In silico analysis of the endogenous time-keeping mechanism in citrus

The endogenous time-keeping mechanism is responsible for organizing plant physiology and metabolism according to periodic environmental changes, such as diurnal cycles of light and dark and seasonal progression throughout the year. In plants, circadian rhythms control gene expression, stomatal opening, and the timing component of the photoperiodic responses, leading to enhanced fitness due to increased photosynthetic rates and biomass production. We have investigated the citrus genome databases of expressed sequence tags (EST) in order to identify genes coding for functionally characterized proteins involved in the endogenous time-keeping mechanism in Arabidopsis thaliana. Approximately 180,000 EST sequences from 53 libraries were investigated and 81 orthologs of clock components were identified. We found that the vast majority of Arabidopsis circadian clock genes are present in citrus species, although some important components are absent such as SRR1 and PRR5. Based on the identified transcripts, a model for the endogenous oscillatory mechanism of citrus is proposed. These results demonstrate the power of comparative genomics between model systems and economically important crop species to elucidate several aspects of plant physiology and metabolism.


Introduction
The rotation of the earth on its axis causes alternating cycles of light and dark at a period of approximately 24 h. The physiology, metabolism and behavior of most organisms are drastically affected by these daily environmental changes that lead to apparent biological oscillations known as diurnal rhythms. The conceptual model of circadian clocks consists of an entrainment or input pathway that synchronizes the pacemaker with its environmental surroundings, a central oscillator that generates and sustains rhythmicity, and multiple output pathways that link the oscillator to the organism's physiology and metabolism (Bell-Pedersen et al., 2005).
In plants, the entraining stimulus is mainly light, with a smaller contribution from temperature (Salomé and McClung, 2005;Edwards et al., 2006). Accordingly, the input pathway to the plant clock consists mostly of photoreceptor families (Millar, 2004), ubiquitously found in plants, including economically important species (Santelli and Sivieiro, 2001;Hecht et al., 2005).
The central oscillatory mechanism of all known circadian clocks is dependent on autoregulatory feedback loops (Bell-Pedersen et al., 2005). In plants, the Arabidopsis circadian clock mechanism has been partially elucidated. However, circadian control of metabolism and physiology is evident in most plant species, including citrus, where several economically important features are circadianly regulated, such as flavonoid and terpenoid metabolism regulation and essential oil composition (Lee and Castle, 2001;Lin et al., 2002;Frizzo et al., 2004). Moreover, gene expression and developmental processes in citrus have also been demonstrated to be under control of an endogenous time-keeping mechanism (Smith and Wareing, 1972;Abied and Holland, 1994).
The oscillatory mechanism depends on CCA1-and LHY-binding to the evening element (EE, AAATATCT) in the TOC1 promoter, which represses transcription of the response regulator-like gene (Alabadí et al., 2001). During the day, the levels of CCA1 and LHY decrease, releasing their repression on TOC1 that peaks at dusk when the levels of CCA1 and LHY are at their trough. Subsequently, the increasing levels of TOC1 during the evening induce its function as a positive regulator of CCA1 and LHY expression through an as yet uncharacterized mechanism (Alabadí et al., 2001, Matsushika et al., 2002Mizoguchi et al., 2002;Harmer and Kay, 2005;Hazen et al., 2005). Several other gene products are required for CCA1 and LHY expression, including GIGANTEA (GI), ELF3, ELF4, and LUX (Park et al., 1999;Doyle et al., 2002;Mizoguchi et al., 2002;Hazen et al., 2005); thus, generating the interlocked feedback loops. The second proposed loop consists of TOC1 transcriptional activation by GI, which in turn, is repressed by TOC1 (Locke et al., 2005). A third autoregulatory loop is hypothesized to consist of CCA1 and LHY as positive regulators of TOC1-family members PRR5, PRR7, and PRR9 (Farré et al., 2005;Harmer and Kay, 2005). Currently, distinct lines of evidence support the idea that PRR5/7/9 are negative regulators of CCA1/LHY: (i) CCA1 and LHY transcripts accumulate in prr7 and prr7 prr9 mutants (Farré et al., 2005) and, (ii) CCA1 is constitutively transcribed in the arrhythmic prr5 prr7 prr9 triple mutant (Nakamichi et al., 2005b). Moreover, PRR5/7/9 and TOC1 are thought to be mutually repressive and partially redundant (Farré et al., 2005;Nakamichi et al., 2005a;Salomé and McClung, 2005).
Post-transcriptional regulation is involved in several processes of circadian timing in plants. CCA1 and LHY are phosphorylated by casein kinase II (CK2) in a manner similar to the pace-control mechanism of the mammalian circadian clock (Sugano et al., 1998, Sugano et al., 1999. The robustness of LHY oscillations has been attributed to posttranslational modifications and proteasome-mediated degradation (Kim et al., 2003;Song and Carré, 2005). A second type of posttranslational modification, implied in clock function, is protein acetylation; mutants lacking the N-acetylglucosamine transferase activity of SPINDLY (SPY) exhibit altered leaf movement rhythms (Tseng et al., 2004). Recent evidence has demonstrated the crucial role of light-and clock-controlled proteolysis for the plant endogenous clock. The novel family of putative photoreceptors ZEITLUPE (ZTL), LOV KELCH PROTEIN 2 (LKP2) and FLAVIN-BINDING KELCH REPEAT F-BOX 1 (FKF1) provides a direct link between the central oscillator and ubiquitin-mediated protein degradation (Nelson et al., 2000;Somers et al., 2000;Schultz et al., 2001). ZTL is a component of the Skp1-Cullin-F-box (SCF) complex that recruits TOC1 for proteasomal degradation Más et al., 2003b;Han et al., 2004). Day length is measured by the integration of circadian clock and light perception information at the level of CONSTANS (CO) protein expression and activity (Suárez-López, 2001;Yanovsky and Kay, 2002;Imaizumi et al., 2003;Valverde et al., 2004;Imaizumi et al., 2005). CO directly regulates the expression of FLOWERING LOCUS T (FT) in Arabidopsis and of its homolog HEADING DATE 3a (HD3a) in rice (Putterill et al., 1995;Kobayashi et al., 1999;Samach et al., 2000;Onouchi et al., 2000;Izawa et al., 2002;Kojima et al., 2002), which promotes the expression of meristem-identity genes triggering the transition from vegetative to reproductive development at the shoot apical meristem. The expression and activity of CO is regulated by the internal circadian clock of the plant, so that CO mRNA transcription is restricted to 12 h after dawn until around dawn of the following day (Suárez-Lopez et al., 2001;Valverde et al., 2004). Under long day conditions, CO mRNA is therefore present during the day, and when the CO protein is translated it is stabilized in the nucleus of the cell and will activate the expression of other genes Valverde et al., 2004). Under short day conditions, CO mRNA is only transcribed during the night because the days are shorter than 12-h long, and the dark translated CO protein is rapidly broken down, and unable to activate gene expression (Valverde et al., 2004). Thus, early flowering of Arabidopsis occurs in the long days of spring and early summer through the coincidence of circadian clock control of CO mRNA expression and direct exposure of the plant to light (Yanovsky and Kay, 2002;Valverde et al., 2004). In rice, CO gene homolog, called HEADING DATE 1 (HD1) also has a key role in flower timing control. However, the function of HD1 in rice is reversed, thus, promoting flowering on short days. The rice CO homolog protein is activated at the end of a long day; but instead of activating other genes as it does in Arabidopsis, HD1 represses gene activity and flowering under longer photoperiods (Yano et al., 2001;Izawa et al., 2002). In contrast, under short day conditions HD1 repression is absent, allowing HD3a to promote flowering Izawa et al., 2002;Kojima et al., 2002). In temperate forest trees, the CO/FT regulatory module has been demonstrated to control of flowering time in response to variations in daylength; and, unexpectedly, it also controls the short-day-induced growth cessation and bud set occurring in the fall (Böhlenius et al., 2006). Thus, annual species and trees appear to have a common time keeping mechanism to control development during the progression of the seasons. The conserved functionality of key meristem identity genes between Arabidopsis and citrus has been demonstrated (Endo et al., 2005;Pillitteri et al., 2004;Pena et al., 2001). However, the regulatory mechanisms of their induction, including light quality and photoperception, temperature and day length changes remain elusive.
Recent evidence has demonstrated that photoperiodic induction of synchronous vegetative bud break or flowering is also common in tropical plants (Borchert et al., 2005). At tropical latitudes, changes in day length that are large enough to affect plant development occur around the equinoxes. Therefore, a single annual period of synchronous flowering after the autumn equinox indicates the induction of flowering in`short-day plants' in response to declining day length; whereas,`long-day plants' flowering and vegetative bud break is induced by an increasing day length that occurs after the spring equinox (Borchert et al., 2005). The molecular mechanism leading to day length measuring in the tropics has been hypothesized to involve the integration of photoperception and the endogenous clock as observed in temperate species (Borchert et al., 2005).
The present work was designed to investigate the circadian temporal programming and its integrative pathways to physiology and metabolism in citrus, employing combined in silico EST profiling and domain structural data analysis. The results presented here demonstrate that the vast majority of the gene products involved in circadian timing in Arabidopsis are present in citrus species. Moreover, genetic distance and domain structure analyses have uncovered extensive amino acid sequence conservation, providing indications of functional equivalence of Arabidopsis and citrus transcripts. Taken together, the results have demonstrated the feasibility of incorporating in silico analyses for gene discovery in non-model species.

Database searches and alignments
Homologs of the Arabidopsis thaliana circadian clock were identified in BLAST searches (Altschul et al., 1997) against citrus EST databases (CitEST), which consist of approximately 180,000 ESTs obtained from the sequencing of 53 specific libraries. Data validation was performed by tBLASTx and tBLASTn searches of the retrieved sequence against the locally built GenBank database at CitEST. Sequences failing to retrieve the original sequence used to query the database were eliminated from the projects. The resulting alignments were filtered by a threshold e-value of 1e-15 and the hits were further analyzed according to functional domain description. Validated sequences were translated and protein (deduced amino acid) alignments were performed using ClustalX (Thompson et al., 1997). When necessary, alignments were manually adjusted using Lasergene MegAlign (DNASTAR, Madison, WI, USA).

Motif analysis and in silico characterization
The identified homologs were further investigated for the presence and sequence conservation of recognizable functional domains described in several protein analysis and gene function databases (European Bioinformatics Institute-European Molecular Biology Laboratory -EMBL-EBI; Expert Protein Analysis System -ExPaSy from Swiss Institute of Bioinformatics -SIB; Gene Ontology database -GO; Protein Families database -Pfam).

Phylogenetic analysis
The functionality of citrus genes in comparison to their Arabidopsis counterparts was assessed by genetic distance and phylogenetic studies. Phylogenetic analyses were performed using distance and parsimony methods in the software PAUP* 4.0b10, using the software default parameters. Resampling bootstrap trees containing 1000 random samples were constructed using PSIGNFIT software. Modular functional domains were employed for genetic distance studies for genes previously characterized as having divergent regions and conserved blocks.

In silico gene expression analysis
Qualitative gene expression profiling was performed by in silico analyses of the afore mentioned citrus EST databases using virtual northern blot analyses. The gene of interest was used in queries against reference sequence databases, generating an alignment of the input gene to its paralogs. The resulting alignment was then used to find sequences in the entire mRNA input that are specific to the gene (probe). The resulting alignments were collectively used to query the EST database again using BLAST. This heuristic attempts to avoid false-positives, or ESTs from a paralog of the input gene rather than the gene itself. The identity numbers of the ESTs matching the probes were recovered and CitEST was used to find the names of the libraries from which those ESTs were derived. The frequency of reads of each EST contig in a given library was calculated and normalized according to the total number of reads from the investigated library and the total number of reads in all libraries. A correlation matrix between EST contigs and libraries was then generated and gene expression patterns among ESTs and libraries were obtained by hierarchical clustering based on a Spearman Rank correlation matrix using Cluster v.2.11 software (Eisen et al., 1998), by substituting the clusters by their average expression pattern. Graphic outputs were generated using Tree View v.1.6 software and presented in grayscale.

Results and Discussion
The essential components of the circadian clock input pathway, gating mechanism, central oscillator, the output and photoperiod integrative pathways from Arabidopsis thaliana, and to a lesser extent Oryza sativa, were compiled and the protein sequences of the main components were employed to search citrus EST databases. The identified hits were ranked according to the deduced amino acid sequence identity to the functionally characterized protein used as bait and further analyzed for their functionality.
This way, 81 citrus ESTs and EST contigs that are similar to components of the endogenous time-keeping mechanism and its ancillary regulation loops were identified and functionally assigned to gene ontology classes ( Figure 1). From this total, 20 transcripts correspond to input pathway photoreceptors, proteins involved in the gating mechanism and in ancillary entrainment loops in plants (Table 1). Twenty-nine citrus sequences showed extensive sequence conservation to components of the central oscillator mechanism (Table 2) and 32 resemble proteins involved in the integration between the pacemaker and plant metabolism and physiology (Table 3). Interestingly, some important components of the time keeping-mechanism from Arabidopsis are absent from citrus sequence databases, such as the input gene SRR1 (AT5G59560; Staiger et al., 2003) and the TOC1/PRR1-quintet member PRR5 (AT5G24470, Nakamichi et al., 2005a). Moreover, citrus and Arabidopsis gene families, including the DOF transcription factor CDF1 and MADS-box CONSTANS, appear to have undergone differential expansion events.

Input pathway Light entrainment pathway
Four of the five Arabidopsis phytochromes (PHYA, PHYB, PHYD and PHYE) function in an additive manner in the red-light (R) input to the clock . CRYPTOCHROME 1 (CRY1) acts as the main clock input photoreceptor under high and low fluences of blue light (B) whereas both CRY1 and CRY2 redundantly function at intermediate fluences of B Somers et al., 1998). Surprisingly, clock entrainment under white light seems to be dependent on the physical interaction of CRY2 and PHYB and their localization in nuclear speckles . Moreover, cry1 mutant plants also show altered entrainment under red light suggesting that CRY1 is required for PHYA signaling to the clock in both R and B . Quadruple Arabidopsis mutants lacking phyA, phyB, cry1 and cry2 are able to maintain circadian rhythmicity (Yanovsky et al., 2000), suggesting that additional components participate in the light input pathway to reset the clock. The rhythmic expression of PHY and CRY genes (Tóth et al., 2001) has been hypothesized to contribute to the gating mechanism on the input to the clock, although the bulk PHY protein levels do not oscillate (Sharrock and Clack, 2002).
Loss-of-function ztl mutants show fluence-rate dependent period alteration, indicating that ZTL has a role in the light input pathway . ZTL mRNA abundance is not clock regulated, but its protein levels peak around dusk and reach trough levels around dawn (Kim et al., 2003). The rate of proteasome-mediated degradation of ZTL varies during the course of the day: ZTL is more stable at dusk, close to its peak value, and is more rapidly degraded at dawn when it reaches its trough. F-box proteins provide specificity to proteasomal degradation pathways by interaction and polyubiquitination-mediated degradation of determined targets. ZTL has been demonstrated to recruit the oscillator component TOC1 for degradation Más et al., 2003b;Han et al., 2004). In ztl mutants, TOC1 levels are elevated and only weakly rhythmic, demonstrating that ZTL is critical for TOC1 proteolysis and rhythmicity. Correspondingly, ZTL overexpression induces a dosage-dependent period shortening (Han et al., 2004). Citrus photoreceptor families and the components of their proteolytic degradation pathway have been extensively discussed in the accompanying paper on photoperception and light signal transduction.

Gating mechanism
In Arabidopsis, light-conditional arrhythmic behavior is observed in the absence of ELF3 (EARLY FLOWERING 3) and TIC (TIME FOR COFFEE) proteins (Hicks et al., 1996;Hicks et al., 2001;McWatters et al., 2000;Covington et al., 2001;Liu et al., 2001;Hall et al., 2003). Interestingly, elf3 null mutants are arrhythmic at different phases of the oscillation when compared to tic mutants and double elf3 tic mutants lack rhythmicity under light and dark conditions (Hall et al., 2003). Thus, these Quecini 797 proteins are likely to be responsible for the gating mechanism at different times of the day and are termed Zeitnehmer (from German, time taker). The ELF3 protein contains no recognizable domains and the TIC gene remains to be identified Hall et al., 2003). In citrus transcriptome, two EST contigs from Citrus sinensis and C. reticulata, one from C. aurantium and one from Poncirus trifoliata, sharing moderate sequence similarity to ELF3, were identified ( Table 2). The lack of functionally characterized domains in Arabidopsis ELF3 prevents further considerations about the role of the citrus homologs.

Temperature entrainment pathway
Temperature signaling to the clock is much less well defined; however, gene expression and cotyledon movement have been demonstrated to be effectively entrained by temperature cycles in Arabidopsis Salomé et al., 2002;Salomé and McClung, 2005). The pseudo-response regulators PRR7 and PRR9 are essential for temperature-mediated clock synchronization, as prr7 prr9 double mutants fail to entrain to temperature cycles that are effective to entrain wild type plants (Salomé and McClung, 2005). Citrus genome has three cDNAs similar to PRR7, two from C. sinensis and one from P. trifoliata, and two EST contigs highly similar to PRR9 (Table 3). In Arabidopsis and rice, these proteins participate in the central oscillator mechanism as well (Farré et al., 2005;Murakami et al., 2005;Nakamichi et al., 2005b). The citrus PRR homologs are discussed in-depth in the following section. Interestingly, the photoperiodic transcriptional activator, GIGANTEA (GI) and the MADS-box protein FLC have been assigned to a quantitative trait locus (QTL) responsible for a substantial portion of natural variation in temperature-mediated circadian clock entrainment (Ed-798 Citrus circadian clock  , 2005). Five citrus transcripts, whose deduced amino acid sequences share significant homology to GI and GI-like proteins, and three FLC homologs were identified (Table 2 and Table 3, respectively). Thus, the Arabidopsis components of the temperature-mediated clock entrainment pathway are present in citrus transcriptome databases.

Central oscillator
The simplified model for the transcriptional/translational negative-feedback loop of the Arabidopsis circadian clock consists of the reciprocal regulation between the highly similar MYB-type transcription factors CCA1/LHY in the negative limb and TOC1 and the novel protein ELF4 Quecini 799 in the positive limb (Strayer et al., 2000;Kikis et al., 2005;McClung, 2006). In citrus transcriptome databases, nine homologs to the TOC1/PRR1 family of pseudo-response regulators, nine cDNAs whose deduced amino acid sequences are similar to the MYB-type transcription factors CCA1 and LHY, and three similar to ELF4 were present (Table 3). Moreover, in silico analyses have identified four homologs of the LUX/PCL1 family, which are hypothesized to function on an interlocked loop to activate CCA1 and LHY transcription (Hazen et al., 2005;Onai and Ishiura, 2005), four homologs of CK2, a casein kinase responsible for CCA1 phosphorylation (Daniel et al., 2004) Putterill et al., 1995, Suárez-López et al., 2001 COL1 and COL family f  and two SPY-like sequences (Table 3). Thus, at least two interlocked feedback regulatory transcription/ translation loops of the Arabidopsis central oscillator and two post-translational modification systems are present and likely to be functional in a citrus endogenous time-keeping mechanism.
Citrus transcriptome analyses revealed the presence of 13 cDNAs corresponding to Arabidopsis MYB-type of transcriptional regulators involved in the circadian clock central oscillator mechanism; namely, CCA1, LHY and the LUX/PCL1 family (Table 3, Figure 2). CCA1/LHY-like sequences are highly similar in C. sinensis, C. reticulata and P. trifoliata, sharing up to 62.2% of sequence identity (Figure 2A). Moreover, the sequences are highly similar to their Arabidopsis counterparts (Table 2, Figure 2A). Interestingly, the single-MYB LUX/PCL1 family in citrus is more divergent and appears to have undergone a differential expansion process ( Figure 2B), as the highest sequence identity to Arabidopsis LUX proteins was of 30.7% and only four members were found, in contrast to the five members identified in Arabidopsis (Hazen et al., 2005). In silico transcriptional analysis of citrus MYB-like genes with similarities to clock components demonstrated a high frequency of CCA1/LHY and LUX homologs in developing fruits and a similar expression pattern (Figure 2). However, the normalized levels of LUX transcripts were approximately 1.5 times smaller than the levels observed for CCA1/LHY (Figure 3).
The Arabidopsis PRR family consists of TOC1/PRR1 and four other members (PRR3, PRR5, PRR7 and PRR9) whose structural designs are very similar (Matsushika et al., 2000) and include a receiver-like (or pseudo-receiver) domain at their N-terminus followed by a long specific intervening sequence, followed by another motif of about 50 amino acids at the very C-terminal end that is shared by the PRR family, termed the CCT mo-tif (CONSTANS, CONSTANS-like and TOC1) ( Figure  4A) . In citrus, eight transcripts sharing extensive sequence similarity to Arabidopsis and Oryza sativa PRR were identified ( Table  3). The identified transcript sequences are highly conserved at the CCT region and more divergent at the response regulator domain ( Figure 4B, Figure 4C). Citrus PRR family is highly represented in fruit developmental stage libraries and shows an expression profile similar to CCA1/LHY and LUX (Figure 3). Interestingly, the transcripts of PRRs begin to accumulate after subjective dawn in the following order PRR9-PRR7-PRR5-PRR3-PRR1, separated by 2-3 h intervals (Matsushika et al., 2000). In Oryza sativa, PRR family also consists of 5 members that present sequential transcriptional induction from subjective dawn and with approximately 2-3 h interval between 802 Citrus circadian clock  them (Murakami et al., 2005). Such sequential induction is preceded by CCA1 transcriptional activation, resulting in waves of sequential expression of CCA1-PRR9/7/5/3/1 . These phenomena were referred to as "circadian waves of the PRR quintet"; however, the biological function of the coordinated induction of CCA1 and PRR transcription remains elusive. The presence of the "circadian waves of the PRR quintet" in citrus remains to be investigated. TOC1/PRR1 and PRR9 have been demonstrated to participate in phytochromemediated photomorphogenesis regulation (Makino et al., 2001;Ito et al., 2003;Más et al., 2003a;Ito et al., 2005). The expression of citrus PRR in developing fruits suggests that they might be involved in light-regulated secondary metabolism, although further functional analysis is required.

Post-translation modification inducers
The post-translational modifications involved in the circadian mechanism in Arabidopsis are protein phosphorylation, proteolytic degradation and acetylation (McClung, 2006). Four CK2-homologous transcripts have been identified in citrus transcriptome ( Table 3). As observed for other oscillator components investigated, CK2 homologs are frequent in libraries obtained from fruit developmental stages and present a unique expression pattern in comparison to other putative pacemaker components; they are highly induced in developing fruits in the first and second developmental stages (Figure 3). The ZTL/FKF1/ LKP2 family, responsible for TOC1 ubiquitin-mediated degradation and regulation of photoperiodic responses, is also represented in citrus, as discussed in the accompanying paper on photoperception. Two singlets highly similar to the N-acetylglucosamine transferase SPY protein were identified in citrus (Table 3). Thus, it is likely that the post-translational mechanisms involved in circadian clock regulation in Arabidopsis are functional in citrus.

Clock outputs and integrative pathways
The endogenous time-keeping mechanisms allow organisms to adjust the timing of crucial developmental transitions to occur at the most favorable time of the year, leading to seasonal adaptation. The mechanism of day length measuring in higher plants is based on the integration of circadian clock and light perception, at the level of CO protein expression and activity (Suárez-López et al., 2001;Yanovsky and Kay, 2002;Imaizumi et al., 2003;Valverde et al., 2004;Imaizumi et al., 2005).
Arabidopsis CO and its rice homolog HD1 directly regulate the expression of FT and HD3a, respectively (Putterill et al., 1995;Kobayashi et al., 1999;Samach et al., 2000;Onouchi et al., 2000;Izawa et al., 2002;, which promote the expression of meristemidentity genes triggering the transition from vegetative to reproductive development at the shoot apical meristem (Komeda, 2004). Fifteen CO/COL homologs were identified in citrus transcriptome analysis (Table 3). The characteristic bipartite B-box Zn finger DNA binding domain is present at the N terminus of all citrus CO homologs ( Figure 5A), although sequence divergence is observed in IIIa and IIIb sub-groups ( Figure 5). A single EST contig from citrus lacked the B1 sub-domain and was clustered in the sub-group where the second B-box is replaced by a more divergent zinc-finger domain (Griffiths et al., 2003) ( Figure 5B). Similarly, the signature CCT domain is highly conserved in groups I, II and IIIa and more divergent in group IIIb ( Figure 5B).
Flowering requires the integration of temporal and spatial information to restrict its initiation to specific locations. Recently, Abe et al. (2005) and Wigge et al. (2005) have demonstrated that the interaction between FT expressed in the leaves and the transcription factor FLOWERING LOCUS D (FD) occurs exclusively in the shoot apex and triggers the expression of floral identity genes in the new primordial. Moreover, Huang et al. (2005) have shown that upon CO-mediated photoperiodic activation, FT mRNA is transported from the leaves to the shoot meristem via the plant vascular tissue. Therefore, CO action in the phloem activates FT expression in leaves in a cell-autonomous manner, according to the observation that CO activates FT expression and promotes flowering only when expressed under the control of phloem-specific promoters in the leaf, but not under control of apex-specific promoters in the shoot Ayre and Turgeon, 2004;Valverde et al., 2004). These results indicate that the activity of CO is crucial for the generation of the mobile signal that originates in the leaf and is perceived in the apex at onset of flowering. Thus, leaf-produced FT activates the transcription of floral identity genes, such as APETALA1 (AP1), at the shoot apex through its interaction with a new bZIP transcription factor, FD that is expressed preferentially at the shoot apex in the region where new primordia are being generated (Abe et al., 2005;Wigge et al., 2005). Multiple lines of evidence from these studies suggest a model by which FD provides the spatial framework or timely activation of flowering by FT (Blázquez, 2005). FT and FD have been shown to physically interact in vitro and in vivo and are hypothesized to act together to activate downstream targets (Abe et al., 2005). Ectopic expression of FD caused up-regulation of AP1 expression in leaves only under conditions that increase FT expression, such as transfer of plants from short to long days (Wigge et al., 2005). The presence of transcripts sharing extensive sequence homology to Arabidopsis flowering integrators suggest that a similar mechanism may operate in citrus as demonstrated for other woody species (Böhlenius et al., 2006).

Concluding Remarks
A model for the endogenous oscillatory mechanism in citrus, based on comparative transcriptome and sequence conservation analysis, is proposed ( Figure 6). This preliminary survey of citrus circadian clock-associated genes has provided basic information for in-depth studies relating to the endogenous time-keeping mechanism in this important plant group. Further studies of the genes identified here will also give a new perspective on the circadian regulation of economically important metabolic processes, such as secondary metabolism control. Moreover, functional characterization of the identified pathways will help to uncover the molecular basis for seasonal responses in tropical tree species. The results highlight the potential and the power of comparative genomics and in silico transcriptional profiling in perennial species.
Quecini 805 Figure 6 -Schematic overview of the endogenous time-keeping mechanism in citrus based on comparative genomic analysis. Solid boxes with italics lettering represent genes. Proteins are represented by capital letters inside oval or oblong shapes. Transcription and translation proteins are indicated by dashed lines. Protein activity is indicated by solid lines with arrowheads for positive action and perpendicular lines for negative action. The core CCA1/LHY/TOC1 feedback loop is represented by the innermost cycle whereas the ancillary loops are represented as outer loops. Phosphorylation of LHY and CCA1 by CK2 is indicated with solid circled Ps. The lower half of the figure represents activities peaking during the subjective night (moon schematic representation), whereas the superior part indicates activities peaking during the subjective day (sun schematic representation). Table 1S-Citrus reads constituting contigs and singlets with homology to genes involved in the input pathway to the circadian clock in Arabidopsis thaliana. Table 2S -Citrus reads constituting contigs and singlets with homology to genes involved in the central oscillator of the circadian clock in Arabidopsis thaliana. Table 3S -Citrus reads constituting contigs and singlets with homology to genes involved in the integration of the circadian clock mechanism and the generation of output responses in Arabidopsis thaliana.

Supplementary Material
This material is available as part of the online article from http://www.scielo.br/gmb.

Associate Editor: Alessandra Alves de Souza
Quecini 809 Table 1S -Citrus reads constituting contigs and singlets with homology to genes involved in the input pathway to the circadian clock in Arabidopsis thaliana.