1. Introduction
Growth and development in multicellular eukaryotes involves the integration of diverse cell types whose specification is under strict spatial and temporal control. This fate specification is the product of precise gene expression patterns established through a balance between transcriptional activation and repression. On a broad scale, this balance ensures that developmental programs are invoked at appropriate stages in an organism’s life cycle. While eukaryotes employ many mechanisms to mediate this control, chromatin modification is especially well-suited to establish and sustain a particular transcriptional state. For example, Polycomb group (PcG) and Trithorax group (trxG) proteins are chromatin modifiers that play critical roles by stably repressing or activating transcription, respectively. Two important PcG multi-protein complexes include PRC2, which trimethylates histone H3K27, and PRC1, which ubiquitylates histone H2A to compact chromatin and silence gene expression [
1]. Conversely, trxG proteins antagonize PcG repression through mechanisms that include trimethylation of histone H3K4 and ATP-dependent chromatin remodeling [
1].
In higher plants, reproductive flowers arise late in development, with floral-specific genes being repressed throughout vegetative growth. In the model angiosperm
Arabidopsis thaliana, chromatin modifiers are key to this repression, as their mutation results in ectopic gene expression and numerous developmental defects. For example, mutation of the histone methyltransferase
CURLY LEAF (
CLF), a PRC2 component, causes derepression of floral organ genes and results in upward-curled leaves, small rosettes, early flowering, short inflorescence stems, and homeotic conversions of floral organs [
2]. Antagonizing
CLF function are trxG factors such as
ARABIDOPSIS HOMOLOG OF TRITHORAX1 (
ATX1) and
ULTRAPETALA1 (
ULT1), mutations of which suppress
clf defects [
3,
4]. Consistent with this antagonistic relationship,
Arabidopsis plants overexpressing
ULT1 display similar phenotypic abnormalities as
clf loss-of-function mutants [
4]. Apart from floral organ gene repression, different
Arabidopsis PRC2 complexes control diverse growth programs, ranging from endosperm and seed development to vernalization-dependent floral transition [
5].
Since organismal growth programs can be strongly influenced by dynamic internal and external cues, they are often plastic in nature [
6,
7]. It follows that the chromatin regulation underlying these processes is similarly flexible. Post-translational protein modification is an effective strategy for conferring such plasticity, and chromatin regulators are commonly subject to these modifications, including phosphorylation [
8]. Phosphorylation can influence many protein properties such as subcellular localization, enzymatic activity, protein–protein interactions, and chromatin association [
9]. For example, phosphorylation of the PRC2 methyltransferase Enhancer of Zeste homolog 2 (EZH2) prevents histone binding and compromises its catalytic activity, resulting in gene derepression [
10]. Human orthologs of the trxG SNF/SWI complex are also phosphorylated during mitosis, resulting in their exclusion from chromatin [
11].
Phosphorylated chromatin modifiers often harbor kinase recognition sequences that are acidic in nature, implicating casein kinase 2 (CK2), a ubiquitous Ser/Thr protein kinase, as a prevalent regulator [
8]. Indeed, the
Drosophila PRC2 component Extra sex combs (Esc) and its mammalian ortholog Embryonic Ectoderm Development (EED) are phosphorylated by CK2, which promotes protein homodimerization and influences overall complex formation [
12,
13]. Similarly, CK2-mediated phosphorylation of the PRC1 subunit Cbx2 in mouse changes its binding affinity for modified histone H3 [
14].
In the present work, we describe the isolation of an activation-tagged mutant in Arabidopsis we named taco leaf-1D (tco-1D) due to its strongly upward-curled leaves. Numerous properties of tco-1D are consistent with disrupted chromatin regulation, including its developmental abnormalities, misregulated genes, and genetic interactions with mutants of PcG and trxG homologs. Additionally, the nuclear localization of TCO often displays a “speckled” distribution reminiscent of numerous eukaryotic chromatin regulators. The expression pattern of TCO and defects associated with tco loss-of-function mutants implicate this factor in the regulation of seed development. Finally, TCO is bound and phosphorylated by CK2, suggesting that a posttranslational regulatory mechanism broadly used to control the activity of chromatin modifiers may also influence TCO function.
3. Discussion
The development of multicellular organisms relies on complex gene expression patterns established through transcriptional activation and repression. We have identified
TCO, a putative transcriptional regulator in
Arabidopsis, whose overexpression causes broad misregulation of reproductive genes in the vegetative stage, resulting in patterning abnormalities. This collection of defects, which includes curled leaves and early flowering, resembles mutants of chromatin-modifying factors [
2,
31,
38], implicating a role for TCO in this mode of transcriptional regulation. This is further supported by the nuclear localization of TCO, which tends to accumulate in distinct foci or “speckles” reminiscent of known chromatin regulators [
4,
30,
31,
32,
33]. Moreover, the enhancement and suppression of
tco-1D developmental defects by
clf and
ult1, respectively, suggests that TCO may regulate a shared set of target genes with these well-characterized chromatin modifiers. While these genetic interactions were apparent in plants ectopically expressing
TCO (
Figure 4A-K;
Figure S5A–D), functional overlap may also occur in their normal domains of expression. Specifically, reporter gene analyses demonstrated that
TCO acts during embryo/seed development, and both
CLF and
ULT1 are expressed in developing seeds [
34,
39]. TCO may also interact with other chromatin modifiers to influence gene expression in these tissues. Indeed, a recent unpublished report describes a physical interaction between TCO (therein named EFC) and Multicopy Suppressor of Ira1 (MSI1) [
40], a core component of
Arabidopsis PRC2 complexes [
5]. While these observations support a role for TCO in chromatin-dependent transcriptional regulation, the mechanism by which TCO participates in these processes is unclear, given its small size and lack of conserved functional domains.
Our determination that the protein kinase CK2 interacts with and phosphorylates TCO provides some insight into the nature of its activity. Phosphorylation of TCO could potentially affect its function in a variety of ways, including by altering its subcellular localization, its interaction with other proteins, and/or its association with chromatin [
9]. Mutation of TCO serine residues targeted by CK2, however, did not alter its nuclear localization or speckle formation, suggesting that CK2-mediated phosphorylation influences other aspects of TCO activity. Similarly, overexpression of most TCO variants with either phospho-deficient or phospho-mimic mutations did not produce phenotypes that differed from plants overexpressing wild-type TCO. Here, functional differences between these various transgenes may have been obscured by their high expression levels. Support for this notion comes from analyses of the
Arabidopsis bZIP transcription factor TGA2, which is also targeted by CK2. While mutation of TGA2 phosphorylated residues did not affect its nuclear localization, it compromised its ability to bind DNA [
41]. Even so, overexpression of wild-type and mutated TGA2 activated target genes to the same extent in planta, leading to the suggestion that high protein levels masked functional disparity between the TGA2 variants [
41]. That CK2-mediated phosphorylation of TCO is biologically relevant is suggested by our inability to isolate
Arabidopsis transgenic lines overexpressing mTCO with all four targeted serine residues mutated to alanine (S11A,S12A,S73A,S74A), despite efficient transient expression of this same construct in tobacco cells. Although the reason for this is unclear, it is possible that phosphorylation of TCO modulates its transcriptional activity, and that overexpressing phospho-deficient TCO broadly disrupts gene expression to an extent that compromises plant viability.
TCO may interact with CK2 not only to mediate its own phosphorylation, but to recruit CK2 to protein complexes to target other factors. The use of protein–protein interaction as a means to modulate kinase activity has been previously proposed for CK2 [
42,
43], and the reported interaction between TCO and the PRC2 component MSI1 [
40] could facilitate this type of mechanism. CK2-mediated phosphorylation of PRC factors has been reported to influence their protein–protein interactions, histone-binding properties and enzymatic activities [
12,
13,
14,
44], raising the possibility that mutation of
TCO indirectly affects the ability of CK2 to target chromatin regulators. If the Enhancer of Zeste (E(z)) protein CLF is one such chromatin regulator, this could explain why
tco-1D phenocopies
clf. In a similar manner, phosphorylation of the human homolog of E(z), EZH2, interferes with its ability to bind and methylate histones, resulting in derepression of target genes [
10].
Contrasting the broad gene misregulation in
tco-1D seedlings,
tco loss-of-function alleles affected gene expression to a lesser extent when tested in young silique tissue (
Figure 5C,D). While this can be explained by the restricted domain of expression naturally displayed by
TCO, the reason that both gain- and loss-of-function alleles displayed gene upregulation is unclear. Removing
TCO function from its natural domains of expression could disrupt protein complexes that rely on TCO and/or CK2 activity, resulting in gene derepression. Strong, ectopic expression of
TCO could similarly affect gene expression if this causes TCO to interfere with the formation of protein complex(es). Indeed, overexpression of factors that operate in protein complexes can mimic loss-of-function defects due to antimorphic effects [
45]. Based on the reported binding of TCO to MSI1 [
40], it is possible that high levels of TCO interfere with the ability of MSI1 to associate with other binding partners, including LIKE HETEROCHROMATIN PROTEIN 1/TERMINAL FLOWER 2 (LHP1/TFL2), which performs a Pc-like role in
Arabidopsis [
46,
47]. Consistent with this scenario,
tco-1D plants phenotypically resemble
lhp1/
tfl2 loss-of-function mutants [
31,
48] and misexpress genes that rely on LHP1/TFL2-mediated repression such as
AG and
SEP3 [
49].
Notably, the most striking developmental defects caused by
TCO overexpression arose in tissues that do not naturally express
TCO, such as vegetative leaves. Growth responses that normally operate in the absence of TCO may, therefore, be particularly sensitive to artificial exposure to TCO activity. In these cases, TCO may engage in non-specific protein interactions, interfere with the composition of transcriptional regulatory complexes, and/or broadly impinge on chromatin topology, as suggested by widespread gene misregulation in
tco-1D seedlings (
Figure 3F;
Figure S4A–D). In contrast,
TCO overexpression did not appear to affect tissues that normally express
TCO, namely the seed (
Figure 5A,B). Specifically,
tco-1D and 2x35Sp::TCO-GFP transgenic backgrounds lacked obvious seed abnormalities, even though such defects were present in the loss-of-function allele
tco-1 (
Figure S7C,D,H,L). These observations indicate that growth processes that already involve TCO activity, such as seed development, may be able to accommodate further increases in TCO levels. Conversely, these same processes are likely to become compromised if
TCO activity is reduced or abolished, as displayed by
tco-1 mutant seeds.
A future challenge will be to better clarify the specific endogenous role of
TCO in
Arabidopsis development. Notably,
TCO has been identified as a candidate imprinted gene that displays seed-specific expression and hypomethylation in the endosperm [
50]. Many imprinted genes play roles in endosperm/seed development [
51], including a number of chromatin regulators [
52]. It is plausible that
TCO also contributes to this mode of transcriptional regulation during seed development, as our experimental data implicates
TCO in the regulation of chromatin.
TCO-like genes are also present in the genomes of other
Brassicaceae species (
Figure S2A–D), and their functional characterization will provide further insight into the role of
TCO and the extent to which this role is conserved in other plant species.