• flowering;
  • photoperiod

To ensure that reproduction occurs during optimal conditions, plants adapted to seasonal environments often coordinate their flowering in response to photoperiodic cues. Early plant biologists developed two longstanding mechanistic models to explain how plants recognize and respond to day length: the external coincidence model; and the florigen hypothesis (Kobayashi & Weigel, 2007). The former postulates that certain photoperiods induce flowering because a light-sensitive floral promoter is under the control of the circadian clock, such that it is only expressed during daylight hours in certain photoperiods. The latter model postulates the existence of a mobile signal that is produced under inductive photoperiods in the leaf and travels to the shoot apex where it promotes reproductive development. Recent molecular studies in Aradiposis thaliana have identified the genes that constitute the mechanistic bases for these models: CONSTANS (CO) and FLOWERING LOCUS T (FT ). Notably, these two mechanisms are connected because CO induces FT expression. The transcription of CO is regulated by the circadian clock, such that peak expression occurs late in the day under long days (e.g. 16 h light : 8 h dark), but after dark in short days (e.g. 8 h light : 16 h dark). Because the CO protein is only stable in the light, it only accumulates under inductive long days. The CO protein leads to the upregulation of FT in the vasculature of leaves; FT protein is then transported to the meristem to initiate reproductive development (Kobayashi & Weigel, 2007). A major unanswered question is exactly how CO functions to regulate FT. Although early circumstantial evidence suggested that CO might act as a typical transcription factor, more recent evidence indicates that CO may instead function as a co-activator. A report by Tiwari et al. in this issue of New Phytologist (pp. 57–66) provides fresh insight into this problem and suggests that CO actually has both transcription factor and co-activator activities.

‘Add to this variety the potential of CO and COL proteins to form homodimers or heterodimers and the regulatory plasticity of the system becomes vast.’

When CO was first identified, it was proposed to be a transcription factor because it contains two zinc-finger motifs that were originally thought to be homologous to those of the GATA transcription factors (Putterill et al., 1995). B-box zinc-finger motifs were subsequently described, however, and it is now accepted that the CO zinc fingers are homologous to these domains that mediate protein–protein interactions, but not DNA binding, in other proteins (Khanna et al., 2009). Nevertheless, several observations still supported the possibility that CO protein might function as a transcription factor with FT as a direct target. Overexpression of CO causes early flowering and increased FT expression; conversely, co mutants flower late and do not express FT during long days (Suárez-López et al., 2001). Also, FT is the only gene upregulated in wild-type plants exposed to a single long day that is not also upregulated in co mutants exposed to a single long day (Wigge et al., 2005). Finally, a screen for early CO targets had been conducted. A steroid-inducible version of CO was used to identify genes that are rapidly upregulated upon induction of CO activity and FT emerged as one of the few early targets identified in this screen (Samach et al., 2000). Because this screen was performed in the presence of cyclohexamide, which blocks translation, the upregulation of FT by CO appears to be direct, at least in the sense that it does not require the synthesis of new proteins.

Despite the circumstantial evidence suggesting that CO acts directly to regulate FT, evidence that CO can actually bind FT promoter sequences has been lacking, and more recent studies have found evidence that CO may instead function as a co-activator. First in tomato and then in A. thaliana, CO homologs were determined to physically interact with components of the NUCLEAR FACTOR Y (NF-Y)/HEME ACTIVATOR PROTEIN (HAP) complex through their CONSTANS-CONSTANS LIKE-TIMING OF CAB 1 (CCT) motif (Ben-Naim et al., 2006; Wenkel et al., 2006; Cai et al., 2007). Because this complex has been shown to bind DNA at CCAAT elements in yeast and mammalian systems, it was proposed that this complex recruits CO to the FT promoter where it acts as a co-activator. This model received further support when NF-YB/HAP3 proteins were demonstrated to bind the FT promoter in vitro (Kumimoto et al., 2008). Thus, the pendulum had swung to favor CO acting as a co-activator in an NF-Y/HAP complex.

The new study from Tiwari and colleagues provides evidence that CO acts both as a typical transcription factor and as a co-activator (Fig. 1). First, consistent with a previous study of a tomato CO homolog (Ben-Naim et al., 2006), the authors demonstrated that the glutamine-rich region of CO is sufficient to drive transcriptional activation in protoplasts. The authors then tested the ability of CO to activate transcription from a series of FT promoter fragments fused to a reporter gene. Notably, even when all the CCAAT motifs were deleted, CO still drove transcription above basal levels, indicating that CO could activate transcription independently of canonical NF-Y/HAP-binding sites. By further winnowing down the FT promoter to a minimal region responsive to CO activation, the authors were able to define two candidate CO-responsive elements (COREs). The importance of these CORE sequences was then confirmed by demonstrating that CO could activate transcription from minimal promoters containing multimerized CORE sequences, but not from mutated promoters containing several base changes in the COREs. To determine if this transcriptional activation might be a result of the direct binding of CO to CORE sequences, the authors performed in vitro gel-shift assays. These experiments showed that CO does indeed bind to COREs and that the CCT domain is responsible for DNA binding. In addition to demonstrating a novel NF-Y/HAP-independent role for CO in the activation of FT, Tiwari et al. also provide evidence for the importance of NF-Y/HAP in the activation of FT by CO. When multimerized and added to a minimal promoter, CORE sequences are sufficient to drive CO-dependent expression of FT; however, FT expression is further enhanced by the addition of a CCAAT sequence.


Figure 1.  Schematic drawing of the FLOWERING LOCUS T (FT) promoter. CONSTANS (CO) can bind the FT promoter directly at CO-responsive element (CORE) sites or through interactions with the NF-Y/HAP (NUCLEAR FACTOR Y/HEME ACTIVATOR PROTEIN) complex at CCAAT sites.

Download figure to PowerPoint

It is worth noting that the emerging model for gene regulation by CO and NF-Y/HAPs provides incredible potential for the formation of complexes with different specificities; Arabidopsis contains 17 CO-like (COL) proteins and 26 HAP3 and HAP5 proteins. Add to this variety the potential of CO and COL proteins to form homodimers or heterodimers and the regulatory plasticity of the system becomes vast. Tiwari and colleagues provide hints that differential affinities of CO/COL proteins for particular COREs may also shape gene-regulatory specificity. CO and COL15 produce similar levels of activation from reporter constructs containing CORE1 or CORE2. By contrast, COL9 shows little activation of CORE1, but strongly activates CORE2.

These results raise additional intriguing questions about the evolution of the COL family and the regulation of FT by CO. For instance, the finding that in some instances different COL proteins cannot bind the same CORE element is just as interesting as the finding that in some instances different COL proteins can bind the same CORE element. Competing interactions with different COL proteins that recruit different binding partners may foster regulatory innovation and complexity. In addition, this redundancy creates the opportunity for this regulatory interaction to experience developmental system drift (True & Haag, 2001), meaning that, over time, evolutionary distant COL proteins may replace one another as the predominant FT regulator. Such an evolutionary process may explain observations from tomato and morning glory suggesting that members of the COL family most closely related to CO in these organisms may not directly regulate FT (Ben-Naim et al., 2006; Hayama et al., 2007). The definition of a CORE sequence may also prove to be a boon for studies of natural variation. Regulatory variants in FT have been implicated in natural variation in flowering in several plant species, but the causal nucleotide changes have yet to be pinpointed, partly because of the absence of defined functional regulatory elements (e.g. Schwartz et al., 2009; Takahashi et al., 2009; Blackman et al., 2010). Hence, polymorphisms in CORE-like sequences of FT homologs are genetic variants ripe for evaluation by functional studies, particularly in rice where variation in the CCT domain of a CO homolog is also associated with flowering time variation (Takahashi et al., 2009).


  1. Top of page
  2. Acknowledgements
  3. References

B.K.B. is supported by NSF grant DBI-0905958. S.D.M. is supported by NIH grant GM075060.


  1. Top of page
  2. Acknowledgements
  3. References