Gibberellin signaling is mediated by interactions between GA and the GIBBERELLIN INSENSITIVE DWARF1 (GID1) GA receptors (Schwechheimer, 2014). GA-binding triggers the proteasomal degradation of DELLA proteins, negative regulators of GA signaling, via the E3 ubiquitin ligase SCF^{SLY 1} (SKP1 – CULLIN – F-BOX PROTEIN with the F-box protein SLEEPY1 [SLY1]) or related complexes (Dill et al., 2004). DELLAs interfere with the activities of other proteins, mainly transcription factors such as PHYTOCHROME INTERACTING FACTORS (PIFs). The GA-induced degradation of DELLAs relieves – in the case of the PIFs – their repressive interactions and allows PIFs to bind DNA (de Lucas et al., 2008; Feng et al., 2008). Studies on the role of GNC and GNL in GA signaling were instigated by the observation that their transcription is repressed by GA signaling. This transcriptional regulation of the two B-GATAs could be explained by the DELLA-dependent control of PIFs, notably PIF3, which directly binds to GNC and GNL promoter elements (Figure 2; Richter et al., 2010). Since PIF activity is not only negatively regulated by DELLA interactions but also by light, the previously reported light-induced transcription of GNL may be explained by the regulation of GNL by PIFs (Naito et al., 2007; Figure 2).

Mutants and overexpressors of GNC and GNL have a number of phenotypes that can be explained by defects in GA signaling in that they promote greening and hypocotyl elongation but repress germination and flowering (Figure 4). When compared to mutants with a strong GA pathway defect, the contribution of GNC and GNL to plant growth regulation is comparatively subtle. For example, the strong flowering time delay of the ga1 mutant is only partially suppressed in ga1 gnc gnl (Richter et al., 2010, 2013a). In qualitative terms, this suppression is comparable to the suppression of the ga1 phenotype by DELLA gene mutants from Arabidopsis. There, the loss of individual members of the five member DELLA gene family only partially suppresses ga1 phenotypes, whereas the loss of multiple DELLA genes results in a strong genetic suppressions (Cheng et al., 2004; Cao et al., 2005). Taking into account that there are six presumably functionally redundant LLM-domain containing B-GATAs in Arabidopsis (Figure 3; Behringer et al., 2014), it could be envisioned that a stronger suppression of ga1 can be achieved when all six LLM-domain B-GATAs are mutated in ga1.

GNC and GNL are also transcriptionally repressed by AUXIN RESPONSE FACTOR2 (ARF2; Richter et al., 2013b). ARF2 belongs to the family of ARF transcription factors that have been subdivided into ARF+ that can bind the auxin-labile AUX/IAA repressors and ARF- that do not engage in such repressive interactions (Vernoux et al., 2011). AUX/IAA repressor abundance is negatively regulated by auxin through a specific SFC-type E3 ubiquitin ligase (Dharmasiri et al., 2005a,b). This regulation, however, only affects the activity of ARF+ and not that of ARF- such as ARF2.

arf2 mutants share a number of phenotypes with GNL and GNL overexpression lines such as seed size, chlorophyll biosynthesis, stamen length, floral organ abscission, and senescence (Ellis et al., 2005; Okushima et al., 2005; Richter et al., 2013b; Figure 4). Interestingly, these arf2 phenotypes are partially or fully suppressed in the presence of gnc and gnl loss-of-function mutants (Richter et al., 2013b). Thus, arf2 mutant phenotypes may be explained by increased GNC or GNL transcript levels in arf2 and the repressive activities of the GATAs in this mutant background. Indeed, GNC and GNL transcription is elevated in arf2 mutants and ARF2 directly binds to the GNC and GNL promoters (Richter et al., 2013b).

Although ARF2 is an auxin regulation-independent ARF-, the transcriptional repression of GNC and GNL can be modulated by auxin. This suggested that also auxin-responsive ARF+ and AUX/IAAs may regulate GNC and GNL expression. Indeed, loss-of-function mutants of the ARF+ proteins ARF7 and its paralog ARF19 as well as gain-of-function mutants of their interacting AUX/IAA SLR1 (SOLITARY ROOT1) are phenotypically similar to GNC and GNL overexpressors. In line with a direct activity of ARF7 on the B-GATA promoters, an auxin-modulated binding of ARF7 to the GNC and GNL promoters could be demonstrated. Thus, GNC and GNL transcription is under the control of auxin- and AUX/IAA-independent (ARF2) as well as auxin- and AUX/IAA-dependent (ARF7) transcription factors (Richter et al., 2013b; Figure 2).

The observation that the two phytohormones, GA and auxin, repress the transcription of GNC and GNL suggested that modulation of the expression of the two GATAs would allow for a transcriptional cross-talk between these two pathways. In fact, several phenotypes of the arf2 mutant could be suppressed by GA treatments or in the presence of a spy (spindly) mutation, which phenotypically mimics the phenotypes of plants with constitutively active GA signaling (Richter et al., 2013b). Thus, GA and auxin signaling converge on the transcriptional regulation of GNC and GNL and these two signals control at least in part the same growth responses (Figure 2).

The analysis of this GA-auxin cross-talk also resulted in the identification of two feedback regulatory mechanisms that contribute to the regulation of GNC and GNL expression (Richter et al., 2013b). First, ARF2 autoregulates its own transcription and thereby negatively feeds back on its own transcription as well as GNC and GNL regulation, and second, GA promotes ARF2 abundance by controlling ARF2 translation or by controlling the stability of a de novo synthesized and unknown GA-responsive protein involved in regulating ARF2 abundance (Richter et al., 2013b).

GNC was isolated based on the greening defect of its loss-of-function mutant (Bi et al., 2005). Although such as greening defect is not visible (but quantifiable) in the gnl mutant, it is enhanced in the gnc gnl double mutant. GNC and GNL thus redundantly regulate greening, possibly together with other LLM-domain containing B-GATAs (Figure 4; Richter et al., 2010; Behringer et al., 2014). The greening phenotype of GNC and GNL overexpression lines correlates with the increased expression of the chloroplast localized GLUTAMATE SYNTHASE, HEMA, GENOMES UNCOUPLED4, and PROTOCHLOROPHYLLIDE OXYDOREDUCTASE genes as well as that of PDV2 (PLASTID DIVISION2; Richter et al., 2010; Hudson et al., 2011). At the same time, overexpression of GNC, GNL, or that of other B-GATAs induces a strong greening phenotype in tissues that normally do not contain significant numbers of chloroplasts such as the lower hypocotyl, the upper part of the root, and epidermal cells of cotyledons, and the hypocotyl (Richter et al., 2010; Chiang et al., 2012; Behringer et al., 2014). The role of the LLM-domain containing B-GATAs in the control of greening is conserved across species since the overexpression of LLM-domain containing B-GATAs from barley, tomato, or rice induces similar phenotypes when tested in Arabidopsis or rice, respectively (Hudson et al., 2011; Behringer et al., 2014). Taken together, LLM-domain containing B-GATAs are at least in some tissues sufficient to strongly promote greening.

B-GATAs may control greening by promoting chlorophyll biosynthesis, chloroplast formation, or chloroplast size. In this regard, it is important to note that CK, which induces GNC and GNL expression, can promote greening in multiple developmental contexts (Kiba et al., 2005; Figure 5). CK induces chloroplast division by activating the expression of the chloroplast division regulators PDV1 and PDV2 in a manner that is dependent on the CK-induced regulator CRF2 (CYTOKININ RESPONSE FACTOR2; Okazaki et al., 2009). Although this increase in chloroplast division correlates with a reduction of chloroplast size, CK-treated plants have elevated chlorophyll levels (Okazaki et al., 2009). Furthermore, CK can promote greening ectopically in tissue that normally does not contain many chloroplasts including the upper part of the root (Kobayashi et al., 2012). Along these lines, the strong greening phenotype of the GNC overexpressors can be explained by an increased number of chloroplasts that is accompanied by the reduction of chloroplast size as it is typical for CK-treated seedlings (Chiang et al., 2012; Figure 5). Although the number of chloroplasts is not reduced, gnc gnl mutants have smaller chloroplasts in the hypocotyls and reduced chlorophyll levels in seedlings (Richter et al., 2010; Chiang et al., 2012). Additionally, ectopic expression of GNC promotes the differentiation of etioplasts from proplastids in dark-grown seedlings, which also can be correlated with an accelerated greening when etiolated seedlings are exposed to light (Chiang et al., 2012). CK treatment induces the expression of GNL but is less efficient in inducing the expression of GNC (Naito et al., 2007; Chiang et al., 2012). The type-B response regulators ARABIDOPSIS RESPONSE REGULATOR1 (ARR1) and ARR12 are important for this regulation since CK-induced gene expression of GNL is strongly compromised in arr1 arr12 mutants where CK effects on chloroplast division are also compromised (Argyros et al., 2008; Chiang et al., 2012; Figure 5). Furthermore, gnc gnl mutants exhibit reduced CK sensitivity in chloroplast division (Chiang et al., 2012). Taken together these findings suggest that the greening defect of gnc gnl mutants is caused by their reduced CK-responsiveness and that this CK response requires the induction of GNL and possibly other B-GATAs through type-B ARRs. This adds LLM-domain containing B-GATAs to the list of transcription factors that can promote greening downstream from CK such as the previously mentioned CRF2 but also GLK2 (GOLDEN LIKE2; Fitter et al., 2002; Okazaki et al., 2009; Kobayashi et al., 2012). Although this has not been studied in detail, it may be that the effects of GA and auxin signaling on greening (Richter et al., 2010; Richter et al., 2013b) are, at least in part, also a consequence of the role of GNC and GNL on chloroplast division as demonstrated for CK signaling.

GNC and GNL are flowering repressors. The contribution of these two GATAs to flowering time control can be observed in the GA-deficient late flowering mutant ga1 where loss of GNC and GNL function promotes the flowering of ga1 by about a month (Richter et al., 2013a). A dedicated analysis has placed these B-GATAs in the network around the flowering time regulator SOC1 (SUPPRESSOR OF THE OVEREXPRESSION OF CONSTANS1; Richter et al., 2013a). The MADS-box transcription factor SOC1 is a major regulator of flowering time in Arabidopsis thaliana. SOC1 expression is under control of a number of flowering promoting inputs and SOC1 expression is essential for floral induction in long day conditions (Samach et al., 2000; Yoo et al., 2005). In short-day conditions, SOC1 is a major integrator of flowering time stimulation by GA (Blázquez and Weigel, 1999; Moon et al., 2003). Based on the central role proposed for SOC1 in flowering time regulation, SOC1 also qualified as a possible target of flowering time control downstream from GNC and GNL. Indeed, the promoter of SOC1 is recognized by both GATAs and SOC1 expression is strongly downregulated when the GATAs are overexpressed (Figure 6). In turn, when SOC1 expression is uncoupled from GNC and GNL control in a SOC1 overexpression line, the flowering repressive effects of GNC and GNL overexpression are suppressed. Thus, GNC and GNL act upstream of SOC1 in flowering time control.

Curiously, the respective genetic interaction experiments also indicated that there may be an inverse relationship between SOC1 and the GATAs in the control of other B-GATA-regulated responses that are not directly related to flowering time control (Figure 6). In fact, the genetic interaction experiments between SOC1, GNC, and GNL indicated that two other phenotypes of soc1 mutants, enhanced greening and decreased cold tolerance, are suppressed in the absence of the GNC and GNL regulators (Richter et al., 2013a). Thus, cross-repressive interactions between these B-GATAs and SOC1 govern distinct biological processes.

During embryo development, HAN is required for the proper positioning of the proembryo boundary (Figure 7). han mutant embryos have several developmental defects, including a vacuolation of the lower tier cells of the embryo and a decrease in suspensor cell divisions (Nawy et al., 2010). The expression domains of embryonic markers for the suspensor and the lower tier are shifted apically in globular stage han embryos and this fits, in the case of the suspensor marker-positive cells, to their morphological resemblance to suspensor cells. The apical shift of auxin distribution in han mutant embryos and a broadening of the expression of the auxin response marker DR5:GFP (DR5:GREEN FLUORESCENT PROTEIN) are further indications for defects in proembryo boundary positioning in these mutants. Since lower tier cells normally give rise to the root, root formation is impaired in han embryos. han mutants fail to form an embryonic root and they are unable to undergo an essential cell division of the uppermost suspensor cell, the hypophysis that produces the quiescent center (QC). Interestingly, most han mutants can later recover from this defect and produce a root independently of the hypophysis at a later stage of embryo development (Nawy et al., 2010). It is thought that this rescue is the consequence of the coincidental expression of several prerequisites for QC formation: a local auxin maximum and the expression of the root regulators PLETHORA, SHORT ROOT and SCARECROW (Nawy et al., 2010). In addition to root development phenotypes, han mutants have also defects in cotyledon growth and initiation. han mutants sometimes have up to four cotyledons (Zhao et al., 2004). Ectopic root formation and altered PLETHORA expression were also observed in a han allele that strongly enhances the phenotype of mutants of the leaf development regulator ANGUSTIFOLIA3 (Kanei et al., 2012).

Auxin is an actively transported hormone and its distribution within the embryo is mediated by auxin efflux carriers such as PIN1 (PIN-FORMED1) and PIN7 (Friml et al., 2003). In the wild type, auxin initially accumulates in the apical part and shifts to the suspensor preceding hypophyseal cell specification. This shift in auxin distribution correlates with a shift of PIN7 in the suspensor from the apical to the basal plasma membrane and a shift of PIN1 from being non-polarly distributed to being polarly distributed in the provascular cells of the proembryo (Friml et al., 2003). Both, the expression domain of PIN1 as well as that of PIN7 is shifted apically in han mutants and the ectopic expression of the PINs, at least that of PIN1, can find its explanation in a possibly direct transcriptional regulation of PIN1 by HAN (Nawy et al., 2010). Thus, the han mutant phenotype may have its molecular cause in a misexpression of the PINs and consequently altered auxin distribution.

HAN was first described based on the han mutants with altered shoot meristem morphology (Zhao et al., 2004; Figure 7). When compared to the wild type, han loss-of-function mutants have small flat shoot meristems, reduced numbers of floral organs in all four whorls as well as fused sepals. HAN is expressed between the meristem and between newly initiated floral organ primordia and in the boundaries between the different floral organ whorls. HAN overexpression, on the other side, results in delayed plant growth, disturbed cell divisions, and a loss of meristem activity. Taken together these findings suggest that HAN acts as a repressor of cell proliferation and that loss of this repressive function could lead to the reduced meristem size, which may be the cause for the reduced floral organ numbers and fused floral organs seen in its mutants (Zhao et al., 2004).

HAN expression surrounds the floral meristem cells and HAN interacts strongly with the CLV (CLAVATA) pathway (Zhao et al., 2004). In Arabidopsis, shoot meristem size is determined on the one side by the plasma membrane-resident receptor proteins CLV1 and CLV2 that are co-expressed in the outer layer of the shoot meristem as well as their putative peptide ligand CLV3 that is expressed in the underlying tissue layers. Defects in any of the three CLV genes results in enlarged shoot and floral meristems and the formation of an increased number of floral organs. The expression of CLV3 is negatively regulated by the homeobox-type transcription factor WUSCHEL (WUS) and CLV3 expression overlies the expression domain of WUS suggesting, in combination with evidence from mutant analyses, that WUS controls meristem size by restricting CLV3 expression as an essential ligand for the CLV1 and CLV2 receptor proteins. A han mutation combined with clv gene mutations resulted in increased inflorescence fasciation and increased floral abnormalities (Zhao et al., 2004). It has been proposed that the HAN gene is required to control WUS expression and reduced WUS expression as well as ectopic WUS expression may have a role in controlling floral meristem growth and repress floral organ primordium initiation (Figure 7). Alternatively, it may be envisioned that defects in nutrient or signal transport hinder meristem growth and floral organ development since HAN is also expressed early in provascular cells (Zhao et al., 2004). Interestingly HAN is also expressed in the boundaries between different whorls and between different floral organs suggesting that HAN could also act as a repressor of cell divisions. In this regard, there are some interesting parallels to the role proposed for HAN-related B-GATAs in bract suppression in monocots as will be discussed below.

Molecular analysis for transcription factor targets identifies HAN as a repression target of JAGGED (Schiessl et al., 2014). Genes acting downstream of HAN were also searched for using translational fusions between HAN and the glucocorticoid receptor, which allows for the glucocorticoid hormone-induced translocation of the HAN-GR fusion protein from the cytoplasm to the nucleus (Zhang et al., 2013). This analysis identified a range of floral development regulators as well as phytohormone-related genes as targets of HAN and suggested that HAN can act as a transcriptional activator and repressor. Interestingly, amongst the phytohormonal target genes are genes of those phytohormonal pathways that are known to regulate the expression of the LLM-domain B-GATAs GNC and GNL such as the DELLA genes of the GA pathway, ARR genes of the CK pathway, and ARF and AUX/IAA genes of the auxin pathway. Although the chosen experimental approach would have permitted to test for direct transcription targets by blocking de novo protein synthesis this possibility was not exploited. In conjunction with the fact that rather long time points (4 h up to 72 h) after glucocorticoid treatment were used for the sampling of the material and that many transcription factor genes were found to be regulated downstream from HAN in this experiment argues that the majority of these downstream genes could represent indirect rather than direct targets of HAN.

Among the genes that were found to be HAN-regulated according to this experiment were also HANL2, GNC as well as GNL. The transcriptional repression of these three genes suggested that their downregulation may be part of a negative feedback mechanism that serves to control B-GATA levels. Indeed, HAN was found to be able to bind to its own promoter as well as the promoter of GNC. Furthermore, genetic interaction studies using mutants of these B-GATAs found that mutant combinations of han with hanl2, gnc, and gnl mutations resulted in a strong decrease in the number of petals formed in these mutants, sepal fusion defects, fertility defects, as well as carpel abnormalities (Zhang et al., 2013). Also during embryogenesis, the combination of B-GATA mutations renders the previously described han embryogenesis defects more severe and embryos frequently terminate differentiation and form only clusters of cells (Zhang et al., 2013). Although the respective mutant analyses suffer from the weakness that mutations in the Columbia and Landsberg erecta backgrounds were combined and some of the observed defects may therefore be the result of these combinations, the genetic interplay between HAN and the other B-GATAs is also supported by the fact that HAN can homodimerize and interact with HANL2 as well as with GNC and GNL in yeast two-hybrid system (Zhang et al., 2013).