WEREWOLF , a Regulator of Root Hair Pattern Formation, Controls Flowering Time Through the Regulation of FT mRNA stability

A key floral activator FT integrates stimuli from long-day, vernalization, and autonomous pathways, and triggers flowering by directly regulating floral meristem identity genes. Since small amount of FT transcript is sufficient for flowering, the FT level is strictly regulated by diverse genes. In this study, we show that WEREWOLF ( WER ), a MYB transcription factor regulating root-hair pattern, is another regulator of FT . The mutant wer flowers late in long days but normal in short days, and shows a weak sensitivity to vernalization, which indicates that WER controls flowering time through the photoperiod pathway. The expression and double mutant analyses showed that WER modulates FT transcript level independent of CO and FLC . The histological analysis of WER shows that it is expressed in the epidermis of leaves where FT is not expressed. Consistently, WER regulates not transcription but stability of FT mRNA. Our results reveal a novel regulatory mechanism of FT which is non-cell autonomous.

Here, we report the late-flowering phenotype of the wer mutant which has been previously reported to have hairy roots. Delayed flowering occurred in long days but not in short days, thus wer can be classified into the photoperiod pathway mutant. The transcript level of FT was reduced in wer mutant in long days, which was independent of the CO and FLC. In addition, such decrease of FT transcript level is due not to altered transcription but reduced stability of mRNA. This study suggests that WER in epidermis modulates FT transcript level in phloem through a novel mechanism.

WER Regulates Flowering Time through Photoperiod Pathway
While exploring if WER acts in other developmental process, we observed the flowering time of wer loss-of-function mutant and WER overexpression transgenic plants is changed. The wer-1 allele in Columbia (Col) background and wer-3 allele in Wassilewskija(Ws) background have nonsense mutations in the region of the second MYB domain, thus both alleles apparently produce nonfunctional proteins (Lee and Schiefelbein, 1999). Under long-day (16 h light/8 h dark) conditions, both wer-1 and wer-3 plants produced more rosette leaves than wild type at the time of bolting (Fig.   1A). In contrast, the transgenic plants containing WER genomic DNA under the control of the strong cauliflower mosaic virus 35S promoter (35S::WER) flowered earlier than wild type (Fig. 1A). To verify that the late-flowering phenotype of wer-1 was caused by loss of WER function, the WER genomic construct containing 5 kb WER genomic fragment that includes 2.5 kb upstream sequence was transformed into wer-1. Most of the resulting transformants showed comparable flowering time to wild type, indicating that late flowering is caused by the loss of WER (Fig. 1B). In addition, the heterozygous lines obtained from the cross between Col and wer-1 showed similar flowering time with Col, confirming that wer is recessive late-flowering mutant (Supplemental Fig. S1).
The flowering time in Arabidopsis is regulated by four major pathways, photoperiod, vernalization, autonomous and GA pathways (Boss et al., 2004;Baurle and Dean, 2006;Oh and Lee, 2007). To determine in which pathway WER regulates flowering, the flowering characteristics of wer were checked in response to photoperiod and vernalization. Unlike in long days, wer-1 flowered similarly to wild type in short days ( Fig. 1C). After 8 weeks of vernalization treatment, all genotypes showed acceleration of flowering compared to non-vernalization treated control. Although wer-1 showed slight acceleration of flowering, the responsiveness was weaker than wild type (Fig. 1D).
These observations demonstrated that loss-of-function in WER results in a delay of flowering only under long days and weak sensitivity to vernalization, which is similar flowering characteristics to photoperiod pathway mutants as co and gi (Koornneef et al., 1991

WER Expression in Root Does Not Affect Regulation of Flowering Time
Floral evocation occurs in shoot apex by inducing floral initiation genes such as AP1and LFY (Weigel and Nilsson, 1995;Hempel et al., 1997). However, it was previously reported that WER is expressed mainly in roots (Lee and Schiefelbein, 1999).
To determine whether a root-derived signal induced by WER affects flowering time, graft chimeras among wer-1, 35S::WER and the wild type were produced by a transverse cut grafting method described before (Turnbull et al., 2002). All grafts were denoted as scion/rootstock genotypes. Self-grafted plants, in which the scion and rootstock were from the same genotype, were also produced as controls. These plants appeared to flower slightly earlier than ungrafted plants probably due to mechanical stress (Fig. 2). The results showed that grafting failed to rescue the late flowering of wer-1scions whichever Col or 35S::WER was used as rootstocks. All grafted plants showed similar flowering time as the plants used as scion (Fig. 2). These results suggest that WER expressed in root does not affect flowering time.

Expression of WER
To examine the tissue expression pattern of WER, RT-PCR analysis was performed with total RNA extracted from various tissues. Although WER was highly expressed in roots as previously reported (Lee and Schiefelbein, 1999), transcripts were also detectable in young leaves, shoot apices, adult rosette leaves, stems, and inflorescences including floral buds and mature flowers (Fig. 3A).
To inspect the spatial pattern of WER expression, WERp::GUS transgenic plants, in which GUS reporter is driven by the WER promoter with a 4 kb DNA fragment upstream of the WER coding sequence, were used for histochemical GUS staining. This transgenic line was used before for the spatial expression analysis because GUS staining faithfully followed the endogenous expression in roots (Lee and Schiefelbein, 1999). In 6-day-old seedlings, GUS expression was most notable in the hypocotyl and the shoot apex as well as in the root tip, whereas weak GUS expression was detected in the margin of the cotyledons ( Fig. 3B; a). GUS was also detected in petiole, stem, stigma, and siliques ( Fig. 3B; b-g). In shoot and root apices, GUS was detected from embryo stage ( Fig. 3B; h). Interestingly, GUS expression was gradually disappeared while leaves get mature ( Fig. 3B; b  longitudinal sectioning of WERp::GUS revealed that WER is concentrated in the epidermis of leaf primordia ( Fig. 3B; i and j). Immunohistochemistry using WERp::MYC-WER transgenic plants, in which WER protein fused with MYC epitope is driven by the WER promoter, was shown that WER proteins are also expressed along the epidermis ( Fig. 3B; k).
The temporal changes in WER transcript level were further determined by RT-PCR to investigate how WER expression is regulated during the flowering process. WER expression in the aerial part of seedlings peaked in 3 days; afterwards it was reduced and became steady under both long days and short days (Fig. 3C). Such temporal expression was not affected by photoperiod. Moreover, WER expression did not show any daily rhythm under long days (Fig. 3D).
When vernalization effect on WER expression was checked in Col:FRI SF2 plants, which has strong flowering response to vernalization, there was no change (Fig. 3E). In addition, GA treatment did not change the expression level (Fig. 3F). These results indicate that the vernalization or GA-dependent pathway does not affect WER expression.

WER Regulates FT Transcript Level Independent of CO and FLC
WER expression was further studied in various flowering-time mutants to elucidate the possible involvement of WER in the previously known flowering pathways.
Although wer showed the flowering characteristics similar to the mutants of the long day pathway, WER expression was unchanged in these mutants as ft-1, co-101, and gi-2 ( Fig. 4A). In addition, the expression was not affected by the mutations in autonomous pathway or flc, indicating that WER does not act downstream of the previously known flowering pathways.
To examine the molecular basis of the late flowering of wer, we checked the expression of the genes in long day pathway or genes acting on the shoot meristem as CO, FT, FD, and SOC1 (Fig. 4B). The real-time quantitative RT-PCR (qRT-PCR) showed that only FT expression was reduced in wer mutants under long days whereas the levels of CO, FD, and SOC1 were similar to wild type. In short days, there was no difference between wild type and wer in the transcript level of all four genes. Such under long days. It also strongly suggests that WER regulates FT independent of CO.
The double mutant analyses also support this hypothesis such that 35S::FT completely suppresses but 35S::CO partially suppresses the late-flowering phenotype of wer-1 (Fig. 4,C and D). Double mutant analysis also shows that flc and wer are additive (Fig. 4F), suggesting that WER acts independent of FLC. It is noteworthy that the molecular basis of the early flowering of 35S::WER was a little complicated because ectopic overexpression of WER caused increase of both FT and SOC1 (Fig. 4B). Thus, neither ft nor soc1 mutation was epistatic to 35S::WER (Fig. 4, C and E).
Because FT is known to be regulated by the circadian clock, we wondered whether the daily rhythm of FT is affected by wer. To address this, FT expression was analyzed every 4 h over a 24-h period under long days by RT-PCR. The result showed that the daily rhythm of FT expression is not affected by wer although the amplitude is lower than wild type (Fig. 4,G and H).

wer Mutation Affects the FT mRNA Stability
Although WER is expressed in the epidermis ( Fig. 3B; i and j), FT is expressed in the vasculature (Takada and Goto, 2003). Thus, it is not likely that WER directly regulates FT. However, it is still possible that WER acts non-cell autonomously through mRNA or protein transport. To check if wer mutation affects transcription of FT, the expression of GUS reporter gene driven by FT promoter in wer-1 was analyzed (Fig. 5A). The result showed that GUS expression level in wer-1 was similar to that in Col, whereas endogenous FT level in wer-1 was less than that in Col. Since initiation of transcription is achieved by binding of RNA polymerase II to the promoter DNA in eukaryotes, we checked if wer mutation affects the binding capacity of RNA polymerase II to the proximal region of the FT promoter by chromatin immunoprecipitation (ChIP) assay. In wer-1 mutants, the enrichment of RNA polymerase II at the FT promoter was similar to that in Col, demonstrating that wer mutation does not affects FT transcription (Fig. 5B).
Unlike at the FT promoter, RNA polymerase II enrichment was reduced by wer-1 approximately 40% at promoters of GL2 and AP1, which are direct target of WER and a downstream gene of FT, respectively. It suggests that the transcription rate of these two genes is reduced in wer-1. Taken together, these results strongly suggest that FT transcript level is reduced in wer-1 by post-transcriptional regulation such as mRNA

decay.
To directly assess whether reduced FT mRNA level in wer-1 was a result of altered mRNA stability, the half-life of FT mRNA was compared between wild type and wer-1 after actinomycin D, a transcription inhibitor, was treated. Total RNA of 7-day-old seedlings grown under continuous light was isolated after incubation with actinomycin D for 0, 2, 4, or 8 hours. After transcriptional block with the treatment of actinomycin D, FT mRNA abundance in wer-1was more rapidly reduced than that in wild type (Fig. 5C).
However, the half-life of control mRNA, TUB2, was not different in wild type and wer-1, suggesting that the wer mutation does not cause a general RNA instability. Therefore, our results suggest that WER regulates the stability of FT mRNA.

Mutation Affects the FT mRNA Stability
For the root hair pattern formation, WER forms a transcriptional protein complex with TTG1 and GL3/EGL3 and positively regulates the transcription of GL2 and CPC (Lee and Schiefelbein, 2002;Schellmann et al., 2002;Wada et al., 2002;Zhang et al., 2003;Koshino-Kimura et al., 2005;Ryu et al., 2005). We wondered if the genetic tool kit regulating root-hair pattern is also involved in the determination of flowering time.
Interestingly, ttg1-13 and gl3-2 egl3-1, mutants of components in the WER protein complex, flowered as late as wer mutants (Fig. 6A), suggesting that WER complex regulates flowering time also. In contrast, gl2 and cpc showed similar flowering-time to wild-type (Fig. 6B)

DISCUSSION
Myriad of genes are reported to regulate transcript level of FT since FT turns out to be a 'florigen' as well as a key integrator of flowering signals. However, most of the case is not clear if it is regulated at transcriptional level or posttranscriptional level. In this study, we clearly show that wer, producing ectopic root hairs, is a typical photoperiod pathway mutant and WER regulates FT mRNA stability by non-cell autonomous way. Therefore, our study provides a novel mechanism regulating FT transcript level.

WER Regulates FT Non-Cell Autonomously
Although WER expression was mostly detected in root, grafting analysis showed that flowering time was not affected by WER in root (Fig.2). This result indicates that WER expressed in aerial parts are involved in the regulation of flowering time. Consistent with this, WER expression was observed in diverse ranges of aerial parts as young leaves, stems, flowers, and siliques (Fig. 3, A and B). In addition, it is expressed in young, developing leaves where FT is expressed ( Fig. 3B; Takada and Goto, 2003). The leaf consists of three distinct tissues as mesophyll, vascular bundle, and epidermis.
Histological analysis showed that WER is expressed in leaf epidermis whereas FT is expressed in vascular bundles (Fig. 3B;Takada and Goto, 2003). Therefore, WER is most likely to regulate FT non-cell autonomously. Alternatively, WER may be transported to vascular bundles for the regulation of FT. However, it is not likely because it produces protein complex as discussed below and our preliminary result showed that WER protein is located in the leaf epidermis.
It is noteworthy that root hair pattern formation shares the same genetic tool kit with trichome formation in the leaves (Schiefelbein, 2003). The many glabrous mutants with no trichome have defects in root hair formation, too (Masucci et al., 1996;Payne et al., 2000;Ohashi et al., 2002;Bernhardt et al., 2003;Zhang et al., 2003). In addition, such genes are expressed in both root and leaf epidermis. Therefore, it provides good evidence implicating that trichome and root hair are evolutionarily homologous organs as suggested before (Kellogg, 2001 regulates root hair pattern formation and flowering in roots and leaves, respectively, the downstream factors involved in each process are different (Fig. 6). Thus, the divergence occurs at the downstream target genes. Future analysis searching the factors mediating signals between epidermis and vascular bundles for the FT regulation would be interesting.
Non-cell autonomous regulation of FT is not unprecedented. It has been reported that phytochrome B located in mesophyll suppresses FT expression through a downstream gene PHYTOCHROME AND FLOWERING TIME 1 (PFT1) (Cerdan and Chory, 2003;Endo et al., 2005). Interestingly, the PFT1 mediating phyB signaling also regulates FT independent of CO, similar to WER. It may indicate that non-cell autonomous regulation of FT in the leaf is a common process. Therefore, it is possible to identify the inter-tissue signals regulating FT which is critical to understand the florigen entity.

WER Regulates FT Posttranscriptionally
CO is known to directly regulate FT transcription by binding to the promoter (Samach et al., 2000;Tiwari et al., 2010). Although CO protein does not have conspicuous domain for transcription factor, many compelling evidences support that it plays as transcriptional coactivator (Samach et al., 2000;Hepworth et al., 2002;Wenkel et al., 2006). In addition, FT is transcriptionally regulated by FLC, a central flowering repressor. FLC protein binds directly to the first intron of FT to prevent the induction of FT transcription (Helliwell et al., 2006). Therefore, transcriptional regulation of FT is relatively well studied but posttranscriptional regulation is poorly studied. Here, we revealed that WER positively regulates FT by controlling mRNA stability at the posttranscriptional level (Fig. 5). It implicates that a FT mRNA decay pathway is involved in the regulation of flowering time. Transcript abundance is determined by the equilibrium between the rate of mRNA synthesis and the rate of degradation; yet in majority of gene expression analysis, mRNA decay process has not been considered seriously. However, recent advances provide some knowledge about mRNA decay pathways such that the mRNAs involved in the regulatory processes have shorter halflives than those involved in metabolic pathway in Arabidopsis (Gutierrez et al., 2002; Belostotsky and Sieburth, 2009). As was in yeast and human studies, it implies that rapid mRNA decay process is required for strict regulation of developmental process. This is consistent with the result of which FT transcript was less stable than TUB2 in this study (Fig.5C). Because FT protein is considered as florigen, the FT protein level is directly linked to flowering. Therefore, it is probable that FT transcripts must be carefully monitored to produce FT protein inappropriate amount. The regulation of FT mRNA stability proposed in this study may provide a new mechanism to control FT transcripts. Since WER encodes a transcription factor, WER would not be directly involved in the regulation of FT mRNA stability. Thus, it is likely that WER activates stabilizing factors or inhibits destabilizing factors for FT mRNA. The genetic components regulating FT mRNA stability will be pursued.

Plant Materials and Genotyping
The wer-1 in Col and wer-3 in Ws background were used (Lee and Schiefelbein, 1999). The 35S::WER is a transgenic line with WER genomic DNA (inserting from the start to the stop codon) driven by CaMV35S promoter in wer-1 mutants (Lee and Schiefelbein, 1999). To confirm that genomic WER rescues flowering phenotype of wer-1, 5 kb genomic fragment including 2.5 kb of the upstream sequence of WER was cloned into the binary vector pPZP221 and transformed into wer-1. The WERp::GUS transcriptional reporter construct was previously reported (Lee and Schiefelbein, 1999). fca-9, fve-3, soc1-2, and ld-1 were in the Col background as described before (Lee et al., 1994;Fowler et al., 1999;Kardailsky et al., 1999;Kobayashi et al., 1999;Page et al., 1999;Takada and Goto, 2003;Ausin et al., 2004;Lee et al., 2006). The flc-3 is originally the line obtained from fast neutron mutagenesis of Col:FRI SF2 , a Columbia line with FRIGIDA (FRI) from San Feliu-2 (SF2) by eight times of backcross (Michaels and Amasino, 1999;Lee et al., 2000). However, the FRI SF2 allele has been eliminated from the flc-3 by backcrossing several times into Col. The FTp::GUS Collinecontaining8.9kb upstream sequence from the start codon fused to GUS protein was used (Takada and Goto, 2003). The FTp::GUS wer-1 was obtained from the cross between FTp::GUS Col and wer-1, and kanamycin-resistant F2 seedlings were genotyped with CAPS markers. To check the genotype of double mutants, the F2 plants were checked using polymerase chain reaction (PCR)-based markers, SSLP markers, and CAPS and dCAPS markers, of which information is described in Supplemental   Table S1.

Growth Conditions
Seeds were sterilized by 75% ethanol with 0.05% Triton X-100, then rinsed twice

Grafting
The transverse cut grafting was performed as a method previously described (Turnbull et al., 2002). Grafting experiment was accomplished under a microscope using 5-day-old seedlings grown on sucrose-free media. Horizontal cuts were made in the upper region of the hypocotyl with small blades (Dorco T-300). For 5 days after grafting, grafts were monitored whether they formed good union without bending or any other growth problems. All successful grafts were transplanted to soil. The final proportion of successful grafts which grow normally until flowering was over 70%. At least 10 plants were used to measure the flowering time of each graft.

Analysis of Gene Expression
Total RNA extraction and RT-PCR were performed as described before (Lee et al., 2008). The RT-PCR analysis was repeated at least three times using separately harvested samples. The information of each primer for PCR is described in Supplemental Table   S2. For Semi-Quantitative RT-PCR analysis, RT-PCR products were analyzed using the each primer for PCR are described in Supplemental Table S3. Reaction conditions were as follows: 5 min at 95ºC , 40 cycles of PCR (30 sec at 95ºC , 30 sec at 60ºC , 30 sec at 72ºC), and a dissociation from 60ºC to 95ºC . Data was collected at 72ºC in each cycle, and TUB2 was used as the reference gene. Exceptionally, half-life of mRNA was referenced by 18S rRNA levels. The qRT-PCR analysis was biologically repeated three times, each consists of three technical replicates.

Plasmid construction
To make the chimeric genes of MYC-tagged WER, the DNA fragment for MYC epitope was inserted in frame into the 5' end of the PCR-amplified coding region of WER genomic DNA. The insertion of this chimeric MYC-WER between a 2.4 kb 5' flanking region DNA fragment and a 1.1 kb 3' flanking region DNA fragment from the WER gene resulted to generate the WERp::MYC-WER construct.

GUS Staining and Histological Analysis
GUS staining and histological analysis were performed following standard methods described before (Choi et al., 2007). Embedded samples in paraffin were sectioned at a thickness of 8 µm with a microtome (Leica, RM2135). Photographs were taken by the digital-microscope (Dimis M) or the digital camera (Photometrics) connected to a microscope (Zeiss, Axioskop 2 Plus).

Chromatin Immunoprecipitation (ChIP) Assay
The 1 g of Col and wer-1 seedlings grown under long days for 11 days was used for ChIP. Procedures for ChIP were followed the method described before (Lee et al., 2007;Lee et al., 2008), and antibody for the C terminal domain of the RNA polymerase II (Abcam; AB817) was used. Four microliter of ChIP products resuspended in 100μl of TE was used for real-time quantitative PCR (qPCR). In qPCR analysis, expression levels were normalized against the expression in Col. The information of the primer pairs for ChIP-qPCR is presented in Supplemental Table S4.

Accession Numbers
Sequence data from this article can be found in the GenBank EMBL/GenBank data  Genotype notation for stock is shown below X-axis and that of scion is above.