© 2008 The Author(s).
This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (http://creativecommons.org/licenses/by-nc/2.0/uk/) which permits unrestricted non-commercial use, distribution, and reproduction in any medium, provided the original work is properly cited.
This paper is available online free of all access charges (see http://jxb.oxfordjournals.org/open_access.html for further details)
RESEARCH PAPER |
Meristem identity gene expression during curd proliferation and flower initiation in Brassica oleracea
Department of Horticultural Sciences, New York State Agricultural Experiment Station, Cornell University, Geneva, NY 14456-0462, USA
* To whom correspondence should be addressed. E-mail: dvd2{at}cornell.edu
Received 14 September 2007; Revised 21 November 2007 Accepted 26 November 2007
 |
Abstract |
|---|
 Top Abstract IntroductionMaterials and methodsResultsDiscussionReferences |
|---|
The regulation of reproductive development in cauliflower (Brassica oleracea var. botrytis DC) and broccoli (B. oleracea L. var. italica Plenck) is unusual in that most enlargement occurs while development is arrested at a distinct stage. Cauliflower and broccoli curds are composed of inflorescence meristems and flower buds, respectively. To determine whether this arrest is maintained by altered expression of the genes that specify these steps in Arabidopsis, the expression of each copy of their homologues (MADS-box genes BoAP1-a, BoAP1-c, BoCAL, BoFUL-a, BoFUL-b, BoFUL-c, and BoFUL-d; and non-MADS-box genes BoLFY, AP2, UFO, and BoTFL1) and the cauliflower curd-specific genes CCE1 and BoREM1 were measured simultaneously in heads that were arrested at different developmental stages by varying temperature, but had a common genotype. Transcript abundance of BoFUL paralogues and BoLFY was highest at the cauliflower stage of arrest, consistent with these genes initiating inflorescence meristems. The expression of other genes was the same regardless of the developmental stage of arrest. The expected models can therefore be excluded, wherein maintenance of arrest at the inflorescence meristem is a consequence of suppression of BoCAL, BoAP1-a, or BoLFY, or failure to suppress BoTFL1. Floral primordia and floral buds were normal in boap1-a boap1-c bocal triple mutants; therefore, other meristem identity genes can specify floral initiation (A-function) in B. oleracea. BoTFL1, a strong repressor of flowering in Arabidopsis, did not suppress the formation of the floral primordium in B. oleracea. Initiation of floral primordia and enlargement of floral buds in broccoli and cauliflower is not controlled solely by homologues of the genes that do so in Arabidopsis.
Key words: Brassica oleracea, broccoli, cauliflower, developmental arrest, flowering, meristem identity genes, temperature
|
Introduction |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionReferences |
|---|
Brassica oleracea is a species with a remarkable variety of cultivars and a variety of edible forms. Reproductive development determines the value of the crop, yet key steps in this phase of growth remain physiologically and genetically poorly understood. The curd phenotype in cauliflower (Brassica oleracea var. botrytis) corresponds to inflorescence meristems that share characteristics of both the vegetative and reproductive apices (Sadik, 1962). In broccoli (Brassica oleracea var. italica), the arrest occurs before anthesis and the head is composed of flower buds (Fujime and Okuda, 1996). Several studies have tried to elucidate the genetic control of developmental arrest in B. oleracea by identifying and characterizing homologues of the Arabidopsis floral homeotic genes. In Arabidopsis, an increase in LEAFY (LFY) expression, and consequent suppression of TERMINAL FLOWER 1 (TFL1), initiates flowering by up-regulation of APETALA 1 (AP1) and CAULIFLOWER (CAL). The homologues in B. oleracea are not consistently expressed in a manner that directly parallels their Arabidopsis functions. The onset of BoLFY expression does not correlate with the initiation of the floral primordium in cauliflower (Anthony et al., 1993; Jordan et al., 1994).
Several lines of evidence support a model for developmental arrest in which BoAP1 and BoCAL are primary regulators. The effect of a mutant CAL allele is present in cauliflower and causes a cauliflower phenotype in Arabidopsis ap1 mutants (Bowman et al., 1993; Kempin et al., 1995). Kempin et al. (1995) proposed that this mutation is responsible for the cauliflower phenotype in B. oleracea. The wild-type BoCAL allele was later found in broccoli (Carr and Irish, 1997), strengthening the parallel. BoAP1-a and BoCAL had additive roles in the stage of arrest in a segregating population of doubled haploid lines from a cross of broccoli and cauliflower (Smith and King, 2000). The phylogenetic distribution of functional and non-functional alleles is consistent with both BoCAL and BoAP1-a being necessary for floral development in B. oleracea (Lowman and Purugganan 1999). Finally, BoAP1-a accumulates at the site of floral initiation in broccoli (Anthony et al., 1995; Carr and Irish, 1997).
There is also evidence that the developmental arrest is more complex, suggesting that BoCAL and BoAP1, while involved in curding, do not predict the phenotype in other populations. The mutant allele at BoCAL is present in some broccoli and non-heading Brassica accessions (Purugganan et al., 2000; Smith and King, 2000), and the wild-type BoCAL allele can occur in cauliflower (Smith and King, 2000). A subsequent survey of broccoli and cauliflower accessions found only a weak association between the boap1-a and bocal-a mutant alleles and the cauliflower phenotype (Labate et al., 2006). Furthermore, curd-related traits, such as days to budding and days from budding to flowering, among others, are affected by as many as 86 quantitative trait loci (Lan and Paterson, 2000), suggesting that the cauliflower arrest is under multigenic control.
The aim of the present study was to identify genes that control the arrest of the reproductive development in heading B. oleracea. The thorough description of genetic control of flowering in Arabidopsis thaliana identified a set of candidate genes for the molecular basis of the arrest of curd and head formation in confamilial cauliflower and broccoli. Candidate genes selected for this study were meristem identity genes (MIGs) that in Arabidopsis are involved in time to flower, the transition from inflorescence meristems to floral primordium, and floral organ identity: LEAFY (LFY), APETALA 2 (AP2), UNUSUAL FLORAL ORGANS (UFO), and the MADS-box genes APETALA 1 (AP1), CAULIFLOWER (CAL), and FRUITFULL (FUL) (Bowman et al., 1993; Shannon and Meeks-Wagner, 1993; Ferrandiz et al., 2000), as well as TERMINAL FLOWER 1 (TFL1), the floral repressor potentially responsible for maintaining arrest in cauliflower. Two other genes have been specifically associated with the cauliflower curd: CAULIFLOWER CURD EXPRESSION 1 (CCE1) and REPRODUCTIVE MERISTEM 1 (BoREM1). CCE1 could potentially be a gene involved in maintaining the arrest (as TFL1 does in Arabidopsis) (Palmer et al., 2001), while BoREM1 could break the arrest (as LFY does in Arabidopsis) to allow continued reproductive development (Franco-Zorrilla et al., 1999). The candidate gene approach permits the use of quantitative real-time PCR, a technique sensitive enough to detect low-abundance genes such as BoTFL1 and to distinguish paralogues with high sequence similarity.
This study is novel because it investigates which genes change expression as different stages of arrest are overcome by simultaneously measuring the expression of several genes having overlapping function using tissue of a single genotype arrested at different developmental stages. Furthermore, B. oleracea has several copies of the key MIGs, allowing determination of whether duplication was followed by subfunctionalization, degeneration, or retained redundant functions.
Curd development and arrest are affected by temperature. Cool temperatures promote flower development, while warm temperatures inhibit reproductive development from proceeding. In production of broccoli and cauliflower, these effects are a serious economic concern. Premature floral buds in cauliflower cause a ‘ricy’ head at low temperatures (Fujime and Okuda, 1996; Grevsen et al., 2003), whereas excessive heat inhibits flower bud development in broccoli and promotes bract and bracteole formation in both forms (Booij and Struik, 1990; Fujime and Okuda, 1996; Björkman and Pearson, 1998; Grevsen et al., 2003, Kop et al., 2003). Here, advantage was taken of temperature effects to induce arrest at specified developmental stages (Labate et al., 2006).
Studies in both Arabidopsis and B. oleracea have indicated that temperature interacts with MIGs. In Arabidopsis, expression of AP1 and LFY transcripts increases with high temperature (Bowman et al., 1993). In cauliflower, heat causes a decline of BoLFY and BoAP1 expression in the shoot apex (Anthony et al., 1996). Also, the BoAP1-a locus has been found to interact with temperature in bracting. Even though the allelic state of BoAP1-a had the greatest influence in bracting (the boap1-a allele increases bracting), high temperatures increased bracting in plants with the same genotype (Kop et al., 2003). This study clarifies whether MIG expression responds directly to temperature, or is associated with the developmental stage.
|
Materials and methods |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionReferences |
|---|
Plant material and growth conditions
Experiment 1: gene expression at different developmental stages and the same genotype
Seeds of the F1 hybrid cauliflower cv. Green Harmony (Known-You Seed Company Ltd, Taiwan) were sown into modular trays of 50 cells (4.5 cm x 4.5 cm) containing Cornell Mix A (Boodley and Sheldrake, 1977). Trays were placed in a greenhouse at day/night temperatures of 22–24 °C/16–18 °C. Seedlings were transplanted into pots containing Cornell Mix A and Osmocote 14–14–14 (slow-release fertilizer) and subsequently fertilized as required with water-soluble fertilizer (EXCEL Cal-Mag 15–5–15, The Scotts Company, Marysville, OH, USA). The reproductive transition was considered to have occurred in the planting when the dissected apical meristem measured 500 µm in sample plants, and the apical meristem under observation presented cauline leaf primordium with incipient axillary meristems. At this time (
28 d after sowing), plants were moved into growth chambers (Conviron E15, Winnipeg, Manitoba, Canada) at three different day/night temperature regimes: 16 °C/12 °C, 22 °C/17 °C, and 28 °C/22 °C with 14 h photoperiod, 75% humidity, and light intensity of 500 mmol m–2 s–1. Plants were kept in the growth chambers until curds reached harvest maturity, right before the curd started to separate as a result of bolting.
Experiment 2: gene expression at different temperature regimes and at the same developmental stage
In this experiment, a long-term and a short-term temperature treatment were performed. For the long-term treatment, experiment 1 was repeated with a purple cauliflower (intermediate stage of arrest) from southern Italy (Cavolfiore Violetto di Sicilia), accession ‘HRI 5295’ [University of Warwick, Genetic Resources Unit of Warwick Horticulture Research International (HRI), UK], less sensitive to temperature changes than ‘Green Harmony F1’. Plants were transferred to the growth chambers right at the initiation of the reproductive stage (determined by dissection and measurement of the apical meristem as described for experiment 1), 39 d after sowing. For the short-term temperature treatment, ‘Green Harmony F1’ plants were raised in the greenhouse (as indicated for experiment 1) until curd diameter was between 2 cm and 3 cm. At this time, plants arrested at the same developmental stage were transferred into growth chambers, under the same three temperature regimes and conditions described above, for a 24 h period. Before moving the plants into the growth chambers the following morphological traits were annotated: number of leaves, curd diameter, and curd characteristics. This information was used later to group the plants in three sets, each of them including three plants, one for each temperature treatment.
Experiment 3: effect of BoAP1 and BoCAL genotype on expression of BoFUL and BoTFL1
Seeds from the double haploid (DH) lines (Table 1) generated from anther culture of the F1 from the cross between the DH cauliflower parent N (CA25: DH line produced from an F1 cauliflower variety, Nedcha) and the recurrent inbred Calabrese broccoli parent B (BI87053), kindly provided by Graham King and Graham Teakle (Warwick HRI, Wellesbourne, Warwick, CV 35, UK), were kept in the greenhouse (the same conditions described for experiment 1) until curds reached 2 cm in diameter, and tissue samples were taken. Plants were characterized phenotypically at the time of sampling. The arrest stages were coded into one of four phenotypic classes: broccoli, composed of fully developed floral buds; intermediate curd, composed of small floral buds and floral primordia; ricy cauliflower, composed of inflorescence meristem and some initial primordia; and cauliflower, composed of inflorescence meristems.
|
RNA extraction and reverse transcription
Tissue was isolated in each case from the whole surface of individual curds. Total RNA of each sample (50 mg of tissue) was extracted using Trizol reagent (Invitrogen, Carlsbad, CA, USA) following the protocol described by the manufacturer. Contaminating genomic DNA was removed using a RNase-free DNase solution according to Ausubel et al. (1994). Total RNA quality was evaluated on a 1.3% formaldehyde gel, and concentration was determined by measuring A260 using a BioMate series 3 spectrophotometer (Thermo Spectronic, Waltham, MA, USA).
Reverse transcription-PCR (RT-PCR) was done in two steps. RNA reverse transcription was performed using the RETROscript kit (Ambion, Inc., Austin, TX, USA) with random decamers according to the manufacturer's protocol. The absence of contaminating genomic DNA was determined by PCR with a set of primers designed around an intron sequence that amplifies the constitutive 
-subunit of the translation elongation factor 1 gene (EF1-) (Vidal et al., 1996).
Gene expression
Quantitative real-time PCR was performed using SYBR GREEN in the Bio-Rad iCycler iQ Real-Time Detection System (Bio-Rad, Hercules, CA, USA). All reactions were run in triplicate, and a non-template control for the primer set under study was included in each run. Deviations of Ct (cycle threshold) values for replicates of the same cDNA were not larger than 0.2 cycles, and 18S rRNA was used as the normalizer gene. For each 20 µl of final reaction, the following were added: 10 µl of IQ SYBR Green Supermix 2X (Bio-Rad), 2 µl of cDNA (0.25 µg of initial RNA), 1.5 µl of reverse and forward primers (final concentration of 0.3 µM each), and 6.5 µl of distilled sterile water. For PCRs using 18S rRNA-specific primers, the cDNA sample was diluted in a 2:1000 ratio of which 1 µl was used in the reaction mix. Amplification of cDNAs involved a 3 min denaturation step at 95 °C, followed by 40 cycles with 95 °C denaturation for 15 s and 60 °C for 30 s. Reaction products were analysed by performing a ‘melting curve’: from 55 °C to 95 °C at 0.5 °C per 7 s (80 cycles). Ratio calculations were obtained using the mathematical model proposed by Pfaffl (2001) that includes efficiency correction for target and normalizer (reference) genes. All the data incorporated in this equation (Equation 1) were an average from two or three runs per sample and the corresponding biological replications done in each experiment. Efficiency of the PCR for each individual run was estimated based on Ramakers et al. (2003).
 | (1) |
Ct is the Ct value difference between control or reference sample (gene expression at day temperature 22 °C) and sample (gene expression at day temperatures 16 °C and 28 °C) of target (TG) and normalizer gene (or reference gene) (NG) transcripts. Ct is the point at which fluorescence increased above a fluorescence threshold above the background fluorescence. The same threshold was used for all samples. Primer pairs were designed using Primer-Express Software 1.5, Primer3 Software (http://frodo.wi.mit.edu/) (Rozen and Skaletsky, 2000), or by hand after alignment of sequences of interest using Clustal W (1.82), EMBL-EBI, (http://www.ebi.ac.uk/clustalw/) (Chenna et al., 2003), following standards recommended for the real-time RT-PCR technique (Bustin, 2000) (Table 2). The specificity of the primers designed was evaluated in different ways. First, amplified PCR products for each set of primers were analysed by agarose electrophoresis, purified using a Mermaid Spin kit (Q-BIOgene BIO101 Systems, USA), cloned into a plasmid using the p-GEM-T Easy Vector kit (Promega, Madison, WI USA), and sequenced in the BioResource Center (BRC), Cornell University, using the Applied Biosystems Automated 3730 DNA Analyzer with Big Dye Terminator chemistry and AmpliTaq-FS DNA polymerase. Secondly, calculation of the melting temperature expected for each amplicon was performed and compared with melting peaks obtained in quantitative real-time PCR. Criteria for the choice of the best primer set included: no primer dimer (checked by melting curves), highest Ct value, and highest PCR efficiency.
|
Genotyping
Leaf disks 10 mm in diameter were taken from ‘Green Harmony F1’, ‘HRI 5295’, and DH plants in order to genotype the following loci: BoAP1-a, BoAP1-c, and BoCAL. DNA was extracted using the protocol of Doyle and Doyle (1987). PCR assays included homozygous mutants (parent line CA25: DH cauliflower parent N), homozygous wild-type (parent line BI87053: inbred broccoli parent B), and heterozygous (F1 progeny from NxB cross) controls.
The allelic variation of the BoCAL and BoAP1-a locus was assayed as described by Labate et al. (2006). BoCAL genotyping was based on the presence of the SpeI restriction enzyme recognition site in the mutant allele at BobCAL [premature stop codon (Kempin et al., 1995)]. BoAP1-a allelic variation was determined by an amplicon length polymorphism in the 100 bp upstream of the start codon (Smith and King, 2000). For genotyping the BoAP1-c locus, a simple sequence repeat (SSR) polymorphism assay was used [AP1cSSRF (5'-CGAGCTTATCACGCTGTTGT-3') and AP1cSSRR (5'-CCTCTCATAGCGTTCCAGTA-3')]. The assay detects a 12 bp difference between the mutant and the wild-type alleles given by the presence of a microsatellite at the 3' end of intron 1 (Smith, 1999).
Statistical analysis
All statistical analyses were performed using SAS (version 8; SAS Institute Inc., Cary, NC, USA). Statistical analyses of differentially expressed genes for experiment 1 and experiment 2 were done using analysis of variance (ANOVA) with PROC GLM. Assumptions of normal distribution and equal variances were tested in each case, and transformation of the data (log base 10) was used when necessary. When the F-test was significant, multiple comparisons of the means among the three different stages of development (cauliflower, intermediate, and broccoli) in experiment 1, or between the different day/night temperature regimes (16 °C/12 °C, 22 °C/17 °C, and 28 °C/22 °C) in experiment 2 were done using Tukey's studentized range [honestly significant difference (HSD)]. For the short-term temperature experiment, every set of plants with similar morphological characteristics was considered a block. When block effect was not significant, it was removed from the model and the main effect (temperature) re-evaluated.
To estimate the assay variation, SEMs were calculated assuming statistical independence of the quantities using propagation of error formulae that allow the inclusion of experimentally determined errors for each efficiency and Ct value (Ku, 1966).
In experiment 3, the statistical analysis of differentially expressed genes was done with ANOVA using PROC MIXED. BoAP1 and BoCAL genotypes were considered fixed effects, and plant line was considered a random effect. Assumption of normal distribution and equal variances were tested in each case, and transformation of the data was used when necessary (log base 10). A 
2 test of independence was performed to test the association between the categorical variables: BoAP1 and BoCAL genotype with respect to phenotype. This test was necessary to determine the variables that needed to be included in the model for the ANOVA.
|
Results |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionReferences |
|---|
Genotyping results
Genotyping results showed that ‘Green Harmony F1’ is homozygous mutant at BoAP1-a, BoAP1-c, and BoCAL, while ‘HRI 5295’ is homozygous mutant at BoAP1-a, and segregates for BoAP1-c and BoCAL. Genotyping results for NxB DH lines are shown in Table 1.
Experiment 1: gene expression at different developmental stages and the same genotype
‘Green Harmony F1’ is highly sensitive to changes in temperature, thus, by using different temperature regimes at the time of reproductive initiation, it was possible to create phenotypic variation in the stage at which the curd arrested (Fig. 1).
|
The relative expression profiles of all the genes studied, between the reference sample (also called control sample: intermediate stage of arrest) and the other two samples (cauliflower and broccoli stages of arrest), normalized to 18S, are shown in Fig. 2. BoAP1-a and BoAP1-c transcripts accumulated significantly (P=0.0006 and P=0.0018, respectively) once the floral primordium initiated. In floral primordium, up-regulation of BoAP1-a and BoAP1-c transcripts increased by a factor of 23 and 21, correspondingly, with respect to transcript levels at the inflorescence meristem stage. Levels of expression were maintained at the floral bud stage of arrest (Fig. 2A). BoCAL had its maximum expression at the inflorescence meristem–floral primordium stages and thereafter declined significantly (P=0.02) by a factor of 2.3 as the reproductive meristem formed floral buds (Fig. 2A). AP2 transcripts reached the highest expression levels at the floral bud stage. Abundance was higher by a factor of 2 (P=0.024) with respect to the intermediate and inflorescence meristem stages of arrest (Fig. 2A). BoLFY was expressed equally in inflorescence meristem and floral primordium stages. Expression was lower by a factor of 8.5 (P <0.0001) in the floral buds. BoLFY reached its maximum expression in the initial stage of reproductive development, the vegetative to reproductive transition (Fig. 2B). BoTFL1 decreased expression by a factor of 2.1 (P=0.008) from floral primordium to floral bud (Fig. 2B).
|
‘Green Harmony F1’ forms floral buds and complete flowers (Fig. 3), even though it is a boap1-a boap1-c bocal triple mutant. Candidate complementary genes specifying floral initiation are FUL or UFO. All four BoFUL paralogues were expressed at all stages of arrest. BoFUL-c and BoFUL-d were the most abundant transcripts. BoFUL-b, BoFUL-c, and BoFUL-d had maximum expression at the inflorescence meristem stage. Their transcript levels decreased significantly (P=0.0001) by a factor of 6.4, 3.66, and 4.23, respectively, from inflorescence meristem to floral primordium. From floral primordium to floral bud, the decrease in the transcript abundance was significant for the three genes [P=0.0007 (BoFUL-b), P=0.0004 (BoFUL-c), and P=0.0017 (BoFUL-d)] by a factor of 2.49, 3.15, and 3.01, respectively (Fig. 2C). BoFUL-a had a slightly different pattern. Its maximum transcript levels were reached at inflorescence meristem stage and it was maintained until plants developed floral buds. At this stage, the gene was down-regulated significantly (P=0.035) by a factor of 2 with respect to its expression levels at the intermediate stage of arrest (Fig. 2C). UFO was expressed at very low levels at all developmental stages, with no significant differences among them (Fig. 2D).
|
Two genes specifically associated with cauliflower curd phenotype were analysed. CCE1 transcript levels were equally high in inflorescence meristem and floral primordium. However, it was lower by a factor of 110 (P <0.0001) in floral buds (Fig. 2D). BoREM1 expression was highest at the inflorescence meristem stage, and transcripts were detectable in floral primordium and floral bud stages. BoREM1 expression was 4.2-fold lower (P=0.0006) in floral primordium than in inflorescence meristem, and a further 4.9-fold lower in floral buds (P=0.0003) (Fig. 2D).
Experiment 2: gene expression at different temperature regimes and at the same developmental stage
As a control to determine whether the temperature exposure used to generate different developmental stages directly caused differences in gene expression, two tests were carried out. First, expression was measured in cauliflower plants from the accession ‘HRI 5295’, which had the same phenotype of floral primordia at all three temperature regimes used during reproductive growth (comparable with the intermediate stage represented in Fig. 1B). Most of the genes under study had no significant differences in expression (Fig. 4). Only BoAP1-a had a significant (P=0.0098) 2.4-fold increase in the abundance of transcripts between the 22 °C/17 °C and 16 °C/12 °C treatments (Fig. 4A). Second, ‘Green Harmony F1’ was exposed to the three temperature regimes for 24 h after heads formed, and development arrested at floral primordia–floral bud stages (set 1, small size floral buds; set 2, very small floral buds; set 3, medium size floral buds.). There were no statistically significant differences in the expression of the genes (Fig. 5). While BoAP1-a transcript abundance increased 4-fold in set 2 and set 3 between the 28 °C/22 °C and 22 °C/17 °C treatments, this difference was not statistically significant.
|
|
Experiment 3: effect of BoAP1 and BoCAL genotype on expression of BoFUL and BoTFL1
This experiment was designed to test whether BoFUL accumulates to compensate for the absence of BoCAL and BoAP1 or whether it is developmentally regulated independently of these genes. It also tests whether BoTFL1 expression is greater when BoCAL is mutant, as would be expected if the primary role of CAL is to repress TFL1. Expression of the genes was measured in a DH population segregating for BoAP1-a, BoAP1-c, and BoCAL, where individual DH lines were selected for representative inflorescence phenotypes (Table 1).
The BoCAL genotype did not affect expression of the four BoFUL paralogues (Fig. 6). The observed variation was not statistically significant (Table 3). The phenotype was also independent of the BoCAL genotype (2=2.598, df=1, P=0.1078). The BoAP1 genotype, in contrast, was significantly correlated (Table 3) with increased expression of BoFUL-b, BoFUL-c, and BoFUL-d (Fig. 6). The variation in BoFUL-b, -c, and -d expression due to BoAP1 genotype could not be distinguished from the developmental stage of the curd because the phenotype was dependent on the BoAP1 genotype (2=21.55, df=1, P <0.0001).
|
|
BoTFL1 transcript abundance was not associated with BoCAL genotype (Fig. 7). The two lines with the highest expression of BoTFL1, L475 (CCAA), and L17 (ccAA), were arrested at the floral primordium stage, while the other lines were arrested at the floral bud stage (Table 1).
|
|
Discussion |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionReferences |
|---|
Conservation of the sequence in homeotic genes from several species does not necessarily indicate conservation in function. What should be expected from species in the same family that share flower morphology, such as Arabidopsis and B. oleracea? This study tested a model of arrest in B. oleracea that incorporated homologues of the key genes involved in the Arabidopsis floral transition: AP1, CAL, FUL, LFY, UFO, AP2, and TFL1, as well as the cauliflower curd-specific genes CCE1 and BoREM1. Since some of these genes are present in multiple copies in B. oleracea, the expression pattern of all the genes and their paralogues was examined in combination and in tissue of the same genotype that was arrested at different developmental stages by varying the temperature regime.
Is the initiation of floral primordium development specified by an increase in BoLFY expression in B. oleracea? Does BoTFL1 maintain arrest in cauliflower by suppressing genes that may be involved in floral primordium specification (BoAP1, BoCAL, and BoLFY)?
Maintenance of the inflorescence meristem in cauliflower could be caused by BoLFY expression being below the threshold required to trigger floral primordium initiation, expression being repressed by high activity of BoTFL1. In Arabidopsis, LFY is responsible for the initial step in the reproductive transition specifying inflorescence meristem identity and floral primordium (Mandel et al., 1992; Kempin et al., 1995; Blazquez et al., 1997), TFL1 is a repressor of flowering and maintainer of indeterminate shoot meristem (Shannon and Meeks-Wagner, 1993; Bradley et al., 1997; Ratcliffe et al., 1999), and their expression is restricted to different zones through mutual suppression. In B. oleracea, BoLFY expression did not increasee in concert with a decline in BoTFL1 expression, as would be expected if mutual suppression resulted in the spatial separation observed in Arabidopsis. If the cauliflower stage of arrest was maintained by suppression of BoLFY expression, there would have been significantly higher BoLFY transcript levels in the intermediate stage, and if BoTFL1 prevented floral identity genes from being expressed, then its transcript abundance should have been lower at the intermediate stage than in the inflorescence meristem stage. That model can therefore be rejected (Fig. 8). The presence of BoLFY transcript at all stages is consistent with BoLFY being necessary for inflorescence meristem initiation, floral primordium specification, and B- and C-function [stamen and carpel formation (Coen and Meyerowitz, 1991)], as it is in Arabidopsis. BoTFL1 expression is consistent with a role in the enlargement of the curd rather than maintenance of the arrest.
|
Another model tested is whether the reproductive transition occurs because BoTFL1 is repressed by BoAP1 or BoCAL. In Arabidopsis, AP1 and CAL suppress the repressor gene TFL1 (Fig. 8) (Shannon and Meeks-Wagner, 1993; Bradley et al., 1997; Ratcliffe et al., 1999). If this regulation occurs in B. oleracea, the absence of BoCAL would allow greater BoTFL1 expression, leading to the arrest in cauliflower curds. However, whether the developmental stage was varied by environment (Fig. 2B) or by genotype (Fig. 7), the pattern of expression of BoTFL1 was consistently associated with the developmental stage of the curd and independently of the BoCAL genotype. Furthermore, the pathway that up-regulates BoAP1-a and initiation of floral primordia occurs, yet BoTFL1 is expressed in all stages of development. None of these results is consistent with BoTFL1 being regulated by BoAP1 or BoCAL, or with the transition being a consequence of BoTFL1 suppression.
The expression of BoTFL1 also differs from that observed in Brassica napus, in which the gene is expressed predominantly in flowers (Mimida et al., 1999). In both Brassica species the expression pattern is inconsistent with TFL1 being a repressor of flowering as it is in Arabidopsis.
Is the initiation of floral primordium development specified by an increase in BoAP1 and BoCAL expression in B. oleracea?
Genetic evidence for BoAP1 and BoCAL in specifying floral primordium initiation is ambiguous. One element of such specification would be that conditions promoting floral initiation would cause up-regulation. In mutant genotypes in which the mutation affects the post-transcription of the gene, as in BoAP1 and BoCAL, such up-regulation would be without direct effect, but indicative of the upstream regulatory pathways. In B. oleracea, the pattern of expression of BoAP1-a and BoAP1-c is consistent with a regulatory mechanism that triggers BoAP1 expression to initiate floral primordia and with both paralogues being under the same regulatory mechanism. The presence of BoAP1-c transcripts in floral primordia is novel, since this gene has previously only been identified with bract formation, not with flower bud development (Smith and King, 2000; Kop, 2003). Expression of BoCAL (Fig. 2A) was not consistent with up-regulation to initiate floral primordium. The different expression patterns of BoCAL and BoAP1 suggest that they are under independent regulatory control (Fig. 8).
Is BoAP1 associated with an A-function gene in B. oleracea? Can floral primordia develop when both BoAP1 and BoCAL are mutant, presumably by the action of other redundant genes?
AP1 is an A-function gene controlling the differentiation of sepals and petals in the two outer whorls of new flowers (Irish and Sussex, 1990; Meyerowitz et al., 1991; Mandel et al., 1992; Liljegren et al., 1999). The presence of BoAP1-a and BoAP1-c transcripts in the inflorescence meristem is inconsistent with a classic A-function. In Arabidopsis, AP1 is not expressed in the inflorescence meristem (Fig. 8). The A-function concept has been already questioned by several authors (Gutierrez-Cortines and Davies, 2000; Litt and Irish, 2003). SQUA in Antirrhinum (Huijser et al., 1992) and VAP1 in grapevine (Calonje et al., 2004) do not have the A-function. Functional analysis of BoAP1-a and BoAP1-c in Arabidopsis is clearly warranted to determine whether these genes have A activity.
In Arabidopsis, ap1 mutant flowers are partially converted to inflorescence shoots, and sepal and petal development is disrupted (Bowman et al., 1993). In the boap1-a boap1-c bocal triple mutant cultivar Green Harmony F1, plants form floral primordia and floral buds (Fig. 3). The organ-specifying function of these genes must be fulfilled by some other MIGs. A prime candidate for the complementary MIG is BoFUL.
Is the initiation of floral primordium development specified by an increase in BoFUL expression in B. oleracea? Are BoFUL paralogues redundant? Is BoFUL expression associated with BoAP1 or BoCAL genotype?
In Arabidopsis, FUL has a redundant role with AP1, CAL, and LFY in promoting flowering time and floral primordium initiation. It is also involved in cauline leaf morphology and carpel/fruit development (Gu et al., 1998; Ferrandiz et al., 2000). Transcripts are abundant in the organs where it is active (Fig. 8) (Mandel and Yanofsky, 1995; Gu et al., 1998; Ferrandiz et al., 2000). Since FUL is a multifunctional gene in Arabidopsis, multiple copies of this gene in other species could increase specialization through subfunctionalization. The presence of four BoFUL paralogues could support even more. However, BoFUL-b, BoFUL-c, and BoFUL-d had identical patterns of expression, while BoFUL-a had a slightly different one (Figs. 2C, 8).
The steady decreases in expression from inflorescence meristem towards floral bud imply that these genes act in a threshold manner, as do BoLFY and BoAP1, requiring higher levels for inflorescence specification and lower levels for floral primordium development. For BoFUL to overcome the absence of BoAP1 and BoCAL, the level of expression should have been higher or equal in inflorescence meristem at the onset of floral primordium. The only paralogue with that pattern was BoAP1-a. The expression pattern for all four paralogues differed from the Arabidopsis model (Fig. 8) by being present in floral primordium and floral bud.
BoFUL expression may be suppressed by BoAP1 or BoCAL. In Arabidopsis, AP1 negatively regulates FUL in emerging flower primordia (Mandel and Yanofsky, 1995), and in proliferating meristems of ap1 cal double mutants FUL is ectopically expressed, allowing the plants to form flowers (Ferrandiz et al., 2000). In order to test this hypothesis, BoFUL expression was quantified in DH lines with different BoAP1/BoCAL genotypic combinations (Table 1). BoCAL genotype did not have any significant association with the expression of any of the four BoFUL paralogues. BoAP1 genotype, on the other hand, was significantly correlated with the expression pattern observed for BoFUL-b, BoFUL-c, and BoFUL-d (Table 3); however, BoAP1 genotype was highly correlated with the phenotype. In ‘Green Harmony F1’ plants (Fig. 2C), BoFUL expression changed with the stage of the arrest, suggesting that expression of BoFUL paralogues was associated with the developmental stage of the curd and not with BoAP1 or BoCAL.
Have AP2 and UFO conserved their roles as both floral primordium promoters and organ identity genes in B. oleracea as they do in Arabidopsis?
There are additional candidate genes for initiating floral primordium and specifying sepal identity. In Arabidopsis, AP2 (Irish and Sussex, 1990; Bowman et al., 1993; Shannon and Meeks-Wagner, 1993) and UFO (Levin and Meyerowitz, 1995; Wilkinson and Haughn, 1995; Samach et al., 1999) act redundantly with the MADS-box gene AP1 in the specification of floral primordium. If these genes have conserved this role in B. oleracea, then both genes should increase their expression at the floral primordium stage for the flower meristem to initiate (Fig. 8). However, the expression pattern observed in the mutant plants ‘Green Harmony F1’ does not support AP2 or UFO as alternative genes to specify floral meristem identity. In B. oleracea, the AP2 pattern is consistent with a role for the gene in organ identity, but not in flower initiation (Fig. 2A).
Are CCE1 or BoREM1 associated with arrest at the inflorescence meristem stage?
A gene that prevents floral initiation and maintains the cauliflower curd would be expressed preferentially in the meristems of the curd, decreasing expression after resumption of floral development. This study shows clearly that CCE1 is not the gene responsible for maintaining the cauliflower arrest since its down-regulation was observed later in development, when floral buds developed. CCE1 could have a putative role in the suppression of class B- and C-function genes. How CCE1 fulfils this role is an interesting question since this protein belongs to a new, so far unidentified, family of transmembrane receptors (Palmer et al., 2001).
The BoREM1 pattern (Fig. 2D) indicates that its function is probably threshold dependent, like BoFUL and BoLFY, and seems to be involved in both inflorescence meristem specification and floral primordium initiation. Like LFY in Arabidopsis, BoREM1 may also be a transcription factor that up-regulates floral meristem genes since the protein sequence has a potential leucine zipper and a motif that resembles a localization signal found in proteins transported to the nucleus (Franco Zorilla et al., 1999).
It is possible that the changes in gene expression are not the result of regulatory processes downstream of the temperature-sensing process, but are direct temperature effects on these genes. To account for that possibility, expression was analysed after varying the temperature without variation in the developmental stage. Temperature was varied throughout reproductive growth in a non-responsive genotype (Fig. 4), or for 24 h in the same genotype (Fig. 5). There was no variation in the expression of the genes studied, resulting in only small significant differences in BoAP1-a. The major differences in gene expression observed in experiment 1 are therefore associated with the developmental stage of the curd.
The long-term exposure resulted in a small decrease in the abundance of BoAP1-a transcripts between the middle and cool temperature (Fig. 4A). Between the same temperatures when they caused a developmental change, there was no expression change to account for. ‘Green Harmony F1’ is homozygous mutant at BoAP1-a, BoAP1-c, and BoCAL, while ‘HRI 5295’ is a homozygous mutant at BoAP1-a, but segregates for BoAP1-c and BoCAL. It is possible that one of these genes could mask a temperature-dependent pathway. In the short-term experiment, BoAP1-a transcript abundance did not change significantly (Fig. 5). The numerical difference between the high and middle temperatures was only 20% of that between the same temperatures when they caused a developmental change. Therefore, even the largest change in this control fails to support direct temperature regulation.
In B. oleracea, bracteole initiation (‘fuzzy’ and ‘leafy’ heads) caused by high temperatures has been associated with a decline of BoLFY and BoAP1 expression in the shoot apex (Anthony et al., 1996) resembling the effect caused by lfy mutations in Arabidopsis plants (partial conversion of the floral primordium to inflorescence meristem accompanied by bracts and cauline leaves) (Weigel et al., 1992; Bowman et al., 1993; Kempin et al., 1995; Liljegren et al., 1999). In this study, BoLFY expression was not affected by high temperature.
Future work should determine with more certainty whether temperature directly regulates BoAP1-a but not BoAP1-c. If so, the regulation of these genes could provide a good model for discovering regulatory pathways sensitive to moderate temperatures.
Even though Arabidopsis and B. oleracea are members of the same family and share flower morphology, reproductive development in B. oleracea cannot be explained with the Arabidopsis model. In this model, TFL1 maintains the indeterminate inflorescence fate, LFY specifies inflorescence meristem, the MADS-box genes AP1, CAL, and FUL redundantly with the non-MADS-box genes LFY, AP2, and UFO specify floral development, and AP1 with AP2 are A-function genes involved in formation of sepals and petals. All of these gene functions are inconsistent with the expression of the Brassica homologues (Fig 8). The model must be expanded to identify additional genes. Of particular interest as candidate genes for the control of floral primordium initiation are the recently discovered genes in Arabidopsis: LATE MERISTEM IDENTITY genes (LMIs) (Saddic et al., 2006).
|
Acknowledgements |
|---|
The authors would like to thank Drs Graham Teakle and Graham King for supplying seeds of the NxB DH population lines, and Dr Graham King for providing the assay used for genotyping BoAP1-c. Also, we acknowledge Dr Jason Soderblom for providing the efficiency calculation program. Special thanks to Joseph Shail for much help in growing the plants and constant advice and support, and Drs John Barnard and Francoise Vermeylen for their help in the statistical analyses of these data. This work was funded by USDA Hatch project NYG-632412.
|
Abbreviations |
|---|
Ct, threshold cycle; DH, double haploid; HSD, honestly significant difference; MIG, meristem identity gene; NG, normalizer gene; SEMs, standard error of the means; TG, target gene.
|
References |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionReferences |
|---|
Anthony RG, James PE, Jordan BR. Cloning and sequence analysis of a FLO/LFY homologue isolated from cauliflower (Brassica oleracea L. var. botrytis). Plant Molecular Biology (1993) 22:1163–1166.[CrossRef][ISI][Medline]
Anthony RG, James PE, Jordan BR. The cDNA sequence of a cauliflower apetala-1/squamosa homolog. Plant Physiology (1995) 108:441–442.[CrossRef][ISI][Medline]
Anthony RG, James PE, Jordan BR. Cauliflower (Brassica oleracea var. botrytis L.) curd development: the expression of meristem identity genes. Journal of Experimental Botany (1996) 47:181–188.
Ausubel FM, Brent R, Kingston RE, Moore DD, Seidman JG, Smith JA, Struhl K. Current protocols in molecular biology (1994) Vol. I. Brooklyn, NY: John Wiley & Sons, Inc.
Björkman T, Pearson KJ. High temperature arrest of inflorescence development in broccoli (Brassica oleracea var. italica L.). Journal of Experimental Botany (1998) 49:101–106.
Blazquez MA, Soowal LN, Lee I, Weigel D. LEAFY expression and flower initiation in Arabidopsis. Development (1997) 124:3835–3844.[Abstract]
Boodley JB, Sheldrake R. Cornell peat-lite mixes for commercial plant growing. In: Cooperative Extension Division Informative Bulletin 43 (1977) Cornell University.
Booij R, Struik PC. Effects of temperature on leaf and curd initiation in relation to juvenility of cauliflower. Scientia Horticulturae (1990) 44:201–214.[CrossRef]
Bowman JL, Alvarez J, Weigel D, Meyerowitz EM, Smyth DR. Control of flower development in Arabidopsis thaliana by APETALA1 and interacting genes. Development (1993) 119:721–743.[Abstract]
Bradley D, Ratcliffe O, Vincent C, Carpenter R, Coen E. Inflorescence commitment and architecture in Arabidopsis. Science (1997) 275:80–83.
Bustin SA. Absolute quantification of mRNA using real-time reverse transcription polymerase chain reaction assays. Journal of Molecular Endocrinology (2000) 25:169–193.[Abstract]
Calonje M, Cubas P, Martinez-Zapater JM, Carmona MJ. Floral meristem identity genes are expressed during tendril development in grapevine. Plant Physiology (2004) 135:1491–1501.
Carr SM, Irish VF. Floral homeotic gene expression defines developmental arrest stages in Brassica oleracea L. vars. botrytis and italica. Planta (1997) 201:179–188.[CrossRef][ISI][Medline]
Chenna R, Sugawara H, Koike T, Lopez R, Gibson TJ, Higgins DG, Thompson JD. Multiple sequence alignment with the Clustal series of programs. Nucleic Acids Research (2003) 31:3497–500.
Coen ES, Meyerowitz EM. The war of the whorls: genetic interactions controlling flower development. Nature (1991) 353:31–37.[CrossRef][Medline]
Doyle JJ, Doyle JL. A rapid DNA isolation procedure from small quantities of fresh leaf tissues. Phytochemical Bulletin (1987) 19:11–15.
Ferrandiz C, Gu Q, Martienssen R, Yanofsky MF. Redundant regulation of meristem identity and plant architecture by FRUITFULL, APETALA1 and CAULIFLOWER. Development (2000) 127:725–734.[Abstract]
Franco-Zorrilla JM, Fernández-calvín B, Madueño F, Cruz-Alvarez M, Salinas J, Martínez-Zapater JM. Identification of genes specifically expressed in cauliflower reproductive meristems. Molecular characterization of BoREM1. Plant Molecular Biology (1999) 39:427–436.[CrossRef][ISI][Medline]
Fujime Y, Okuda N. The physiology of flowering in Brassicas, especially about cauliflower and broccoli. Acta Horticulturae (1996) 407:247–254.
Grevsen K, Olesen JE, Veierskov B. The effects of temperature and plant developmental stage on the occurrence of the curd quality defects ‘bracting’ and ‘riciness’ in cauliflower. Journal of Horticultural Science and Biotechnology (2003) 78:638–646.
Gu Q, Ferrandiz C, Yanofsky MF, Martienssen R. The FRUITFULL MADS-box gene mediates cell differentiation during Arabidopsis fruit development. Development (1998) 125:1509–1517.[Abstract]
Gutierrez-Cortines ME, Davies B. Beyond the ABCs: ternary complex formation in the control of floral organ identity. Trends in Plant Science (2000) 5:471–476.[CrossRef][ISI][Medline]
Huijser P, Klein J, Lonnig WE, Meijer H, Saedler H, Sommer H. Bracteomania, an inflorescence anomaly, is caused by the loss of function of the MADS-box gene squamosa in Antirrhinum majus. EMBO Journal (1992) 11:1239–1249.[ISI][Medline]
Irish VF, Sussex IM. Function of the APETALA-1 gene during Arabidopsis floral development. The Plant Cell (1990) 2:741–753.
Jordan BR, Anthony RG, James PE. Control of floral morphogenesis in cauliflower (Brassica oleracea L. var. botrytis). In: Molecular and cellular aspects of plant reproduction—Scott R, Stead AD, eds. (1994) Cambridge: Cambridge Academic Press. 17–29.
Kempin SA, Savidge B, Yanofsky MF. Molecular basis of the cauliflower phenotype in Arabidopsis. Science (1995) 267:522–525.
Kop EP, Teakle GR, McClenaghan R, Lynn JR, King GJ. Genetic analysis of the bracting trait in cauliflower and broccoli. Plant Science (2003) 164:803–808.
Ku H. Notes on the use of propagation of error formulas. Journal of Research of National Bureau of Standards—C. Engineering and Instrumentation (1966) 70C:263–273.
Labate J, Robertson L, Baldo A, Björkman T. Inflorescence identity genes alleles are poor predictors of inflorescence type in broccoli and cauliflower. Journal of the American Society for Horticultural Science (2006) 131:667–673.
Lan TH, Paterson AH. Comparative mapping of quantitative trait loci sculpting the curd of Brassica oleracea. Genetics (2000) 155:1927–1954.
Liljegren SJ, Gustafson-Brown C, Pinyopich A, Ditta GS, Yanofsky MF. Interactions among APETALA1, LEAFY, and TERMINAL FLOWER1 specify meristem fate. The Plant Cell (1999) 11:1007–1018.
Litt A, Irish VF. Duplication and diversification in the APETALA1/FRUITFULL floral homeotic gene lineage: implications for the evolution of floral development. Genetics (2003) 165:821–833.
Lowman AC, Purugganan MD. Duplication of the Brassica oleracea APETALA1 floral homeotic gene and the evolution of domesticated cauliflower. Journal of Heredity (1999) 90:514–520.
Mandel MA, Gustafson-Brown C, Savidge B, Yanofsky MF. Molecular characterization of the Arabidopsis floral homeotic gene APETALA1. Nature (1992) 360:273–276.[CrossRef][Medline]
Mandel MA, Yanofsky MF. The Arabidopsis AGL8 MADS-box gene is expressed in inflorescence meristems and is negatively regulated by APETALA1. The Plant Cell (1995) 7:1763–1771.[Abstract]
Meyerowitz EM, Bowman JL, Brockman LL, Drews GN, Jack T, Sieburth LE, Weigel D. A genetic and molecular model for flower development in Arabidopsis thaliana. Development (1991) (Supplement 1):157–167.
Mimida N, Sakamoto W, Murata M, Motoyoshi F. TERMINAL FLOWER 1-like genes in Brassica species. Plant Science (1999) 142:155–162.
Palmer JE, Dikeman DA, Fujinuma T, Kim B, Jones JI, Denda M, Martinez-Zapater JM, Cruz-Alvarez M. A Brassica oleracea gene expressed in a variety-specific manner may encode a novel plant transmembrane receptor. Plant and Cell Physiology (2001) 42:404–413.
Pfaffl MW. A new mathematical model for relative quantification in real-time RT-PCR. Nucleic Acids Research (2001) 29:e45.
Purugganan MD, Boyles AL, Suddith JI. Variation and selection at the CAULIFLOWER floral homeotic gene accompanying the evolution of domesticated Brassica oleracea. Genetics (2000) 155:855–862.
Ramakers C, Ruijter JM, Deprez RH, Moorman AF. Assumption-free analysis of quantitative real-time polymerase chain reaction (PCR) data. Neuroscience Letters (2003) 339:62–66.[CrossRef][ISI][Medline]
Ratcliffe OJ, Bradley DJ, Coen ES. Separation of shoot and floral identity in Arabidopsis. Development (1999) 126:1109–1120.[Abstract]
Rozen S, Skaletsky HJ. Primer3 on the WWW for general users and for biologist programmers. In: Bioinformatics methods and protocols: methods in molecular biology—Krawetz S, Misener S, eds. (2000) Totowa, NJ: Humana Press. 365–386.
Saddic LA, Huvermann B, Bezhani S, Su Y, Winter CM, Kwon CS, Collum RP, Wagner D. The LEAFY target LMI1 is a meristem identity regulator and acts together with LEAFY to regulate expression of CAULIFLOWER. Development (2006) 133:1673–1682.
Sadik S. Morphology of the curd of cauliflower. American Journal of Botany (1962) 49:290–297.[CrossRef][ISI]
Samach A, Klenz JE, Kohalmi SE, Risseeuw E, Haughn GW, Crosby WL. The UNUSUAL FLORAL ORGANS gene of Arabidopsis thaliana is an F-box protein required for normal patterning and growth in the floral meristem. The Plant Journal (1999) 20:433–445.[CrossRef][ISI][Medline]
Shannon S, Meeks-Wagner DR. Genetic interactions that regulate inflorescence development in Arabidopsis. The Plant Cell (1993) 5:639–655.
Smith LB. The molecular genetics of curd morphology and domestication of cauliflower (Brassica oleracea L. var. botrytis L.) (1999) PhD thesis, University of Warwick.
Smith LB, King GJ. The distribution of BoCAL-a alleles in Brassica oleracea is consistent with a genetic model for curd development and domestication of the cauliflower. Molecular Breeding (2000) 6:603–613.[CrossRef][ISI]
Vidal JR, Kikkert JR, Donzelli BD, Wallace PG, Reisch BI. Biolistic transformation of grapevine using minimal gene cassette technology. Plant Cell Reports (2006) 25:807–814.[CrossRef][ISI][Medline]
Weigel D, Alvarez J, Smyth DR, Yanofsky MF, Meyerowitz EM. LEAFY controls floral meristem identity in Arabidopsis. Cell (1992) 69:843–859.[CrossRef][ISI][Medline]
Wilkinson MD, Haughn GW. UNUSUAL FLORAL ORGANS controls meristem identity and organ primordia fate in Arabidopsis. The Plant Cell (1995) 7:1485–1499.[Abstract]
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||