Challenges in studies on flowering time: interfaces between phenological research and the molecular network of flowering genes

Abstract

Flowering time is a well-studied subject in ecology, evolution and molecular biology. Long-term phenological studies have shown relationships between flowering time and environmental and endogenous factors in many species. In contrast, molecular studies using model plants have revealed a complex regulatory network of flowering. We propose that flowering would be a model trait for the integrated study of ecology, evolution and molecular biology. We introduce briefly the flowering regulatory pathways of Arabidopsis thaliana, followed by molecular techniques such as transgenic manipulation, quantitative real-time PCR and detection of differentially expressed genes that could facilitate the study of ‘nonmodel’ species of ecological interest but with little available genome information. Application of the flowering gene network to wild species will be illustrated by two examples: modeling and prediction of the expression of flowering genes in Arabidopsis halleri, and the latitudinal cline of bud set and cessation in Populus. Finally, we discuss the challenges in integrating knowledge of the regulatory network on flowering into ecologically unique flowering phenomena such as synchronous intermittent mass flowering—the topic of this special issue.

Appendices

Appendix 1

Photoperiod pathway

In the photoperiod pathway, flowering is promoted by exposure to inductive photoperiod. Plants sense night length and use it as a flowering cue (reviewed in Imaizumi and Kay 2006; Kobayashi and Weigel 2007). Arabidopsis thaliana is classified as a quantitative long-day plant, meaning that it can flower more rapidly in long-day conditions, while flowering is delayed in short-day conditions. The genes involved in the photoperiod pathway were isolated using mutants defective in day-length-specific acceleration of flowering (Koornneef et al. 1991). In other words, mutants in the photoperiod pathway flower similarly under long-day and short-day conditions. The key components of the photoperiod pathway are the CONSTANS (CO) gene (Putterill et al. 1995) and the FLOWERING LOCUS T (FT) gene. Expression of the CO gene is regulated by circadian clock genes and shows a circadian pattern (Suárez-López et al. 2001; Yanovsky and Kay 2002). In addition, activity of the CO protein is also regulated in a light-dependent manner through the functions of photoreceptors, and thus can induce expression of the downstream gene, FT, only in long-day conditions (Valverde et al. 2004). This provides the crucial mechanism of day length measurement. Expression of the FT gene occurs in leaves, but FT protein moves to the shoot apical meristem (SAM) in which flower development occurs, acting as a ‘florigen’ (Corbesier et al. 2007; Tamaki et al. 2007). Once FT protein arrives at the SAM, it interacts with another protein, FD, and activates downstream floral meristem identity genes such as APETALA 1 (AP1) (Abe et al. 2005; Wigge et al. 2005), and other floral promoters such as SUPPRESSOR OF OVEREXPRESSION OF CONSTANS 1 (SOC1) (Michaels et al. 2005; Yoo et al. 2005).

Vernalization pathway

Many individuals of A. thaliana grown in northern Europe need prolonged exposure to the cold of winter to render them competent to flower (Nordborg and Bergelson 1999). This process is called vernalization and ensures that they flower in favorable spring conditions. The vernalization requirement in A. thaliana is conferred by a floral repressor gene, FLOWERING LOCUS C (FLC) (Michaels and Amasino 1999; Sheldon et al. 1999). Without vernalization, highly expressed FLC represses flowering by suppressing the expression of the flowering activators FT, FD and SOC1 (Hepworth et al. 2002; Helliwell et al. 2006; Searle et al. 2006). However, as the plants are exposed to prolonged cold, FLC expression is gradually decreased, and the plants become ready to flower (Michaels and Amasino 1999; Sheldon et al. 1999).

Light quality pathway

Along with the duration of light (photoperiod), light quality is also an important flowering cue. Plants recognize their growing conditions from the light quality, for example, under canopy shade conditions or in the proximity of other plants, and then show acceleration of flowering. This process is mediated by photoreceptors called phytochromes, which function upstream of the FT gene and regulate flowering time (Cerdán and Chory 2003; Halliday et al. 2003).

Thermosensory pathway

Another important cue is ambient temperature. Small changes in ambient temperature regulate flowering time through FT in a thermosensory pathway involving the floral repressor gene, SHORT VEGETATIVE PHASE (SVP) (Lee et al. 2007).

Autonomous pathway

The autonomous pathway is composed of genes that cause late flowering in both long-day and short-day conditions when they are mutated. Therefore, the term “autonomous pathway” is generally used as the pathway that is independent of environmental cues, especially photoperiod. As expected from this definition, the autonomous pathway is not a linear pathway. Rather, it is comprised of subpathways of genes with different biochemical functions (Chan and Struhl 1997; Macknight et al. 1997; Schomburg et al. 2001; He et al. 2003; Simpson et al. 2003; Ausin et al. 2004; Lim et al. 2004; Noh et al. 2004). However, these genes have a common feature in that they control flowering time by mediating the expression level of the floral repressor FLC (Michaels and Amasino 2001).

Gibberellin pathway

Gibberellins (GAs) are a class of plant hormones. In A. thaliana, GAs promote flowering (Wilson et al. 1992; Reeves and Coupland 2001; Willige et al. 2007). GA₄ is the most active form of GA in floral induction, and the endogenous levels of GA₄ in the SAM are reported to increase up to nearly 100-fold preceding floral initiation (Eriksson et al. 2006). This physiological change in the endogenous GA levels functions through the upregulation of genes such as LEAFY (LFY) and SOC1 (Blázquez and Weigel 2000; Moon et al. 2003).

Age pathway

Plants also sense their age. Plants experience a gradual transition from the juvenile to adult phase as they grow, and this phase transition renders them competent to flower. The abundance of SQUAMOSA PROMOTER BINDING-LIKE (SPL) proteins acts as a key factor in this developmental pathway. A gradual increase in these proteins following the phase transition regulates genes such as FT, SOC1, LFY and AP1 in a direct or indirect manner (Wang et al. 2009a; Wu et al. 2009; Yamaguchi et al. 2009).

Appendix 2

Application of transcriptome profiling to the identification of differentially expressed genes

A central interest in molecular biology has been to identify differentially expressed genes through comparisons of transcription levels before and after specific biological processes. Microarrays have been used widely for this purpose. Microarray platforms provide a reliable, rapid and cost-effective technology with which to analyze genome-wide gene expression patterns by quantifying cDNA by hybridization to probes. This technique has facilitated large-scale transcriptome analysis in many organisms (for example, Schmid et al. 2003; Zeller et al. 2009). However, it has been limited mostly to model organisms with known genome sequences because gene sequences are required to design probes on microarrays.

On the other hand, in nonmodel organisms, expressed sequence tag (EST) sequencing with traditional Sanger capillary sequencers has been used successfully to analyze the transcriptome (for example, Ma et al. 2009; Pirooznia et al. 2010). The EST is a short subsequence of a transcribed cDNA sequence from a known tissue source. The example of strawberry discussed in the main text is one example of EST sequencing (Mouhu et al. 2009). However, this technique is time consuming and costly.

Alternative methods for transcriptomics using high-throughput sequencers have appeared recently (reviewed in Matsumura et al. 2008; Wang et al. 2009b; Wilhelm and Landry 2009). One of the major approaches is RNA-Seq. RNA-Seq provides both the sequence information of transcripts and their expression levels simultaneously. The millions of short reads generated by high-throughput sequencers are either aligned to a reference genome or assembled de novo to infer from which genes they were transcribed. The expression level of each gene can then be measured by counting the number of reads. RNA-Seq has the advantages of higher sensitivity and greater dynamic range of gene expression than microarrays if enough sequencing depth is achieved. Furthermore, it is more rapid and cost-effective than EST sequencing. Owing to advances in methods of de novo assembly, it can be applied to nonmodel organisms with little prior knowledge of their genome sequence (see, e.g., Barakat et al. 2009; Birzele et al. 2010; Ekblom et al. 2010; Rodriguez et al. 2010). RNA-Seq has already been applied to some nonmodel plant species to identify genes expressed differentially in response to several phenomena. For example, several candidate genes responsible for resistance to fungus infections have been revealed through the analyses of differentially expressed genes between fungus-sensitive American chestnut (Castanea dentata) and fungus-resistant Chinese chestnut (Castanea mollissima) (Barakat et al. 2009). As another example, differentially expressed genes in the resurrection plant Craterostigma plantagineum have been characterized during dehydration and rehydration (Rodriguez et al. 2010).

Although transcriptomics using high-throughput sequencers provide useful tools, the high run cost restricts the analysis to a small number of samples. When compared with high-throughput sequencers at the same resolution, microarrays still have an advantage in terms of cost. Therefore, a combination of both microarrays and high-throughput sequencers has been attempted for transcriptomics in nonmodel organisms. Once custom microarrays based on the sequence data generated by high-throughput sequencers have been designed, the microarray serves as a cost-effective tool with which to analyze a large number of samples. There are now some examples of the successful utilization of this system to identify differentially expressed genes (e.g., Bellin et al. 2009; Schmid et al. 2010).