-function approaches, we show here that this STM downregulation is important for primordium initiation (Fig. 3). We further show that STM downregulation is mediated by MP indirectly, via its target FIL andin a parallel pathwayby a pair of repressive ARFs, ETT and ARF4 (Figs. 1, 2, 4, 5). Finally, FIL and the class B ARFs jointly recruit the histone deacetylase HDA19 to silence STM (Fig. 7f). Removal of acetyl groups from lysine on histone H3 tails promotes chromatin compaction, a chromatin environment unfavorable for transcription 42  gene, BP, is regulated in much the same way as STM (Fig. S5). On the basis of prior studies, BP downregulation also contributes to reproductive primordium initiation 51 . STM is known to be a target of Polycomb Repressive Complex 2 (PRC2) [52][53][54] . In our genetic enhancer tests, we did not observe enhanced flower initiation defects when we combined mp-S319 with a PRC2 mutant ( Supplementary Fig. 11). Polycomb repression frequently drives long-term silencing 55 and downregulation of STM in the reproductive primordium is not permanent. Once flowers have reached stage 2 of development, STM is upregulated when the floral meristem forms 26 .
It has been proposed that class B ARFs may inhibit activity of class A ARFs by competing for binding sites at the same target genes 56 . We show here that class B and class A ARFs can act cooperatively in the same pathway by directly repressing and activating different classes of genes (this study and references 9-13 ). Evidence is accumulating for unique DNA sequence binding preferences of ETT and MP 41,57-59 . It is likely that multiple different types of interactions between the evolutionarily conserved classes of ARFs 14,60,61 enable unique transcriptional and developmental outcomes in response to a single hormonal cue, auxin. Additional specificity is provided by ARF interactions with other classes of transcription factors, such as the YABBY transcription factor FIL we identify here.
Auxin promotes FIL accumulation in incipient reproductive primordia by triggering ubiquitin-mediated degradation of Aux/ IAA proteins complexed with MP, thus enabling MP to recruit chromatin remodelers to activate FIL expression 8,9 . How auxin modulates ETT/ARF4 accumulation or activity to promote flower initiation is not clear. In a pathway parallel to MP, auxin may regulate ETT and ARF4 message abundance transcriptionally or post transcriptionally via the transacting siRNAs that target these class B ARFs 18 . Alternatively, ARF4 activity may be inhibited by Aux/IAA proteins in conditions of low auxin. In this scenario, activity of ETT, which lacks the Aux/IAA protein interacting domain, could be blocked by Aux/IAA proteins as a DNA-binding domain mediated heterodimer with other ARFs, such as ARF4 16,58 . Since ETT activity is modulated by changes in auxin concentration 36,59 , auxin might instead control ETT interaction with FIL and/or HDAC-containing co-repressor complexes.
In summary, our study implicates epigenetic silencing of a pluripotency factor in initiation of reproductive primordia, and elucidates the mechanism for this silencing. Our findings moreover suggest that in plants, like in animals 62 , organogenesis requires both activation of lineage-specific gene expression programs and silencing of pluripotency genes.
For activation of p35S:STM-GR in mp-S319, plants were treated with 1 µM dexamethasone (DEX, Sigma) plus 0.015% Silwet from 12 days of age onwards, the stage when the first reproductive primordium initiates 68 . Treatment was continued for 9 days, because new reproductive primordia initiate continuously and the synthetic steroid rapidly decays in plants 69 . For expression studies, 9-day-old MS agar-grown seedlings were sprayed with 10 µM Dex and seedlings were harvested after 4 h. Mock treatment consisted of 0.1% DMSO and 0.015% Silwet. For the cytokinin treatment, mp-S319 inflorescences were treated with 1 µM cytokinin (trans-Zeatin, Sigma) in 0.015% Silwet from 11 days of age onwards every 2 days, for a total of eight treatments. Mock treatment consisted of 0.1% DMSO and 0.015% Silwet. For deacetylase inhibitor treatment, soil-grown plants were sprayed with 5 µM of Trichostatin A (TSA, Sigma) plus 0.015% Silwet or with 50 µM of Suberoylanilide hydroxamic acid (SAHA, Sigma) plus 0.015% Silwet every 3 days from 11 days of age onwards for a total of six treatments. After manifestation of pin inflorescence phenotypes, spraying was resumed once weekly for 2 weeks. Mock treatment consisted of 0.1% DMSO plus 0.015% Silwet. For NPA treatment, fil ett arf4 inflorescences were treated with 500 nM of NPA plus 0.015% Silwet every 2 days on from 11 days of age onwards for a total of total eight treatments. Mock treatment consisted of 0.05% DMSO plus 0.015% Silwet. Because treated plants are less vigorous and make fewer flowers then untreated plants (compare WT in Fig. 1b to Fig. 3b), phenotypes of treated plants were always compared to treated control plants. The number of flowers formed was counted in 55-65-days-old plants, once development had ceased, to ensure an accurate count of all flowers or filamentous structures initiated.
Transgenic plants. p35S:STM-GR 70 was transformed into mp-S319/+. In the T1 generation, plants heterozygous for mp-S319 and hemizygous for STM-GR were selected by genotyping and growth on selective medium, respectively. In the T2 generation, seedlings were germinated on selective medium to recover plants hemi or homozygous for the transgene and transferred to soil. Plants homozygous for WT or mp were identified by genotyping. Steroid treatment and phenotypic analyses were conducted in the T2 generation on WT and mp-S319 siblings.
p35S:amiR STM 35 was transformed into mp-S319/+ ett-7/+ arf4-2. In the T1 generation, plants carrying the transgene (amirSTM) were identified by BASTA treatment and plants heterozygous for mp and ett by genotyping. In the T2 generation, T1 progeny was sprayed with BASTA to identify plants hemi or homozygous for amiRSTM. mp ett arf4 triple mutants were identified visually (they form very few leaves) and confirmed by genotyping. amiRSTM arf4 plants were identified by visual inspection and genotyping from the same T1 progeny. Phenotypic analyses were conducted in the T2 generation on arf4 and mp-S319 ett arf4 siblings. The arf4 single mutant is indistinguishable from the WT 20 .
To generate ETTr, the tasi-RNA target sites of ETT 19 were mutagenised by sitedirected mutagenesis using primer sets previously described 19 . The ETTr fragment was amplified by PCR and cloned into pENTR/D Topo (Thermo Fisher Scientific). To generate ARF4r, the full length ARF4 coding sequence was amplified and  Supplemental Table 1 for primer information) were recombined to pUC-SPV-NE GW and pUC-SPV-CE GW 71 for BiFC. pMP:MP-6xHA 9 was transformed into mp-12/ + . Lines which showed full phenotypic rescue were selected for ChIP. To construct gFIL-6xHA, the genomic locus of FIL, including the 5780 bp upstream intergenic region plus the coding region with introns was amplified as a single module and assembled together with 6xHA and tNOS terminator using the BsaI-HF enzyme (New England BioLabs) in an entry Gateway compatible vector. Following LR reaction (Invitrogen), the pFIL:FIL-6xHA-tNOS fragment was recombined into a modified Gateway-compatible version of the pPZP222 vector. To increase the cut-ligate reaction efficiency, site-specific mutagenesis was used to mutate two endogenous BsaI site in the FIL locus. The final destination vectors were transformed into in the fil_049 mutant background. Constructs used for stable plant transformation were transformed into Agrobacterium tumefaciens strain GV3101 by electroporation, followed by plant transformation by floral dip.
Phylogenetic shadowing. Genomic sequence covering from AT1G62370 to the STM 3′UTR was used as a query for finding sequences of closely related species.
The genomic sequences of Brassicaceae species were obtained by running BLASTn discontinuous megablast. The mVISTA program (http://genome.lbl.gov/vista/ index.shtml) was employed to identify regions of conservation. Clustal omega (https://www.ebi.ac.uk/Tools/msa/clustalo/) was used to identify conserved cis motifs in the conserved regions of the STM locus.
Chromatin immunoprecipitation. ChIP was performed as described in our detailed published protocol 72 . For inflorescence ChIP,~0.5 g of young inflorescences from plants with 0~5 cm bolts were harvested. All flowers from stage 6 onwards were removed. Anti-GFP (Thermo Fisher Scientific, A11122, Lot 2015992; 1:200 dilution) and Anti-HA (Roche, 12CA5, Lot 15782900; 1:40 dilution) were used for IP. Antibody quality was validated by confirming published ChIP data for LFY-GFP 73 and FIE-HA 71 . In addition, Anti-H3K27ac antibodies (Abcam, ab4729, lot GR3216173-1; 1:500 dilution), anti-H3K9ac antibodies (Active Motif, 39137, lot 9811002; 1:500 dilution) and anti-H3 antibodies (Abcam, ab1791, lot GR3198255-1, 1:500 dilution) were used for IP. The antibodies were validated by the manufacturers. For ChIP in wild-type leaves and ap1 cal inflorescences,~2 g of fully expanded Columbia leaves and 0.5 g of ap1-1 cal-1 74 inflorescence tissue was collected and fixed for analysis. Antibody amounts were 5 µg for anti GFP and HA, and 2 µg for anti H3, H3K27ac and H3K9a. The TA3 retrotransposon (At1g37110) served as negative control locus (NC) 9 . The LFY locus region e (LFYe) 10 served as positive control for MP ChIP and as a negative control for ETT ChIP. Throughout, ChIP was also performed using the same antibodies in the WT to control for nonspecific DNA enrichment. To estimate the DNA enrichment, the qPCR value of the ChIP product was normalized over input DNA. Primer sequences are listed in Supplementary Table 1.
Expression analysis. In situ hybridization of STM, CLV3, MP and ETT was performed as described in our published protocol 75 with minor modifications. The antisense probes for STM, FIL and ARF4 were generated using in vitro transcription from the full-length STM (1149 base pair), FIL (690 base pairs) and ARF4 (2367 base pairs) coding sequences cloned into pGEM-T (Promega). The full-length (516 base pair) CLV3 coding sequence and 1273 base pair of the ETT coding sequence (551-1824 base pairs) were cloned into pGEM-T easy for probe synthesis. The MP probe has been described 10 . Inflorescence tissue was harvested at 21-28 days of age for WT and ett arf4 and at 28 days of age for genotypes showing pin phenotypes (mp-S319 and mp-S319 ett arf4). Seedling tissue was grown on MS agar medium and harvested at 10 days of age. Pictures were taken of 8 μ sections using a brightfield microscope (Olympus BX51). For qRT-PCR expression analysis from 1-5 cm bolt inflorescences and fully expanded adult leaves, RNA was extracted by TRIzol (Thermo Fisher Scientific) and further purified using the RNeasy mini kit (Qiagen). First strand cDNA was synthesized using the superscript III kit (Thermo Fisher Scientific) from 1 µg of RNA. Quantitative real time PCR was performed using Power SYBR Green master mix (Thermo Fisher Scientific). Relative expression value was determined using a cDNA standard curve and normalized over that of the housekeeping gene UBQ10 (AT4G05320).
Protein interaction. For test of interaction between ETT, ARF4, FIL and HDA19 in yeast, full length clones for all four proteins, as well as ETT-N (amino acids 1-389) and ETT-C (amino acids 390-608) were cloned into pENTR/D TOPO and recombined into pDEST22 and pDEST32. These constructs were co-transformed into competent AH109 yeast cells using the Frozen-EZ Yeast Transformation II Kit (Zymo research). To examine yeast growth, serially-diluted yeast cells were spotted on media lacking Leucine and Tryptophan. To test for interactions, yeast cells were  spotted on media lacking Leucine, Tryptophan and Histidine and containing 3-amino-1,2,4-triazole (3-AT) as competitive inhibitor. 3-AT concentrations used were 1 mM and 0.1 mM. Pictures were taken after 3 days of growth. Bimolecular fluorescence complementation (BiFC) in Arabidopsis leaf protoplasts was performed as previously described 76 . Only protoplasts prepared and transfected at the same time were used for direct comparison. Three-week-old soil-grown Columbia plants were either sprayed with 10 µM of indole-3-acetic acid (IAA, Sigma) plus 0.015% silwet or with 0.01% DMSO plus 0.015% Silwet nine hours prior to protoplasting. To avoid message depletion by tasiRNAs, tasiRNA resistant versions of ETT (ETTr) and ARF4r version were employed. Protoplasts  were counted per replicate and sample. Confocal imaging was conducted using a Leica TCS SP8 confocal microscope and a 40X water immersion objective.
Scanning electron microscopy. For scanning electron microscopy, 10-day-old seedlings were fixed overnight in FAA (formaldehyde 3.7%, glacial acetic acid 5%, ethanol 50%) and dehydrated in a serial dilution of ethanol. The samples were critical point dried, mounted and gold coated. Images were taken with a Zeiss Supra 55VP Field emission scanning electron microscope using an acceleration voltage of 3 kV.
Statistical analyses and replication. Throughout, error bars shown represent the standard error of the mean for all numerical values. For frequencies, the error bars are proportional to the standard error of the pooled percentage computed using binomial distribution 9 . For qRT-PCR and ChIP-qPCR, one representative of three experiments is shown. Box and whisker plots: Lower vertical bar = sample minimum. Lower box = lower quartile. Black Line = median. Upper box = upper quartile. Upper vertical bar = sample maximum. A nonparametric Mann Whitney U test was used for phenotypic data, which frequently are not normally distributed. Unpaired t-tests were used for qRT-PCR and ChIP-qPCR data, which were normally distributed based on Kolmogorov-Smirnov tests. We employed one-tailed tests to examine differences between groups in a specific direction, and two-tailed tests to probe differences between groups in any direction. Sample size was determined by expedience and based on prior studies that found significance using similar samples sizes 9,10,71,73 .
Reporting summary. Further information on experimental design is available in the Nature Research Reporting Summary linked to this article.

Data availability
The authors declare that the data supporting the findings of this study are available within the paper and its supplementary information files. The data underlying Fig. 2c,  Fig. 6b