The bZIP Transcription Factor PERIANTHIA: A Multifunctional Hub for Meristem Control

As sessile organisms, plants are exposed to extreme variations in environmental conditions over the course of their lives. Since plants grow and initiate new organs continuously, they have to modulate the underlying developmental program accordingly to cope with this challenge. At the heart of this extraordinary developmental plasticity are pluripotent stem cells, which are maintained during the entire life-cycle of the plant and that are embedded within dynamic stem cell niches. While the complex regulatory principles of plant stem cell control under artificial constant growth conditions begin to emerge, virtually nothing is known about how this circuit adapts to variations in the environment. In addition to the local feedback system constituted by the homeodomain transcription factor WUSCHEL (WUS) and the CLAVATA signaling cascade in the center of the shoot apical meristem (SAM), the bZIP transcription factor PERIANTHIA (PAN) not only has a broader expression domain in SAM and flowers, but also carries out more diverse functions in meristem maintenance: pan mutants show alterations in environmental response, shoot meristem size, floral organ number, and exhibit severe defects in termination of floral stem cells in an environment dependent fashion. Genetic and genomic analyses indicate that PAN interacts with a plethora of developmental pathways including light, plant hormone, and meristem control systems, suggesting that PAN is as an important regulatory node in the network of plant stem cell control.


INTRODUCTION
In contrast to most animals, plants continue to form new organs throughout their lives. This remarkable capacity is dependent on the continuous presence of undifferentiated and self-renewing stem cells over long periods of time. These stem cells reside at the growing points of a plant, the tips of roots and shoots, and are embedded into specialized structures called meristems (Barton, 2010).
Several genes affecting meristem and stem cell function have been identified by mutant screens in Arabidopsis thaliana. Most notably WUSCHEL (WUS) and SHOOTMERISTEMLESS (STM ) are required for the maintenance of the shoot meristem (Barton and Poethig, 1993;Laux et al., 1996;Long et al., 1996;Mayer et al., 1998). Their inactivation causes premature differentiation and the eventual exhaustion of the stem cell pool, leading to the termination of the shoot meristem. Another group of genes, the CLAVATA (CLV ) genes, have an opposite effect on meristems and if defective, shoot meristems overproliferate and expand inappropriately (Clark et al., 1993(Clark et al., , 1995Kayes and Clark, 1998).
With the exception of CLV2, all genes mentioned above are expressed in small domains in the shoot apical meristem (SAM). Elegant genetic studies have shown that WUS and CLV3 are connected by a negative feedback loop to control the size of the stem cell pool. WUS, which is expressed in the organizing center, induces the expression of CLV3 in the overlying true stem cells, which in turn signals back to the organizing center to keep WUS expression in check (Brand et al., 2000;Schoof et al., 2000). In addition to these local regulatory interactions, meristem function is affected by global hormone signaling pathways, including auxin and cytokinin circuitries. While STM mediates cytokinin biosynthesis (Jasinski et al., 2005;Yanai et al., 2005) to allow cell proliferation in the meristem, its expression is repressed by auxin (Furutani et al., 2004), which in turn allows organ initiation on the flanks of the SAM. In contrast, WUS does not interfere with cytokinin biosynthesis, but directly regulates Atype ARABIDOPSIS RESPONSE REGULATORS (ARRs; Leibfried et al., 2005;Busch et al., 2010) that act in the negative feedback regulation of cytokinin response . This feedback system of cytokinin signal transduction is also connected to auxin signaling and ARR7 and ARR15 are directly repressed by the AUXIN RESPONSE FACTOR5/MONOPTEROS transcription factor . A-type ARRs execute important meristematic functions (Leibfried et al., 2005;Buechel et al., 2010;Zhao et al., 2010) by so far undiscovered mechanisms (Leibfried et al., 2005;Zhao et al., 2010).
Cells that leave the shoot meristem during the initial, vegetative phase of the life-cycle give rise to leaves and meristems of axillary shoots. After the transition to the reproductive phase, meristems that newly arise at the flanks of the SAM will develop into flowers instead. This is due to the redundant activity of meristem identity genes such as LEAFY (LFY ) and APETALA1 (AP1). In contrast to the shoot apex, which is indeterminate, flowers are determinate and stem cell activity ceases after a fixed number of organs have been formed. In plants that lack LFY activity, flowers are converted into partially indeterminate shoot-like structures (Weigel et al., 1992).
One set of genes that is directly controlled by the LFY transcription factor includes homeotic genes that specify the fate of the different floral organs (Parcy et al., 1998;Busch et al., 1999). We have previously shown that LFY acts together with WUS, which also encodes a transcription factor, to contribute to the transcriptional activation of the homeotic gene AGAMOUS (AG) in the center of young flowers. AG in turn, not only specifies the fate of the floral reproductive organs, but also terminates stem cell maintenance by negative feedback on WUS expression (Lohmann et al., 2001). The bZIP transcription factor PERIANTHIA (PAN) is expressed in the SAM, as well as in developing flowers, where it overlaps with STM, WUS, the CLV transcripts, and AG, respectively (Chuang et al., 1999). Loss-of PAN function leads to an increase in the number of perianth organs, the sepals and petals, while on a gross morphological level the SAM seems unaffected (Running and Meyerowitz, 1996). In flowers, PAN genetically interacts with ABC homeotic genes, however these interactions appear mostly additive (Running and Meyerowitz, 1996). PAN protein expression was shown to be independent of the meristematic regulators CLV1 and CLV3 as well as of floral meristem identity genes, such as LFY or AP1, demonstrating that PAN also acts in parallel to these factors (Chuang et al., 1999). It has been shown that PAN interacts with the NPR1-like proteins BLADE ON PETI-OLE 1 (BOP1) and BOP2 in yeast and that bop mutants share some of pan mutant features (Hepworth et al., 2005). However, their expression domains only overlap marginally, suggesting that PAN primarily acts together with other co-factors. It was shown that PAN plays important roles in the activation of AG (Das et al., 2009;Maier et al., 2009), which are strikingly modified in various day-length settings. While PAN brings about the termination of floral stem cell fate by the direct transcriptional activation of AG, its function in the SAM, where it is also strongly and specifically expressed, remains poorly understood.

RESULTS AND DISCUSSION
Since we had noted before that the floral functions of PAN are strongly dependent on the environment (Maier et al., 2009), we carefully analyzed vegetative phenotypes of wild-type Columbia and pan mutant plants under various growth conditions and found that day-length had a substantial impact on the penetrance of pan related defects. In contrast to the reproductive phase, where pan mutants showed the most dramatic aberrations under short-day conditions, pan plants at the early vegetative stage were largely undistinguishable from wild-type in short days (SD; Figures 1A,D). Conversely, pan mutants exhibited pleiotropic phenotypes when exposed to long days (LD), including elongated petioles, curled leaves, and a twisted rosette (Figures 1B,E). Under continuous light (CL), Col and pan phenotypes were less distinct, but pan plants continued to show more extreme leaf-curling and rosette twisting. In addition to the morphological traits, we observed that pan mutants flowered slightly early and on average formed 1.5 or 2.5 rosette leaves less than wild-type under LD or CL, respectively ( Figure 3A; n = 50). Furthermore, we realized that pan mutants are extremely sensitive to variations in diverse environmental conditions, including water and nutrient availability, as well as biotic and abiotic stress (data not shown). Taken    together, these phenotypes indicated that PAN might act to stabilize the developmental program of the shoot apex and thus buffers the impact of diverse environmental inputs.
Since the activity of the SAM is mainly determined by the WUS-CLV feedback system, which acts on the stem cell population, as well as the repression of differentiation throughout the meristem provided by STM, we investigated their regulatory and genetic interaction with PAN. Using in situ hybridization on serial histological sections, we first analyzed in detail the mRNAexpression patterns of PAN in the inflorescence meristem and found that, consistent with a buffering function, PAN mRNA is most highly expressed in a ring-shaped domain surrounding the stem cells (Figures 2A-D). We detected weaker signals throughout the center of the SAM, suggesting that PAN might execute slightly different functions depending on expression levels. Similar to the situation identified for WUS, which was shown to bind to distinct cis-regulatory motifs with different affinity (Busch et al., 2010), these functions could be mediated by distinct sets of PAN downstream targets. However, in situ detection of PAN protein on sections of the SAM did not show the ring domain, but rather suggested that PAN is found throughout the meristem (Chuang et al., 1999). Unfortunately, we were unable to resolve whether these Frontiers in Plant Science | Plant Physiology    differences were of technical nature, or reflected relevant biology.
Hybridizations on cross sections demonstrated that PAN mRNA is strongly reduced even in early organ primordia (Figures 2E-H). We next investigated how the SAM regulatory system is affected by the loss-of PAN function. First, we noticed that the SAM was significantly increased in size (Figures 2I,M) and that the WUS expression domain is substantially wider compared to the wildtype situation (Figures 2J,N). Interestingly, the stem cell domain marked by CLV3 expression remained largely unaffected despite the expanded stem cell niche (Figures 2K,O), suggesting that the regulatory interaction between WUS and CLV3 is partially uncoupled in pan mutants. In line with the enlarged meristem, we found expanded STM expression in pan apices (Figures 2L,P) and the absence of STM transcripts from emerging organ primordia was less pronounced in pan when compared to wild-type. Taken together, these results demonstrate that PAN function is required for normal SAM development, which might be mediated by its effects on the expression of the canonical meristem regulators.
To address how PAN is integrated into the regulatory network of the SAM, we analyzed its expression in wus and clv3 mutants, which represent the extremes in meristem dis-regulation. Since wus mutants rarely form inflorescence meristems, we focused our analysis on the seedling stage and found accumulation of PAN mRNA mostly in the center of the SAM in wild-type. In addition, we detected weaker signals on the periphery of the meristem and at the adaxial sides of young leaves ( Figure 2Q). Consistent with the loss-of a fully developed SAM in wus, we were unable to detect PAN transcripts in central tissue of this mutant, however, strong expression was found in leaf-primordia and young leaves ( Figure 2R). While Chuang et al. (1999) had reported that PAN protein expression is mostly independent of CLV3, we observed that PAN transcripts accumulated throughout the SAM, with a ring of strong expression toward the base with weaker signals toward the top of the expanded clv3 meristem (Figures 2S,T).
Having shown that PAN is more tightly connected to the regulatory system of the SAM than previously anticipated, we extended our analysis to test the functional interaction of PAN with CLV3, WUS, and STM using genetics. Plants that carry mutations in CLV3 are characterized by an enlarged SAM, an increase in the number of lateral organs developing from the SAM and over-proliferation of floral meristems. When we combined the clv3-7 loss-of-function allele with pan, we observed a substantial enhancement of the clv3 phenotype ( Figure 3A). Compared to clv3 single mutants, SAMs of pan clv3 double mutants were even further enlarged (arrowheads in Figures 3C,D) and developed even more lateral organs (Figures 3C,D). Consistent with an enhancement of meristem phenotypes by the pan mutation, we observed a drastic reduction SAM function when we combined wus and pan ( Figure 3F). In contrast to wus mutants, which develop a bushy stature because of the stop and go phenotype of the meristem (Laux et al., 1996), stem cell activity in wus pan double mutants ceased after the formation of leaves and elongated shoots were never formed. Since CLV3 and WUS act in the same pathway and both showed synergistic genetic interactions with PAN, we next wondered how PAN would interact with STM, whose activity is independent of the WUS-CLV system. To our surprise we found that the stm phenotype was partially suppressed in pan stm double mutants, which developed a substantially larger number of lateral organs and shoots compared to stm plants ( Figure 3G). In some cases we even observed flowers with a regular arrangement of floral organs, however these flowers remained sterile. Thus, while in the case of WUS and CLV3 PAN behaved as a molecular buffer, which is able to stabilize SAM function in the absence of other meristem regulators, this function was not observed when pan was combined with stm, suggesting that they have antagonistic activities.
To elucidate some of the mechanisms that could underlie these complex meristematic functions of PAN, we recorded the molecular phenotype of pan single mutants by transcript profiling. Wild-type and pan mutants were grown in LD for 25 days before we sampled two independent pools of 50 inflorescence meristems of each genotype by removing developing flowers older than stage 8. After Affymetrix Ath1 profiling we applied GC-RMA to normalize the data and derive expression values (Wu et al., 2004) followed by Rank Products to identify differentially expressed genes at a false discovery rate of 0.05 (Breitling et al., 2004). One hundred sixty transcripts showed increased abundance (Table 1), while 120 mRNAs were found to be significantly reduced in inflorescence apices of pan mutants compared to wild-type ( Table 2). To obtain a first insight into the potential function of PAN downstream genes we used Gene Ontology (GO) analysis on the level of the annotation of biological function, as well as using molecular function as a readout. Interestingly, we found the "response to stimulus" category as highly enriched among the genes with increased as well as reduced expression. Among the increased Frontiers in Plant Science | Plant Physiology  mRNAs we found diverse functional sub-categories indicating that PAN plays a role in stress and environmental response (Figure 4). A prominent example was GIGANTEA (GI ), whose expression is controlled by the circadian clock and whose activity is necessary for normal clock function and promotion of flowering under LD (Fowler et al., 1999;Park et al., 1999). To test if GI plays a relevant role as PAN downstream gene, we created pan gi double mutants and compared them to the respective parental genotypes. Strikingly, we found that loss-of PAN function was able to fully suppress the late flowering phenotype of gi mutants in LD (Figure 5), demonstrating that GI and PAN act in the same pathway.
In contrast to the rather diverse GO categories observed in the list of genes with increased expression, the reduced transcripts revealed a much more specific developmental signature. Among them we identified a substantial overrepresentation of genes with annotated functions in hormone signaling, specifically for gibberellin, ethylene, auxin and, most prominently, cytokinin response (Figure 6). This developmental signature was also apparent in the GO analysis for molecular functions with "transcription regulator activity" and "two-component response regulator activity"as the most overrepresented annotation terms (Figure 7). Twocomponent response regulators build the backbone of cytokinin signal transduction and response, with B-type ARRs acting as cytokinin dependent transcription factors directly upstream of Atype ARRs as immediate early cytokinin response genes with roles in negative feedback regulation (Werner and Schmülling, 2009). Strikingly, only the expression of A-type ARRs was affected in pan mutants and ARR4, ARR5, ARR6, ARR7, ARR15, and ARR16, were among the transcripts with significantly reduced abundance, a result which we independently confirmed using quantitative realtime RT-PCR (data not shown). In addition to cytokinin response genes, we identified two cytokinin oxidases, CKX3 and CKX5, as genes with reduced expression. Since CKX proteins irreversibly degrade cytokinin (Mok and Mok, 2001;Werner et al., 2003) and because A-type ARRs counteract cytokinin signaling , a reduction of their expression in pan mutants suggests that PAN acts to limit cytokinin activity in the SAM. This interpretation is consistent with the finding that SAM size is increased in pan mutants reminiscent of plants with increased cytokinin levels (Bartrina et al., 2011). In addition, we had previously identified  ARR5, ARR6, ARR7, and ARR15 as direct transcriptional targets of WUS, connecting these cytokinin response genes to the core regulatory system of the SAM. While from the list of genes with reduced expression an antagonistic interaction of PAN and cytokinin could be deduced, it also suggested that PAN acts to stimulate auxin signaling, since it contained YUCCA1 and YUCCA4, two genes coding for important auxin biosynthesis enzymes (Zhao et al., 2001). Since auxin directly represses transcription of ARR7 and ARR15 via the Auxin Response Factor MONOPTEROS  in the SAM, PAN could act on the expression of A-type ARRs in multiple independent pathways. Strikingly, WUS was identified among the transcriptional regulators with reduced expression, confirming that PAN is intimately connected to the SAM regulatory network.
Having identified cytokinin and auxin signaling as major downstream effector pathways of PAN we next addressed the functional relevance of these regulatory interactions using genetics. We focused our analysis on ARR7 and ARR15, since both of them were shown to have important meristematic functions (Leibfried et al., 2005;Zhao et al., 2010), and combined these mutants (Figures 8D,E) with pan ( Figure 8B) and clv3 (Figure 8C) in double and triple mutant combinations. While single A-type arr mutants have no phenotypes or very mild ones (Figures 8D,E; To et al., 2004), combination of arr7 and arr15 with pan lead to severe growth retardation (Figures 8G,H). Interestingly, while removing CLV3 function in the pan background lead to massive over-proliferation and meristem expansion beyond the regular clv3 defect (Figures 3B-E), this phenotype was completely suppressed in the pan clv3 arr7 combination (Figures 8F-I). However, the growth retardation was only transient and pan arr15 as well as pan arr15 clv3 plants recovered after about 2 weeks and developed plants with pentameric flowers, which closely resembled pan clv3 mutants. This capacity to overcome A-type ARR related Frontiers in Plant Science | Plant Physiology developmental defects was also observed in plants carrying an over-activated form of ARR7 (Leibfried et al., 2005) and suggest that the cytokinin signaling system has a strong ability to adapt to perturbations. Mutation of multiple A-type ARRs, such as in an arr7 arr15 double mutant did not cause the phenotypes observed in the pan arr combinations (Figure 8J) underlining the important role of PAN in the SAM. Having observed a strong genetic interaction of PAN with components of the cytokinin response, we next tested its ability to modify auxin related defects. To this end we analyzed the interaction of PAN with PINFORMED-1 (PIN1), the major auxin efflux carrier responsible for generating local auxin maxima at the periphery of the SAM and thus organ initiation during shoot development (Gälweiler et al., 1998;Reinhardt et al., 2000). While pin1 mutants rarely developed flowers under our growth conditions (Figures 9A,C), pin1 pan double mutants exhibited a significantly increased number of flowers ( Figures 9B,C), which were deformed and generally sterile. Again, as in the case of cytokinin signaling, these results demonstrated that PAN is able to modulate auxin dependent developmental functions, in line with the hypothesis that PAN might act as a multifunctional hub for diverse meristematic functions.

SUMMARY AND OUTLOOK
Taken together, we have shown here by molecular phenotyping and genetics that PAN is connected to a plethora of diverse input pathways and may act as an integrator to buffer shoot meristem activity. PAN inputs include pathways for environmental sensing, such as day-length and other abiotic factors, as well as hard-wired developmental circuitries, such as the WUS-CLV system. Strikingly, the same holds true for the PAN output network, which we found to include components of the circadian clock and stress response as examples for modulating environmental interactions. Furthermore, PAN downstream genes showed a strong developmental signature, which was most apparently represented by a number of plant hormone signaling systems. Based on our results we suggest that PAN might act as a node between cytokinin and auxin signaling pathways, with cytokinin outputs being repressed and auxin activity being induced by PAN. PAN is a member of the D-class of bZIP transcription factors (Jakoby et al., 2002) and thus groups with the TGA regulators, which are involved in mediating pathogen defense (Zander et al., 2010). The sequence similarity of PAN and TGA pathogen response regulators suggests that PAN function might have evolved from an environmental surveillance activity, which was enhanced to include developmental roles to give rise to an integrated buffering system.

MICROARRAY EXPERIMENTS
Pools of 50 microscopically dissected inflorescence apices of pan mutants and wild-type both carrying the KB14 AG::GUS reporter gene (Busch et al., 1999;Lohmann et al., 2001) were grown for 25 days in LD conditions and profiled in duplicate using the Affymetrix ATH1 platform. RNA extraction and microarray analyses were performed as described (Schmid et al., 2005;Buechel et al., 2010). Expression estimates were derived by GC-RMA (Wu et al., 2004) at standard settings implemented in R. We determined significant changes on a per-gene level by applying the Rank products algorithm (Breitling et al., 2004) using 100 permutations and a false discovery rate cut-off of 5%. GO analysis was carried out using AgriGO (Du et al., 2010).

QUANTITATIVE REAL-TIME PCR
Total RNA was extracted from apices of plants grown in an independent experiment using RNeasy Mini columns with on-column DNAse digestion (Qiagen). Reverse transcription was performed with 1 μg of total RNA, using a Reverse Transcription Kit (Fermentas). PCR amplification was carried out in the presence of the double-strand DNA-specific dye SYBR Green (Molecular Probes) using intron spanning primers (Andersen et al., 2008). Amplification was monitored in real-time with the Opticon Continuous Fluorescence Detection System (MJR). BETA-TUBULIN-2 transcript levels served to normalize mRNA measurements.