A putative CCAAT-binding transcription factor is a regulator of flowering timing in Arabidopsis

Flowering at the appropriate time of year is essential for successful reproduction in plants. We found that HAP3b in Arabidopsis ( Arabidopsis thaliana ), a putative CCAAT-binding transcription factor gene, is involved in controlling flowering time. Overexpression of HAP3b promotes early flowering while hap3b , a null mutant of HAP3b , is delayed in flowering under a long-day photoperiod. Under short-day conditions, however, hap3b did not show a delayed flowering compared to wild-type based on the leaf number, suggesting that HAP3b may normally be involved in the photoperiod-regulated flowering pathway. Mutant hap3b plants showed earlier flowering upon gibberellic acid or vernalization treatment, which means that HAP3b is not involved in flowering promoted by GA or vernalization. Further transcript profiling and gene expression analysis suggests that HAP3b can promote flowering by enhancing expression of key flowering time genes such as FT and SOC1 . Our results provide strong evidence supporting a role of HAP3b in regulating flowering in plants grown under long day conditions. plants (wt) or the normalized transcript level of overexpression plants ( P actin :HAP3b ) by that of C1 ( P actin ) plants. The data are means ± SE of three independent experiments. The pattern was reproducible in each experiment.


Introduction
Flowering time in plants is controlled by environmental stimuli such as day length (photoperiod pathway), light quality, exposure to low temperatures (vernalization pathway), and internal factors such as plant age or stage of development (autonomous and gibberellic acid (GA) pathways). These different pathways converge to regulate a small set of genes, such as FT (FLOWERING LOCUS T, a small protein with similarity to RAF-kinase inhibitor) and SOC1 (SUPPRESSOR OF OVEREXPRESSION OF CONSTANS 1, a MADS transcription factor). For example, the photoperiod pathway promotes flowering through CO (CONSTANS, a zinc finger transcription factor) to upregulate FT and SOC1. However, a similar upregulation of FT and SOC1 through the vernalization and the autonomous pathways occurs through a mechanism that suppresses a floral suppressor, FLC (FLOWERING LOCUS C, a MADS transcription factor) (reviewed by Mouradov et al., 2002;Putterill et al., 2004;Amasino, 2005).
Recent studies indicated that several members in a large transcription factor family called HAP are involved in regulating flowering timing. HAP in plants encodes a putative CCAAT-binding transcription factor (CBF) similar to Heme Activator Protein (HAP) or nuclear factor-Y (NF-Y) in yeast and vertebrate (Lotan et al., 1998;Mantovani, 1999). In yeast and mammalian systems, it is known that HAPs form a complex and regulate gene expression. The complex forms through an initial interaction between HAP3 and HAP5, which then allows the formation of a heterotrimer with HAP2. The heterotrimer then binds to a CCAAT element with very high specificity and affinity (Romier et al., 2003). There is seemingly only one gene for each HAP factor in yeast and animals (Mantovani, 1999). In plants, however, these factors are all encoded in a gene family (Edwards et al., 1998;Yang et al., 2005). There are at least 10 annotated members in each family in Arabidopsis and also multiple members in each HAP family in rice (Kwong et al., 2003;Yang et al., 2005). To distinguish different members in the same family in Arabidopsis, each member is labeled alphabetically. For instance, in the HAP3 family, At2g38880 is named as HAP3a, and At5g47640 as HAP3b (Wenkel et al., 2006).

Several HAP members have been studied in plants. LEC1 encodes a HAP3 in
Arabidopsis. Mutant lec1 is defective in embryo development during the seed development stage, causing a leafy cotyledon phenotype. Ectopically over-expressing LEC1 can induce embryo development in vegetative cells (Lotan et al., 1998). Recently, another HAP3 (L1L) which is closely related to LEC1, was also found to be involved in embryo development (Kwong et al., 2003). In rice, an OsHAP3 is involved in chloroplast biogenesis. Suppression of gene expression of OsHAP3 using RNAi reduced expression of the gene and affected chlorophyll and chloroplast development (Miyoshi et al., 2003).
In regulating flowering timing, Ben-Naim et al. reported that overexpression of a tomato HAP5a in Arabidopsis caused early flowering (Ben-Naim et al., 2006). However, overexpression of Arabidopsis HAP3a and HAP5a delayed flowering (Wenkel et al., 2006). These conflicting results from overexpression hinder a clear conclusion for the roles of these HAPs in controlling flowering timing due to lack of supporting evidence from loss-of-function mutants. In this study, we provide evidence from both loss-offunction mutant and overexpression plants to support the role of HAP3b in controlling flowering. Up-regulation of HAP3b promotes early flowering, whereas a hap3b knockout results in delayed flowering in a long day photoperiod.

Altered flowering timing in mutant hap3b and HAP3b-overexpression plants
A genetic approach was taken to examine the role of a group of HAP in plant growth/development and response to stress. Among the insertional mutants identified from the SALK T-DNA insertion collection (http://signal.salk.edu), an insertion mutant for HAP3b showed delayed flowering phenotype compared to its wild-type (wt) plants grown at a long day (16h/8h light/dark) photoperiod (Figs. 1A and 1D). The mutant plants developed on average about four more leaves than wild-type plants before flowering (i.e., about a 33% delay which equals approximately 7 days). The hap3b mutant (SALK_025666) has a T-DNA insertion at 9 bp after the first ATG (Fig. 1B) and no full-length transcript was detected using RT-PCR, suggesting a loss of function mutation (Fig. 1C). A null mutation was further confirmed by the microarray data (see below), which showed no evidence for the accumulation of a truncated HAP3b transcript.
To confirm that the mutant phenotype was not an artifact of a second site mutation, we set up a complementation test by expressing a wild-type copy of the HAP3b cDNA under the control of CaMV 35S promoter. When the hap3b mutant was transformed with the P 35S :HAP3b-GFP vector, the delayed flowering phenotype was reversed, indicating that HAP3b-GFP fusion protein was functional and capable of rescuing the loss-of-function hap3b mutant (Supplemental Fig. 1).
Not only did the HAP3b-GFP overexpression lines show a reversal of the hap3b mutant phenotype, they also provided evidence that the up-regulation of the HAP3b gene could promote premature flowering (Supplemental Fig. 1). An overexpression of HAP3b in wild-type plants promotes early flowering even more. As shown in Figure 1D The predicted HAP3b protein (191 amino acid) consists of three domains (Lee et al., 2003). The central domain, comprised of more than 90 amino acids, showed significant sequence identity with HAP3 or subunit B of NF-Y in yeast and rat that is required for DNA binding and interactions with other HAP proteins (Lee et al., 2003). A HAP3b-GFP fusion protein was predominately localized in nuclei of leaf (Fig. 2a) and root ( Fig. 2b) cells, consistent with HAP3b's predicted role as a transcription factor.
Using a uidA reporter gene (encoding the reporter enzyme glucuronidase or GUS) fused to the predicted promoter region from HAP3b, we observed that the HAP3b promoter is active in leaves, vascular tissues, flower stem, cauline leaves and flowers, which support the information in public databases (the Genevestigator database). In addition, we also revealed some more detailed expression patterns, such as in leaf trichome, filaments, and transmitting tissues in the style (Supplemental Fig. 2A-E).
Interestingly, HAP3b expression is highly upregulated by salt and osmotic (mannitol) stress in both leaves and roots of Arabidopsis 3h after treatment (Kreps et al., 2002). The upregulation of HAP3b by stress is supported by the information in public databases where HAP3b transcript levels are also upregulated by drought, heat and abscissic acid (ABA) treatment in addition to salt and an osmotic stress (Supplemental Table I, the TAIR gene expression database, http://www.arabidopsis.org). Nutrient deficiency and UV treatments also moderately upregulated the transcript level of HAP3b (data not shown, the Genevestigator database, https://www.genevestigator.ethz.ch/at/index.php).

Effect of photoperiod, GA and vernalization on flowering timing of hap3b
Many genes are known to regulate flowering timing through their activity in four major pathways, i.e. photoperiod, vernalization, GA, and autonomous pathways. In order to understand whether HAP3b is related to these known pathways, wild-type, hap3b, and A function for HAP3b in the vernalization pathway was also excluded, since the hap3b mutant and HAP3b-overexpression plants showed earlier flowering after vernalization treatment (Fig. 3B). Wild-type plants did not show significant response to vernalization treatment, while C1 showed significant reduction in leaf number by vernalization. That vernalization treatment has no effect on flowering timing in Columbia wild-type was also reported before (Lim et al., 2004).

Transcript profiling of hap3b and HAP3b-overexpression plants using microarray
To identify potential candidate genes that are affected by HAP3b, a microarray experiment using the Arabidopsis whole genome array (Affymetrix ATH1 chip) was carried out with hap3b knockout and HAP3b-overexpression plants. The arrays were used here as a discovery tool, with significant changes independently confirmed by quantitative RT-PCR. To identify potentially important changes, we grouped genes that show an opposite response (at least 25% change in gene expression level based on mean values of signal intensity) in hap3b and HAP3b-overexpression plants in comparison with wild-type plants. We identified 15 candidate genes that were downregulated in hap3b but upregulated in overexpression plants ( Table I) Table II, and are either expressed at very low levels (signal absent or marginal) or showed no consistent opposite pattern (supplemental Table II) in hap3b and the overexpression plants.
Also in the list (Table I) are a putative cell wall protein gene (At2g20870), a putative cytochrome P450 gene (At3g10570) and a GDSL-motif lipase/hydrolase family protein gene (At5g33370) which all showed expression predominantly in floral organs.
A vegetative storage protein 1 gene (At5g24780), a jacalin lectin family protein gene (At2g39330) and a UDP-glucose 4-epimerase gene (At1g12780) were expressed at the highest level in floral organs as well as in stem apex or cauline leaves (the Genevestigator database). These results suggest that the majority of the genes affected by HAP3b in the list are involved in reproductive growth.
Several major flowering genes were selected for quantitative PCR analysis. SOC1 was upregulated in overexpression plants and downregulated in hap3b, confirming the expression pattern in the array analysis. FT which was not detected on the array in wildtype and mutant plants was detected by qPCR and showed the same pattern as SOC1 ( Fig. 4). This supports a model in which HAP3b normally promotes flowering through a pathway involving the up-regulation of SOC1 and FT. Expression levels of other major flowering-related genes, TOC1 (TIMING OF CAB1), CO and FLC, were not significantly affected (supplemental Fig. 3).

Discussion
In this study we provide genetic evidence for the function of HAP3b, which encodes a CCAAT-binding transcription factor, in controlling flower timing.

HAP3b regulates flower timing through a photoperiod pathway
Evidence presented here from hap3b mutants and HAP3b-overexpression clearly shows that HAP3b contributes to the regulation of flower timing under long day photoperiod conditions. We found no evidence to link HAP3b to flower timing under short day conditions. Similar long day specific phenotypes have also been observed for co and gi mutants, which are the key players in the photoperiod pathway. Since hap3b

Model for HAP3b Mode of Action
In yeast and animal systems, HAPs form a heterotrimer for transcription activation. Wenkel et al. (2006) showed that HAP3a and HAP5a in Arabidopsis were able to interact in vivo. Thus, it is very possible that HAPs in plants work in a similar way in animal and yeast, i.e. by forming a HAP complex during transcription activation.
However, a binding of plant HAP complex to the CCAAT element has not been demonstrated. Plant promoters don't have a consistent location for CCAAT elements (Lotan et al., 1998;Wenkel et al., 2006). In animals, HAPs regulated genes usually have a CCAAT-box located at the -60 to -100 location in the promoter (Mantovani, 1998).
Our analysis of the top ~20 HAP3b-affected genes from the array experiment also showed a random pattern of CCAAT distribution in the promoters, even though a majority of these genes have one or two CCAAT boxes within -1 to -500 bp (data not shown). Thus, HAPs in plants may bind to an element or sequence that differs from CCAAT.
Overexpression of HAP3a and HAP3b, two members in the same family, resulted in opposite results in flowering timing control, raising an interesting question of how these different HAP3s achieve an opposite effect. One of the possibilities is that HAP3a and HAP3b form different complexes with their own HAP5 and HAP2 so that the complexes function differently. Alternatively, HAP3a and HAP3b form of a complex involving the same HAP5 and HAP2, since they both can interact with CO and COL in the yeast-two hybrid system. In this case, a competition of HAP3a with HAP3b for binding CO will decrease the number of CO-HAP3b-containing complexes and delay flowering. Thus, a fine balance of HAP3a and HAP3b will determine the flowering timing in plants, which may represent a novel mechanism in regulating flowering timing in the photoperiod pathway.
In conclusion, our results provide strong genetic evidence supporting a model in which  Fig. 2, also see the Genevestigator database), to CO (An et al., 2004). Since CO is known to activate FT expression mainly in the leaf phloem companion cells (Takada and Goto, 2003;An et al., 2004), the co-localization of CO and HAP3b may be required for the interaction of these two proteins which will further activate expression of FT or other genes identified in Table 1 from the array analysis. It needs to be mentioned that some genes such as At2g20870 and At3g10570 which are downregulated in hap3b in Table 1 are also downregulated in ft mutants, while others such as At2g39330 and At1g12780 which are downregulated in hap3b mutants are not affected in ft mutants (the Genevestigator database). Thus, some of the genes affected by the hap3b mutation in Table 1 are potentially regulated through FT while some may be regulated by HAP3b in a different mechanism. Future work also is needed to address whether or which HAP2/HAP5 are involved in forming a HAP complex with HAP3b to regulate flowering in vivo. Since several environmental stresses up-regulate HAP3b, these results raise a possibility that HAP3b provides a pathway by which abiotic stress response pathways may help promote early flowering. Insertion mutant information was obtained from the SIGnAL website at http://signal.salk.edu and verified by PCR and RT-PCR methods. Only SALK_025666 was confirmed as a true insertional mutant.

Plasmid Constructs and Plant Transformation
The promoter (1.5 kb before 5'UTR) or the transcribed portion including 5'UTR and 3'UTR of HAP3b were PCR-amplified from Arabidopsis genomic DNA separately and cloned into the Zero Blunt PCR Cloning vector (Invitrogen, Carlsbad, CA). All PCR amplifications were carried out with high-fidelity DNA polymerase (PfuUltra DNA polymerase, Stratagene, La Jolla, CA). The sequence of the cloned promoter or

GUS Staining and GFP Localization:
T1 and T2 transgenic plants carrying the P HAP3b :GUS construct were assayed for GUS color reaction following a method described by Stangeland and Salehian (2002).
For HAP3b-GFP localization, roots and leaves of young seedlings were used for examination using a laser scanning confocal microscope (Bio-Rad MRC 1024, BioRad, CA).

Genomic DNA Extraction and T-DNA Insertional Mutant Screening:
Homozygous T-DNA insertional mutants were identified by following the protocol described at the SALK Insertion Sequence Database (http://signal.salk.edu/tabout.html) using a PCR method. Leaf tissues of soil-grown seedlings were first collected from individual plants. Genomic DNA was extracted using a quick CTAB method (Rogers and Bendich, 1988) and used for PCR reactions with the primers recommended in the SALK protocol.

Microarray and Gene Expression:
Seeds of wild-type, hap3b mutant and HAP3b-overexpression transgenic plants were germinated in the same flat containing well-watered potting mix. Leaves of 18-d old plants grown in soil under a 16h/8h light/dark photoperiod were harvested 6 h after lights were on. RNA was extracted using Tri-reagent (Ambion, Austin, TX). The array labeling, hybridization, scanning and initial data processing were conducted as a service by the Center of Integrated BioSystems at Utah State University. A total of five arrays (Affymetrix ATH1 chip, Cat#:900385) were processed: two chips for wild-type plants, two for mutant plants (hap3b) and one for overexpression plants (P actin :HAP3b). RNA used for the chip experiment was from five independent biological samples from two independent experiments. Each sample represented a collection of leaves from 12 plants.
To confirm expression of selected genes from the microarray experiments using a quantitative PCR, seeds of wild-type, hap3b mutant, HAP3b-overexpression transgenic plants and overexpression control plants (C1= P actin :GUS) were germinated in a single MS-Phytagel plate. Fifteen-day-old seedlings were harvested for RNA extraction. A quantitative PCR method was performed by following a method described by Wang et al.

Supplemental Materials
Supplemental        samples (or arrays), and showed at least 25% change in transcript level in both hap3b and overexpression plants. " A " indicates the signal on arrays is labeled as "Absent".   Long day pathway