HUA2 is required for the expression of floral repressors in Arabidopsis thaliana


  • Mark R. Doyle,

    1. Department of Biochemistry, College of Agricultural and Life Sciences, University of Wisconsin, Madison, WI 53706 1544, USA
    Search for more papers by this author
  • Colleen M. Bizzell,

    1. Department of Biochemistry, College of Agricultural and Life Sciences, University of Wisconsin, Madison, WI 53706 1544, USA
    Search for more papers by this author
  • Melissa R. Keller,

    1. Department of Biochemistry, College of Agricultural and Life Sciences, University of Wisconsin, Madison, WI 53706 1544, USA
    Search for more papers by this author
  • Scott D. Michaels,

    1. Department of Biochemistry, College of Agricultural and Life Sciences, University of Wisconsin, Madison, WI 53706 1544, USA
    Search for more papers by this author
  • Judong Song,

    1. Kumho Life and Environmental Science Laboratory, 1 Oryong-Dong, Puk-Gu, Kwangju 500 712, Korea
    2. Environmental Biotechnology National Core Research Center, Gyeongsang National University, Jinju 660 701, Korea
    Search for more papers by this author
  • Yoo-Sun Noh,

    1. Kumho Life and Environmental Science Laboratory, 1 Oryong-Dong, Puk-Gu, Kwangju 500 712, Korea
    2. Environmental Biotechnology National Core Research Center, Gyeongsang National University, Jinju 660 701, Korea
    Search for more papers by this author
  • Richard M. Amasino

    Corresponding author
    1. Department of Biochemistry, College of Agricultural and Life Sciences, University of Wisconsin, Madison, WI 53706 1544, USA
    Search for more papers by this author

For correspondence (fax +1 608 262 3453; e-mail


The HUA2 gene acts as a repressor of floral transition. Lesions in hua2 were identified through a study of natural variation and through two mutant screens. An allele of HUA2 from Landsberg erecta (Ler) contains a premature stop codon and acts as an enhancer of early flowering 4 (elf4) mutants. hua2 single mutants, in the absence of the elf4 lesion, flower earlier than wild type under short days. hua2 mutations partially suppress late flowering in FRIGIDA (FRI )-containing lines, autonomous pathway mutants, and a photoperiod pathway mutant. hua2 mutations suppress late flowering by reducing the expression of several MADS genes that act as floral repressors including FLOWERING LOCUS C (FLC ) and FLOWERING LOCUS M (FLM ).


Proper timing of the switch from vegetative to reproductive growth in flowering plants is critical for achieving maximal reproductive fitness. Most plants have an optimal time of year to flower and set seed. In species with a large geographic distribution, the mechanisms that underlie the control of flowering must be adaptable in order to adjust to factors that may be encountered in different habitats.

Two environmental factors used by plants to assess seasonal change are prolonged periods of cold temperatures and relative changes in photoperiod. The promotion of flowering in response to a prolonged period of cold is known as vernalization. Vernalization often requires several weeks of cold (0–6°C) for a maximum response (Michaels and Amasino, 2000). This helps ensure that the winter season has passed and allows for rapid flowering in the spring. Floral induction in response to lengthening days also leads to flowering in the spring. Plants with such a response, including Arabidopsis thaliana, are known as long day plants (LDPs). Short day plants (SDPs), on the other hand, flower in the autumn in response to shortening days and lengthening nights (Thomas and Vince-Prue, 1997). The response to daylength in both SDP and LDP is closely associated with the plant circadian clock.

The geographic range of Arabidopsis includes areas with vast differences in daylength and ambient temperature throughout the year. Not surprisingly, populations of Arabidopsis from different regions often display differences in the regulation of flowering. Studies of natural variation in Arabidopsis have revealed loci involved in the regulation of flowering time (Alonso-Blanco et al., 1998; Kowalski et al., 1994). Variation between two rapid-flowering accessions, Landsberg erecta (Ler) and Cape Verdi Islands (Cvi), has been studied using recombinant-inbred lines (RILs) (Alonso-Blanco et al., 1998). One of the flowering quantitative trait loci (QTL) encodes the blue light receptor CRYPTOCHROME2 (CRY2). The Cvi allele of CRY2 contains a single amino acid change that leads to a more stable protein. This change affects daylength perception as Ler plants containing an introgressed CRY2 from Cvi are daylength-insensitive (El-Din El-Assal et al., 2001). This same collection of RILs was used in a separate experiment to isolate QTL that function in the control of circadian period. The circadian clock is closely linked to the control of flowering and many of the circadian QTL found in this study co-localized with known flowering-time genes (Swarup et al., 1999).

The most striking difference in flowering time among natural populations of Arabidopsis is seen between rapid-flowering and winter-annual accessions. Winter-annual accessions are late flowering unless vernalized. Crosses between plants with these different flowering habits revealed that dominant alleles of two loci, FRIGIDA (FRI ) and FLOWERING LOCUS C (FLC ) make a major contribution to the delay in flowering in winter annuals (Burn et al., 1993; Clarke and Dean, 1994; Koornneef et al., 1994; Lee et al., 1993, 1994b). FRI acts to upregulate FLC, a MADS-box transcription factor that is sufficient to repress flowering (Michaels and Amasino, 1999; Sheldon et al., 1999). Vernalization leads to a stable decrease in FLC levels. Molecular analysis of FRI and FLC in rapid-flowering accessions has shown that many contain null mutations in FRI (Gazzani et al., 2003; Johanson et al., 2000; Le Corre et al., 2002), and some contain mutations in FLC that result in lower expression (Gazzani et al., 2003; Michaels et al., 2003).

Plants that contain an active allele of FRI have elevated levels of FLC. Flower promotion via FLC repression occurs through a number of pathways. Vernalization turns off FLC through chromatin remodeling (Bastow et al., 2004; Sung and Amasino, 2004). Another group of genes, the autonomous pathway, acts to repress FLC independently of daylength and temperature. Mutations in members of the autonomous pathway are late flowering due to elevated FLC levels (Michaels and Amasino, 2001). Some members of the autonomous pathway including FLD and FVE are required for de-acetylation of FLC chromatin (Ausin et al., 2004; He et al., 2003). Many of the other members of the autonomous pathway are genes believed to function in RNA binding and processing. FY encodes a 3′ end-processing factor (Simpson et al., 2003), and FPA, FCA, and FLK are all predicted to bind mRNA (Lim et al., 2004; MacKnight et al., 1997; Schomburg et al., 2001). The homeodomain of LUMINIDEPENDENS (LD) may act in RNA or DNA binding (Lee et al., 1994a).

The HUA2 gene plays a role in RNA processing (Chen and Meyerowitz, 1999; Cheng et al., 2003). In this study we identified HUA2 as a gene involved in the control of flowering by two independent mutant screens and by studies of natural variation. HUA2 is a repressor of flowering and appears to function by enhancing the expression of several genes that delay flowering including FLC.


The Arabidopsis accession Landsberg erecta contains an enhancer of elf4-1

The EARLY FLOWERING 4 (ELF4) gene encodes a small protein that is essential for circadian clock function (Doyle et al., 2002). elf4-1 mutants in the Ws accession are early flowering in short days (SD) (Figure 1a) and display aberrant circadian outputs. As part of a genetic analysis, crosses were made between elf4 and mutants in light perception and circadian regulation. Certain mutants were not available in Ws, and thus several crosses were made between elf4-1 and mutants in Ler. All F2 populations resulting from crosses to Ler contained elf4 mutants that flowered extremely early (Figure 1b), although the extremely early flowering plants did not correspond to the double mutants. The frequency of the extreme phenotype was approximately 1/16, suggesting that the phenotype was the result of a single recessive locus contributed by Ler. Therefore, we crossed elf4-1 to wild-type Ler. The resulting F2 population also contained extremely early flowering plants indicating that Ler contained an enhancer of elf4, which we referred to as the eel locus.

Figure 1.

hua2-5 enhances the elf4-1 phenotype.
(a) Short day-grown Ws and elf4-1.
(b) A plant with the eel phenotype and a non-eel elf4-1 plant from the same segregating F2 population.
(c) Two plants with the eel1 genotype: elf4/elf4; hua2-5/hua2-5. An extra copy of HUA2 from Ws (pMDH2Ws) has been added to the plant on the left.
(d) Schematic showing the exon/intron structure of HUA2 and the position of various mutations within the gene.

The elf4 eel phenotype was first observed in F2 plants grown in SD under high-pressure sodium lamps. Under these conditions the elf4 eel phenotype was extremely early flowering (Figure 1b). When this same F2 was grown in SD under fluorescent lamps, the elf4 eel plants flowered slightly later but were still much earlier than elf4 single mutants. This shift to later flowering is observed in wild-type plants as well. All flowering and mapping data presented below were collected from plants grown under fluorescent lights.

The eel locus encodes HUA2, a gene involved in the regulation of a floral homeotic gene

A backcross of an elf4/eel plant from the original F2 population to elf4-1 in Ws produced F2 populations that segregated 3 elf4-like:1 elf4/eel. Analysis of 600 F2 plants from such a population enabled us to narrow the eel locus to a 40-kb region that spanned two BAC clones, MYJ24 and MKD15. This region contains the HUA2 gene (Chen and Meyerowitz, 1999). An HUA2 clone from the Ws accession was able to rescue the eel phenotype (Figure 1c). To further confirm the identity of HUA2 as the elf4 enhancer, two additional alleles of HUA2 were crossed into the elf4-1 background: the original hua2-1 allele which disrupts a splice junction at the sixth intron (Chen and Meyerowitz, 1999), and hua2-4, a T-DNA allele from Columbia (Figure 1d). In the presence of elf4-1, both hua2 alleles result in an eel phenotype (Table 1).

Table 1.  Flowering phenotype of eel plants in short day (SD) conditions
 Total leaf number
  1. Values represent total leaf number ± standard deviation. All plants were grown under SD conditions (8 h light:16 h dark). Total leaf number is equal to the number of rosette leaves plus the number of cauline leaves. At least eight plants were analyzed for each entry. The elf4-1 (Ws/Col) and elf4-1 hua2-4 (Ws/Col) represent the two clear early-flowering classes that appeared in this F2. The frequency of early flowering plants, however, was less than expected suggesting that Col may contain a suppressor of elf4.

Ws38.0 ± 1.8
Ler47.3 ± 3.0
elf4-1 (Ws/Ler)16.6 ± 3.1
elf4-1 hua2-5 (eel) (Ws/Ler)8.7 ± 1.9
elf4-1 hua2-1 (Ws/Ler)6.4 ± 0.5
Col64.3 ± 3.1
elf4-1 (Ws/Col)17.2 ± 2.4
elf4-1 hua2-4 (Ws/Col)9.0 ± 1.3

The HUA2 gene was first characterized in a screen for enhancers of ag-4, a weak allele of the floral homeotic gene AGAMOUS (AG). The original hua2 allele was isolated in combination with a mutation in another gene, hua1. The enhancer of ag-4 phenotype caused by mutations in hua2 is more apparent in a hua1 hua2 ag4 mutant background. However, hua2-1 ag-4 double mutants do display petaloid stamens and slightly enlarged anthers when compared with hua2-1 and ag-4 single mutants (Chen and Meyerowitz, 1999). The HUA2 protein sequence contains an RPR domain, a motif found in proteins that function in RNA metabolism, and HUA2 affects AG pre-mRNA processing in certain genetic backgrounds (Cheng et al., 2003).

The eel lesion in HUA2 is not present in all strains of Ler

HUA2 was sequenced from both the Ws and Ler ecotypes. The sequence data revealed two polymorphisms that resulted in differences between the Ws and Ler amino acid sequences. The Ws sequence contained a 12-bp deletion relative to Ler and Col resulting in the removal of amino acids 967–970. The Ler allele of HUA2 contained a premature stop codon. This lesion is located in exon 8 and results in the removal of 280 amino acids from the C-terminal end (Figure 1d). Given the recessive nature of the Ler allele, we concluded that the premature stop codon was the cause of the eel phenotype.

To clarify further discussion, our HUA2 allele containing the premature stop codon will be referred to as hua2-5. The truncated protein in hua2-5, if it is produced, contains several motifs including a PWWP domain, the RPR domain, and four putative nuclear localization sequences. The 280 C-terminal amino acids missing in hua2-5 do not contain any recognizable protein motifs.

The hua2-1 lesion disrupts a splice junction at the sixth intron (Chen and Meyerowitz, 1999). Interestingly the screen that produced the hua2-1 mutant was carried out in a Ler genetic background. Given that we detected a stop codon in the HUA2 gene in Ler, it was surprising that hua2-1 was isolated in Ler. We sequenced exon 8 from hua2-1 and did not find the stop codon. This indicates that the parental Ler line of hua2-1 contains a polymorphism with our Ler strain containing hua2-5. The hua2-5 base change was not present in any other Arabidopsis accessions that were examined (Table 2). However, we did find examples of Ler lines that contained the hua2-5 lesion including the Ler used in the Ler × Cvi RIL population (Table 2) (Alonso-Blanco et al., 1998).

Table 2.  Genotype of various Ler lines at HUA2
  1. *Lines believed to be those described in Koornneef et al. (1991) Mol. Gen. Genet.229, 57–66.

  2. Individual recombinant-inbred lines from the Ler × Cvi population that are homozygous Ler at HUA2 (Alonso-Blanco et al., 1998).

  3. Genotype of various Ler lines at the HUA2 locus.

Ler lines containing hua2-5 (premature stop codon)
 gi-3 (fb)*, fd*, ft*, co-3*, fy*, fwa*, fpa*, CVL135, CVL40
Ler lines containing wild-type HUA2 (no hua2-5 lesion)
 fca*, fve*, clf-2 (Goodrich et al., 1997) ag-4 (Chen and Meyerowitz, 1999)
Accessions containing wild-type HUA2 (no hua2-5 lesion)
 Cvi, Col, Ws, Ema-1, Seattle-0, Limeport, H55, Petergof, Litva, Oy-1, Bla-2, Nd-1, Ber, Condara, Cnt-1, Di-G, ENF, Li-5, Je54, M3385S, Est, Berkeley, Da(1)-12, Abd-0, Gr3, Co, Shahdara, Wei-0, LIN,

hua2 mutants flower early in the absence of elf4-1 mutations

hua2-5 was identified due to its ability to enhance the elf4-1 phenotype. Thus, it was of interest to see whether this effect on flowering time was elf4-specific or whether mutations in hua2 affected flowering in the absence of elf4. A Col insertion allele, hua2-4, flowered early under both LD and SD relative to wild type. Both hua2-1 and hua2-5 in Ler also flowered early in SD but flowered similar to wild type in LD (Figure 2a). hua2-1 flowered earlier than hua2-5 suggesting that hua2-5 is a weaker allele. hua2-1 plants were crossed to both hua2-5 and Ler. F1 plants from hua2-1 × hua2-5 flowered with an intermediate number of leaves. F1 plants from the hua2-1 × Ler flowered similar to Ler (Figure 2a). Thus, hua2-1 is recessive when paired with a wild-type allele in Ler. Other strong hua2 alleles in Ws and Col exhibit a heterozygous phenotype (see below). The lack of a heterozygous phenotype with hua2-1 in Ler may be due to the Ler genetic background or a unique feature of the hua2-1 allele.

Figure 2.

Effect of hua2 mutants on flowering in long day (LD) and short day (SD) conditions.
(a) All hua2 alleles tested flower earlier than the respected wild type: hua2-4 (Col), hua2-1 and hua2-5 (Ler), hua2-6 (Ws). White bars represent the total leaf number (rosette + cauline) in plants grown in LD. Black bars represent the total leaf number in plants grown in SD. Error bars represent ±standard deviation.
(b) Distribution of a population that segregates for hua2-6 grown in SD. Black, gray, and white bars represent hua2-6 homozygotes, heterozygotes, and wild type, respectively.

In addition to isolating hua2 as an elf4 enhancer, a HUA2 mutant, hua2-6, was recovered in a screen for early-flowering mutants in the Ws background. Plants homozygous for the hua2-6 mutation flowered earlier than wild type under both LD and SD, but the difference was more pronounced under SD. Plants heterozygous for hua2-6 flower with an intermediate number of leaves when compared with hua2-6 homozygous and wild-type plants (Figure 2b).

HUA2 plays a role in the regulation of the floral repressor FLC

Columbia plants containing an active FRI gene from the San Feliu-2 ecotype (FRI-Col) show a delay in flowering unless vernalized (Lee et al., 1994a). In addition to our identification of one hua2 allele (hua2-5) as an enhancer of elf4 and another (hua2-6) as a mutant that flowers early in SD, two additional hua2 alleles were uncovered as suppressors of FRI-mediated late flowering. hua2-2 contains a T-DNA in exon 8 (Figure 1d). hua2-3 was recovered from a fast neutron population and contains an insertion in exon 3. A wild-type copy of the HUA2 gene restored FRI-like late flowering when introduced into the hua2-2 and hua2-3 mutant backgrounds (data not shown).

hua2-2 and hua2-3 vary in their ability to suppress FRI-mediated late flowering. hua2-3 suppressed FRI to a greater extent than hua2-2 (Figure 3a). The hua2-3 lesion occurs near the N-terminal end of the protein whereas hua2-2 contains a mutation in exon 8 and may produce a truncated but partially functional protein product. The relative strength of the hua2-3 and hua2-2 lesions with respect to FRI suppression is reminiscent of the relative strength of hua2-1 and hua2-5 in Ler, which also truncate the protein at different places.

Figure 3.

Suppression of FLC-mediated late flowering by mutations in HUA2.
(a) hua2 mutants suppress late flowering in FRI-containing plants as well as autonomous pathway mutants.
(b) hua2 mutations reduce the steady-state level of FLC mRNA in an FRI background. The blot was probed first for FLC and then with 18S rRNA as a loading control.
(c) hua2 mutations in Ler reduce steady-state levels of FLC mRNA. Ubiquitin (UBQ) was used as a control.
(d) The hua2-5 mutation suppresses the effect of FRI in a Ler background. Error bars represent ±standard deviation. Plants in (a) and (d) are in the Col and Ler backgrounds, respectively. All plants were grown under long days.

Plants containing FRI are late flowering due to elevated levels of FLC (Michaels and Amasino, 2001). The fact that hua2 mutations partially suppress FRI-mediated late flowering suggests that hua2 mutations may alter FLC levels. RNA blot analysis revealed that FLC levels are in fact reduced in FRI plants containing hua2 mutations (Figure 3b). There was also reduced expression of FLC in Ler lines containing hua2 lesions. In Ler, FLC levels are low due both to the lack of FRI activity (Johanson et al., 2000) and to the insertion of a transposable element in the first intron of FLC (Michaels et al., 2003). Thus, RT-PCR was used to determine the effect of hua2 mutations on FLC levels in Ler. FLC was reduced in hua2-1 compared with wild type. Consistent with the flowering phenotype, hua2-5 mutants showed an intermediate level of FLC (Figure 3c).

Genes in the autonomous pathway act to repress FLC expression. Mutations in any member of this pathway result in late flowering due to elevated levels of FLC (Michaels and Amasino, 2001). Mutants in two members of the autonomous pathway, fpa and ld, were crossed to the hua2-2 mutants to evaluate whether hua2 could suppress the late flowering effect of autonomous-pathway mutants. In both cases, late flowering was partially suppressed (Figure 3a).

The effect of hua2-5 on FRI-mediated late flowering in Ler was tested genetically. A previously reported Ler line containing an introgressed copy of FRI is also homozygous for the hua2-5 lesion (Lee et al., 1994a). This line was crossed to the strain of Ler with a lesion-free HUA2 gene and to hua2-5. Due to the dominant nature of FRI, F1 plants could be evaluated for flowering time. Plants containing a wild-type allele of HUA2 (FRI/fri; HUA2/hua2-5) flowered later than plants homozygous for hua2-5 (FRI fri; hua2-5 hua2-5) (Figure 3d).

HUA2 affects flowering independently of FLC

Mutations in hua2 lead to a decrease in FLC, which likely contributes to the early flowering phenotype of hua2 mutants. Although flc null mutants flower slightly earlier than wild type in SD (Michaels and Amasino, 2001), hua2 mutants flower even earlier than flc null mutants (Figure 4a). In addition, hua2 mutations suppress the late flowering of co mutants in a dose-dependent manner (Figure 4b). Mutations in flc have no effect on the co mutant phenotype (Michaels and Amasino, 2001). Thus, the overall effect of HUA2 on flowering time cannot be entirely explained through its interaction with FLC. HUA2 must also interact with other components of flowering time regulation.

Figure 4.

Interaction between HUA2 and floral regulators other than FLC.
(a) hua2 mutants flower earlier than an flc null mutant in both long day and short day plants.
(b) hua2 mutations suppress the effect of co mutations.
(c) hua2 mutations in Ler reduce the mRNA levels of FLM, SVP, and MAF2. The two bands seen with MAF2 represent two splice variants. UBQ was used as a control. Bands for SOC1 represent basal levels of gene expression. The tissue was harvested prior to the increase in SOC1 seen at the time of floral transition.

An additional candidate for HUA2 interaction is FLOWERING LOCUS M (FLM/MAF1). FLM is a MADs box gene that is closely related to FLC both at the amino acid level and in exon/intron gene structure (Ratcliffe et al., 2001; Scortecci et al., 2001). In addition, FLM, like FLC, acts as a negative regulator of the floral transition. Like hua2 mutants, flm mutants have a pronounced early-flowering phenotype in SD compared with wild type, and flm mutations, like hua2 mutations, partially suppress co (Scortecci et al., 2003). FLM RNA levels were examined in Ler, hua2-1, and hua2-5 seedlings grown under SD. Like FLC, FLM was reduced in the hua2-1 mutant compared with wild type. hua2-5 mutants displayed an intermediate level of FLM expression (Figure 4c). Thus, early flowering in hua2 mutants is due, at least in part, to the combined reduction in FLC and FLM expression. Expression of a third member of the FLC clade, MAF2, is also reduced by hua2 mutations in a manner similar to FLC and FLM (Figure 4c).

In addition to FLC, FLM, and MAF2, the expression of two additional MADS genes involved in flowering time regulation were studied. SHORT VEGETATIVE PHASE (SVP ) is a repressor of flowering (Hartmann et al., 2000). Like mutants in flm, svp mutants are early flowering, and the early-flowering phenotype is most apparent in SD. hua2-5 mutations did not appear to have an effect on SVP levels. SVP levels were, however, reduced in a hua2-1 background although the degree of reduction was less than that seen for genes in the FLC clade (Figure 4c). SUPRESSOR OF OVEREXPRESSION OF CONSTANS 1 (SOC1) is a promoter of flowering (Borner et al., 2000; Lee et al., 2000; Onouchi et al., 2000). Mutations in hua2 did not alter the expression of SOC1 (Figure 4c).


The switch from vegetative growth to flowering results from the accurate perception of specific environmental signals such as photoperiod and prolonged periods of cold temperature. The environmental parameters that define the optimal time to flower can vary substantially throughout the geographic range of a species, especially one as vast as that of Arabidopsis. Arabidopsis accessions collected from a variety of habitats provide a wealth of variation that can be used for gene discovery.

Using natural variation as a tool for gene discovery has revealed several genes involved in regulating flowering time. For example, crosses between winter-annual and rapid-flowering accessions led to the identification of FRI and FLC, two loci required for the winter-annual habit (Michaels and Amasino, 2000). Although the loss of FRI results in a major switch in reproductive strategy (Johanson et al., 2000), most polymorphisms in flowering-time genes among natural populations are likely to result in a far subtler effect.

Here we identified an elf4-1 enhancer locus as HUA2 in a cross between the elf4 mutant in Ws and one strain of Ler. The ability to isolate the hua2-5 lesion was facilitated by evaluating natural variation in the elf4 mutant background because hua2-5 caused extremely early flowering when combined with elf4-1. However, hua2-5 mutants alone flower earlier than wild type, and thus, the effect of the hua2-5 lesion on flowering in the absence of elf4 could be detectable by standard QTL analyses. In fact, the Ler strain used to create the Ler × Cvi RILs contains hua2-5, and a flowering QTL identified from these lines, FLG, co-localizes with HUA2 (Alonso-Blanco et al., 1998).

In an elf4 background, hua2-5 homozygous mutants may be more apparent because, by attenuating the photoperiod response, the elf4 mutation decreases the overall number of genes contributing to floral regulation. None of the genes affected by HUA2 are major factors in the regulation of circadian rhythms and thus the eel phenotype results from the attenuation of ELF4-independent pathways. A hua2 mutant may be more pronounced in an elf4-1 background because an ELF4-dependent pathway can no longer compensate for the lack of HUA2. Therefore, studies of natural variation that begin with a mutant background could serve to enhance the effect of QTL or reveal additional loci that might not be apparent in wild-type backgrounds. On the other hand, such an approach might result in the loss of loci detectable in a cross between wild types. If, for example, genetic differences between Ws and Ler affect an ELF4-dependent process, one might expect the effect of this polymorphism to be masked in an elf4 background.

Whether hua2-5 resulted from natural variation or an induced mutation is not known. The erecta mutation is the result of a gamma ray-induced mutagenesis and thus the hua2-5 lesion could have been induced. Alternatively the original Landsberg strain was not isogenic (Rédei, 1992) and could have contained both HUA2 alleles. If the original erecta strain was heterozygous for the hua2-5 mutation, the two alleles of HUA2 could have been fixed in different single seed lineages. It is also possible that the hua2-5 lesion has arisen more than once and was inadvertently selected due to its subtle affect on the regulation of flowering time. To date, we have not found other accessions of Arabidopsis that contain the hua2-5 lesion, although we only analyzed a subset of available accessions. However, natural variation at a different site within the HUA2 gene has been uncovered in another study (V. Grbic, University of Western Ontario, Ontario, Canada, personal communication) suggesting that a HUA2 variant may have been a target for natural selection at least one other time.

The HUA2 gene was originally isolated in a screen for enhancers of a weak allele of AG, ag-4 (Chen and Meyerowitz, 1999). The HUA1 gene, which encodes a putative RNA-binding protein, was also isolated in this screen. hua1-1 hua2-1 ag-4 triple mutants resemble strong ag alleles, but hua2-1 alone in an ag-4 background has only a slight enhancing effect. In the absence of ag-4, hua2-1 mutants do not alter floral morphology. Thus, the only phenotype of a hua2 single mutant reported to date is the alteration of flowering time.

The AG gene contains a 3-kb intron. In wild type improperly spliced AG transcripts containing regions of this large intron can be seen at low levels (Cheng et al., 2003). In hua1-1 hua2-1 double mutants these aberrant transcripts accumulate to higher levels than in wild type. Thus, HUA1 and HUA2 appear to have a role in processing AG pre-mRNA. The gene structure of AG has several similarities with the floral repressor FLC and other members of the FLC clade. These genes encode type II MADS-box transcription factors and contain one large intron. However, unlike the case with AG, FLC transcripts containing regions of the large intron are not found in wild type or in the hua1 hua2 double mutant background (Cheng et al., 2003). Thus, there is no evidence that HUA1 and HUA2 affect FLC splicing, although it possible that aberrantly spliced FLC transcripts are turned over too rapidly to be detectable.

In the hua1 hua2 double mutant background, both FLC and AG levels are reduced, but FLC is reduced to a much greater extent (Cheng et al., 2003). We show that hua2 single mutants have a large effect on FLC mRNA levels and thus the decrease in FLC levels previously reported in the hua1 hua2 double mutant is due primarily to hua2. In contrast, the hua2 single mutant does not appear to affect AG mRNA levels (Cheng et al., 2003). In addition, hua1 mutants in combination with elf4 do not result in an eel phenotype indicating that HUA1 does not have a role similar to that of HUA2 in the control of flowering time (M.R. Doyle and R.M. Amasino, unpublished data). Thus, HUA2 may affect AG and FLC expression by entirely different mechanisms.

Downregulation of FLC only partially explains the early-flowering phenotype in hua2 mutants. hua2-4, a T-DNA mutant in Columbia, flowers earlier than an flc null mutant indicating that mutations in hua2 do more than simply reduce FLC levels. FLM is a repressor of flowering that is closely related to FLC in amino acid sequence and has a similar gene structure containing a large first intron (Ratcliffe et al., 2001; Scortecci et al., 2001). As with FLC, hua2 mutations also result in decreased FLM expression. This is also the case with MAF2, another gene in the FLC clade, although the differences in MAF2 expression between different hua2 alleles appear subtler than those seen for FLC and FLM. Another MADs box gene that represses flowering, SVP, is not affected in a hua2-5 background, but does appear to be downregulated in the hua2-1 background. Whether the early flowering of hua2 mutants might result from effects on genes in addition to those identified din this study remains to be determined.

The MADS genes FLC, FLM/MAF1, SVP, and MAF2 all act as repressors of the floral transition; however, these genes act in different pathways of floral regulation. FLC is involved primarily in the vernalization response (Michaels and Amasino, 1999; Sheldon et al., 1999). flc mutants are early flowering in non-inductive SD, but the effect is not as great as that seen in flm and svp mutants (Hartmann et al., 2000; Michaels and Amasino, 2001; Scortecci et al., 2001). MAF2 represses flowering in response to short periods of cold temperatures (Ratcliffe et al., 2003). Because hua2 mutations affect the expression of all these genes, HUA2 either interacts with these genes directly or interacts with an unidentified factor or factors that regulate this group of transcription factors.

The genes that appear to be affected the most by HUA2 (FLC, FLM, and MAF2) are closely related to one another. Thus, although the specific function of these genes in floral regulation has diverged over time, they still share some of the same regulatory components such as HUA2 and members of a PAF-complex that mediate gene activation (He et al., 2004). The related primary amino acid sequence and gene structure of FLC, FLM, and MAF2 indicate that these genes arose from a common ancestral gene that most likely required HUA2 for expression. Thus, these genes have maintained some regulatory pathways in common despite their functional divergence into different flowering pathways, and mutations in common regulators of these genes such as hua2 will have a broad-flowering phenotype.

Several other genes have been described that affect flowering via both FLC-dependent and FLC-independent pathways. These genes include a putative RNA-binding protein, EARLY FLOWERING 5 (ELF5) (Noh et al., 2004); the putative chromatin remodeling proteins PHOTOPERIOD-INDEPENDENT EARLY FLOWERING1 (PIE1) and EARLY FLOWERING IN SHORT DAYS (EFS) (Noh and Amasino, 2003; Soppe et al., 1999; S.D. Michaels and R.M. Amasino, unpublished data); and a SUMO protease, EARLY IN SHORT DAYS 4 (ESD4) (Murtas et al., 2003; Reeves et al., 2002). It is of interest to determine whether the FLC-independent role of these genes in flowering-time regulation is similar to that of HUA2. Like HUA2, these genes may affect the expression of several floral suppressors including FLM, MAF2, and SVP.

Experimental procedures

Plant materials and growth conditions

The elf4-1 mutant is in the Ws background and has been previously described (Doyle et al., 2002). The hua2-1 strain and the strain referred to as Ler have been described previously (Chen and Meyerowitz, 1999). Other accessions were obtained from the Arabidopsis stock center ( was isolated from the Salk Institute Genome Analysis Laboratory (SIGnAL) T-DNA collection (reference number:SALK_032281; The hua2-3 fast neutron allele and the hua2-2 T-DNA allele were isolated from mutagenized populations in a Columbia background containing FRI. hua2 mapping and flowering time analyses were conducted on plants grown in either LD or SD, 16 h light:8 h dark and 8 h light:16 h dark, respectively, under cool-white fluorescent light unless mentioned otherwise. Flowering time data are presented as total leaf number (rosette + cauline) as no difference in the ratio of rosette to cauline leaves was observed in any line.

Cloning and genotyping of HUA2

HUA2 was cloned from the Ws ecotype using PCR. The clone consisted of 2 kb of upstream sequence, the HUA2 genomic sequence, and 0.8 kb of downstream sequence. This sequence was cloned in two fragments using an internal BamHI site. Primers used to amplify these two fragments were as follows: HUA2L1: 5′-AAAAGCTTCGCTATATGCCACTGCTTTG-3′, HUA2R1: 5′-CTAATTTGGGGAAGCAAGGA-3′, HUA2L2: 5′-CTTTGGGCGATGAGGATTC-3′, HUA2R2: 5′-AAACTCGAGGCAGCGAGACATAACTT-3′. Restriction sites existing at the 5′ends of HUA2L1 and HUA2R1 were used in conjunction with the native BamHI site to clone both fragments into the pPZP221B binary vector (Kang et al., 2001).

The premature stop codon in hua-5 was detected using the following oligo sequences: HUALerL: 5′-CTTCACAATCATTAACAACTCAG-3′ and HUA LerR: 5′-TGCTGCATAGATCCTGGGTA-3′. These primers produced a 110-bp fragment spanning the region containing the polymorphism for which the sequence was determined.

Expression analysis

Total RNA was isolated from 7-day-old seedlings using TRI Reagent (Sigma, St Louis, MO, USA). Complementary DNA was synthesized using 2 μg of total RNA, an oligo dT primer and Superscript II reverse transcriptase (Gibco, Carlsbad, CA, USA).FLC, FLM, MAF2, SVP, SOC1, and UBQ10 were detected by RT-PCR as previously described (Doyle et al., 2002; Michaels et al., 2003; Ratcliffe et al., 2003; Scortecci et al., 2001). RNA blots contained 15 μg of total RNA per lane on a denaturing formaldehyde gel (1% agarose). RNA was transferred to a nylon membrane and probed with FLC.


We thank Xuemei Chen for supplying Ler, hua2-1, and hua1 hua2 mutant seed and the ABRC at Ohio State for providing a T-DNA insertion line, hua2-4. This work was supported by the College of Agricultural and Life Sciences and the Graduate School of the University of Wisconsin, and by National Science Foundation grants 0133663 and 0209786 to R.M.A. Work in Y.S.N.'s laboratory was supported by the Molecular and Cellular BioDiscovery Research Program (M1-0311-09-0000) grant from the Ministry of Science and Technology of Korea and by a grant from the Korea Science and Engineering Foundation to the Environmental Biotechnology National Core Research Center (R15-2003-012-01002-0).