Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

Gene expression patterns to define stages of post-harvest senescence in Alstroemeria petals

Gene expression patterns to define stages of post-harvest senescence in Alstroemeria petals <h1>Introduction</h1> The primary function of large and colourful floral structures (the corolla) on plants is to attract insects or other pollinators. Once pollination has occurred, the role of the flower is over and the corolla is rapidly lost from the plant. The final stages of floral development can take a number of different forms ( van Doorn, 2001 ). The petals of some flowers abscise from the plant with no obvious signs of deterioration, indicating that little remobilization has occurred. The petals of other flowers show extensive wilting (turgor loss, increased translucency and slow desiccation) before they are abscised from the plant. This could indicate that degradation and remobilization of its cellular constituents has taken place. The majority of large monocotyledonous flowers, such as those of the Liliales order including those of Alstroemeria , show this type of senescence. Alstroemeria is a member of the Alstroemeriaceae family and is an important cut flower in Northern Europe. It has a relatively long vase life, taking around 10 days under optimal conditions after harvest to reach full senescence. A time frame for floral senescence including the measurement of biochemical and physiological parameters such as protein degradation has been developed and described previously ( Leverentz et al ., 2002 ; Wagstaff et al ., 2001, 2003 ), and gene expression patterns for a small number of genes during senescence have been analysed ( Wagstaff et al ., 2002 ). Molecular studies of leaf senescence have shown that gene expression patterns change dramatically as senescence progresses, with a large number of genes either showing reduced or increased expression ( Buchanan-Wollaston et al ., 2003 for review). Novel gene expression is essential for senescence to occur. Many different biochemical events occur during leaf senescence, macromolecules are degraded and soluble nutrients are remobilized. Thus, the purpose of leaf senescence is one of redistribution of resources. The study of gene expression changes during flower senescence has not been extensive, although a number of senescence-enhanced genes have been isolated from a variety of different flowers ( Rubinstein, 2000 ). Many of these encode potential catabolic enzymes such as proteases and nucleases, and genes encoding enzymes related to ethylene biosynthesis have also been identified. A more global overview of genes expressed in petals was obtained by DNA sequence analysis of nearly 2000 individual cDNA clones from RNA isolated from rose petals at four different stages of development including senescent petals ( Channeliere et al ., 2002 ). A few of these genes were shown to exhibit enhanced expression in senescing petals. In many species, the time to petal senescence is regulated by ethylene, and a manipulation of the levels of ethylene biosynthesis has been effective in delaying the rate of senescence in several flower types such as the carnation ( Michael et al ., 1993 ) and Torenia ( Aida et al ., 1998 ). However, flowers of some species, such as many of the monocotylenonous plants including Alstroemeria , are not dependent on ethylene for senescence, even though the final abscission of the wilted petal may be ethylene-dependent. Genes that show enhanced expression in senescing monocotyledon flowers have been identified in a number of different ways including differential display (in daylily, Panavas et al ., 1999 ) and subtractive hybridization (in daffodil, Hunter et al ., 2002 ). In each case a small number of genes was identified. We have used a suppressive subtractive hybridization technique to produce several cDNA libraries enriched for genes expressed at particular stages of Alstroemeria petal senescence. Genes showing both up and down-regulation during senescence have been identified, and microarrays were used to determine the overall changes in gene expression patterns that occur during post-harvest senescence. This is a novel approach to the analysis of flower senescence; it has identified extensive collections of genes that reflect the complex processes occurring during petal senescence. A selection of these genes may represent useful markers to determine the stage of flower senescence in this and other floral crops. <h1>Results</h1> <h2>Construction of subtracted libraries and gene identification</h2> Eight different stages of Alstroemeria flower development and senescence have been identified by clearly defined visible changes ( Figure 1 ). Subtracted libraries were constructed to clone and identify genes showing differential expression between these stages. Initially, four libraries were made and these were enriched for: (1) genes expressed in Stage 0 and not in Stage 2, (2) genes expressed at Stage 2 and not at Stage 0, (3) genes expressed in Stage 2 and not in Stages 4 + 5 combined, and (4) genes expressed in Stages 4 + 5 and not in Stage 2. Over 80% of the clones in library (4) represented the same gene family (see below) and so a fifth library was constructed that was enriched for genes expressed in Stage 3 and not in Stage 2. DNA sequence analysis followed by database searches for each subtracted clone identified genes with a wide range of different potential functions. For each library, between 200 and 400 clones were sequenced and, once poor and short sequences and sequences of structural RNA had been removed, a total of 991 EST sequences were generated. In order to identify ESTs that represented the same gene, all the sequences obtained were entered into the alignment program S eqman (DNAS tar ) and common sequences were grouped together as contigs. This resulted in 500 different groups, each representing a separate Alstroemeria gene (Supplementary Table S1A). Of the 991 EST clones characterized in this way, 297 were singletons, i.e. a single representative of a particular gene. All the other sequences occurred at least twice, with the most common gene occurring 21 times. The redundancy of the EST collection was calculated to be 70% (number of ESTs in clusters/total number of ESTs, Sterky et al ., 1998 ). Therefore any new sequence has a 70% chance of already being represented in the collection. The contigs represented by two or more sequences were given an ALF TC number ( Alstroemeria Flower Tentative Contigs). Singletons were given an ALF number. There was likely to be some redundancy in the TC and EST sets, because sequences must have a matching overlap at a certain percentage identity (95%) to be combined, and different parts of the same gene may be represented separately. The potential function of each gene was identified by database searching with the TC or single EST sequence, using B last N and TB last X databases. For each cloned Alstroemeria gene, its likely function and the most similar rice gene or Arabidopsis gene are shown in Supplementary Table S1A. Rice and Arabidopsis were used to identify the closest matched gene, since the genomes of these two plants are almost completely sequenced and annotated. Each of the sequences has been submitted as an EST sequence to G en B ank and the G en B ank codes of the ESTs found in each contig are shown in Supplementary Table S1B. <h2>Differential gene expression determined by subtractive hybridization</h2> Five different subtracted libraries were generated as described above. Up to 400 clones were sequenced from each library and genes encoding a wide range of different functions were obtained (Supplementary Table S1). The genes identified were assigned to potential functional groups according to their sequence similarity and an overall summary of these functions in the EST collection is shown in Figure 2A . A high proportion (29%), of the clones sequenced represented genes with no matches in the database or which showed a similarity to genes of unknown function. Metal binding proteins (mostly metallothionein-like proteins) were the most highly represented class of genes with known potential function (19%). A comparison of the potential functions of the genes identified in each of the subtracted libraries could give an indication of the processes taking place in the petals through post-harvest development. Each subtracted library contained clones representing a different collection of genes, and these genes are listed in Supplementary Table S2. The five different subtracted libraries showed a considerable variation in the range of genes they represented and the potential function of each gene was used to classify them into groups. The relative proportions of different gene functions in each library are illustrated in Figure 2B , and show a considerable variation between the libraries. This graph shows the relative proportions of each functional class of gene in the five different libraries and may be a reflection of the changes that are taking place as the petals are undergoing post-harvest senescence. The libraries are ordered on this graph to show the progression of senescence, the first two (S0–S2 and S2–S4 & 5) should be enriched for genes that are decreasing in expression levels as senescence progresses. The other three libraries (S2–S0, S3–S2 and S4 & 5–S2) should be enriched for genes that are increasing in expression as senescence progresses. From the graph it appears that there are considerable differences in the transcripts that are present at different stages. Cell wall related genes are only present in the first two libraries and also genes involved in metabolism are present at a higher proportion in the earlier stages. As has been described above, genes encoding metal binding proteins (mostly metallothionein-like) were the major component of the subtracted libraries that were constructed to identify senescence enhanced genes, i.e. 4/5–2 and 3–2, and also there were a significant number of metallothionein clones in the library 2–0. However, metallothionein genes were not represented at all in the libraries enriched for genes showing decreasing expression during senescence (i.e. S0–2 and S2–4/5). This indicates that the metallothionein transcripts were subtracted in these libraries and suggests that the subtraction procedure was successful. <h2>Metallothionein genes</h2> Grouping the EST sequences together illustrated the extent of the redundancy in the EST collection. Many genes were represented only once or twice while the largest group consisted of 21 different sequences. The SEQMAN assembly was carried out with a high stringency (95% sequence identity) resulting in separation of different members of the same gene family. For example, the alignment of the many metallothionein sequences showed that there were at least 38 different genes represented, many of which were present multiple times in the EST collection (Supplementary Table S1). The protein sequences for 17 of these different groups of metallothionein genes (those that were each represented four times or more in the EST collection and therefore had a reliable consensus sequence) were predicted from the consensus DNA sequence and compared in an alignment programme using C lustal W, to determine the differences between them ( Figure 3A ). Two main classes of protein were apparent from this alignment ( Figure 3B ). The majority of genes appeared to encode a protein of around 60 amino acids, and this class can be divided into two subclasses, 1A and 1B, as shown in Figure 3A,B . These proteins had between 100 and 90% identity (Class 1A) and 75–67% identity (class 1B) to the ALF 1 protein. The second class of metallothionein (Class 2) was illustrated by two of the TC sequences (ALF TC41 and ALF TC42, each derived from six individual EST clones); these appeared to encode a much smaller protein of around 40 amino acids and showed around 30% identity to the ALF1 protein. Many different metallothionein genes have previously been identified from a variety of plant species. The Alstroemeria genes in class 1A and 1B showed the closest similarity to the metallothionein 1 genes from Fritillaria agrestis , another member of the Liliales order ( Figure 3C ). The most similar match to this protein in Arabidopsis was shown by the Met3 gene (At3g15353), a single copy gene in the Arabidopsis genome ( Figure 3D ). The Alstroemeria Class 2 metallothionein proteins are closely related to the C terminal part of a metallothionein protein (79 amino acids) from Typha latifolia (common cattail) but are most similar in size to the Mt1a protein from Arabidopsis (44 amino acids). The C terminals of these proteins showed similarity, but the N terminal half of the proteins were very different. All the putative metallothionein proteins have characteristic conserved cysteine residues and these are shown in Figure 3A and C . <h2>Gene expression patterns from microarray analysis</h2> All the EST clones generated in the subtractive library screening (1532 clones) were used to generate a microarray. Each slide carried two replicates of an array that contained three copies of each probe. Therefore, each DNA fragment was present six times on a slide and each hybridization was carried out four times with reciprocal labelling. Thus the final data points are the result of hybridization to 24 copies of each gene. Microarrays were hybridized with labelled cDNA made from RNA from four different stages of petal development (Stages 0, 2, 3 and 4/5 mixed together). Some of the genes placed on the microarray were not sequenced until after the hybridization had indicated that they showed potentially interesting expression patterns. For the G ene S pring analysis, data was only included from clones that had been shown by sequence analysis to contain potential gene transcripts. Unsequenced clones, clones containing no inserts and clones representing untranslated RNA were excluded from the analysis. A representative clone for each contig was then selected at random and data for 500 cDNA clones, each representing a different gene, was analysed. From the 500 cDNA clones used for the analysis, 251 showed an expression pattern that was significantly up or down-regulated ( P < 0.05) for at least one point over the time course. Normalized data indicated the relative expression level of each gene in comparison to all the other genes at that stage of development. The t -test P -value showed how significantly this normalized figure differed from the average (i.e. a value of 1). The expression level of each gene at each time point ( y -axis value) was calculated taking account of both the total expression of all genes at that time point to standardize each hybridization, and also the changes in expression level of each gene across the four time points. This allowed the changes in expression levels of each gene over time to be assessed. The changes in expression of this group of genes is shown in Figure 4A (G ene S pring graphs). Only genes that showed at least a 2.5-fold change in normalized expression level between Stages 0 and 4/5 were then selected for further analysis. From this group, two separate clusters were identified, those showing high expression at Stage 0, decreasing at later stages and those showing low expression levels at Stage 0 increasing at Stages 3 or 4/5 ( Figure 4B,C ). To confirm the expression patterns of these selected sets of genes, the experiment was repeated by hybridizing array slides carrying the same cDNA clones, with RNA isolated from a different batch of petals harvested at a different time. At least two slides were hybridized with each RNA sample (i.e. 12 data points, n = 4) and the data analysed. The expression patterns of genes in the two clusters identified as described above were analysed in the new experiment, and in general the patterns of expression were highly comparable. However, a few of the genes did not show a significant change in expression pattern in the repeat experiment and these were removed from the list. Supplementary Table S3 shows the resulting lists of genes that were significantly altered in expression in both experiments and presents the normalized expression level for each gene at the different time points. From this data, the relative expression change between Stages 0 and 4/5 can be calculated. The data in Supplementary Table S3 is summarized in Table 1 , which shows the potential functions of the genes that are present in the up and down-regulated gene lists. There are considerably more genes represented in the up-regulated list than in the list of down-regulated genes. This may be a reflection of the changes in metabolic activity in the petals. In the early stages, 0 and 2, the petals are already fully formed and are not undergoing dramatic metabolic changes, while in later stages a whole new set of genes involved with the degradation and mobilization typical of senescing tissues is expressed ( Table 1 ). <h2>Northern hybridization</h2> Microarray experiments are extremely useful for showing the relative changes in expression levels of a large number of different genes during development. However, the data from microarrays can be subject to considerable experimental variation, and conclusions from these experiments should be tested using techniques such as Northern hybridization to confirm expression patterns for a select number of genes. Northern hybridization analysis was carried out with a selection of genes from Supplementary Table S3. A range of genes was selected from the lists shown in Supplementary Table S3 and these showed the expected expression pattern in Northern hybridization experiments ( Figure 5 ). The MtN19 -like gene, represented by ALF TC139, was identified in the list of genes showing highest expression in Stages 3 and 4/5. The Northern hybridization results support this and show that the expression of this gene continues to increase at later stages of development ( Figure 5B ). Similarly, the Cer1 gene, represented by ALF TC46, appeared in the list showing significant increased expression in Stages 3 and 4/5. The Northern hybridization mirrored this expression pattern exactly ( Figure 5C ). However, unlike ALF TC139, the expression of this gene then decreased at later stages of senescence. The S-adenosyl-L-methionine:salicylic acid carboxyl methyltransferase like gene, represented by the contig ALF TC141, was also present in the list of genes showing significantly higher expression in Stage 4/5. The Northern blotting data also supported this observation and showed that the expression of this gene peaked at Stage 5 and then started to fall in later developmental stages ( Figure 5D ). The 2-oxoglutarate dehydrogenase gene was present in the list of genes showing high expression at Stage 0 with reduced expression later in development. The Northern hybridization results from this gene confirmed this expression ( Figure 5E ). Interestingly, the expression pattern found for the MET1 gene (ALF TC1) was very different to that seen on the microarrays. This gene and most of the other MET contigs were not represented in the senescence enhanced gene list in spite of the fact that the subtraction data suggested that members of this gene family were very highly expressed in the later stages of development. The Northern data confirms the subtraction results – the expression levels detected on the Northern blot were very high in late developmental stages; the autoradiograph shown in Figure 5 was the result of a film exposure of only 3 h. This shows very high levels of transcript at stages 3, 4, 5, 6 and 7 ( Figure 5F ). A longer exposure showed that some transcript was detectable at Stage 2, a very small amount at Stage 1, but none at all was detected at Stage 0 (data not shown). Although the probe was made using the ALF TC 1 cDNA, it is likely that the hybridization seen is also to many of the other MET contigs of Class 1A and 1B, which were very similar in sequence. The DNA sequence of the Class 2 represented by ALF TC 42 was divergent enough to make cross-hybridization unlikely. The expression pattern of this gene was completely different, with detectable transcript at all stages increasing to Stage 4 and then decreasing ( Figure 5G ). The expression level of this gene was also quite high, but weaker than that of ALF TC1 (the autoradiograph shown in Figure 5G was also the result of a 3 h film exposure). <h1>Discussion</h1> <h2>Characterization of subtracted libraries</h2> The suppressive subtractive hybridization technique ( Diatchenko et al ., 1996 ) has been used extensively for the identification of differentially expressed genes in many different organisms. The technique was developed to enrich for differentially expressed genes and at the same time normalize the relative abundance of the different messages in the target population allowing more of the rare transcripts to be cloned. A reported potential drawback of this method is the presence of background clones representing non-differentially expressed genes that can make up a large proportion of the library ( Rebrikov et al ., 2000 ). Five subtracted libraries were constructed using RNA isolated from different stages of petal senescence. From these libraries good sequences were obtained for between 100 and 320 different clones (Supplementary Table S2). Analysis of the DNA sequences of the collections of genes in each library revealed a wide range of different potential gene functions. The most obvious difference between the libraries was the proportion of clones representing a metallothionein-like protein which was often identified in the libraries that were enriched for senescence enhanced genes (Stages 2–0, 4/5–2 and 3–2) but very rarely in the other two libraries that were potentially enriched for genes decreasing in expression during senescence (Stages 0–2 and 2–4/5). This comparison indicated that the subtraction procedure had at least been successful for this gene. However, the very high proportion of metallothionein-like genes (83% and 55%, respectively) in the senescence-enhanced libraries Stages 4/5–2 and 3–2 indicated that the suppression of abundant transcripts for this class of gene was not successful. In fact, this proportion of metallothionein clones exceeds that seen in unsubtracted (and unsuppressed) cDNA libraries made from RNA from these developmental stages where metallothioneins make up about 30% of the identified genes (C. Wagstaff, unpublished results). It therefore appears that the very high level of metallothionein transcripts present in the older tissues have been preferentially amplified and cloned in this SSH procedure. This has presumably prevented the detection of the many other transcripts that are present in this tissue in these libraries, as shown by the microarray and Northern hybridization analyses. The cDNA cloning method has, however, provided an extensive collection of genes that are transcribed during post-harvest petal senescence and the analysis of the potential functions of these can give an indication of the processes that are occurring. The content of the first three libraries, that do not contain many metallothionein transcripts, provides a snapshot of genes that are transcribed in the petals at Stages 0 and 2 ( Figure 2B ). Overall, the genes identified in the different libraries reflect similar functional classifications to those found in other developmental sequencing projects in plants such as Lotus japonicus ( Asamizu et al ., 2000 ) and Citrus sinensis ( Bausher et al ., 2003 ). For example, genes relating to cell wall synthesis, protein synthesis, metabolism and signalling were most abundant in the younger developmental stages. Channeliere et al . (2002 ) carried out a similar analysis on rose petals; they cloned and sequenced 1794 ESTs and showed a range of potential functions comparable to those found in the Alstroemeria petals. <h2>Analysis of the microarray data</h2> More information about the genes expressed at the different developmental stages can be obtained from the microarray data, which provides an expression pattern for each gene during development. Many of the genes on the array did not show significant changes in expression across the four time points, indicating that they were expressed at a relatively constant level in all petal stages. Two clusters were identified containing genes either strongly down-regulated during senescence or showing significantly increased expression between Stage 0 and Stages 3 and 4/5. ( Figure 4 , Tables S3 and 1). Analysis of the functions of these genes can give an idea of the processes that are occurring at the different stages. The small number of genes that were only expressed during the early stages, and were down-regulated as senescence commenced, mainly encoded genes involved in lipid and amino acid biosynthesis as well as genes involved in the TCA cycle and in photosynthesis. The presence of these genes indicated that biosynthetic processes are occurring in the young petal and energy is being produced via the TCA cycle and photosynthesis. Petals in Stage 0 are quite green and this is lost in the later stages ( Figure 1 ). The rapid reduction in expression levels of 2-oxoglutarate dehydrogenase, a TCA cycle enzyme, implied that the production of energy via this pathway only occurred in young petals. Many more genes showed increased expression during the later stages of senescence ( Table 1 ). This indicated that novel pathways are induced during senescence, and the enhanced expression of a wide range of transcription factors and other signalling factors supports the evidence from many plant systems showing that senescence is an active process requiring new transcription and translational events in order to recycle components to other parts of the plant ( Thomas et al ., 2003 ). Many of the genes identified as being senescence-enhanced in Alstroemeria petals have previously been implicated in senescence in other plants and tissues. Protein degradation is an important process that occurs during senescence, and the enhanced expression of genes such as the ubiquitin conjugating enzyme, the aspartic protease and the Clp proteinase show that this process is occurring in the petal tissues. Similar genes have all been shown to have senescence-enhanced expression in other plant systems ( Buchanan-Wollaston et al ., 2003 ) and an aspartic protease was identified in senescing daylily petals ( Panavas et al ., 1999 ). Cytosolic glutamine synthetase has a role in the mobilization of nitrogen from senescing leaves ( Kamachi et al ., 1992 ) and the presence of this gene in Stage 4/5 Alstroemeria petals indicates that nitrogen mobilization is occurring from the petals, presumably to the developing gynoecium ( Nichols and Ho, 1975 ; Nichols, 1976 ). Genes involved in carbohydrate metabolism such as endoxyloglucan transferase and sucrose synthase have also been identified as showing enhanced expression in senescing Arabidopsis leaves ( Park et al ., 1998 ; V.B.-W., unpublished data). These genes may have a role in mobilizing C from cell wall components and converting it to sucrose for transport. The senescence-enhanced expression of a sugar transporter has also been reported previously ( Quirino et al ., 2001 ). Stress related genes such as chitinase and glutathione peroxidase are senescence-enhanced in other systems ( Hanfrey et al ., 1996 ; Page et al ., 2001 ). The gene encoding S-adenosyl-L-methionine:salicylic acid carboxyl methyltransferase, which catalyses the methyl esterification of salicylic acid, producing methylsalicylate, has been implicated in defence responses in Brassica ( Zheng et al ., 2001 ). Moreover, methylsalicylate is a component of the scent of flowers such as Clarkia breweri and the activity of the enzyme encoded by this gene is enhanced in mature flowers of this plant, declining several days after anthesis ( Dudareva et al ., 1998 ). The gene encoding terpene synthase may also have a role in scent production ( Dudareva et al ., 2003 ). The role of these enzymes in the unscented Alstroemeria flowers is not clear, but the coordinated expression of both genes during flower development may indicate the presence of a scent production pathway in Alstroemeria . Although commercial hybrid Alstroemerias such as var. Rebecca used in the work are not scented, native Brazilian cultivars are; it is therefore likely that at least some of the scent producing metabolic pathway is present. In fact, the failure of Alstroemeria flowers to be scented could be related to the mutation of a single gene. Thus the potential for generating scented Alstroemeria flowers, either by conventional breeding or through genetic manipulation, would appear possible. The Northern hybridization results generally confirmed the expression patterns seen in the microarrays. A notable exception to this was the result for the metallothionein genes which showed little change in expression on the microarrays, but were strongly up-regulated according to the Northern hybridization. The reasons for this discrepancy are not obvious. There may be some difference in the stringency of hybridization in the two different systems resulting in more background hybridization on the microarray, or it is possible that the large numbers of different metallothioneins in the subtracted libraries, and consequently the many copies on the microarrays, reduced the hybridization intensity to each individual spot. However, the similarity in sequences between the genes would result in hybridization of the labelled probe to all the similar transcripts in the Northern hybridization, indicating a high expression level. <h2>Potential role of metallothioneins</h2> By far the largest group of genes (19%) found in the libraries encoded a Type 3 metallothionein-like protein defined by Cobbett and Goldsbrough (2002 ). The Northern hybridization experiment showed that an extremely high level of this transcript was present in the older petal stages. The large number (> 20) of different members of this gene family in Alstroemeria may be a reflection of the duplication in the genome in this species. Genes encoding metallothionein-like proteins have also been found in large numbers in the EST collection from rose petals (10.7%) although these were mainly of Type 2 ( Channeliere et al ., 2002 ). Of 401 cDNA sequenced as showing differential expression during strawberry fruit development, only one clone was identified as a Type 3 metallothionein and was represented 27 times ( Aharoni et al ., 2000 ). A Type 3 metallothionein was also the most abundant transcript found in a recent SAGE analysis of rice leaves (2.83%) and accounted for 3.17% of ESTs in a TIGR rice EST collection ( Gibbings et al ., 2003 ). Thus the high proportion of these genes represented in our libraries is not unusual, although the levels are higher than in other systems. Metallothioneins generally contain two cysteine-rich domains, which are able to bind to a variety of metals through mercaptide bonds, and plant metallothioneins can be divided into four types based on amino acid sequence and distinct arrangements of the cysteine residues ( Cobbett and Goldsbrough, 2002 ). Both Class 1A and Class 1B Alstroemeria metallothioneins are members of the Type 3 family of metallothioneins. The Alstroemeria Class 2 metallothionein-like genes cannot be classified, since they do not possess the N-terminal cysteine rich region found in all other plant metallothioneins. Thus, they either represent a novel class of metallothionein or may represent a related gene. The function of metallothioneins in plants is not clear-cut. The isolation of metallothionein proteins from plants has proven difficult due to their instability in the presence of oxygen, and thus their function has largely been inferred from gene expression. Type 3 metallothionein transcripts are highly expressed in ripening fleshy fruits, e.g. in banana ( Clendennen and May, 1997 ), apple ( Reid and Ross, 1997 ) and kiwi ( Ledger and Gardner, 1994 ), and in plants producing non-fleshy fruits, like Arabidopsis , they are also expressed at high levels in senescing leaves ( Cobbett and Goldsbrough, 2002 ), however, no direct link to metal binding has been demonstrated for this type of metallothionein. This is the first report of an abundance of Type 3 metallothioneins in senescent petals, and provides an interesting link between processes occurring in fruit ripening and Alstroemeria petal senescence. Metallothionein genes have been reported in petals from other species, e.g. daffodil ( Hunter et al ., 2002 ) and rose ( Channeliere et al ., 2002 ), however, in both cases the metallothioneins reported were of Type 2. Increased expression of metallothioneins during organ senescence has been reported in Brassica napus leaves ( Buchanan-Wollaston, 1994 ), elder leaves ( Coupe et al., 1995 ) and rice leaves ( Hsieh et al ., 1995 ), however, again these were not Type 3 metallothioneins. Plant senescence is known to generate an increase in free radicals from membrane degradation ( Voisine et al ., 1993 ), and may also generate an increase in free metals such as copper from pigment complexes. Interestingly, metallothioneins are up-regulated in the DAF2 strain of C. elegans that has an increased life-span ( Murphy et al ., 2003 ), where it may be playing a role in preventing or repairing oxidative damage and thereby increasing longevity. In addition, a metallothionein transcript was the most up-regulated transcript in mouse brain tissue that had suffered stroke damage ( Trendelenberg et al ., 2002 ). Therefore, there may be a common protective role for metallothioneins in stress responses in both animals and plants. <h1>Conclusions</h1> The application of a suppressive subtractive hybridization technique for the cloning of petal transcripts, combined with the use of microarrays to show the expression pattern for each gene has proved a powerful method for the identification of differentially expressed transcripts. However, the very high abundance of the metallothionein transcripts caused a serious problem in the use of the SSH procedure, since few other transcripts were identified in the senescence enhanced libraries. Moreover, the metallothionein genes did not appear to be senescence enhanced in the microarray analysis, even though the Northern analysis showed that this gene family was very strongly up-regulated during petal senescence. This indicates that none of the methods are without potential pitfalls and shows that the application of a combination of techniques is essential to obtain informative results. It is clear from the gene expression analysis in this paper that many senescence-related processes are taking place during the post-harvest senescence of Alstroemeria petals. It is likely that macromolecule degradation and mobilization are occurring in these organs, since genes related to these functions are expressed. Ultrastructural analysis of developing Alstroemeria petals has shown that many structural changes occur before flower opening ( Wagstaff et al ., 2003 ) and many degradative processes may be underway, even by the Stage 0 defined in this paper. These findings are consistent with those shown in other floral systems such as Iris ( van der Kop et al ., 2003 ), although in others such as Sandersonia ultrastructural changes occurred only after flower opening ( O’Donoghue et al ., 2002 ). Therefore, future work will include the analysis of gene expression patterns in younger bud stages. As in medical diagnosis, where expression profiling is proving a powerful tool in identifying marker genes to distinguish, for example, between ovarian and colon cancers ( Nishizuka et al ., 2003 ), we believe that these techniques offer the opportunity to identify genes that are indicative of floral quality. A comparison of the expression profiles of floral genes from a range of genotypes with different post-harvest performance could identify biochemical processes causatively linked to ethylene-independent senescence and thus provide targets for chemical or genetic manipulation. Furthermore, based on the results of the present study, it is now possible to design a microarray that would be informative regarding the ageing of flowers even where visible symptoms are not apparent. A similar approach, using expression profiling, is being proposed in several areas of medicine, for example for the detection of micrometastatic breast cancer ( Baker et al ., 2003 ). A diagnostic chip for floral quality could be used to identify cultivars with improved post-harvest performance or to identify flowers in which the potential vase life has been reduced due to poor handling practices. This could provide retailers with the opportunity to adjust their vase life guarantees to the consumer to reflect the quality of material whilst helping wholesalers and growers obtain premium prices for better quality produce. <h1>Experimental procedures</h1> <h2>Plant material</h2> Alstroemeria flowers (cv. Rebecca) were harvested at Stage 0 of bud development (see Figure 1 ) from Oak Tree Nursery, Egham, UK and transported dry back to the laboratory. Stems were rehydrated in water for c . 30 min before individual cymes were isolated and maintained in vials of dH 2 O. The upper two petals of each developmental stage from bud opening to senescence were harvested, immediately frozen in liquid nitrogen and powdered in a mortar and pestle. Ground, frozen tissue was stored at −80 °C until required for RNA extraction. <h2>RNA isolation</h2> RNA was extracted from 0.5 g aliquots of frozen, ground petal tissue using 5 mL Trireagent (Sigma) according to the manufacturer's protocol but with the addition of two phenol : chloroform : isoamylalcohol separations after resuspension of the first pellet in water. The final aqueous layer was ethanol precipitated overnight at −80 °C and after centrifugation the final pellet was resuspended in RNAse-free water. Total RNA was further purified using an RNAeasy purification column (Qiagen) followed by DNAse treatment with RQ1 DNAse (Promega). <h2>Subtracted library construction</h2> Five subtracted libraries were made using RNA from the following stages of petal development: Stage 0 vs. 2; Stage 2 vs. 0; Stage 3 vs. 2; Stage 4 & 5 vs. 2 and Stage 2 vs. 4 & 5. First strand cDNA was synthesized from 3 µg total RNA using the Smart cDNA synthesis Kit (Clontech). Synthesis of second strand cDNA by LD PCR was optimized to ensure that the ds cDNA was in the exponential phase of amplification. For all templates, amplification was carried out over 17 cycles. PCR-Select cDNA subtraction (Clontech) was performed using Smart ds cDNA according to the manufacturer's protocol, except for the final amplification step, which was carried out over 11 cycles. A subtraction efficiency test was performed for each subtraction as described in the Clontech handbook. The subtracted cDNAs were cloned into the pGEM-T-vector (Promega) and the library was then transformed into E. coli JM109 (Promega). Colony PCR was performed on between 192 and 480 colonies from each subtracted library in order to amplify the inserts for sequencing using M13 forward and reverse primers. DNA was amplified using standard procedures (35 cycles, annealing at 55 °C). PCR products were purified using Millipore MANU 03050 plates and then sequenced using BigDye version 2 (Applied Biosystems) and analysed on an Applied Biosystems 373 sequencer. Database searches were carried out using the B last network service (NCBI). EST sequences were trimmed and assigned to contigs using SeqMan (DNAS tar ). Alignments of protein sequences were carried out using the VectorNTI A lign X programme. <h2>Arrays</h2> Purified PCR products were dried down under vacuum and resuspended in 50% DMSO to give a final DNA concentration of approximately 0.2 µg/µL. Microarrays were printed on CMT-GAPS coated slides (Corning) using a BioRobotics Microgrid II robot. Each slide carried two replicates of an array which itself contained three copies of each target DNA. Slides were baked for 4 h at 80 °C and then stored with dessicant at room temperature. <h2>Probes</h2> Probes were prepared in duplicate with Cy3 and Cy5 reciprocal labelling. Total RNA was treated with RQ1 DNase (Promega), purified with an RNeasy column, vacuum dried and resuspended in ddH 2 O to a concentration of 3.33 µg/µL. For each probe, 20 µg total RNA was reverse transcribed to give either Cy3-dUTP or Cy5-dUTP labelled first-strand cDNA. Briefly, 2 µg oligo pd(T) 12−18 (Invitrogen) was annealed to the RNA by heating to 70 °C for 10 min and then cooled on ice for 1 min. A master mix containing (final concentrations in a total volume of 19 µL) 1× First Strand buffer (Invitrogen); 1 m m DTT (Invitrogen); 1 m m each dATP, dCTP and dGTP (Invitrogen); 0.2 m m dTTP (Invitrogen); 3 nmols Cy3- or Cy5-dUTP (Amersham Biosciences) and 50 Units SuperScript II reverse transcriptase (Invitrogen) was added and the probe incubated at 42 °C for 1 h. A further 1 µL (50 Units) of SuperScript II was then added and the sample incubated at 42 °C for an additional 1 h. The reaction was stopped using 1.5 µL 20 m m EDTA, and the template RNA degraded by the addition of 1.5 µL 500 m m NaOH and heating to 70 °C for 10 min. Samples were neutralized by adding 1.5 µL 500 m m HCl. The Cy3 and Cy5 labelled probes were then combined and purified using a QiaQuick PCR clean-up column (Qiagen), vacuum dried and re-dissolved in 20 µL ddH 2 O. Microarray slides were pre-hybridized in 5× SSC, 0.1% SDS and 1% BSA for 45 min at 42 °C in a Coplin jar and then washed in ddH 2 O for 1 min, rinsed with isopropanol and air-dried. For hybridization, 2 µL yeast tRNA (4 µg/µL) (Invitrogen) and 1 µL polyA (8 µg/µL) (Amersham Biosciences) were added to the purified probe and denatured at 95 °C for 3 min. The probe (23 µL) was then mixed with 23 µL 2× hybridization buffer containing 50% formamide, 10× SSC and 0.2% SDS pre-warmed to 42 °C and applied to the microarray under a HS-60 22 × 60 mm cover slip (Hybrislip, Grace Biolabs). The array was placed in a hybridization chamber (Corning) and incubated for 16–20 h at 42 °C. After hybridization, the slides were washed by agitation successively in 1× SSC/0.2% SDS for 2 × 4 min; 0.1× SSC/0.2% SDS for 1 × 4 min and 0.1% SDS for 4 × 4 min. The arrays were then dried by centrifugation at room temperature and immediately scanned. <h2>Slide analysis</h2> The slides were scanned using an Affymetrix 428 array scanner at 532 nm (Cy3) and 635 nm (Cy5). Scanned data was quantified using I magene version 4.2 software (BioDiscovery) using the following settings: Spot quality labelling (flags) was defined for empty spots with a signal strength threshold of 1.68 and for poor spots with a threshold of 0.27. Background measurements were taken for each spot and were set to 4.0 pixels for the background buffer and 3.0 pixels for the background width. Signal intensity range was set between 19 and 90% and the background intensity range was set from 4 to 90%. Imagene data files for signal median, signal standard deviation, background mean and flags were analysed using G ene S pring version 5.1 (Silicon Genetics). Background values were subtracted from each spot value and the data was then normalized according to the standard 1 colour scenario in the following order; measurements less than 0.0 were set to 0.0, whole chip data was normalized to the 50th percentile and data for each gene was normalized to the median. Thus the total fluorescence for each slide was used to normalize the data to allow comparisons to be made between slides. Each slide carried three replicates of each gene contained within two separate arrays. The scores for the three replicates within each array were immediately averaged in G ene S pring , and the two arrays on each slide were analysed as separate replicates. Therefore if four slides were analysed per treatment, the data obtained was the result of eight replicate arrays. Therefore n = 8 in the statistical analysis carried out using G ene S pring . Genes were identified that showed a significantly different ( P < 0.05) expression from the majority in at least one stage of development ( n = 8). From this list of genes, cluster analysis (K-means) was performed to identify two lists of genes that showed either decreasing expression levels over time (higher expression in Stage 0), or increased expression over time (higher expression at Stages 3 and 4/5) <h2>Northern hybridization</h2> This was carried out as described by Buchanan-Wollaston and Ainsworth (1997 ). DNA probes were PCR amplified EST products labelled with 32 P dCTP using the RediPrime random priming labelling kit (Amersham BioSciences). http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Plant Biotechnology Journal Wiley

Gene expression patterns to define stages of post-harvest senescence in Alstroemeria petals

Loading next page...
 
/lp/wiley/gene-expression-patterns-to-define-stages-of-post-harvest-senescence-6a7vMMf0AY

References (45)

Publisher
Wiley
Copyright
© 2004 Blackwell Publishing Ltd
ISSN
1467-7644
eISSN
1467-7652
DOI
10.1111/j.1467-7652.2004.00059.x
pmid
17147607
Publisher site
See Article on Publisher Site

Abstract

<h1>Introduction</h1> The primary function of large and colourful floral structures (the corolla) on plants is to attract insects or other pollinators. Once pollination has occurred, the role of the flower is over and the corolla is rapidly lost from the plant. The final stages of floral development can take a number of different forms ( van Doorn, 2001 ). The petals of some flowers abscise from the plant with no obvious signs of deterioration, indicating that little remobilization has occurred. The petals of other flowers show extensive wilting (turgor loss, increased translucency and slow desiccation) before they are abscised from the plant. This could indicate that degradation and remobilization of its cellular constituents has taken place. The majority of large monocotyledonous flowers, such as those of the Liliales order including those of Alstroemeria , show this type of senescence. Alstroemeria is a member of the Alstroemeriaceae family and is an important cut flower in Northern Europe. It has a relatively long vase life, taking around 10 days under optimal conditions after harvest to reach full senescence. A time frame for floral senescence including the measurement of biochemical and physiological parameters such as protein degradation has been developed and described previously ( Leverentz et al ., 2002 ; Wagstaff et al ., 2001, 2003 ), and gene expression patterns for a small number of genes during senescence have been analysed ( Wagstaff et al ., 2002 ). Molecular studies of leaf senescence have shown that gene expression patterns change dramatically as senescence progresses, with a large number of genes either showing reduced or increased expression ( Buchanan-Wollaston et al ., 2003 for review). Novel gene expression is essential for senescence to occur. Many different biochemical events occur during leaf senescence, macromolecules are degraded and soluble nutrients are remobilized. Thus, the purpose of leaf senescence is one of redistribution of resources. The study of gene expression changes during flower senescence has not been extensive, although a number of senescence-enhanced genes have been isolated from a variety of different flowers ( Rubinstein, 2000 ). Many of these encode potential catabolic enzymes such as proteases and nucleases, and genes encoding enzymes related to ethylene biosynthesis have also been identified. A more global overview of genes expressed in petals was obtained by DNA sequence analysis of nearly 2000 individual cDNA clones from RNA isolated from rose petals at four different stages of development including senescent petals ( Channeliere et al ., 2002 ). A few of these genes were shown to exhibit enhanced expression in senescing petals. In many species, the time to petal senescence is regulated by ethylene, and a manipulation of the levels of ethylene biosynthesis has been effective in delaying the rate of senescence in several flower types such as the carnation ( Michael et al ., 1993 ) and Torenia ( Aida et al ., 1998 ). However, flowers of some species, such as many of the monocotylenonous plants including Alstroemeria , are not dependent on ethylene for senescence, even though the final abscission of the wilted petal may be ethylene-dependent. Genes that show enhanced expression in senescing monocotyledon flowers have been identified in a number of different ways including differential display (in daylily, Panavas et al ., 1999 ) and subtractive hybridization (in daffodil, Hunter et al ., 2002 ). In each case a small number of genes was identified. We have used a suppressive subtractive hybridization technique to produce several cDNA libraries enriched for genes expressed at particular stages of Alstroemeria petal senescence. Genes showing both up and down-regulation during senescence have been identified, and microarrays were used to determine the overall changes in gene expression patterns that occur during post-harvest senescence. This is a novel approach to the analysis of flower senescence; it has identified extensive collections of genes that reflect the complex processes occurring during petal senescence. A selection of these genes may represent useful markers to determine the stage of flower senescence in this and other floral crops. <h1>Results</h1> <h2>Construction of subtracted libraries and gene identification</h2> Eight different stages of Alstroemeria flower development and senescence have been identified by clearly defined visible changes ( Figure 1 ). Subtracted libraries were constructed to clone and identify genes showing differential expression between these stages. Initially, four libraries were made and these were enriched for: (1) genes expressed in Stage 0 and not in Stage 2, (2) genes expressed at Stage 2 and not at Stage 0, (3) genes expressed in Stage 2 and not in Stages 4 + 5 combined, and (4) genes expressed in Stages 4 + 5 and not in Stage 2. Over 80% of the clones in library (4) represented the same gene family (see below) and so a fifth library was constructed that was enriched for genes expressed in Stage 3 and not in Stage 2. DNA sequence analysis followed by database searches for each subtracted clone identified genes with a wide range of different potential functions. For each library, between 200 and 400 clones were sequenced and, once poor and short sequences and sequences of structural RNA had been removed, a total of 991 EST sequences were generated. In order to identify ESTs that represented the same gene, all the sequences obtained were entered into the alignment program S eqman (DNAS tar ) and common sequences were grouped together as contigs. This resulted in 500 different groups, each representing a separate Alstroemeria gene (Supplementary Table S1A). Of the 991 EST clones characterized in this way, 297 were singletons, i.e. a single representative of a particular gene. All the other sequences occurred at least twice, with the most common gene occurring 21 times. The redundancy of the EST collection was calculated to be 70% (number of ESTs in clusters/total number of ESTs, Sterky et al ., 1998 ). Therefore any new sequence has a 70% chance of already being represented in the collection. The contigs represented by two or more sequences were given an ALF TC number ( Alstroemeria Flower Tentative Contigs). Singletons were given an ALF number. There was likely to be some redundancy in the TC and EST sets, because sequences must have a matching overlap at a certain percentage identity (95%) to be combined, and different parts of the same gene may be represented separately. The potential function of each gene was identified by database searching with the TC or single EST sequence, using B last N and TB last X databases. For each cloned Alstroemeria gene, its likely function and the most similar rice gene or Arabidopsis gene are shown in Supplementary Table S1A. Rice and Arabidopsis were used to identify the closest matched gene, since the genomes of these two plants are almost completely sequenced and annotated. Each of the sequences has been submitted as an EST sequence to G en B ank and the G en B ank codes of the ESTs found in each contig are shown in Supplementary Table S1B. <h2>Differential gene expression determined by subtractive hybridization</h2> Five different subtracted libraries were generated as described above. Up to 400 clones were sequenced from each library and genes encoding a wide range of different functions were obtained (Supplementary Table S1). The genes identified were assigned to potential functional groups according to their sequence similarity and an overall summary of these functions in the EST collection is shown in Figure 2A . A high proportion (29%), of the clones sequenced represented genes with no matches in the database or which showed a similarity to genes of unknown function. Metal binding proteins (mostly metallothionein-like proteins) were the most highly represented class of genes with known potential function (19%). A comparison of the potential functions of the genes identified in each of the subtracted libraries could give an indication of the processes taking place in the petals through post-harvest development. Each subtracted library contained clones representing a different collection of genes, and these genes are listed in Supplementary Table S2. The five different subtracted libraries showed a considerable variation in the range of genes they represented and the potential function of each gene was used to classify them into groups. The relative proportions of different gene functions in each library are illustrated in Figure 2B , and show a considerable variation between the libraries. This graph shows the relative proportions of each functional class of gene in the five different libraries and may be a reflection of the changes that are taking place as the petals are undergoing post-harvest senescence. The libraries are ordered on this graph to show the progression of senescence, the first two (S0–S2 and S2–S4 & 5) should be enriched for genes that are decreasing in expression levels as senescence progresses. The other three libraries (S2–S0, S3–S2 and S4 & 5–S2) should be enriched for genes that are increasing in expression as senescence progresses. From the graph it appears that there are considerable differences in the transcripts that are present at different stages. Cell wall related genes are only present in the first two libraries and also genes involved in metabolism are present at a higher proportion in the earlier stages. As has been described above, genes encoding metal binding proteins (mostly metallothionein-like) were the major component of the subtracted libraries that were constructed to identify senescence enhanced genes, i.e. 4/5–2 and 3–2, and also there were a significant number of metallothionein clones in the library 2–0. However, metallothionein genes were not represented at all in the libraries enriched for genes showing decreasing expression during senescence (i.e. S0–2 and S2–4/5). This indicates that the metallothionein transcripts were subtracted in these libraries and suggests that the subtraction procedure was successful. <h2>Metallothionein genes</h2> Grouping the EST sequences together illustrated the extent of the redundancy in the EST collection. Many genes were represented only once or twice while the largest group consisted of 21 different sequences. The SEQMAN assembly was carried out with a high stringency (95% sequence identity) resulting in separation of different members of the same gene family. For example, the alignment of the many metallothionein sequences showed that there were at least 38 different genes represented, many of which were present multiple times in the EST collection (Supplementary Table S1). The protein sequences for 17 of these different groups of metallothionein genes (those that were each represented four times or more in the EST collection and therefore had a reliable consensus sequence) were predicted from the consensus DNA sequence and compared in an alignment programme using C lustal W, to determine the differences between them ( Figure 3A ). Two main classes of protein were apparent from this alignment ( Figure 3B ). The majority of genes appeared to encode a protein of around 60 amino acids, and this class can be divided into two subclasses, 1A and 1B, as shown in Figure 3A,B . These proteins had between 100 and 90% identity (Class 1A) and 75–67% identity (class 1B) to the ALF 1 protein. The second class of metallothionein (Class 2) was illustrated by two of the TC sequences (ALF TC41 and ALF TC42, each derived from six individual EST clones); these appeared to encode a much smaller protein of around 40 amino acids and showed around 30% identity to the ALF1 protein. Many different metallothionein genes have previously been identified from a variety of plant species. The Alstroemeria genes in class 1A and 1B showed the closest similarity to the metallothionein 1 genes from Fritillaria agrestis , another member of the Liliales order ( Figure 3C ). The most similar match to this protein in Arabidopsis was shown by the Met3 gene (At3g15353), a single copy gene in the Arabidopsis genome ( Figure 3D ). The Alstroemeria Class 2 metallothionein proteins are closely related to the C terminal part of a metallothionein protein (79 amino acids) from Typha latifolia (common cattail) but are most similar in size to the Mt1a protein from Arabidopsis (44 amino acids). The C terminals of these proteins showed similarity, but the N terminal half of the proteins were very different. All the putative metallothionein proteins have characteristic conserved cysteine residues and these are shown in Figure 3A and C . <h2>Gene expression patterns from microarray analysis</h2> All the EST clones generated in the subtractive library screening (1532 clones) were used to generate a microarray. Each slide carried two replicates of an array that contained three copies of each probe. Therefore, each DNA fragment was present six times on a slide and each hybridization was carried out four times with reciprocal labelling. Thus the final data points are the result of hybridization to 24 copies of each gene. Microarrays were hybridized with labelled cDNA made from RNA from four different stages of petal development (Stages 0, 2, 3 and 4/5 mixed together). Some of the genes placed on the microarray were not sequenced until after the hybridization had indicated that they showed potentially interesting expression patterns. For the G ene S pring analysis, data was only included from clones that had been shown by sequence analysis to contain potential gene transcripts. Unsequenced clones, clones containing no inserts and clones representing untranslated RNA were excluded from the analysis. A representative clone for each contig was then selected at random and data for 500 cDNA clones, each representing a different gene, was analysed. From the 500 cDNA clones used for the analysis, 251 showed an expression pattern that was significantly up or down-regulated ( P < 0.05) for at least one point over the time course. Normalized data indicated the relative expression level of each gene in comparison to all the other genes at that stage of development. The t -test P -value showed how significantly this normalized figure differed from the average (i.e. a value of 1). The expression level of each gene at each time point ( y -axis value) was calculated taking account of both the total expression of all genes at that time point to standardize each hybridization, and also the changes in expression level of each gene across the four time points. This allowed the changes in expression levels of each gene over time to be assessed. The changes in expression of this group of genes is shown in Figure 4A (G ene S pring graphs). Only genes that showed at least a 2.5-fold change in normalized expression level between Stages 0 and 4/5 were then selected for further analysis. From this group, two separate clusters were identified, those showing high expression at Stage 0, decreasing at later stages and those showing low expression levels at Stage 0 increasing at Stages 3 or 4/5 ( Figure 4B,C ). To confirm the expression patterns of these selected sets of genes, the experiment was repeated by hybridizing array slides carrying the same cDNA clones, with RNA isolated from a different batch of petals harvested at a different time. At least two slides were hybridized with each RNA sample (i.e. 12 data points, n = 4) and the data analysed. The expression patterns of genes in the two clusters identified as described above were analysed in the new experiment, and in general the patterns of expression were highly comparable. However, a few of the genes did not show a significant change in expression pattern in the repeat experiment and these were removed from the list. Supplementary Table S3 shows the resulting lists of genes that were significantly altered in expression in both experiments and presents the normalized expression level for each gene at the different time points. From this data, the relative expression change between Stages 0 and 4/5 can be calculated. The data in Supplementary Table S3 is summarized in Table 1 , which shows the potential functions of the genes that are present in the up and down-regulated gene lists. There are considerably more genes represented in the up-regulated list than in the list of down-regulated genes. This may be a reflection of the changes in metabolic activity in the petals. In the early stages, 0 and 2, the petals are already fully formed and are not undergoing dramatic metabolic changes, while in later stages a whole new set of genes involved with the degradation and mobilization typical of senescing tissues is expressed ( Table 1 ). <h2>Northern hybridization</h2> Microarray experiments are extremely useful for showing the relative changes in expression levels of a large number of different genes during development. However, the data from microarrays can be subject to considerable experimental variation, and conclusions from these experiments should be tested using techniques such as Northern hybridization to confirm expression patterns for a select number of genes. Northern hybridization analysis was carried out with a selection of genes from Supplementary Table S3. A range of genes was selected from the lists shown in Supplementary Table S3 and these showed the expected expression pattern in Northern hybridization experiments ( Figure 5 ). The MtN19 -like gene, represented by ALF TC139, was identified in the list of genes showing highest expression in Stages 3 and 4/5. The Northern hybridization results support this and show that the expression of this gene continues to increase at later stages of development ( Figure 5B ). Similarly, the Cer1 gene, represented by ALF TC46, appeared in the list showing significant increased expression in Stages 3 and 4/5. The Northern hybridization mirrored this expression pattern exactly ( Figure 5C ). However, unlike ALF TC139, the expression of this gene then decreased at later stages of senescence. The S-adenosyl-L-methionine:salicylic acid carboxyl methyltransferase like gene, represented by the contig ALF TC141, was also present in the list of genes showing significantly higher expression in Stage 4/5. The Northern blotting data also supported this observation and showed that the expression of this gene peaked at Stage 5 and then started to fall in later developmental stages ( Figure 5D ). The 2-oxoglutarate dehydrogenase gene was present in the list of genes showing high expression at Stage 0 with reduced expression later in development. The Northern hybridization results from this gene confirmed this expression ( Figure 5E ). Interestingly, the expression pattern found for the MET1 gene (ALF TC1) was very different to that seen on the microarrays. This gene and most of the other MET contigs were not represented in the senescence enhanced gene list in spite of the fact that the subtraction data suggested that members of this gene family were very highly expressed in the later stages of development. The Northern data confirms the subtraction results – the expression levels detected on the Northern blot were very high in late developmental stages; the autoradiograph shown in Figure 5 was the result of a film exposure of only 3 h. This shows very high levels of transcript at stages 3, 4, 5, 6 and 7 ( Figure 5F ). A longer exposure showed that some transcript was detectable at Stage 2, a very small amount at Stage 1, but none at all was detected at Stage 0 (data not shown). Although the probe was made using the ALF TC 1 cDNA, it is likely that the hybridization seen is also to many of the other MET contigs of Class 1A and 1B, which were very similar in sequence. The DNA sequence of the Class 2 represented by ALF TC 42 was divergent enough to make cross-hybridization unlikely. The expression pattern of this gene was completely different, with detectable transcript at all stages increasing to Stage 4 and then decreasing ( Figure 5G ). The expression level of this gene was also quite high, but weaker than that of ALF TC1 (the autoradiograph shown in Figure 5G was also the result of a 3 h film exposure). <h1>Discussion</h1> <h2>Characterization of subtracted libraries</h2> The suppressive subtractive hybridization technique ( Diatchenko et al ., 1996 ) has been used extensively for the identification of differentially expressed genes in many different organisms. The technique was developed to enrich for differentially expressed genes and at the same time normalize the relative abundance of the different messages in the target population allowing more of the rare transcripts to be cloned. A reported potential drawback of this method is the presence of background clones representing non-differentially expressed genes that can make up a large proportion of the library ( Rebrikov et al ., 2000 ). Five subtracted libraries were constructed using RNA isolated from different stages of petal senescence. From these libraries good sequences were obtained for between 100 and 320 different clones (Supplementary Table S2). Analysis of the DNA sequences of the collections of genes in each library revealed a wide range of different potential gene functions. The most obvious difference between the libraries was the proportion of clones representing a metallothionein-like protein which was often identified in the libraries that were enriched for senescence enhanced genes (Stages 2–0, 4/5–2 and 3–2) but very rarely in the other two libraries that were potentially enriched for genes decreasing in expression during senescence (Stages 0–2 and 2–4/5). This comparison indicated that the subtraction procedure had at least been successful for this gene. However, the very high proportion of metallothionein-like genes (83% and 55%, respectively) in the senescence-enhanced libraries Stages 4/5–2 and 3–2 indicated that the suppression of abundant transcripts for this class of gene was not successful. In fact, this proportion of metallothionein clones exceeds that seen in unsubtracted (and unsuppressed) cDNA libraries made from RNA from these developmental stages where metallothioneins make up about 30% of the identified genes (C. Wagstaff, unpublished results). It therefore appears that the very high level of metallothionein transcripts present in the older tissues have been preferentially amplified and cloned in this SSH procedure. This has presumably prevented the detection of the many other transcripts that are present in this tissue in these libraries, as shown by the microarray and Northern hybridization analyses. The cDNA cloning method has, however, provided an extensive collection of genes that are transcribed during post-harvest petal senescence and the analysis of the potential functions of these can give an indication of the processes that are occurring. The content of the first three libraries, that do not contain many metallothionein transcripts, provides a snapshot of genes that are transcribed in the petals at Stages 0 and 2 ( Figure 2B ). Overall, the genes identified in the different libraries reflect similar functional classifications to those found in other developmental sequencing projects in plants such as Lotus japonicus ( Asamizu et al ., 2000 ) and Citrus sinensis ( Bausher et al ., 2003 ). For example, genes relating to cell wall synthesis, protein synthesis, metabolism and signalling were most abundant in the younger developmental stages. Channeliere et al . (2002 ) carried out a similar analysis on rose petals; they cloned and sequenced 1794 ESTs and showed a range of potential functions comparable to those found in the Alstroemeria petals. <h2>Analysis of the microarray data</h2> More information about the genes expressed at the different developmental stages can be obtained from the microarray data, which provides an expression pattern for each gene during development. Many of the genes on the array did not show significant changes in expression across the four time points, indicating that they were expressed at a relatively constant level in all petal stages. Two clusters were identified containing genes either strongly down-regulated during senescence or showing significantly increased expression between Stage 0 and Stages 3 and 4/5. ( Figure 4 , Tables S3 and 1). Analysis of the functions of these genes can give an idea of the processes that are occurring at the different stages. The small number of genes that were only expressed during the early stages, and were down-regulated as senescence commenced, mainly encoded genes involved in lipid and amino acid biosynthesis as well as genes involved in the TCA cycle and in photosynthesis. The presence of these genes indicated that biosynthetic processes are occurring in the young petal and energy is being produced via the TCA cycle and photosynthesis. Petals in Stage 0 are quite green and this is lost in the later stages ( Figure 1 ). The rapid reduction in expression levels of 2-oxoglutarate dehydrogenase, a TCA cycle enzyme, implied that the production of energy via this pathway only occurred in young petals. Many more genes showed increased expression during the later stages of senescence ( Table 1 ). This indicated that novel pathways are induced during senescence, and the enhanced expression of a wide range of transcription factors and other signalling factors supports the evidence from many plant systems showing that senescence is an active process requiring new transcription and translational events in order to recycle components to other parts of the plant ( Thomas et al ., 2003 ). Many of the genes identified as being senescence-enhanced in Alstroemeria petals have previously been implicated in senescence in other plants and tissues. Protein degradation is an important process that occurs during senescence, and the enhanced expression of genes such as the ubiquitin conjugating enzyme, the aspartic protease and the Clp proteinase show that this process is occurring in the petal tissues. Similar genes have all been shown to have senescence-enhanced expression in other plant systems ( Buchanan-Wollaston et al ., 2003 ) and an aspartic protease was identified in senescing daylily petals ( Panavas et al ., 1999 ). Cytosolic glutamine synthetase has a role in the mobilization of nitrogen from senescing leaves ( Kamachi et al ., 1992 ) and the presence of this gene in Stage 4/5 Alstroemeria petals indicates that nitrogen mobilization is occurring from the petals, presumably to the developing gynoecium ( Nichols and Ho, 1975 ; Nichols, 1976 ). Genes involved in carbohydrate metabolism such as endoxyloglucan transferase and sucrose synthase have also been identified as showing enhanced expression in senescing Arabidopsis leaves ( Park et al ., 1998 ; V.B.-W., unpublished data). These genes may have a role in mobilizing C from cell wall components and converting it to sucrose for transport. The senescence-enhanced expression of a sugar transporter has also been reported previously ( Quirino et al ., 2001 ). Stress related genes such as chitinase and glutathione peroxidase are senescence-enhanced in other systems ( Hanfrey et al ., 1996 ; Page et al ., 2001 ). The gene encoding S-adenosyl-L-methionine:salicylic acid carboxyl methyltransferase, which catalyses the methyl esterification of salicylic acid, producing methylsalicylate, has been implicated in defence responses in Brassica ( Zheng et al ., 2001 ). Moreover, methylsalicylate is a component of the scent of flowers such as Clarkia breweri and the activity of the enzyme encoded by this gene is enhanced in mature flowers of this plant, declining several days after anthesis ( Dudareva et al ., 1998 ). The gene encoding terpene synthase may also have a role in scent production ( Dudareva et al ., 2003 ). The role of these enzymes in the unscented Alstroemeria flowers is not clear, but the coordinated expression of both genes during flower development may indicate the presence of a scent production pathway in Alstroemeria . Although commercial hybrid Alstroemerias such as var. Rebecca used in the work are not scented, native Brazilian cultivars are; it is therefore likely that at least some of the scent producing metabolic pathway is present. In fact, the failure of Alstroemeria flowers to be scented could be related to the mutation of a single gene. Thus the potential for generating scented Alstroemeria flowers, either by conventional breeding or through genetic manipulation, would appear possible. The Northern hybridization results generally confirmed the expression patterns seen in the microarrays. A notable exception to this was the result for the metallothionein genes which showed little change in expression on the microarrays, but were strongly up-regulated according to the Northern hybridization. The reasons for this discrepancy are not obvious. There may be some difference in the stringency of hybridization in the two different systems resulting in more background hybridization on the microarray, or it is possible that the large numbers of different metallothioneins in the subtracted libraries, and consequently the many copies on the microarrays, reduced the hybridization intensity to each individual spot. However, the similarity in sequences between the genes would result in hybridization of the labelled probe to all the similar transcripts in the Northern hybridization, indicating a high expression level. <h2>Potential role of metallothioneins</h2> By far the largest group of genes (19%) found in the libraries encoded a Type 3 metallothionein-like protein defined by Cobbett and Goldsbrough (2002 ). The Northern hybridization experiment showed that an extremely high level of this transcript was present in the older petal stages. The large number (> 20) of different members of this gene family in Alstroemeria may be a reflection of the duplication in the genome in this species. Genes encoding metallothionein-like proteins have also been found in large numbers in the EST collection from rose petals (10.7%) although these were mainly of Type 2 ( Channeliere et al ., 2002 ). Of 401 cDNA sequenced as showing differential expression during strawberry fruit development, only one clone was identified as a Type 3 metallothionein and was represented 27 times ( Aharoni et al ., 2000 ). A Type 3 metallothionein was also the most abundant transcript found in a recent SAGE analysis of rice leaves (2.83%) and accounted for 3.17% of ESTs in a TIGR rice EST collection ( Gibbings et al ., 2003 ). Thus the high proportion of these genes represented in our libraries is not unusual, although the levels are higher than in other systems. Metallothioneins generally contain two cysteine-rich domains, which are able to bind to a variety of metals through mercaptide bonds, and plant metallothioneins can be divided into four types based on amino acid sequence and distinct arrangements of the cysteine residues ( Cobbett and Goldsbrough, 2002 ). Both Class 1A and Class 1B Alstroemeria metallothioneins are members of the Type 3 family of metallothioneins. The Alstroemeria Class 2 metallothionein-like genes cannot be classified, since they do not possess the N-terminal cysteine rich region found in all other plant metallothioneins. Thus, they either represent a novel class of metallothionein or may represent a related gene. The function of metallothioneins in plants is not clear-cut. The isolation of metallothionein proteins from plants has proven difficult due to their instability in the presence of oxygen, and thus their function has largely been inferred from gene expression. Type 3 metallothionein transcripts are highly expressed in ripening fleshy fruits, e.g. in banana ( Clendennen and May, 1997 ), apple ( Reid and Ross, 1997 ) and kiwi ( Ledger and Gardner, 1994 ), and in plants producing non-fleshy fruits, like Arabidopsis , they are also expressed at high levels in senescing leaves ( Cobbett and Goldsbrough, 2002 ), however, no direct link to metal binding has been demonstrated for this type of metallothionein. This is the first report of an abundance of Type 3 metallothioneins in senescent petals, and provides an interesting link between processes occurring in fruit ripening and Alstroemeria petal senescence. Metallothionein genes have been reported in petals from other species, e.g. daffodil ( Hunter et al ., 2002 ) and rose ( Channeliere et al ., 2002 ), however, in both cases the metallothioneins reported were of Type 2. Increased expression of metallothioneins during organ senescence has been reported in Brassica napus leaves ( Buchanan-Wollaston, 1994 ), elder leaves ( Coupe et al., 1995 ) and rice leaves ( Hsieh et al ., 1995 ), however, again these were not Type 3 metallothioneins. Plant senescence is known to generate an increase in free radicals from membrane degradation ( Voisine et al ., 1993 ), and may also generate an increase in free metals such as copper from pigment complexes. Interestingly, metallothioneins are up-regulated in the DAF2 strain of C. elegans that has an increased life-span ( Murphy et al ., 2003 ), where it may be playing a role in preventing or repairing oxidative damage and thereby increasing longevity. In addition, a metallothionein transcript was the most up-regulated transcript in mouse brain tissue that had suffered stroke damage ( Trendelenberg et al ., 2002 ). Therefore, there may be a common protective role for metallothioneins in stress responses in both animals and plants. <h1>Conclusions</h1> The application of a suppressive subtractive hybridization technique for the cloning of petal transcripts, combined with the use of microarrays to show the expression pattern for each gene has proved a powerful method for the identification of differentially expressed transcripts. However, the very high abundance of the metallothionein transcripts caused a serious problem in the use of the SSH procedure, since few other transcripts were identified in the senescence enhanced libraries. Moreover, the metallothionein genes did not appear to be senescence enhanced in the microarray analysis, even though the Northern analysis showed that this gene family was very strongly up-regulated during petal senescence. This indicates that none of the methods are without potential pitfalls and shows that the application of a combination of techniques is essential to obtain informative results. It is clear from the gene expression analysis in this paper that many senescence-related processes are taking place during the post-harvest senescence of Alstroemeria petals. It is likely that macromolecule degradation and mobilization are occurring in these organs, since genes related to these functions are expressed. Ultrastructural analysis of developing Alstroemeria petals has shown that many structural changes occur before flower opening ( Wagstaff et al ., 2003 ) and many degradative processes may be underway, even by the Stage 0 defined in this paper. These findings are consistent with those shown in other floral systems such as Iris ( van der Kop et al ., 2003 ), although in others such as Sandersonia ultrastructural changes occurred only after flower opening ( O’Donoghue et al ., 2002 ). Therefore, future work will include the analysis of gene expression patterns in younger bud stages. As in medical diagnosis, where expression profiling is proving a powerful tool in identifying marker genes to distinguish, for example, between ovarian and colon cancers ( Nishizuka et al ., 2003 ), we believe that these techniques offer the opportunity to identify genes that are indicative of floral quality. A comparison of the expression profiles of floral genes from a range of genotypes with different post-harvest performance could identify biochemical processes causatively linked to ethylene-independent senescence and thus provide targets for chemical or genetic manipulation. Furthermore, based on the results of the present study, it is now possible to design a microarray that would be informative regarding the ageing of flowers even where visible symptoms are not apparent. A similar approach, using expression profiling, is being proposed in several areas of medicine, for example for the detection of micrometastatic breast cancer ( Baker et al ., 2003 ). A diagnostic chip for floral quality could be used to identify cultivars with improved post-harvest performance or to identify flowers in which the potential vase life has been reduced due to poor handling practices. This could provide retailers with the opportunity to adjust their vase life guarantees to the consumer to reflect the quality of material whilst helping wholesalers and growers obtain premium prices for better quality produce. <h1>Experimental procedures</h1> <h2>Plant material</h2> Alstroemeria flowers (cv. Rebecca) were harvested at Stage 0 of bud development (see Figure 1 ) from Oak Tree Nursery, Egham, UK and transported dry back to the laboratory. Stems were rehydrated in water for c . 30 min before individual cymes were isolated and maintained in vials of dH 2 O. The upper two petals of each developmental stage from bud opening to senescence were harvested, immediately frozen in liquid nitrogen and powdered in a mortar and pestle. Ground, frozen tissue was stored at −80 °C until required for RNA extraction. <h2>RNA isolation</h2> RNA was extracted from 0.5 g aliquots of frozen, ground petal tissue using 5 mL Trireagent (Sigma) according to the manufacturer's protocol but with the addition of two phenol : chloroform : isoamylalcohol separations after resuspension of the first pellet in water. The final aqueous layer was ethanol precipitated overnight at −80 °C and after centrifugation the final pellet was resuspended in RNAse-free water. Total RNA was further purified using an RNAeasy purification column (Qiagen) followed by DNAse treatment with RQ1 DNAse (Promega). <h2>Subtracted library construction</h2> Five subtracted libraries were made using RNA from the following stages of petal development: Stage 0 vs. 2; Stage 2 vs. 0; Stage 3 vs. 2; Stage 4 & 5 vs. 2 and Stage 2 vs. 4 & 5. First strand cDNA was synthesized from 3 µg total RNA using the Smart cDNA synthesis Kit (Clontech). Synthesis of second strand cDNA by LD PCR was optimized to ensure that the ds cDNA was in the exponential phase of amplification. For all templates, amplification was carried out over 17 cycles. PCR-Select cDNA subtraction (Clontech) was performed using Smart ds cDNA according to the manufacturer's protocol, except for the final amplification step, which was carried out over 11 cycles. A subtraction efficiency test was performed for each subtraction as described in the Clontech handbook. The subtracted cDNAs were cloned into the pGEM-T-vector (Promega) and the library was then transformed into E. coli JM109 (Promega). Colony PCR was performed on between 192 and 480 colonies from each subtracted library in order to amplify the inserts for sequencing using M13 forward and reverse primers. DNA was amplified using standard procedures (35 cycles, annealing at 55 °C). PCR products were purified using Millipore MANU 03050 plates and then sequenced using BigDye version 2 (Applied Biosystems) and analysed on an Applied Biosystems 373 sequencer. Database searches were carried out using the B last network service (NCBI). EST sequences were trimmed and assigned to contigs using SeqMan (DNAS tar ). Alignments of protein sequences were carried out using the VectorNTI A lign X programme. <h2>Arrays</h2> Purified PCR products were dried down under vacuum and resuspended in 50% DMSO to give a final DNA concentration of approximately 0.2 µg/µL. Microarrays were printed on CMT-GAPS coated slides (Corning) using a BioRobotics Microgrid II robot. Each slide carried two replicates of an array which itself contained three copies of each target DNA. Slides were baked for 4 h at 80 °C and then stored with dessicant at room temperature. <h2>Probes</h2> Probes were prepared in duplicate with Cy3 and Cy5 reciprocal labelling. Total RNA was treated with RQ1 DNase (Promega), purified with an RNeasy column, vacuum dried and resuspended in ddH 2 O to a concentration of 3.33 µg/µL. For each probe, 20 µg total RNA was reverse transcribed to give either Cy3-dUTP or Cy5-dUTP labelled first-strand cDNA. Briefly, 2 µg oligo pd(T) 12−18 (Invitrogen) was annealed to the RNA by heating to 70 °C for 10 min and then cooled on ice for 1 min. A master mix containing (final concentrations in a total volume of 19 µL) 1× First Strand buffer (Invitrogen); 1 m m DTT (Invitrogen); 1 m m each dATP, dCTP and dGTP (Invitrogen); 0.2 m m dTTP (Invitrogen); 3 nmols Cy3- or Cy5-dUTP (Amersham Biosciences) and 50 Units SuperScript II reverse transcriptase (Invitrogen) was added and the probe incubated at 42 °C for 1 h. A further 1 µL (50 Units) of SuperScript II was then added and the sample incubated at 42 °C for an additional 1 h. The reaction was stopped using 1.5 µL 20 m m EDTA, and the template RNA degraded by the addition of 1.5 µL 500 m m NaOH and heating to 70 °C for 10 min. Samples were neutralized by adding 1.5 µL 500 m m HCl. The Cy3 and Cy5 labelled probes were then combined and purified using a QiaQuick PCR clean-up column (Qiagen), vacuum dried and re-dissolved in 20 µL ddH 2 O. Microarray slides were pre-hybridized in 5× SSC, 0.1% SDS and 1% BSA for 45 min at 42 °C in a Coplin jar and then washed in ddH 2 O for 1 min, rinsed with isopropanol and air-dried. For hybridization, 2 µL yeast tRNA (4 µg/µL) (Invitrogen) and 1 µL polyA (8 µg/µL) (Amersham Biosciences) were added to the purified probe and denatured at 95 °C for 3 min. The probe (23 µL) was then mixed with 23 µL 2× hybridization buffer containing 50% formamide, 10× SSC and 0.2% SDS pre-warmed to 42 °C and applied to the microarray under a HS-60 22 × 60 mm cover slip (Hybrislip, Grace Biolabs). The array was placed in a hybridization chamber (Corning) and incubated for 16–20 h at 42 °C. After hybridization, the slides were washed by agitation successively in 1× SSC/0.2% SDS for 2 × 4 min; 0.1× SSC/0.2% SDS for 1 × 4 min and 0.1% SDS for 4 × 4 min. The arrays were then dried by centrifugation at room temperature and immediately scanned. <h2>Slide analysis</h2> The slides were scanned using an Affymetrix 428 array scanner at 532 nm (Cy3) and 635 nm (Cy5). Scanned data was quantified using I magene version 4.2 software (BioDiscovery) using the following settings: Spot quality labelling (flags) was defined for empty spots with a signal strength threshold of 1.68 and for poor spots with a threshold of 0.27. Background measurements were taken for each spot and were set to 4.0 pixels for the background buffer and 3.0 pixels for the background width. Signal intensity range was set between 19 and 90% and the background intensity range was set from 4 to 90%. Imagene data files for signal median, signal standard deviation, background mean and flags were analysed using G ene S pring version 5.1 (Silicon Genetics). Background values were subtracted from each spot value and the data was then normalized according to the standard 1 colour scenario in the following order; measurements less than 0.0 were set to 0.0, whole chip data was normalized to the 50th percentile and data for each gene was normalized to the median. Thus the total fluorescence for each slide was used to normalize the data to allow comparisons to be made between slides. Each slide carried three replicates of each gene contained within two separate arrays. The scores for the three replicates within each array were immediately averaged in G ene S pring , and the two arrays on each slide were analysed as separate replicates. Therefore if four slides were analysed per treatment, the data obtained was the result of eight replicate arrays. Therefore n = 8 in the statistical analysis carried out using G ene S pring . Genes were identified that showed a significantly different ( P < 0.05) expression from the majority in at least one stage of development ( n = 8). From this list of genes, cluster analysis (K-means) was performed to identify two lists of genes that showed either decreasing expression levels over time (higher expression in Stage 0), or increased expression over time (higher expression at Stages 3 and 4/5) <h2>Northern hybridization</h2> This was carried out as described by Buchanan-Wollaston and Ainsworth (1997 ). DNA probes were PCR amplified EST products labelled with 32 P dCTP using the RediPrime random priming labelling kit (Amersham BioSciences).

Journal

Plant Biotechnology JournalWiley

Published: Mar 1, 2004

Keywords: senescence; petal; Alstroemeria ; EST; microarray; gene expression

There are no references for this article.