| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
|
First published online November 11, 2004; 10.1105/tpc.104.026724 © 2004 American Society of Plant Biologists
Characterization of Antirrhinum Petal Development and Identification of Target Genes of the Class B MADS Box Gene DEFICIENS
|
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONRESULTSDISCUSSIONMETHODSREFERENCES |
|---|
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTSDISCUSSIONMETHODSREFERENCES |
|---|
; Jack, 2004). In Antirrhinum, two homeotic genes DEFICIENS (DEF; Sommer et al., 1990; Schwarz-Sommer et al., 1992) and GLOBOSA (GLO; Tröbner et al., 1992), the so-called class B genes, control petal and stamen organogenesis. Their RNA and protein expression patterns overlap and are maintained until late stages of petal and stamen development (Zachgo et al., 1995). The def and glo mutants show identical phenotypes, with petals transformed into sepaloid organs and stamens transformed into carpeloid structures. Additionally, initiation of carpel organogenesis in the center of the flower is dependent on DEF and GLO function because no fourth whorl organs are formed in the respective mutants. The DEF and GLO proteins heterodimerize and were shown to bind in vitro as dimers to short conserved DNA elements, called CArG-boxes (Schwarz-Sommer et al., 1992; Tröbner et al., 1992). CArG-box elements are also located in their own promoters and are presumed to mediate maintenance of late expression by an autoregulatory mechanism (Schwarz-Sommer et al., 1992; Zachgo et al., 1995). The DEF/GLO heterodimer associates with the MADS box protein SQUAMOSA via their C termini, resulting in an increased DNA binding affinity. This ternary complex requires for binding the presence of two CArG-boxes, located, for instance, in the GLO promoter (Egea-Cortines et al., 1999).
Although class B genes from Antirrhinum and various other species have been intensively studied, still little is known about the target genes that realize their regulatory potential during petal and stamen organogenesis. A modest number of putative class B target genes that are preferentially expressed either in petals or stamens was isolated by differential screening strategies from Antirrhinum, Arabidopsis, and Gerbera hybrida (Nacken et al., 1991a, 1991b; Rubinelli et al., 1998; Sablowski and Meyerowitz, 1998; Kotilainen et al., 1999). In vivo proof for direct regulation of NAP (for NAC-LIKE, ACITVATED BY AP3/PI) by the Arabidopsis class B protein APETALA3 (AP3; Jack et al., 1992) was delivered by an engineered steroid-inducible AP3 system (Sablowski and Meyerowitz, 1998). NAP was proposed to participate in the transition from cell division to cell elongation processes in petals and stamens.
Recently, several studies were performed using high-throughput genomic approaches to characterize flower development in roses, Arabidopsis, and Iris hollandica (Channelière et al., 2002; Guterman et al., 2002; Schmid et al., 2003; van Doorn et al., 2003; Zik and Irish, 2003; Wellmer et al., 2004). Arabidopsis expression profiling studies were performed with probes from different floral mutant inflorescences to identify petal- and stamen-specific genes. The number of the identified genes that depend on the activity of the Arabidopsis class B genes AP3 and PISTILLATA (PI; Goto and Meyerowitz, 1994) throughout petal and stamen development varies between 
200 and 1100 (Zik and Irish, 2003; Wellmer et al., 2004). Antirrhinum petals served as a key model system to analyze flower asymmetry, scent, and pigmentation production as well as epidermal cell differentiation leading to conical cell formation contributing to their velvet sheen (Noda et al., 1994; Luo et al., 1999; Dudareva et al., 2000). Whereas growth of early petal stages has been reported from different species to be realized by cell division processes, later growth is mainly achieved by cell elongation (Martin and Gerats, 1993; Ben-Nissan and Weiss, 1996; Rolland-Lagan et al., 2003). These examples illustrate that the formation of different tissues with distinctive functions requires a tight spatial and temporal regulation that is likely reflected by dynamic transcriptome changes throughout petal organogenesis. A better understanding of organ formation would thus profit from a temporally and spatially restricted target gene analysis to avoid blurring of differences in mRNA expression levels. The large size of the Antirrhinum flower allows analysis of dissected stages and organs. Moreover, Antirrhinum offers a large flower mutant collection, forward and reverse genetic tools, and a molecular linkage map, and very recently a large EST collection has been established (for a review, see Schwarz-Sommer et al., 2003).
Here, we present an Antirrhinum genomics approach using macroarrays with >11,600 spotted Antirrhinum unigenes. Toward a better understanding of the molecular mechanisms controlled by DEF during petal organogenesis, we conducted a two-step profiling procedure. First, late petal and sepal development was compared. Class B gene expression in sepals conditions petal organ fate, as shown by the class B mutant phenotypes and by ectopic petal formation in the first whorl upon class B overexpression (Sommer et al., 1990; Davies et al., 1996). More than 500 ESTs were identified representing target genes directly and indirectly controlled by DEF. For determination of more directly controlled target genes, profiling experiments were conducted with petals from the conditional def-101 mutant. More than 100 upregulated and downregulated DEF petal target genes were identified. Further expression analyses of selected target genes corroborated their dependence on the DEF function and reflect the broad spectrum of basic cellular processes contributing to petal development.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTSDISCUSSIONMETHODSREFERENCES |
|---|
). Flowers of def-101 mutants cultivated at the permissive temperature (15°C) display an almost wild type–like morphology, with upper and lower lopes being slightly reduced in size and less strongly folded (Figures 1A and 1B). Def-101 plants grown at the nonpermissive temperature (26°C) resemble DEF null-mutants (Figure 1C). Higher temperature was shown to affect the stability of the DEF/GLO heterodimer, causing its rapid degradation and thus made the def-101 mutant an ideal tool for target gene identification.
|
1 cm (defined as stage 3, see below) and were then grown until anthesis at the nonpermissive temperature. Figure 1D shows a wild type–like def-101 flower before reduction of the DEF function. At this stage, all floral organs are formed and stamens already developed microspores and are about to reach their final length by filament elongation. After 4 d of cultivation at the nonpermissive temperature, carpeloid structures were formed close to the base of the filaments (arrows in Figure 1E), and petal development is disturbed. In comparison with def-101 plants cultivated exclusively in the cold (Figure 1F), reduction of the DEF function for 72 h affected coloration and shape of petals. For instance, formation of greenish sectors was observed, indicating loss of petal identity and transformation toward sepaloid organs (Figure 1G). The shortest time span of DEF function sufficient to restore a morphological aspect of DEF control was determined by double shift experiments. Plants were grown at the nonpermissive temperature, and plants with 1-cm-long inflorescences were shifted transiently for 12, 24, 48, and 72 h to the permissive temperature before cultivation until flower maturation at the nonpermissive temperature. Transient activation of the DEF function for 24 h was sufficient to regulate downstream targets that control meristematic activities in the center of the flower, such that one flower per investigated inflorescence initiated a normal fourth whorl (Figure 1H). By contrast, second and third whorl organs stayed transformed, even after extended periods of DEF upregulation for 72 h (data not shown). Thus, an early and transient DEF function is required to initiate fourth whorl development that subsequently does not depend on DEF activity. By contrast, a continuous DEF function is required until late stages to maintain normal petal and stamen differentiation. These observations emphasize the importance of conducting spatiotemporal-specific studies to identify target genes involved in different regulatory processes. We took advantage of the large Antirrhinum flower and used petal organs from distinct stages to analyze late stages of petal morphogenesis and to identify the genes that realize the regulatory potential of DEF.
Characterization of Late Petal Development in Antirrhinum by Expression Profiling
Macroarrays were produced by spotting 14,186 Antirrhinum ESTs representing 11,615 partially sequenced unigenes from different vegetative and floral organs as double spots on nylon filters. The array filters represent 40% of the estimated total Antirrhinum cDNA number, assuming that the number of expressed genes is comparable between Arabidopsis and Antirrhinum.
First, we characterized petal development by comparing sepal and petal transcriptomes. By this approach, all identified differentially expressed ESTs represent genes whose transcription is directly or indirectly controlled by the DEF/GLO proteins. To increase the sensitivity and resolution, late petal development was divided into four distinct stages (Figures 2A and 2B). In stage 1 (bud length 0.5 to 0.8 cm), petals are still enclosed by sepals, whereas stage 2 (0.8 to 1.0 cm) is characterized by an outgrowth of petals. At stage 3 (1.0 to 2.0 cm), rapid growth of the flower starts, reflected by doubling flower length within 3 d. During stage 4 (2.0 to 4.0 cm), final growth until anthesis of the flower is realized. Petal stages 1 and 2 match the late phase F and stage 3 and 4 the phase G, which were recently defined when a temporal and morphological framework was established for the entire Antirrhinum flower development (Vincent and Coen, 2004). Petal tissue from the four stages was harvested separately. Sepals were pooled and used as one sample for comparison as no morphological differences affecting, for example, their size or trichome development could be observed during the analyzed stages.
|
). After normalization for each gene and experimental condition, the median of the data was calculated, and the significance of variations in RNA levels was assessed (Beissbarth et al., 2000; Fellenberg et al., 2001; see Methods for details). Quality of the data sets was further controlled by determination of the Pearson correlation coefficient, revealing good reproducibility among individual filters in the same experiment (Figures 3A and 3B). Gradual reduction of the coefficient values for the individual experimental comparisons reflects increased transcriptome differences throughout the investigated stages (Figures 3C to 3F).
|
; Lagorce et al., 2003). Independent expression analyses were conducted, confirming that a cutoff value of 1.2-fold change can be reliably applied in this data set for an intensity range that disregards 80% of the spotted genes as not expressed (Figure 6; see supplemental data online). Using this value, 537 ESTs were identified as being differentially expressed between sepals and petals in at least one stage during late flower development (see supplemental data online for list of all genes). This number of ESTs includes up to 22% of genes that were spotted twice on the filter. Figure 2C shows the numbers of ESTs differentially expressed between sepals and the four petal stages. Additionally, numbers of stage-specifically regulated ESTs are indicated. Counting all the activated and repressed nonredundant ESTs revealed 226 repressed and 322 activated ESTs (data not shown). Expression of 45% of the ESTs (104/226) is constantly reduced throughout late petal development, and only 15% of the ESTs (34/226) are repressed specifically during one developmental stage.
|
3% of the ESTs (11/322) are expressed significantly higher throughout all petal stages compared with sepals, whereas 60% (208/322) are upregulated exclusively during one petal stage (Figure 2C). The high value of 104 ESTs, activated in stage 1, includes 47 stage-specifically regulated ESTs and probably reflects the sampling. Because petals younger than stage 1 were not separately analyzed, transcript changes in stage 1 might have accumulated. After stage 1, the number of activated ESTs gradually increases from 85 (stage 2) to 128 (stage 3), and finally up to 174 (stage 4), coinciding with a steady increase in the proportion of the stage-specific ESTs. Whereas the contribution of the stage-specifically regulated genes in stage 2 is only 16% (14/85), this increases during stage 3 up to 26% (33/128) and, most strongly, up to 66% (114/174) during stage 4. By contrast, only <1% (1/156, stage 2), 3% (4/165, stage 3), and 13% (19/161, stage 4) are stage-specifically downregulated, indicating a higher uniformity of the repressed transcription mode compared with the activated mode. In summary, these data reveal a highly dynamic gene expression regulation, with a large proportion of transcripts being stage-specifically upregulated by DEF during late stages of Antirrhinum petal development.
Functional Annotation of Regulated Genes during Petal Development
To determine putative functions of differentially expressed cDNAs, Antirrhinum EST sequences were functionally annotated. A surrogate annotation approach was conducted, using as a basis the existing role categorization from the Munich Information Center for Protein Sequences (MIPS) available for the proteome of Arabidopsis. This commonly applied approach is based on the assumption that conserved sequences reflect functional relationships. Antirrhinum genes with an expect value (E-value) of <2.0 E–12 were grouped into 20 functional categories. Applying this E-value, no homolog could be identified for 36% (4226 genes) of the unigene set, and we assume that this group might comprise genes exerting functions specific for Antirrhinum. They were not included in a functional comparison conducted with the Arabidopsis unigene set (see Methods). Distribution of genes among the different groups is very similar between the two species (Figure 4), indicating that the Antirrhinum EST collection comprises a representative proportion of the totally expressed genes.
|
; Rolland-Lagan et al., 2003).
|
; Knox, 1995; Cassab, 1998). Identification of aquaporins, water channel proteins, and tonoplast intrinsic proteins reflects changing turgor conditions in growing petal cells. Interestingly, the frequency of the unclassified protein group increases throughout petal development. Whereas the available EST set includes 41% of unclassified ESTs, 56% of the ESTs belonging to this group are expressed in petals, and even higher percentages were identified as being stage-specifically expressed. This suggests that a large proportion of genes with a tight expression regulation during petal development exert an intriguing, yet uncharacterized, function during petal morphogenesis. The major group of ESTs that are continuously repressed in developing petals are related to photosynthesis, like ribulose-1,5-bisphosphate carboxylase/oxygenase (Rubisco) small chain, Rubisco activase, and a light harvesting chlorophyll a/b binding protein precursor (see supplemental data online). This is in accordance with the observation that loss of the DEF function at late petal stages still causes activation of the photosynthetic program as indicated by the observed regreening of second whorl organs (Figure 1G).
Identification of DEF Target Genes
Further profiling experiments were conducted with the def-101 mutant to identify genes that are more likely to represent direct targets of the DEF/GLO heterodimer as their expression level is altered upon abolishing the DEF function for different time spans. Petal stage 3 was selected for this target gene screen because the number of novel, stage-specifically upregulated genes increases at this stage, and the DEF function was shown to be still required to maintain normal petal morphogenesis. For petal probe preparation, def-101 plants grown at the permissive temperature until stage 3 were shifted for 0, 24, and 72 h to the nonpermissive temperature. Parallel experiments were performed with wild-type flowers to exclude temperature-sensitive genes from the studies (shown in supplemental data online).
Expression of 125 ESTs was significantly changed after reducing the DEF function for 24 and/or 72 h. Seventeen ESTs needed to be subtracted because they represented generally temperature-sensitive genes. The resulting EST number of 108 represents 101 unigenes, comprising 67 genes that were already recovered by the petal/sepal comparison. Forty-nine genes were identified as downregulated upon DEF reduction, thus representing activated target genes. A nearly equal proportion of 52 genes was upregulated and thus forms the group of repressed genes. Table 2 shows the relation between the representation of genes in the EST collection with those that are expressed in petals at stage 3 and with those specifically regulated in this stage. Ten of the 20 considered categories show similar percentages in all three data sets. In seven categories, target genes are overrepresented, most significantly in the groups metabolism, transport facilitation, energy, and control of cellular organization. As already revealed by the sepal/petal comparison, these data emphasize the diversity of DEF target gene regulation, including a large number of unclassified proteins with putative novel functions during petal morphogenesis.
|
Analysis of Activated and Repressed Target Genes
For a more detailed analysis, hierarchical clustering with the 101 target genes was performed (Figure 5). After reducing the DEF function for 24 h, the expression of 46 genes was significantly altered. The expression of 27 genes was upregulated and that of 19 genes downregulated. After 72 h, 93 genes were identified (45 upregulated and 48 downregulated). Out of these genes, 20 and 17 revealed a continuous upregulated and downregulated expression, respectively. Among the activated genes, we found two transcription factors (Figure 5A). After reducing the DEF function for 24 h, DEF expression was significantly decreased, caused by an ongoing autoregulatory control at this late stage (Zachgo et al., 1995). The other transcription factor also represents a MADS box gene, named DEFICIENSHOMOLOG84 (DEFH84), predominately expressed in petals and stamens of flowers at stage 3 (S. Zachgo, unpublished data). Given that the overall representation of the functional category transcription in the sepal/petal stage 3 comparison experiments was normal (Table 1), we assume that low recovery of transcription factors in the temperature shift experiments is not because of a sensitivity problem of the detection method. Rather, it indicates that DEF might control a broad variety of downstream target genes directly without establishing intermediate regulatory modules. A similar observation was reported for AP3 activity during petal development (Zik and Irish, 2003).
|
- and ß-tubulins that form microtubules (Mayer and Jürgens, 2002). Four different lipid transfer proteins (LTPs), known to mediate the transfer of phospholipids between membranes in vitro (Zachowski et al., 1998), were identified, of which the LTP 018_4_03_c07 responded most strongly toward a reduced DEF function. The expression level of two chitinases decreased after reducing the DEF function for 72 h. Chitinases are enzymes capable of cleaving chitin, a major cell wall constituent of many pathogenic fungi (Davis and Bartnicki-Garcia, 1984), and are thought to mediate disease resistance to chitin containing pathogens. Because their expression response occurred with delay, chitinases are probably not directly controlled by DEF. A group of photosynthesis related genes (Rubisco small chain, Rubisco activase, and photosystem II precursor proteins) identified as being repressed in petals by the sepal/petal comparison was upregulated upon reduction of the DEF function (Figure 5B). Rubisco activase expression especially responded strongly and rapidly to an altered DEF function. This observation is in accordance with the morphological consequences of late reduction of DEF activity (Figure 1G) and strengthens the importance of a continuous DEF function until late stages of petal development to suppress photosynthesis. Other identified genes were more diverse in their functions, including enzymes like ß-glucosidases, catalases, and acyl-CoA oxidases or giberellin-induced proteins and plasma membrane intrinsic proteins.
Expression Analysis of Selected Target Genes Corroborates Dependency on the DEF Function
Twenty-one genes were selected as representatives for different regulated processes to confirm by RT-PCR studies their dependence on the DEF function. For 19 out of 21 selected genes, expression differences detected after abolishing DEF activity for 24 and 72 h correlated with the differences detected using array experiments (nine are shown in Figure 6A in an extended analysis, 12 are shown in supplemental data online). Thus the M-CHiPS analysis (Fellenberg et al., 2002) for array data processing and quality control is a reliable method to detect genes that are regulated by DEF with changes down to 1.2-fold. This value applies to genes expressed above an intensity threshold that disregards 80% of the spotted genes as not being transcribed under the applied experimental conditions. Out of the 19 target genes, nine were selected that responded strongly and rapidly toward a reduced DEF function, including DEF itself. To further test the directness of their regulation by DEF, def-101 flowers were shifted for shorter time spans (0, 4, 8, 12, 24, and 72 h) from the permissive to the nonpermissive temperature at stage 3, and petals were processed for semiquantitative RT-PCR experiments (Figure 6A). Normalization was performed using Ran3 (EST 018_6_06_k11), a small GTPase identified in macroarray experiments as an invariantly expressed gene. Furthermore, RT-PCR experiments with floral and vegetative wild-type organs were conducted to investigate the overall expression pattern of the selected genes (Figure 6B). DEF expression drops already after 4 h of heat treatment and decreases thereafter continuously (Figure 6A). This confirms rapid degradation of DEF proteins upon a short temperature increase and reflects a disturbed autoregulatory expression control (Zachgo et al., 1995). However, some residual expression still remains and appears to be independent of the autoregulatory control. GDSL-lipase expression is no longer detectable after 72 h, reaching the lowest RT-PCR expression value of all investigated genes. LTP expression decreases after 12 h of DEF reduction. LTP and GDSL-lipase are both strongly expressed in petals and slightly weaker in stamens, and the latter one reveals also a weak carpel expression. Their expression patterns overlap with the DEF expression, which is confined to petals and stamens and is also weakly detectable in carpels (Figure 6B), supporting the regulatory function of DEF. Interestingly, expression dynamics of - and ß-tubulins are similar, suggesting coregulation to allow heterodimer formation and thereby microtubule assembly. Expression of a calmodulin-like gene, a putative calcium sensor, and an extensin-like protein, representing cell wall related processes, decreases gradually upon abolishing the DEF function for extended time spans (Figure 6A). Slight fluctuations in the 4, 8, or 12 h expression levels might be caused by expression differences as a result of a circadian control because harvesting at identical time points was not feasible for these short time spans. Expression of these genes is strongest in petals and stamens but also detectable at reduced levels in sepals and carpels, as well as in vegetative organs (Figure 6B). DEF might thus control expression of these genes, but their broad expression range indicates that other factors regulate their expression outside of the petal/stamen context.
The repressed target genes ß-glucosidase and Rubisco activase show enhanced expression levels upon DEF reduction, the latter one revealing a delayed response (Figure 6A). Absence of expression in petals and stamens is supportive for a direct negative regulatory function of DEF. Both genes are weakly expressed in def-101 petals at stage 3 at the permissive temperature (Figure 6A), likely because of the slightly reduced DEF function in the def-101 mutant compared with wild-type flowers (Zachgo et al., 1995). Thus, at 15°C, the DEF activity level seems to be already below the threshold required for their full repression in petals of the def-101 mutant. In wild-type carpels, quantitative requirements for the DEF function to suppress Rubisco activase might be low, and weak DEF expression suffices to repress its expression in the fourth whorl (Figure 6B).
In summary, RT-PCR analyses demonstrate that expression of the selected genes was sensitive to shortened time spans of an altered DEF function. Therefore, these genes represent target genes whose transcription might be directly controlled by DEF.
Analysis of Tissue-Specific Expression of Target Genes
Tissue-specific expression of the selected early target genes was determined by in situ hybridizations on serial cross sections through the bottom and upper part of stage 3 wild-type flowers (Figure 7A). In the basal flower part, DEF is expressed almost uniformly throughout the lower tube formed by five fused petals and in the four stamen filaments (Figure 7B). Expression is stronger in epidermal cell layers than in internal tissues. In agreement with RT-PCR analyses of expression in individual organs, DEF is also expressed in carpels, mainly in valves and ovules. A similar pattern was observed in the upper part of the flower, formed by the upper and lower lobes. However, DEF expression is enhanced in the extensively folded edges of the dorsal petals (Figure 7C).
|
- and ß-tubulin and calmodulin-like genes are more uniformly expressed throughout these organs (Figures 7J to 7O). LTP is strongly expressed in the epidermal cell layer of the dorsal and lateral petal edges that, upon maturation, will be unfolded and exposed to pollinators (Figure 7E). GSDL-lipase expression was exclusively detected in the inner and outer epidermis of petals and filaments (Figures 7F and 7G). A similar expression pattern was detected for the extensin-like gene, however, with an additional weak staining throughout the two whorls. In addition, weak expression in ovules was observed, confirming the RT-PCR analysis (Figures 7H and 7I). These observations emphasize the importance of the epidermal petal cell layers that function not only in scent and sheen production but also make a major contribution to the overall petal shape (Noda et al., 1994; Perbal et al., 1996; Dudareva et al., 2000; Efremova et al., 2001). For the - and ß-tubulins, overlapping expression was detected in petals and stamens (Figures 7J to 7M). Stronger labeling of the dorsal petal edges might indicate an increased requirement of microtubules during the rapid expansion phase of these structures. The calmodulin-like gene, likely involved in cell signaling, is transcribed in all whorls, with a weakly enhanced expression in the folded edges of the petals (Figures 7N and 7O). Supporting the organ-specific RT-PCR experiments, the repressed target genes ß-glucosidase and Rubisco are expressed in sepals, and ß-glucosidase expression was also detected in carpels (Figures 7P and 7Q). Furthermore, in situ studies were conducted on sections from def-101 flowers at stage 3, shifted for 0, 24, and 72 h from the permissive to the nonpermissive temperature (Figure 8). Thereby, tissue-specific expression pattern changes were monitored during the time course of reducing the DEF function. Representative data are shown for one activated and one repressed target gene. Expression of the activated target gene GSDL-lipase gradually decreases in petal and filament epidermal cell layers until no signal is any longer detectable after abolishing DEF activity for 72 h (Figures 8A to 8C). Expression in carpels is not affected. Conversely, expression of the repressed target gene Rubisco activase is enhanced in second whorl organs during prolonged shift periods of reduced DEF activity (Figures 8D to 8F). Increased expression is more pronounced in ventral petals (Figure 8F), which is in accordance with our morphological data showing more severe effects on the ventral part of the flower (Figure 1G).
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTSDISCUSSIONMETHODSREFERENCES |
|---|
Late Stages of Petal Development Reveal Highly Dynamic Transcriptome Changes
High-throughput analysis using macroarrays comprising >11,600 Antirrhinum unigenes was conducted to compare petal and sepal expression profiles. Regulation of functional categories was investigated throughout four petal developmental stages starting at a floral bud size of 0.5 cm until a 4.0-cm-long mature flower was formed 11 d later. As a reference, a mixture of sepals was used. The M-CHiPS method applied for array data analysis has been already shown for other data to be suitable for studying subtle transcriptional changes (Lagorce et al., 2003). More than 90% (19/21) of genes identified by expression profiling as being differentially expressed with changes down to 1.2-fold (applying intensity thresholds excluding 80% of the spotted genes as not expressed) were verified by independent expression studies.
In total, 537 ESTs were found to be differentially expressed in at least one investigated stage, dividing into 322 activated and 226 repressed ESTs, and comprising 11 ESTs assigning into both groups. Considering that these data include approximately one-fifth redundant ESTs as 22% of the unigenes were spotted twice on the arrays and that the ESTs represent only 40% of the total cDNAs, the total number of upregulated genes during late petal stages might be in the order of 600 or more. Using whole flowers from different stages of Arabidopsis mutants and also transgenic inflorescences, 200 (Zik and Irish, 2003) and >1100 genes, including 600 pollen-specific ones (Wellmer et al., 2004), were predicted as being regulated in petals and/or stamens by the Arabidopsis class B genes AP3 and PI. Identification of the significantly larger number of differentially expressed Antirrhinum genes most likely results from using mRNA from distinct organs and stages, which minimizes blurring of transcript differences and thus increases the sensitivity for detection of expression differences. The estimated number of 600 genes represents genes whose expression is differentially regulated between sepals and petals, but they also include genes that are expressed in stamens and carpels, as it was shown for some of the selected DEF target genes (Figure 6B). It will be intriguing to determine by floral organ comparisons a group of exclusively petal-specific genes with this system in the future.
The number of genes that are upregulated during petal development increases steadily from stage 2 on. As organ harvesting for the sepal/petal comparison started in stage 1, the identified 104 upregulated ESTs likely record an accumulating effect of expression differences from earlier petal stages. The functional category cell cycle and DNA processing is highly overrepresented during stage 1, as reflected by increased expression of DNA replication markers, as for instance, histone H4. The level of these transcripts drops to an average level at stage 2; thus, Antirrhinum petal growth is realized mainly by cell division processes until flowers reach a size of 0.8 cm. After stage 2, the number of activated genes continuously increases mainly by strong upregulation of stage-specifically expressed ESTs. In total, 2851 genes showed an expression level high enough in at least one stage to be considered for petal profiling analysis. Thirty-three and 114 of the analyzed ESTs were specifically upregulated in stages 3 and 4, respectively. The highly dynamic transcriptome changes reflect the need for specialized proteins to participate in the final maturation phase of petal development. Analysis of 20 selected functional categories showed that especially genes with a metabolism-related function are strongly upregulated. Similar results were obtained by functional classification of 2100 rose petal genes (Guterman et al., 2002). Closer inspection of functions by individual BLAST analyses showed that during stages 3 and 4 genes involved in processes associated with cell growth are upregulated. Enzymes, such as pectinacetylesterases, modifying cell wall components, or structural cell wall proteins, like extensins, were identified. The observed upregulation of aquaporins, water channel proteins, and tonoplast proteins reflects an alteration of turgor pressure driving cell expansion during the final petal growth phase (Schäffner, 1998). Clearly, there is a requirement for primary metabolites as well as for secondary metabolites to allow final petal growth. Genes involved in secondary metabolite production with specialized function in the epidermal cell layer were identified as being upregulated. For instance, S-adenosyl-L-methionine:benzoic acid carboxyl methyltransferase is the final enzyme in the biosynthesis of the scent compound methyl benzoate (Dudareva et al., 2000), anthocyanidin synthase catalyzes the penultimate step in the biosynthesis of the anthocyanin class of flavonoids (Wilmouth et al., 2002), and CER1 is involved in wax biosynthesis (Aarts et al., 1995).
Surprisingly, the proportion of unclassified proteins was higher in the petal expressed transcripts compared with their general representation in the Antirrhinum EST collection. This group might contain genes with specialized functions in secondary metabolite production because they are highly divergent among plant genomes and therefore might escape from identification by sequence comparison (Pichersky and Gang, 2000).
Altogether, 156 to 180 ESTs were repressed during the four petal stages compared with sepals. In contrast with the upregulated ESTs, only a low number, 34 ESTs, were stage-specifically repressed. The majority of the repressed ESTs exerts a function associated with photosynthesis, in accordance with plastid ontogeny in Arabidopsis, where green chloroplasts are initially formed in all young floral organs. During further petal differentiation, chlorophyll content decreases and redifferentiation to leucoplasts occurs while chloroplasts remain only at the base of petals (Pyke and Page, 1998).
Expression of More Than 100 Genes Is Affected by a Reduced DEF Function at Petal Stage 3
Petal stage 3 was chosen to identify target genes, likely to be under direct control of DEF. This stage is characterized by the onset of rapid petal growth realized mainly by cell elongation processes, and we showed that novel, stage-specific transcripts start to accumulate. Expression changes were investigated with the def-101 mutant after reducing the DEF function for 24 and 72 h by increasing growth temperature. After subtracting generally temperature-sensitive genes detected in control experiments, 101 unigenes were identified that show a significantly altered expression level in at least one condition (see supplemental data online). Sixty-seven of these target genes were also detected in the sepal/petal comparison, indicating a high overlap of the identified genes in the two different experiments but reflecting also differences in culturing conditions and different DEF expression levels between def-101 and wild-type flowers. Expression of 46 genes was affected by reduction of the DEF function for 24 h. The expression of these 27 repressed and 19 activated target genes could be directly controlled by DEF. Prolongation of the reduction period for up to 72 h increased the number of affected genes to 93, with almost equal numbers of activated and repressed genes. Thus, for normal petal development to proceed, the expression of similarly sized groups of genes has to be activated and repressed by DEF. Furthermore, the importance of DEF to act constantly as a repressor of sepal-specific genes during late petal development corroborates our morphological data.
Assuming that the EST collection represents 40% of the total Antirrhinum unigene set, we estimate that >200 target genes might be regulated by DEF at stage 3. As we have shown, highly dynamic transcriptome changes occur during the four late petal developmental stages. Therefore, this number is representing only a subset of the total DEF petal target genes.
Putative Direct DEF Target Genes Are Involved in Different Processes Associated with Cell Growth during Late Petal Development
Functional categorization of DEF target genes showed that mainly genes belonging to the groups metabolism, transport facilitation, energy, and control of cellular organization were strongly overrepresented. For detailed analysis, target genes were selected that are representative for different processes cooperating during petal development. Eight genes that showed a strong expression level change after the DEF function was abolished for 24 h were further analyzed by RT-PCR studies where DEF activity was reduced for shorter time spans (4, 8, and 12 h). For all selected genes, changes in expression levels were detected during the shortened time spans. Four of them responded already after 4 h and two after 8 and 12 h of DEF reduction, respectively. Toward the goal to investigate if quickly responding genes represent direct DEF target genes, we have started to analyze their promoters. Preliminary data show that they contain CArG-box motifs (data not shown). Interestingly, the extensin-like promoter contains two nearby CArG-boxes that could be required, similarly to the two CArG-box motifs in the GLO promoter, for binding a DEF ternary protein complex (Egea-Cortines et al., 1999).
Expression analysis data, together with putative gene functions, are discussed separately for the different target genes.
DEF Regulates a Group of LTPs
Four LTPs were isolated as being regulated by the DEF activity, three as activated, and one as a repressed target gene. We investigated the expression of the activated LTP 018_4_03_c07 that was also found to be strongly upregulated (>12-fold) in the sepal/petal comparison in more detail. Its expression level decreases after reducing the DEF function for 12 h. At stage 3, it is exclusively expressed in petals and stamens, thus representing an interesting candidate for a direct target gene of DEF. Cell-type specific investigation by in situ analysis revealed that this LTP is expressed in restricted areas in petals. A signal was localized in the upper dorsal petals, predominately in parts that, upon unfolding, will be exposed to pollinators. LTPs can mediate the transfer of phospholipids between membranes in vitro (Zachowski et al., 1998). However, an Arabidopsis LTP was localized in the cell wall, questioning their function in catalyzing intracellular lipid transfer between membranes (Thoma et al., 1993). Other plant LTPs have been shown to be expressed in a large variety of different floral tissues, indicating that they are involved in various processes (Kotilainen et al., 1994; Rubinelli et al., 1998). The identification of Antirrhinum LTPs as DEF target genes suggests a function during late petal and stamen development.
GDSL-Lipases
A GDSL-lipase revealed the strongest expression response of all investigated target genes and was no longer detectable by in situ and RT-PCR experiments after 72 h of DEF reduction. Its expression is strongest in petals and slightly weaker in stamens. Similar to DEF, very weak expression was also detected in carpels. In situ analysis shows that the expression of this Antirrhinum GDSL-lipase is restricted to the epidermal cell layers of petals and stamens. The GDSL-lipase might thus function during differentiation of the highly specialized epidermal cells. GDSL-lipases are lipolytic enzymes with an active site, the name-giving conserved GDSL amino acid sequence (Upton and Buckley, 1995). Regulation of GDSL-lipases genes might be conserved between Scrophulariaceae and Brassicaceae class B genes, as two independent studies identified a GDSL-lipase from Arabidopsis as a gene that is not expressed in mutants lacking petals (Zik and Irish, 2003; Wellmer et al., 2004).
Structural Proteins Involved in Cell Morphogenesis
Expression of - and ß-tubulins responded similarly to an altered DEF function. Both genes are expressed ubiquitously throughout the plant, with the highest expression level in petals and stamens. The final cell shape is determined by extent and direction of cell expansion, requiring regulated deposition of cellulose microfibrils in the innermost cell wall. It has been suggested that the orientation of cortical microtubules, mainly composed of - and ß-tubulins forming heterodimers, determines cell wall deposition in elongating cells (Mayer and Jürgens, 2002). The Antirrhinum -tubulin reveals >95% amino acid sequence identity to the Arabidopsis -tubulin 6, 4, and 2 proteins (Kopczak et al., 1992). The double mutant lefty1 lefty2 of the -tubulin 6 and -tubulin 4 genes produces defective microtubules, causing severe abnormalities during hypocotyl, root, and flower morphogenesis (Abe et al., 2004). Mutant flowers produce shorter anther filaments as directional cell growth is impaired and form twisted petals. Our data indicate that DEF coordinates upregulation of - and ß-tubulin expression during the rapid cell elongation phase in Antirrhinum petal and stamen development. In situ analysis showed overlapping expression patterns in stamens and petals with a higher expression level within the dorsal petal edges. Coregulation of - and ß-tubulin expression provides a means to form microtubular structures required for the final growth phase of the petal. Interestingly, the serum response factor, an animal MADS box transcription factor essential for mesoderm formation in mouse embryos, also regulates cytoskeletal proteins, particularly the expression of different actins (Schratt et al., 2002).
Various Genes Participate in Cell Wall Formation
As mentioned above, the cell wall affects cell shape and function. We identified several structural proteins and enzymes participating in cell wall formation as DEF target genes, including pectin modifying enzymes, cell wall proteins, and extensin-like genes that were strongly upregulated during later petal development stages. An extensin-like gene was investigated in more detail, revealing a rapidly decreased expression upon reducing the DEF function. Extensins belong to the HRGP superfamiliy, a major class of structural proteins in the primary cell wall of higher plants (Kieliszewski and Lamport, 1994; Cassab, 1998). Extensins share the massive presence of Pro residues occurring in repetitions of at least two consecutive Pro. These abundant proteins represent the major cell wall scaffolding components together with other HRGP subgroups, such as repetitive Pro-rich proteins, arabinogalactan-proteins, and lectins. An extensin-like gene from tomato (Lycopersicon esculentum) is expressed in the root hair differentiation zone and is involved in oriented cell elongation leading to cellular tip growth of tomato root hair (Bucher et al., 2002). Two LEUCINE-RICH REPEAT EXTENSIN genes (LRX1 and LRX2) were recently cloned from Arabidopsis and shown to be required for cell morphogenesis of root hairs (Baumberger et al., 2001, 2003). The intracellular cytoskeleton provides the means by which vesicles are translocated to the cell wall to deliver components required for cell elongation. Coregulation of these processes by one key transcription factor, such as DEF, provides an effective mechanism to control this interdependency.
Cell Signaling
Calcium signaling orchestrates responses to cellular stimuli. Intriguingly, a calmodulin-like gene was identified as a target gene that is ubiquitously expressed with higher expression levels in petals, stamens, carpels, stems, and roots. Calmodulin serves as a universal calcium sensor in all eukaryotes, mediating calcium action by regulating the function of many targets in diverse cellular pathways (Zielinski, 1998). The requirement of cell-signaling processes for petal organogenesis has been demonstrated by the observation that DEF acts non-cell-autonomously and by quantitative growth modeling, predicting a long-range signal to control petal growth direction (Perbal et al., 1996; Rolland-Lagan et al., 2003).
Repressed DEF Target Genes Comprise a Large Group of Genes Associated with Photosynthesis
Homology comparison showed that a large group of genes whose expression is repressed by DEF exerts functions related to photosynthesis. Genes like Rubisco small chain, Rubisco activase, and photosystem II protein precursor were found to be highly expressed in sepals and upregulated upon a reduction of the DEF function for 24 h. As a representative of this group, we analyzed Rubisco activase in more detail. The enzyme is expressed in sepals, bracts, leafs, and stems. It activates Rubisco, the key enzyme of photosynthesis, by a carbamylation reaction (Werneke et al., 1989). As indicated by comparative petal expression profiling, Rubisco activase is repressed continuously throughout petal development, and we showed that its expression level is increased after 12 h of DEF reduction. Together with the morphologically visible petal to sepal transformations, this demonstrates the rapid and long maintained potential of the second whorl organs to adopt a sepaloid structure. Although chloroplasts redifferentiate during petal development into leucoplasts (Pyke and Page, 1998), they seem to keep the plasticity that allows them to quickly convert into photosynthetically active plastids even at late stages of petal development.
Various other enzymes, involved in different metabolic processes, were identified as repressed genes. Detailed expression studies were conducted with a ß-glucosidase, an enzyme hydrolyzing conjugated ß-glucoside compounds present in plant secondary metabolism. Physiological functions of ß-glucosidases are broad as they depend on the function of the molecule released after hydrolysis. Plant development and growth can be affected by regulating cell division via release of active phytohormones from their respective inactive glycoconjugated forms (Brzobohaty et al., 1993). The enzymes participate in the modification of oligosaccharides in cell walls (Akiyama et al., 1998) and play a role in defense metabolism against pathogens (Sue et al., 2000) and in the production of secondary metabolites such as flavonoids and lignins (Leinhos et al., 1994). Interestingly, ß-glucosidase expression is confined to sepals and carpels only, as expected from a target gene negatively regulated by DEF.
Tissue-Specific Target Gene Expression Indicates Suborgan Specific Regulation of Gene Expression
In situ studies showed that expression of the two putative direct target genes, LTP and GDSL-lipase, is restricted to distinct tissues within petals, in particular to epidermal cell layers. These genes represent good candidates for being directly and exclusively controlled by DEF because their expression patterns overlap with DEF expression domains. A wild-type petal comprises about six cell layers. However, only the epidermal cell layer has been reported to exert specialized functions, such as scent compound production and formation of characteristic conical cells contributing to the velvet sheen of Antirrhinum petals. The LTP and GDSL-lipase target genes might thus participate in specialized functions of epidermal cells. Other investigated target genes show a broader expression throughout petals and stamens with a tendency for higher expression at the edges of dorsal petals where rapid expansion and unfolding during the final maturation phase occurs. Because they were also expressed at lower levels in vegetative tissues, their expression seems to be not exclusively controlled by DEF.
We detected only one other transcription factor, the MADS box gene DEFH84, whose expression is upregulated by DEF. From this observation, we conclude that DEF seems to exert its regulatory control until late stages rather directly, without delegating transcriptional control functions to a large number of other intermediate transcription factors. Showing that expression of some target genes is confined to various subdomains of DEF expression raises the question of how this spatial restricted target gene regulation is achieved. Formation of multimeric complexes has been demonstrated for MADS box proteins and thus provides a mechanism by which different binding specificities for various target genes can be generated (Egea-Cortines et al., 1999; Honma and Goto, 2001; Ferrario et al., 2003). In the future, advancement of proteomics technologies will allow in vivo analysis of MADS box protein complex composition and help to unravel different interactions in the control of distinct target genes.
Former studies have shown a high degree of conservation in target gene regulation throughout the plant kingdom. The Antirrhinum class B gene DEF can functionally substitute the orthologous Arabidopsis class B gene AP3 by complementing the ap3 mutant phenotype (Samach et al., 1997). Developmental profiling and target gene studies of floral organs from model species besides Arabidopsis allow conducting comparative studies. Thereby, the question can be addressed to which extent spatial and temporal expression alteration and/or recruitment of novel target gene functions contributed downstream of the key regulatory genes to generate the floral organ diversity observed today.
|
METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTS |
|---|
