Why so repressed? Turning off transcription during plant growth and development

Summary

To ensure correct patterns of gene expression, eukaryotes use a variety of strategies to repress transcription. The transcriptional regulators mediating this repression can be broadly categorized as either passive or active repressors. While passive repressors rely on mechanisms such as steric hindrance of transcriptional activators to repress gene expression, active repressors display inherent repressive abilities commonly conferred by discrete repression domains. Recent studies have indicated that both categories of regulators function in a variety of plant processes, including hormone signal transduction, developmental pathways, and response to biotic and abiotic stresses.

Introduction

As sessile organisms, plants must perceive and respond to a wide range of biotic and abiotic signals in order to optimize their growth and development. Moreover, cells within a plant rely on positional information from their neighbors in order to adopt proper fates. A large part of these responses involves appropriate regulation of gene expression. To this end, eukaryotes employ a wide repertoire of transcriptional repression mechanisms. In general, such mechanisms can be separated into two main types: active and passive repression. Active repressors display an intrinsic repressive capacity conferred by defined repression domains [1,2]. For example, repression domains of sequence-specific transcription factors can be used to interact with non-DNA-binding proteins such as co-repressors. Co-repressors, in turn, recruit other regulators including chromatin remodeling factors that can promote the formation of a repressive chromatin state. Some of the best characterized of these factors are histone deacetylases (HDACs) which remove acetyl groups from lysine residues of histone amino terminal tails, generally resulting in a tightening of chromatin and gene silencing [3]. Contrasting active repression, regulatory proteins can employ steric hindrance mechanisms to counteract the function of transcriptional activators, such as preventing their binding to DNA. Such proteins that indirectly influence transcription by physically interfering with activators are termed passive repressors [1,2,4]. Interestingly, some transcription factors are able to repress gene expression both passively and actively. For instance, the mammalian retinoblastoma protein Rb passively interferes with E2F transcriptional activators by binding and “masking” their transactivation domain while recruiting histone modifiers such as HDACs to actively repress transcription [2,5]. In this review, we discuss various reports demonstrating that plants use a number of transcriptional repression methods to ensure correct gene expression. While we concentrate on mechanisms involving transcription factors, plants display numerous other strategies to silence genes [for reviews, see 6,7].

Transcriptional Repression in Hormone Signal Transduction

In recent years, a common theme has emerged regarding the induction of gene expression in response to a variety of plant hormones, including auxin, jasmonate (JA) and gibberellin (GA). In these signaling pathways, DNA-binding transcription factors are under the negative regulation of labile repressors. Upon exposure to the relevant hormone, the repressors are targeted for 26S proteosome-mediated degradation by Skp1-Cullin-F-box (SCF)-type E3 ubiquitin ligases. Following this degradation, transcriptional regulators are liberated to activate downstream target genes necessary for mediating the correct hormone response.

In the case of auxin signaling, AUX/IAA repressor proteins bind and negatively regulate AUXIN RESPONSE FACTORs (ARFs), a family of DNA-binding transcription factors involved in auxin-mediated developmental processes [8] (Figure 1a). Auxin relieves this repression by binding to its receptors, the F-box protein TRANSPORT INHIBITOR RESISTANT1 (TIR1) and its close homologs, resulting in increased affinity of SCF^TIR1 for AUX/IAAs which are subsequently targeted for degradation via ubiquitination [9-12]. Repression by AUX/IAAs depends on a short sequence of amino acid residues (LxLxL), termed the ERF-associated amphiphilic repression (EAR) motif, located in their conserved domain I [13]. The motif is so named because it was originally identified as a strong transcriptional repression domain in members of the ethylene response factor (ERF) family [14]. However, the molecular mechanism behind EAR motif-conferred repression has remained unknown until recently. Insight was provided by a yeast 2-hybrid screen that identified IAA12/BODENLOS (BDL), an AUX/IAA which influences root and vascular pattern formation [15,16], as an interactor of the Groucho(Gro)/Tup1-like transcriptional co-repressor TOPLESS (TPL) [17*]. This interaction, which depends on the EAR motif of IAA12/BDL, supports a model whereby AUX/IAAs recruit TPL to actively repress ARF-mediated transcriptional regulation of target genes (Figure 1a).

Figure 1.

Transcriptional repression mechanisms in hormone signaling pathways.

(a) Active transcriptional repression of auxin-responsive genes.

(Left) In the absence of auxin, AUX/IAA repressor proteins bind directly to ARF transcriptional regulators through shared conserved carboxy-terminal domains [8]. AUX/IAAs recruit the transcriptional co-repressor TPL, an interaction that depends on the AUX/IAA EAR repression motif [17*]. As a Gro/Tup1-like co-repressor, TPL is predicted to recruit chromatin remodeling factors such as HDACs to negatively regulate target genes [88].

(Right) High auxin concentrations promote the interaction between AUX/IAAs and the auxin receptor TIR1, an F-box protein of an SCF-type E3 ligase [9-12]. This leads to degradation of AUX/IAAs in a 26S proteosome-dependent fashion and, because TPL is no longer recruited to the DNA, induction of auxin-responsive genes by activating ARFs.

(b) Transcriptional regulation of JA-inducible genes.

(Left) Through their conserved carboxy-terminal Jas domain, JAZ proteins physically interact with transcriptional regulators controlling JA-inducible gene expression such as MYC2 [19**-21]. Currently, the mode of JAZ-mediated repression is unknown. Possible passive mechanisms include JAZs interfering with the ability of MYC2 to bind DNA or to recruit factors involved in transcription initiation at target genes. JAZ proteins may act through an active repression mechanism analogous to AUX/IAA repressors by recruiting transcriptional co-repressors/chromatin remodeling factors to negatively regulate target gene expression.

(Right) Exposure to JA causes an increased association between JAZs and SCF^COI1, resulting in the ubiquitination and degradation of JAZs [19**,22**,23]. This liberates MYC2 to activate primary genes of the JA response.

(c) Passive repression of GA-mediated transcription.

(Left) DELLAs passively repress the transcriptional regulators PIF3 and PIF4 by directly associating with their bHLH domains and preventing their binding to DNA targets, including genes that promote hypocotyl elongation [33**,34**].

(Right) Binding of GA to its GID1 receptors increases their association with DELLAs, resulting in enhanced affinity of DELLAs for the SCF^SLY1/GID2 complex and their 26S proteosome-mediated degradation [26-31]. Consequently, PIF transcription factors can activate the expression of genes responsible for hypocotyl growth.

Similar regulatory modes control the induction of genes by JA signaling, which functions in the defense response to various abiotic and biotic stresses [18]. Members of the JASMONATE ZIM-DOMAIN (JAZ) family of proteins bind and negatively regulate transcriptional regulators, such as MYC2, that confer JA responsive gene expression [19**-21] (Figure 1b). CORONATINE INSENSTIVE1 (COI1), an essential component of the JA receptor, is an F-box protein related to TIR1. In the presence of bioactive JA, COI1 displays an increased affinity for JAZ proteins and promotes their 26S proteosome-dependent degradation [19**,22**,23]. While identification and characterization of the JAZ repressors uncovered a key link between SCF^COI1 activity and JA-inducible gene expression, their mode of transcriptional repression remains to be determined. It has been shown that JAZ3 binds MYC2 at its amino-terminus, which harbors a putative transcriptional activation domain [19**,24]. This suggests JAZs may passively repress transcription by “masking” the ability of activators to recruit the transcriptional machinery (Figure 1b). However, if similarities to auxin signal transduction extend further, JAZ proteins may silence genes by recruiting transcriptional co-repressors. Such recruitment may occur through the conserved ZIM domain of JAZ proteins, as it was recently shown that this domain facilitates protein-protein interactions [21,25].

GA-mediated transcriptional regulation is subject to a repression mechanism involving DELLA domain proteins, a subfamily of the plant-specific GRAS transcriptional regulators. DELLA destabilization occurs upon GA binding to GIBBERELLIN INSENSITIVE DWARF1 (GID1) receptors, which complex with DELLAs and promote their association with the E3 ligase SCF^{SLEEPY(SLY1)/GID2} [26-31]. In Arabidopsis, there are five DELLAs, subsets of which have been implicated in a variety of GA-regulated processes [32].

Two recent reports have uncovered a role for DELLAs in the convergence of light and GA signaling and have described a mechanism of DELLA-mediated transcriptional repression [33**,34**]. In darkness, GA is required to maintain etiolated growth of seedlings, which includes hypocotyl elongation [35]. Phytochrome-interacting factors (PIFs) PIF3 and PIF4 are basic helix-loop-helix (bHLH) transcriptional regulators that also promote hypocotyl growth [36,37]. In response to light, however, PIF3 and PIF4 are degraded in a phytochrome-dependent fashion [33**,38-40]. Work by de Lucas et al. [33**] and Feng et al. [34**] has shown that the PIFs are also inactivated by DELLAs which directly bind the PIF bHLH DNA-recognition domain and prevent their binding to DNA targets (Figure 1c). Moreover, chromatin immunoprecipitation experiments were unable to detect association of affinity-tagged DELLAs with the promoters of GA-responsive genes [34**]. Collectively, these results suggest that DELLA-mediated repression occurs passively through the sequestration of transcription factors such as PIFs from DNA. Under conditions of increased GA levels, DELLAs are destabilized allowing PIF binding to target genes and the promotion of hypocotyl growth [33**,34**] (Figure 1c).

Transcriptional Repression in Developmental Responses

Similar to DELLA-dependent repression, a passive mechanism has been proposed for the regulation of class III homeodomain leucine zipper (HD-ZIPIII) proteins, which regulate shoot apical meristem maintenance and promote adaxial fate in lateral organs [41-43]. Arabidopsis HD-ZIPIII family members contain an HD immediately followed by a conserved leucine zipper domain that facilitates their dimerization, a requirement for DNA-binding of HD-ZIPs [44-47]. Interestingly, small negative regulators of HD-ZIPIIIs, termed LITTLE ZIPPERs (ZPRs), have been independently identified through analysis of genes induced by the HD-ZIPIII protein REVOLUTA (REV) and through a gain-of-function activation tagging screen [48**,49**]. These ZPRs, of which there are four in Arabidopsis (ZPR1-4), contain little more than a leucine zipper domain that facilitates their physical interaction with the structurally similar ZIP domain of HD-ZIPIIIs. Notably, ZPR association with HD-ZIPIIIs is believed to prevent HD-ZIPIII dimerization and disrupt DNA binding (Figure 2a). This hypothesis was strengthened by in vitro gel shift experiments where addition of ZPR3 abrogated the ability of REV to bind a probe containing its consensus recognition site [48**]. Since ZPR expression is positively regulated by HD-ZIPIIIs, the ZPRs appear to establish a negative feedback regulatory loop that dampens HD-ZIPIII activity [48**,49**]. It will be a future challenge to clarify if and how signals specifying cell fate (such as meristem and adaxial identity) influence the composition of HD-ZIPIII dimers, potentially promoting productive HD-ZIPIII/HD-ZIPIII or repressive HD-ZIPIII/ZPR interactions depending on the developmental context.

Figure 2.

Transcriptional repression mechanisms in developmental responses.

(a) Passive repression of HD-ZIPIII target genes.

(Left) ZPR proteins physically interact with HD-ZIPIII transcription factors through shared ZIP domains, preventing HD-ZIPIII/HD-ZIPIII dimerization [48**,49**]. Since HD-ZIPIII/HD-ZIPIII dimerization appears necessary for DNA binding [44-47], target genes are passively repressed.

(Right) In the absence of ZPR association with HD-ZIPIII transcription factors, the latter are able to dimerize, bind DNA, and activate targets, including genes involved in specifying meristem and adaxial identity [41-43].

(b) Passive repression of TALE homeodomain-mediated gene expression.

(Left) KNATM directly binds TALE homeodomain proteins, such as BELL transcription factors, and is proposed to render them inactive and/or sequester them in the cytoplasm [50**].

(Right) In the absence of repressive KNATM interactions, BELL proteins are able to enter the nucleus and/or bind gene targets. This may involve association with members of the KNOX family of transcriptional regulators, which have been shown to dimerize with BELL factors [89-91].

(c) Active repression of KNOX expression.

(Left) DNA recognition sites for AS1/AS2 heterodimers are present at two positions in the promoters of KNOX genes KNAT1/BP and KNAT2. Upon DNA-binding, heterodimers are proposed to associate with one another, likely due in part to the ability of AS1 to bind itself, resulting in a “looping” of the intervening promoter DNA [54,55**]. AS1 can also physically associate with the chromatin remodeling factor HIRA which plays a role in gene silencing [54,62-64], potentially due to interaction with HDACs [65]. This protein complex is predicted to produce a repressive chromatin state in this region of KNOX promoters, leading to the silencing of transcriptional enhancer (En) elements in the vicinity. These events are believed to effectively maintain KNOX gene silencing in domains of AS1/AS2 function, including leaf primoridia.

(Right) In the absence of AS1/AS2 activity, transcriptional enhancer elements in KNOX promoters are able to induce gene expression.

A newly characterized protein resembling KNOTTED1-LIKE HOMEOBOX (KNOX) transcription factors but lacking the conserved three amino acid loop extension (TALE) homeodomain may function similarly to ZPRs to passively repress transcription [50**]. This protein, KNATM, was originally identified from an in silico search for KNOX-related proteins in Arabidopsis [50**]. KNATM interacts with KNAT1/BREVIPEDICELLUS (BP) and BEL1-LIKE (BELL) homeodomain proteins through its amino-terminal acidic coiled-coil and conserved MEINOX (MEIS-KNOX) domains, respectively [50**,51]. Both the MEINOX domain and TALE homeodomain are shared between plant KNOX proteins and animal Myeloid ecotropic viral integration site (MEIS) proteins. Interestingly, isoforms of a mammalian MEIS homolog lacking a complete HD act as dominant-negative regulators of HD-containing variants [52]. KNATM, which is proposed to play a role in leaf proximal-distal patterning, may likewise act as a negative regulator of transcription factors by sequestering them in the cytoplasm and/or titrating them as inactive dimers [50**] (Figure 2b). In support of this hypothesis, bimolecular fluorescence complementation analysis showed that KNATM-BELL dimers preferentially accumulate in the cytoplasm of plant cells. Furthermore, combining overexpression lines of KNATM and the BELL gene SAWTOOTH1 [53] revealed an antagonistic relationship, as phenotypic abnormalities displayed by each individual transgenic line were mutually normalized, restoring a wild-type appearance. However, defining the precise role of KNATM in transcriptional regulation is complicated by the fact that it exhibits transcriptional activation activity [50**]. The isolation and analysis of a knatm loss-of-function allele should help clarify its function in the future.

A novel mechanism has been proposed for the transcriptional repression of the Arabidopsis KNOX meristem genes KNAT1/BP and KNAT2 in leaf primordia [54,55**]. In these developing organs, KNOX gene down-regulation corresponds with the expression of two transcriptional regulators, the MYB-domain factor ASYMMETRIC LEAVES 1 (AS1) and the LOB domain (LBD) protein AS2, which are necessary for maintaining repression and promoting determinate cell fate [56-59]. Chromatin immunoprecipitation experiments identified two distinct regions of both the KNAT1 and KNAT2 promoters bound by AS1, each comprised of a consensus MYB-binding site (motif I) followed by a previously uncharacterized motif (termed motif II) [55**]. Interestingly, in gel retardation experiments, AS1 only bound these regions when co-translated with AS2 in a cell-free expression system. Given that AS1 can physically interact with AS2 [54,60], cooperative association of AS1 and AS2 on motifs I and II is potentially required for DNA binding and repression of KNOX gene targets. Furthermore, since this binding module is repeated in a second position on both KNOX promoters, and since AS1 can homodimerize [54,61], the authors proposed a model in which two DNA-bound AS1/AS2 dimers associate with each other resulting in a looping-out of the intervening promoter region [55**] (Figure 2c). AS1 and its maize homolog ROUGH SHEATH2 (RS2) can physically interact with the chromatin remodeling factor HIRA [54], homologs of which in other eukaryotic systems associate with HDACs and function in gene silencing [62-65]. Since promoter regions in the vicinity of this proposed loop harbor enhancer elements necessary for ectopic KNOX expression, HIRA-mediated remodeling events are proposed to actively maintain KNOX silencing in developing lateral organs by negating enhancer activity. This activity closely resembles that of genetic insulators, which can form repressive chromatin loops that interfere with the ability of enhancer elements to communicate with promoters [66]. Lending support to this model, reduced levels of HIRA, like as1 and as2 mutants, result in ectopic KNOX expression in leaves [54].

HDAC Recruitment Confers Active Repression

HDACs are one of the best-studied classes of proteins recruited to facilitate active transcriptional repression. Histone acetylation is largely correlated with gene expression; therefore, removal of these modifications by HDACs generally leads to repression of transcription [3]. Mutation of the Arabidopsis Rpd3-like class I histone deacetylase HDA19, whose protein product exhibits HDAC activity in vitro [67], results in increased histone acetylation states in planta [68-73]. Furthermore, HDA19 functions cooperatively with co-repressors. For example, the Gro/Tup1-like transcriptional co-repressor LEUNIG (LUG), which shares structural homology with TPL, has been shown to physically interact with HDA19 in vitro [74**]. LUG plays a role in restricting the expression of the gene AGAMOUS (AG), which specifies the fate of the floral reproductive structures, to the inner two whorls of the flower [75,76]. DNA-binding transcription factors appear to recruit LUG to non-coding regulatory regions of AG through the intermediary adaptor protein SEUSS [77-79]. The direct and specific interaction between LUG and HDA19 in vitro implies that LUG negatively regulates genes such as AG by promoting the formation of a repressive chromatin structure. Interestingly, the observation that LUG can directly interact with Arabidopsis homologs of the Mediator complex indicates that LUG may also exert transcriptional repression by influencing RNA polymerase II activity [74**].

In other eukaryotes, class I Rpd3-like HDACs can function as part of multi-protein repressor complexes such as the Sin3 complex [80]. Arabidopsis homologs of some of these components have also been shown to associate with HDA19. For example, the putative transcriptional repressor ERF7 is proposed to function as a negative regulator of abscisic acid (ABA) and drought response by directly binding to ABA-inducible target genes and recruiting AtSIN3 and, in turn, HDA19 [81]. Similarly, the Arabidopsis homolog of Sin3-associated polypeptide of 18kDa (AtSAP18) interacts with the transcription factors ERF3 and AGAMOUS-LIKE 15 (AGL15) (which are expressed in response to salt stress and during embryogenesis, respectively) and is proposed to aid in the recruitment of HDA19 to repress target genes [82,83].

A variety of other reports have demonstrated the importance of HDA19 in regulating gene expression in response to environmental signals. For instance, HDA19 negatively regulates photomorphogenesis, and hda19 mutants exhibit increased levels of histone acetylation on a variety of light-responsive genes [72,73]. Conversely, HDA19 appears to be a positive regulator of plant defense by indirectly influencing the expression of PATHOGENESIS RELATED (PR) genes [71,84**]. For example, HDA19 is strongly induced by wounding, infection by Alternaria brassicicola (a pathogenic fungus), and the stress signals JA and ethylene [71]. Expression of PR genes co-regulated by JA and ethylene are increased in HDA19-overexpressing transgenic lines and decreased in lines with compromised HDA19 function, which show enhanced and weakened resistance to A. brassicicola, respectively [71]. Furthermore, in a recent study, HDA19 was identified as a physical interactor of the type III WRKY transcription factors WRKY38 and WRKY62 [84**]. These factors can activate transcription and are proposed to act on genes that, in turn, negatively regulate aspects of the plant defense response. Overexpression of HDA19, however, was shown to specifically reduce the ability of WRKY38 and WRKY62 to activate a reporter gene target in planta [84**]. Intriguingly, WRKY38 and WRKY62 are actually induced by the stress signal salicylic acid and infection by virulent Pseudomonas syringae strains. In this fashion, these WRKYs possibly prevent over-activation of the defense response at the onset of infection when pathogen levels are low. When a stronger effect becomes needed, HDA19, whose expression displays a delayed response to the stress signal, interacts with DNA-bound WRKYs to repress their target genes. This positively influences the plant defense response, including the induction of PR1 [84**].

Conclusions

It is becoming increasingly clear that plants rely heavily on transcriptional repression to control gene expression, which ensures proper development and responses to numerous environmental cues. Studies have shown that various repressive strategies are employed, including both passive and active mechanisms. It is only recently that active repression domains have been identified in plant proteins. This includes the EAR motif and two newly characterized repression domains [85*,86*]. The large number of transcriptional regulators containing one or more of these domains predicts a significant expansion of the plant repressor field in the near future. This will necessitate the identification and characterization of the co-repressors and/or chromatin remodeling factors that are recruited to confer repression. For example, while roles have been identified for HDA19, there are 17 other HDACs in Arabidopsis, most of which have not been functionally characterized [87]. Indeed, there will be numerous novel regulators that will remain silent no longer.