<%BANNER%>

Effects of Ethylene on Floral Fragrance of Petunia x hybrida 'Mitchell Diploid'


PAGE 1

EFFECTS OF ETHYLENE ON FLORAL FRAGRANCE OF PETUNIA X HYBRIDA MITCHELL DIPLOID By BEVERLY ANN UNDERWOOD A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLOR IDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2003

PAGE 2

Copyright 2003 by Beverly Ann Underwood

PAGE 3

This document is dedicated to my husband, Stuart, my family, and my faithful feline Pookey.

PAGE 4

ACKNOWLEDGMENTS I would like to thank my advisor and mentor, Dr. David G. Clark, for his insight, guidance, and patience through my program. His positive attitude and excitement for science and life have been inspirational. I truly appreciate all that he has done for me. I extend deep appreciation to Dr. Harry Klee in his continued guidance, patience, and generosity through my graduate program. I also would like to thank Dr. Karen Koch, Dr. Don McCarty, and Dr. Jim Barrett. Without their guidance, time, and generosity this work would not have been possible. Your influence will have lasting effects on my life and my future in science. Special thanks to Dr. Kenichi Shibuya for his friendship, lessons, humor, making wonderful cDNA libraries, and showing me together with Holly that 40 RNA gel blot hybridizations are no problem. I thank Holly Loucas for her friendship, all of her help through the years with all aspects of the lab, plant transformations, and interpretations. I thank Kris Barry and Jason Jandrew for their friendship, kindness, and sharing and help with plants. I thank Jenny Davis for her kindness, sharing and help with plants, and leave my best wishes to her in her research with the many petunia sequences, wherever that path may go. I thank Rick Dexter for all of his help with plants, scent panels, lab work, outright sarcasm, and friendship. I especially want to thank Dr. Denise Tieman. I will never forget her thoughtfulness, generosity, and wisdom. She really had a huge impact on me at many levels and I will never forget her true kindness. Also, I wish to thank Dr. Joseph Ciardi iv

PAGE 5

for teaching me from the beginning, showing me how to clone, and to have a lot of fun with all of it. I thank all of the past and present members of Dr. Clarks and Dr. Klees laboratories and to the Floriculture group for their friendship, perspectives, and help with my research. I truly appreciate and am honored to have worked with all of you. I thank my husband Stuart, my Mom, my Dad, and sisters for all of their support, unconditional love, and patience with me when I needed it. I especially thank my faithful cat Pookey for her complete understanding, companionship, and company next to the computer at all hours. v

PAGE 6

TABLE OF CONTENTS page ACKNOWLEDGMENTS .................................................................................................iv LIST OF TABLES .............................................................................................................ix LIST OF FIGURES .............................................................................................................x ABSTRACT .......................................................................................................................xi CHAPTER 1 INTRODUCTION........................................................................................................1 2 LITERATURE REVIEW.............................................................................................4 Ethylene........................................................................................................................4 Ethylene Biosynthesis...................................................................................................4 Regulation of Ethylene Biosynthesis............................................................................5 Ethylene Perception and Signal Transduction..............................................................7 Floral Senescence.......................................................................................................11 Pollination-Regulation of Ethylene Biosynthesis.......................................................12 Prevention of Ethylene Effects...................................................................................15 Gene Expression During Senescence.........................................................................18 Flowers and Pollination..............................................................................................19 Floral Fragrance and Insects.......................................................................................21 Fragrance Composition...............................................................................................24 Fragrance Biochemistry Brief History.....................................................................24 Fragrance Biochemistry Ado-Met Dependent Methyltransferases.........................25 Fragrance Regulation..................................................................................................28 Fragrance Emission in Petunia...................................................................................29 Genetic Engineering for Improved Flower Fragrance................................................29 Petunia........................................................................................................................30 Research Objectives....................................................................................................32 3 MICROARRAY ANALYSIS OF ETHYLENE-INDUCED FLORAL SENESCENCE IN PETUNIA....................................................................................35 Introduction.................................................................................................................35 Results and Discussion...............................................................................................38 Establishment of a Petunia EST Collection........................................................38 vi

PAGE 7

EST Analysis.......................................................................................................39 Microarray Expression Analysis of Ethylene-Treated Flowers..........................41 Materials and Methods...............................................................................................45 Plant Cultural Conditions....................................................................................45 Tissue Collections and RNA Extractions............................................................46 Bioinformatic Analysis........................................................................................47 Functional Categorization of ESTs.....................................................................48 Microarray Fabrication........................................................................................48 Probe Synthesis and Microarray Experimental Procedures................................50 Verification of Microarray Data..........................................................................52 4 ETHYLENE-REGULATED FLORAL VOLATILE SYNTHESIS IN PETUNIA COROLLAS...............................................................................................................61 Introduction.................................................................................................................61 Results.........................................................................................................................66 RNAi PhBSMT Reduces Methyl Benzoate Emission and Changes Floral Fragrance in Petunia........................................................................................66 PhBSMT1 and PhBSMT2 Are Spatially and Temporally Regulated in Petunia Flowers.............................................................................................................67 Substrate Regulation in Response to Pollination and Ethylene Treatments........69 Volatile Emission Is Down-Regulated in Response to Exogenous Ethylene and Pollination.................................................................................................70 Discussion...................................................................................................................71 Transgenic PhBSMT RNAi Plants Have Lower Methyl Benzoate Emission.....72 Ethylene Regulates PhBSMT Expression in Petunia Floral Organs...................73 Pollination and Ethylene Treatments Down-Regulate Floral Volatiles in Petunia.............................................................................................................76 Materials and Methods...............................................................................................79 Plant Material......................................................................................................79 cDNA Isolation....................................................................................................79 Tissue Treatments and Collections......................................................................80 Spatial and Temporal Analysis of mRNA Expression in Flowers......................81 Generation of Transgenic PhBSMT RNAi Petunias............................................82 Volatile Collection and Analysis.........................................................................83 Benzoic Acid & Salicylic Acid Extraction and Quantification...........................84 Human Olfaction Panels......................................................................................84 5 GENERAL DISCUSSION AND CONCLUSIONS..................................................97 Building Fundamental Tools for Genomic Studies....................................................97 Microarray Analysis of Ethylene Regulated Genes....................................................98 Ethylene Regulates Floral Fragrance in Petunia........................................................99 vii

PAGE 8

APPENDIX SUPPLEMENTAL DATA FOR CHAPTER 3......................................102 LIST OF REFERENCES.................................................................................................116 BIOGRAPHICAL SKETCH...........................................................................................130 viii

PAGE 9

LIST OF TABLES Table page 3-1 Sequence characteristics of petunia floral cDNA libraries......................................56 3-2 Number of clones from cDNA libraries not included in the functional analysis.....56 3-3 Contigs from each of the cDNA libraries with the greatest number of clones........56 3-4 Number of cDNAs putatively differentially regulated by ethylene.........................57 3-5 Expression patterns in petunia corollas of differentially regulated cDNAs.............58 4-1 Benzoic acid, salicylic acid, and cinnamic acid levels after ethylene treatment......96 4-2 Benzoic acid, salicylic acid, and cinnamic acid after pollination............................96 A-1 Functional categorization of cDNAs identified from microarray analysis............103 ix

PAGE 10

LIST OF FIGURES Figure page 1-1 Ethylene biosynthesis in plants................................................................................33 1-2 Proposed model of ethylene signal transduction in Arabidopsis thaliana...............34 3-1 Flowchart of petunia EST project and expression studies.......................................53 3-2 Outline of microarray experimental procedures.......................................................54 3-3 Putative functional categories of ESTs....................................................................55 4-1 PhBSMT RNAi reduces MeBA emission and PhBSMT mRNA..............................85 4-2 PhBSMT RNAi reduces MeBA emission only........................................................86 4-3 Volatile emission patterns from MD floral organs...................................................87 4-4 PhBSMT mRNA expression after ethylene treatment..............................................88 4-5 PhBSMT1 mRNA expression in MD and 44568 after pollination...........................89 4-6 PhBSMT2 mRNA expression in MD and 44568 after pollination...........................90 4-7 MeBA emission after ethylene treatment (A) and pollination (B)...........................91 4-8 Rhythmic emission of MeBA and rhythmic expression of PhBSMTs.....................92 4-9 Rhythmic emission of MeBA and rhythmic expression of PhBSMTs.....................93 4-10 Regulation of volatiles from MD and 44568 flowers in response to ethylene.........94 4-11 Regulation of volatiles from MD and 44568 flowers in response to pollination (x) and non-pollination (NP)....................................................................................95 x

PAGE 11

Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy EFFECTS OF ETHYLENE ON FLORAL FRAGRANCE OF PETUNIA X HYBRIDA MITCHELL DIPLOID By Beverly Ann Underwood December 2003 Chair: David G. Clark Major Department: Plant Molecular and Cellular Biology Ethylene is involved with regulating many plant processes including stress responses, fruit ripening, and flower senescence. In these studies, the effects of ethylene on gene expression in Petunia x hybrida Mitchell Diploid (MD) flowers were examined using Petunia cDNA microarrays. The process of developing tools for genomic studies through microarrays is described and is followed by the characterization of two genes isolated from the microarray experiments. Two cDNAs from Petunia hybrida encoding benzoic acid:salicylic acid carboxyl methyltransferases (PhBSMT1 and PhBSMT2) were identified as ethylene down-regulated in petunia flowers. Literature reports show that in vitro these and orthologous carboxyl methyltransferases catalyze the synthesis of methyl benzoate, a ubiquitous floral scent volatile in flowering plants. Expression of the genes in planta was reduced by RNA interference and these plants emitted methyl benzoate less than 10% of wild type xi

PAGE 12

levels, thus demonstrating BSMT function. Human olfaction panels with flowers of these plants demonstrated that the floral fragrance was significantly changed relative to wild type. Wild type and transgenic ethylene-insensitive petunias, 35S::etr1-1, were used to examine ethylene regulation of methyl benzoate emission. The expression of both PhBSMT1 and PhBSMT2 was down-regulated by exogenous ethylene and by pollination-induced ethylene in the floral organs of wild type, but not 35S::etr1-1. Ethylene treatments also reduced emission of other major floral volatiles in wild type, but not emission from 35S::etr1-1 flowers. These results demonstrate a unique role for ethylene in regulation of floral fragrance in petunia. xii

PAGE 13

CHAPTER 1 INTRODUCTION Fragrant and colorful flowers have long attracted the attention of people and animals. In addition to providing an aesthetically pleasing display for humans, they are a food source for many living creatures. Flowers evolved as an efficient means to facilitate sexual reproduction and have evolved into thousands of different forms, using a fascinating diversity of strategies to coordinate pollination and subsequent fertilization. Floral diversity in the plant kingdom is reflected by a continuum of species that vary in fragrance, floral color, gross morphology, and phenology. Plants invest energy in the production of chemicals that make flowers visually attractive and also chemicals that are perceived by olfaction to attract pollinators. Together these characteristics can increase the attractiveness of the plant to pollinators flying in the area, luring them to the flower and thus increasing the chance for a pollination event. One extreme example of a pollinator attaction strategy is observed in many species of orchids. These flowers have heavily reinforced floral organs with thick, waxy cuticles that preserve the flower for many weeks, sometimes months, until the flowers are pollinated (ONeill et al., 1993; Bernhardt, 1999). Flowers of one species of orchid, Grammatophyllum multiflorum, have been recorded to live as long as nine months (Bernhardt, 1999). While most plants do not exhibit the extreme flower longevity observed in some orchid species, there are many plants that exhibit senescence and abscission of organs involved with pollinator attraction in response to pollination. In many plant species the senescence and abscission responses are mediated by the phytohormone ethylene (Stead, 1992). 1

PAGE 14

2 Broad taxonomic studies on plants that exhibit senescence have shown that diverse genera exhibit senescence in response to pollination (Woltering and van Doorn, 1988; van Doorn, 2001; 2002). In these plants, pollination often accelerates senescence through the production and subsequent perception of ethylene in the flower (ONeill, 1997). In response to this ethylene and other pollination signals, the flower is initiated into post-pollination development and transition into fruit and seed development. The purpose of these studies was to investigate processes that are affected by ethylene in petunia flowers. An understanding of these processes is important for understanding plant physiology and identification of ways to circumvent or prevent the effects of ethylene for commercial purposes. This research specifically addressed the effect of ethylene on gene expression in petunia flowers. The process of generating the necessary genetic tools for addressing the effects of ethylene on gene expression is described and data from these experiments are presented. This is followed by characterization of expression of two genes that were ethylene down regulated in the flowers. These two genes were similar to salicylic acid carboxyl methytransferases that have been reported in vitro to catalyze the synthesis of methyl salicylate and methyl benzoate, two ubiquitous floral volatiles. The two petunia genes were named benzoic acid: salicylic acid carboxyl methyltransferase (PhBSMT1 and PhBSMT2) based on in vivo and in vitro data presented here and in Negre et al. (2003). The ethylene down-regulation of PhBSMT1 and PhBSMT2 was verified by examination of expression in flowers of transgenic ethylene insensitive petunias generated by Wilkinson et al (1997) that constitutively express an Arabidopsis thaliana etr1-1 ethylene receptor mutant allele (35S::etr1-1). Corresponding emission patterns of methyl benzoate and other major

PAGE 15

3 floral volatiles emitted by petunia flowers in response to ethylene and pollination are presented.

PAGE 16

CHAPTER 2 LITERATURE REVIEW Ethylene Ethylene is a simple, two-carbon, gaseous plant hormone that is involved in many aspects of plant development. It has roles in regulating seedling growth, vascular differentiation, cell elongation, root development, stress responses, fruit ripening, abscission, and senescence (Abeles et al., 1992; Wang et al., 2002). Since ethylene is involved in such a diverse array of plant processes in many plant species, it has been of considerable interest for biological studies. One of the classical ethylene responses used to identify and study many components of ethylene biosynthesis and signaling is the seedling triple response. The seedling triple response is the development of a short, thick hypocotyl with a pronounced apical hook that develops in the presence of ethylene (Knight et al., 1910). Using the triple response as a genetic screen, mutants in ethylene responses, biosynthesis, and signal transduction have been identified (reviewed in Wang et al., 2002). These mutants largely group into two categories: lack of an ethylene response in the presence of ethylene (ethylene insensitive etr, ein2, ein3) and ethylene response in the absence of ethylene (eto, ctr). Studies on many plant species demonstrate that ethylene biosynthesis, perception, and signaling is conserved and highly regulated to mediate appropriate ethylene responses in the plant life cycle. Ethylene Biosynthesis Ethylene is synthesized in plants in a two-step reaction from S-adenosyl-L-methionine (SAM). SAM is synthesized from methionine and ATP in the methionine4

PAGE 17

5 recycling pathway by SAM synthetase (Adams and Yang, 1979). While SAM is a precursor to ethylene, SAM is also a widely used substrate and cofactor for other reactions in plant cells, such as protein synthesis, polyamine biosynthesis, and ubiquitous methylation reactions (Coruzzi and Last, 2000). The first committed step to ethylene biosynthesis is synthesis of 1-aminocycloproprane-1-carboxylic acid (ACC) and side product 5-methylthioadenosine (MTA) from SAM, which is catalyzed by ACC synthase (ACS) in the presence of a pyridoxal phosphate cofactor (Yu et al., 1979; Sato and Theologis, 1989). The second and final step of ethylene biosynthesis is catalyzed by ACC oxidase (ACO), which oxidizes ACC to ethylene (C 2 H 4 ) and side products of carbon dioxide (CO 2 ) and hydrogen cyanide (HCN) (Hamilton et al., 1991; Spanu et al., 1991). The side products of each of these reactions are further metabolized for sulfur recycling (MTA) and prevention of cellular toxicity (HCN) (reviewed in Wang et al., 2002). Genetic evidence for ACO activity was first shown with tomato plants expressing an antisense ACO transgene (Hamilton et al., 1990). These plants exhibited decreased ethylene biosynthesis resulting in delayed fruit ripening (Hamilton et al., 1990). Regulation of Ethylene Biosynthesis Ethylene synthesis is induced in plants in response to various stresses and developmental processes (Yang, 1987). For example, during the onset of ripening of climacteric fruit there is a rapid increase in ethylene synthesis, which coordinates many of the ripening-associated processes in these fruits (reviewed in Abeles et al., 1992). Ethylene synthesis is both positively and negatively regulated in plants. It is generally controlled through regulation of ACS mRNA levels, enzyme activity, and protein degradation, as well as through conjugation of ACC and control of ACO activity (Wang et al., 2002).

PAGE 18

6 Since the initial cloning of ACS from zucchini (Sato and Theologis, 1989) and ACO from tomato (Hamilton et al., 1990), multiple ACS and ACO genes have been identified, studied in various plant species, and demonstrated to be inducible by various factors. Expression levels of both ACS and ACO are generally correlated with ethylene production levels (Acaster and Kende, 1991), and expression of subsets of ACS and ACO genes have been shown to be ethylene inducible, depending on the tissue and developmental stage (e.g., Woodson et al., 1992; Bui and ONeill, 1998; reviewed in McKeon et al., 1995). The ethylene inducible nature of some ACS and ACO genes illustrates the primary way autocatalytic ethylene synthesis can be induced. Through inducible expression of ethylene biosynthetic genes, ethylene triggers synthesis of itself. However, ethylene synthesis can also be auto-inhibitory. One mechanism for auto-inhibition of ethylene synthesis takes place through inhibition of ACC synthase activity, which restricts ACC supply (Riov and Yang, 1982; Abeles et al., 1992). In addition to ACS and ACO being ethylene-inducible, some ACS and ACO genes are reported to be auxin inducible (e.g., Bui and ONeill, 1998). The differential nature of ACS and ACO gene family regulation likely aids in integration of multiple signals for mediation of appropriate physiological responses at specific times and stages of plant development. Additional post-translational control of ethylene biosynthesis has been demonstrated with the ethylene overproducing mutants (eto mutants) and studies of ACC regulation. Three eto mutants have been identified to overproduce ethylene compared with wild type, except in the presence of ethylene synthesis inhibitors (Guzman and Ecker, 1990; Kieber et al., 1993). Two of the eto mutations identified are disrupted in a key regulatory domain in carboxyl terminal end of two ACS isoforms (eto2 ACS5, eto3

PAGE 19

7 ACS9), resulting in increased ACS stability (Vogel et al., 1998; Chae et al., 2003). This is thought to account for the increased ethylene synthesis phenotype observed in eto2 and eto3 mutant plants, thus implicating post-translational control of ACS in regulation of ethylene synthesis. Whether or not all of the ACS proteins are regulated post-translationally in this manner remains to be examined. Ethylene production is also controlled through regulation of ACC levels. Conjugation of ACC to 1-malonylaminocyclopropane-1-carboxylic acid (MACC) by ACC N-malonyltransferase regulates levels of ACC. (Yang et al., 1987; Peiser and Yang, 1998). This activity creates a sink for excess ACC. This activity also provides an additional level of auto-inhibitory ethylene regulation, as ethylene can increase levels of malonyltransferase (Abeles et al., 1992), and no role for malonylated ACC has been established (McKeon et al., 1995). In summary, there is differential control of ACS and ACO transcript abundance, protein levels, activity, and ACC levels, and these are all potential points of ethylene synthesis regulation in plants. Ethylene Perception and Signal Transduction The genetic components of the ethylene-signaling network have largely been isolated through mutant screens using the seedling triple response in Arabidopsis (reviewed in Wang et al., 2002). The signal transduction mutants fall into two broad categories based on their phenotype: reduced or absence of a response to ethylene (ethylene insensitive phenotype, etr, ers, ein2, ein3) or ethylene responses in the absence of ethylene (constitutive triple response, ctr). Epistasis analysis of these mutants has been used to create a general map of the components in ethylene response signaling networks (Roman et al., 1995).

PAGE 20

8 Ethylene is perceived through a membrane-bound receptor protein dimer, ETR (ethylene response) (Chang et al., 1993; Schaller and Bleecker, 1995). The ethylene receptor was identified through a triple response mutant screen (Bleecker et al., 1988) and subsequent cloning of ETR and mutant alleles by chromosome walking (Chang et al., 1993). ETR was shown to be an ethylene receptor genetically, by conferring ethylene insensitivity in plants transformed with a mutant allele (Chang et al., 1993) and biochemically by generating ethylene binding sites in yeast transformed with ETR (Schaller and Bleecker, 1995). Five confirmed ethylene receptors have been identified in Arabidopsis thaliana (AtETR1, AtERS1, AtETR2, AtEIN4, and AtERS2) (reviewed in Wang et al., 2002) and six in Lycopersicum esculentum, five of which have been confirmed to bind ethylene (LeETR1, LeETR2, NR, LeETR4, LeETR5, and LeETR6) (reviewed in Klee, 2002). Each of the receptor proteins exhibits similarity to bacterial two-component regulators. Bacterial two-component regulators are known signal transducers that consist of a sensor and a response regulator, which function together to regulate cellular responses to environmental stimuli. The amino terminus of ETR, or the sensor, has hydrophobic stretches that are responsible for membrane localization (Schaller et al., 1995), dimerization through covalent disulfide bonds (Schaller et al., 1995), copper co-factor binding (Rodriguez et al., 1999), and ethylene binding (Schaller and Bleecker, 1995). The carboxyl terminus is divergent among the receptors, as some have a conserved histidine kinase domain, while others do not, and in some receptors a response regulator domain is present. The role of each of these domains in ethylene signaling is not yet clear.

PAGE 21

9 Single mutant lines for each of the ethylene receptors in Arabidopsis have all exhibited dominant ethylene insensitivity (reviewed in Wang et al., 2002). One of the mutant alleles, etr1-1, has one missense mutation in the amino terminus, which abolishes ethylene binding (Schaller and Bleecker, 1995). The etr1-1 lesion is a Cys 65 to Tyr missense mutation (Chang et al., 1993), which disturbs coordination of a Cu 2+ cation that is required for ethylene binding (Rodriguez et al., 1999). Plants expressing mutant alleles of the ethylene receptors are disrupted in many responses throughout the life cycle of the plant, including seedling growth, fruit ripening, delayed flower and leaf senescence, and inhibition of adventitious root formation (Lanahan et al., 1994; Wilkinson et al., 1995, 1997; Clark et al., 1999; Gubrium et al., 2000). The dominant nature of the mutation indicates that the receptor negatively regulates ethylene responses. The signal transduction model states that in wild type plants in the absence of ethylene, the ethylene receptor is signaling, repressing an ethylene response. In the presence of ethylene, signaling is inactivated and repression of ethylene responses is relieved. In plants expressing a mutant etr allele, ethylene insensitivity is observed because the mutant receptors continue to signal, thus repressing an ethylene response even in the presence of ethylene. Experiments with receptor loss-of-function mutants in Arabidopsis and antisense lines in tomato have validated this proposed model of receptor action for both plant species (Hua and Meyerowitz, 1998; Tieman et al., 2000). In Arabidopsis, single receptor loss-of-function lines do not show a phenotype, while triple and quadruple loss-of-function mutants exhibit ethylene hypersensitivity (Hua and Meyerowitz, 1998). In cases of single and double loss-of-function mutants, functional redundancy between the receptors likely compensates for the loss of the absent receptors, while triple and

PAGE 22

10 quadruple loss-of-function mutants are no longer able to compensate. Due to decreased levels of the receptors in these plants, signaling through the pathway is presumably reduced to a level that does not repress ethylene responses. Research in tomato uncovered some interesting aspects of receptor regulation. Antisense lines for the NR tomato ethylene receptor exhibit no change in ethylene sensitivity, while antisense LeETR4 lines exhibit increased ethylene sensitivity. When receptor mRNA levels were measured in the two lines, it was discovered that LeETR4 expression increased in the NR antisense line compensating for decreased NR expression and thus wild type levels of ethylene sensitivity were maintained in the antisense NR plants (Tieman et al., 2000). These results indicate that ethylene sensitivity is regulated by receptor levels, thus allowing for fine control of an ethylene response. The ethylene receptor signals to downstream components presumably through a phosphorylation cascade to repress ethylene responses. The next component that has been placed in the pathway is CTR (constitutive triple response), a Raf-like protein-kinase and negative regulator of ethylene responses (Kieber et al., 1993). Plants homozygous for a ctr null mutation exhibit a constitutive ethylene response phenotype even in the presence of ethylene biosynthesis and perception inhibitors (Kieber et al., 1993). Analogous signal cascades in animals and yeast indicate that signaling from CTR may be propagated via a MAP-kinase cascade (Wang et al., 2002) to EIN2 (ethylene insensitive) (Roman et al., 1995). EIN2 mutants are recessive and loss-of-function mutants are completely insensitive to ethylene, indicating EIN2 is required for ethylene responses and that it positively regulates ethylene responses (Alonso et al., 1999). Genetic screens for ethylene insensitivity have also led to the discovery of nuclear factors

PAGE 23

11 that have been shown to regulate ethylene responsive gene expression. EIN3 (Ethylene Insensitive 3) and related EILs (Ethylene Insensitive 3 Like) encode for nuclear localized transcription factors that regulate ethylene responsive genes (Chao et al., 1997; Solano et al., 1998). One of the EIN3 positively regulated genes identified is ethylene response factor1 (ERF1), which encodes a transcription factor that binds to GCC consensus ethylene response elements (ERE) in the promoters of ethylene responsive genes (Solano et al., 1998). Subsequent studies with ERF1 showed that it is induced in response to infection by necrotrophic fungi via ethylene and jasmonic acid signaling. ERF1 in turn regulates expression of a subset of pathogen responsive defense genes (Berrocal-Lobo et al., 2002; Lorenzo et al., 2003). However, ERF1 does not regulate all ethylene responsive genes. For example, in Arabidopsis the ethylene sensitive HOOKLESS1 gene is not induced by ERF1 even though the promoter contains an ERE (Solano et al., 1998). It is likely that there are ERFs regulating subsets of ethylene responsive genes, as the ERFs are a part of a large gene family in Arabidopsis (Riechmann and Meyerowitz, 1998). Floral Senescence Floral senescence is a genetically regulated process in plants that represents the terminal phase of flower development. Since this process affects the visual display of the flower, there have been significant efforts to understand how this process takes place for the goal of increasing flower longevity for commercial purposes. In many plant species, this process is developmentally programmed, but is accelerated by pollination induced ethylene and exposure to ethylene (van Doorn, 1997). Additionally, stress and wounding can cause ethylene synthesis in the flower and therefore induce floral senescence (Ichimura, 1998; reviewed in Rubenstien, 2000). However, in some plants ethylene does

PAGE 24

12 not accelerate floral senescence and developmentally programmed proteolytic events in the flower are predominately responsible for hastening senescence (Beileski and Reid, 1992; Stead, 1992). Pollination-Regulation of Ethylene Biosynthesis One of the first detectable responses to pollination in many flowers is the production of ethylene (ONeill, 1997; Tang and Woodson, 1996). The role of pollination-induced ethylene in mediating post-pollination changes including flower senescence and abscission has been implicated in species from many families of plants (van Doorn, 1997). Studies of ethylene biosynthetic gene expression and corresponding ethylene emission have demonstrated directly that endogenous ethylene synthesis increases following pollination (ONeill, 1997; Bui and ONeill, 1998). Pollination signals and ethylene have a role in coordinating morphological changes in the flower that include stigma closure, ovary maturation, ovule differentiation, and perianth senescence (Zhang and ONeill, 1993; ONeill et al., 1993). There are changes in gene expression as well as post-translational events that take place prior to senescence and are important for mediating these processes (Lawton et al., 1990; Abeles et al., 1992). These pollination induced programmed events are thought to have evolved in some plant genera because after pollination and fertilization the purpose of the petals has been fulfilled. Therefore, maintenance of a floral structure that is attractive to pollinators is no longer necessary and the flower must switch to a phase of fruit and seed development (Rubenstein, 2000). Phalaenopsis, carnation, and petunia flowers all senesce in response to ethylene and exhibit underlying similarities in post-pollination development. In each of these flowers, ethylene synthesis is stimulated shortly after pollination through induction of ACC synthase (ACS) and ACC oxidase (ACO) expression. Both of these enzymes are

PAGE 25

13 encoded by multigene families in these plants and are temporally and spatially regulated in the flowers (ONeill et al., 1993; Bui and ONeill, 1998; Tang et al., 1994; Jones and Woodson, 1999). There are differences in these flowers in the temporal pattern of ethylene production that reflect the spatial patterns of ACS and ACO gene expression. The identity of the initiating signal from the pollination event is still obscure. There is evidence for multiple phytohormones (auxin, ethylene) and hormone precursors (ACC) in initiation of the post-pollination syndrome (ONeill, 1997; Porat et al., 1998). It is possible that the initiating signal is unique to taxonomic groups of plants. Bui and ONeill studied post-pollination responses in Phalaenopsis and proposed a model for inter-organ regulation of the ethylene biosynthetic genes, corresponding enzyme activity, and subsequent morphological and developmental changes (Bui and ONeill, 1998). The Phalaenopsis pollination model is an excellent model system for study because morphological changes in the ovary and perianth are triggered by pollination and not aging. Therefore, there is a clear distinction in changes induced by pollination (ONeill et al., 1993). First, pollination initiates expression of ACS2 in the stigma, which together with basal levels of ACO activity induces ethylene production (ONeill et al., 1993; Bui and ONeill, 1998). Upon the initial synthesis of ethylene, autocatalytic ethylene synthesis initiated as expression of the ethylene responsive ACS1 gene is induced. Pollination also induces expression of ACS3 in the ovary. However, very low levels of ACO activity and ethylene emission have been detected from the ovary (ONeill et al., 1993), so it is thought that ACC produced here is mostly transported to the perianth and labellum (Bui and ONeill, 1998). In the petal organs, ACC together with diffusible ethylene induces ACS1 and ACO1 expression in the labellum and ACO1

PAGE 26

14 expression in the perianth. This stimulates autocatalytic ethylene production from these tissues and signals for senescence. The primary signal inducing ACS2 and ACS3 in Phalaenopsis is thought to possibly be auxin (Bui and ONeill, 1998). Evidence for this was demonstrated with auxin treatments of the stigmas, which induced expression of ACS2 in the stigma and ACS3 in the ovary (Bui and ONeill, 1998). Auxin can also induce ovary growth and differentiation of the ovules in the absence of pollination (Zhang and ONeill, 1993). In summary, the induction of ACS, ACO, and corresponding ethylene is spatially and temporally regulated in Phalaenopsis by pollination, auxin, and ethylene, coordinating post-pollination events in each of the floral organs (Bui and ONeill, 1998). This type of inter-organ regulation of ethylene-sensitive biosynthetic genes demonstrates that ethylene sensitivity is also required for eliciting ethylene responses. However, studies of ethylene receptor regulation in Phalaenopsis have not been published to date. The post-pollination senescence response has also been extensively studied in Petunia. In contrast to Phalaenopsis flowers, petunia flowers exhibit developmental senescence as well as pollination-accelerated and ethylene-induced senescence (Whitehead et al., 1984; Wilkinson et al., 1997; Gubrium et al., 2000). Additionally, pollination-induced ethylene synthesis in Petunia is likely not induced by auxin (Hoekstra and Van Roekel, 1986). After a compatible pollination, germination of the pollen begins as early as one hour post-pollination (PP) (Tang and Woodson, 1996). The stigma produces a large amount of ethylene beginning 2-4 hours PP (Tang and Woodson, 1996; Wilkinson et al., 1997; Jones et al., 2003). Ethylene produced at this time does not hasten petal senescence (Hoekstra and Weges, 1986) and is thought to promote pollen

PAGE 27

15 tube growth in Petunia inflata (Holden et al., 2003). After pollen tubes have begun to grow through the style, another burst of ethylene is produced beginning at 12 hours PP from the stigma+style and ovary. Ethylene is produced from these organs continuously thereafter, peaking around 24 hours PP (Jones et al., 2003), corresponding with the time of fertilization (Tang and Woodson, 1996). These later phases of ethylene production are thought to be responsible for inducing corolla senescence for two reasons. First, treatment of P. hybrida stigmas with an ethylene action inhibitor, 2,4-norbornadiene, prevented ethylene synthesis from the stigma, but did not delay senescence of the corolla (Hoekstra and Weges, 1986). Next, incompatible pollinations in P. inflata do not result in a second peak of ethylene production or senescence (Singh et al., 1992). While the early burst of ethylene does not appear to be responsible for petal senescence, it is clear that there is a wilt-inducing substance that is being transmitted by the stigma+style within the first five hours after pollination (Gilissen and Hoekstra, 1984). Experimental evidence for this was shown when the stigma+styles were removed from the flower at five hours post-pollination and petal senescence was still initiated. Experiments with radiolabeled ACC are suggestive of ACC as the wilt-inducing substance (Reid et al., 1984). It is possible that translocated ACC and diffusible ethylene produced in the stigma+style and ovary are factors in inducing petal senescence in Petunia. Prevention of Ethylene Effects Due to the detrimental effect of ethylene on flower longevity and display, there have been many efforts to find ways to avoid ethylene effects to increase the longevity of flowers. Increasing flower longevity has tremendous commercial value since annual floriculture crop sales total billions of dollars worldwide. Estimated sales of petunia plants in the United States alone in 2002 were approximately 140 million dollars

PAGE 28

16 (Jerardo, 2003). Means of improving floral longevity in ethylene responsive flowers include chemical applications which inhibit endogenous ethylene production or block ethylene perception and genetically engineering flowers for decreased ethylene biosynthesis or perception. Treatments with ethylene biosynthetic inhibitors such as aminoethoxyacetic acid (AOA) and aminoethoxyvinylglycine (AVG) reduce endogenous ethylene production (Serek et al., 1995). However, these treatments do not prevent effects from exogenous ethylene sources and are thus commercially limited. Approaches that block the perception of ethylene have been used to successfully diminish the effects of exogenous and endogenously produced ethylene. Silver thiosulfate (STS) has been used for this purpose for many years (Cameron and Reid, 1982) and works by binding to the ethylene receptors and likely displacing the Cu 2+ cation that is required for ethylene sensitivity, and thus reducing ethylene binding and subsequent responses (Rodriguez, 1999). Compounds that have similar chemical structures to ethylene and competitively bind to the receptor have also been used for blocking ethylene responses. This includes chemicals such as 2,4-norbornadiene (2,4-NBD) (Sisler and Pian, 1973) and 1-methylcyclopropene (1-MCP) (Serek et al., 1994). The latter is a more recent development and has been shown to be advantageous to other receptor binding compounds because it is effective at low concentrations and is non-toxic, whereas 2,4-NBD is highly noxious and STS is a heavy metal and potential ground water contaminant (Sisler and Serek, 1997). While these compounds are effective in blocking existing ethylene receptors, these inhibitors do not block ethylene receptors that are synthesized after the blocking treatments.

PAGE 29

17 Insensitivity to ethylene has also been generated by genetically engineering plants for reduced ethylene biosynthesis or altered ethylene signaling. This approach is advantageous in that no chemical treatments are required and the effects are long lasting. Decreased ethylene biosynthesis has been engineered in carnation through constitutive antisense expression of an ACC oxidase gene (Savin et al., 1995). While these plants synthesize reduced levels of ethylene compared with wild type, the plants were still sensitive to exogenous sources of ethylene. Engineering of ethylene insensitivity through alteration of ethylene signaling has been very successful in increasing flower longevity. While extended floral longevity isnt directly valuable in tomato, plants engineered for ethylene insensitivity at multiple points in the signaling pathway displayed extended flower life in addition to altered fruit ripening (Wilkinson et al., 1997; Tieman et al., 2001; Whitelaw et al., 2002). These results are of interest because they can be applied to ethylene-sensitive floriculture crops. Extended floral longevity through altering ethylene signaling was first demonstrated in tomato, petunia, and tobacco in plants constitutively expressing the Arabidopsis etr1-1 allele (Wilkinson et al., 1997). Similar results were also observed when petunia was transformed with a Brassica oleracea mutant ERS gene (Shaw et al., 2002). In plants expressing heterologous mutant forms of ethylene receptors, flower longevity was increased, but there were negative side effects caused by constitutive ethylene insensitivity (Knoester et al., 1998; Shaw et al., 2002). These plants were significantly reduced in their ability to form adventitious roots (Clark et al., 1999) and were more susceptible to disease and death (Knoester et al., 1998; Shaw et al., 2002). Through flower specific expression of etr1-1, extended floral life was observed in carnation (Bovy et al., 1999). However, horticultural performance characteristics were

PAGE 30

18 not reported for these plants. In tomato plants engineered for reduced EIL (EIN3-like) expression, delayed floral senescence and abscission was observed (Tieman et al., 2001). Since the ethylene signaling pathway seems to be highly conserved in multiple plant species, it is likely that engineering of these points in the pathway will be useful in generating longer lasting flowers. Perhaps future engineering of ethylene insensitivity through tissue specific expression will help to alleviate some of these unfavorable side effects. Horticultural performance studies with these plants will ultimately be required to help answer this question and provide longer lasting flowers to the commercial floriculture markets. Gene Expression During Senescence A number of studies have shown that senescence processes involve changes in gene expression (Buchanan-Wollaston, 1994; Panavas et al., 1999, Quirino et al., 1999). Studies with inhibitors of transcription and translation have demonstrated that synthesis of new RNAs and proteins are required for senescence (Abeles et al., 1992; Borochov and Woodson, 1989; Woodson, 1987; 1993). Through these studies, some differentially regulated genes have been found including cysteine proteases, ribosomal proteins, ethylene biosynthetic genes, nucleases, genes with putative roles in defense, and also genes with unknown function (Lohman et al., 1994; Quirino et al., 1999; Panavas et al., 1999). Many of these studies have focused on leaves for identification of senescence-associated genes (SAGs) and a few of these have examined expression of the SAGs in other senescing organs, like the flowers (Quirino et al. 1999). Quirino et al. (1999) found that many of the genes differentially regulated during leaf senescence were also similarly regulated during flower senescence and in response to pathogen attack. These results present the possibility that there are underlying mechanisms controlling and/or

PAGE 31

19 facilitating the senescence process at the level of gene expression. These results also demonstrate that changes in gene expression are likely important for senescence processes affected by ethylene. Flowers and Pollination Plants have evolved flowers for facilitating sexual reproduction. There is a tremendous diversity of floral forms and strategies that plants use to facilitate pollination and subsequent fertilization. In general, biotic (insect or animal), abiotic (wind), and ambophily (both biotic and abiotic) means of pollination are utilized in flowering plants. In the case of biotic pollination, flowers often have a vivid display of petals and color, secrete nectar, have sticky pollen, and often emit a fragrance (Procter et al., 1996). In contrast, wind pollinated flowers are often less complex and produce large amounts of powdery pollen from flowers situated on the outside of the plant (Culley et al., 2002). Plants capable of both types of mechanisms may be able to take advantage of conditions that may favor one pollination mechanism over the other. An example of this has been observed in Salix spp., where wind pollination was observed to be favored in more open areas and insect pollination was favored in sheltered, forested areas (Vroege and Stelleman, 1990). There are two major views regarding the degree of specialization of biotic pollination in plants. Historically, biotic pollination has been discussed as being a highly specialized relationship that is the result of co-evolution and that there is an evolutionary trend towards specialization of pollinators and plants. In this regard, many plant species have been characterized as having a particular pollination syndrome based on a suite of floral characteristics including morphology, chemical characteristics (fragrance and nectar composition, color), and phenological characteristics (flower opening, temporal

PAGE 32

20 regulation of fragrance) that together attract a particular class or order of pollinators (e.g., beetles or birds; bees or moths). For example, nocturnal moth-pollinated flowers are generally typified as having white, tubular shaped flowers that are heavily scented at night (Knudsen and Tollsten, 1993), while bee-pollinated flowers are usually patterned with UV absorbing chemicals demarcating the nectar and pollen location to pollinating insects and have a weak fragrance (Kevan et al., 1996). More recently, there have been reports that challenge the pollination syndrome dogma with the hypothesis that pollinator systems are more generalized than specialized. One clear, recently published example of this is a study with Isertia laevis (Rubiaceae), which exhibits floral characteristics typical of moth pollination (Wolff et al., 2003). However, both hummingbird species and sphinx moths were observed to visit the flowers, with hummingbird visits being more frequent than the sphingid moths. Hummingbird visits produced a greater number of fruit set than the sphinx moth, but the sphinx moths deposited more pollen than the hummingbirds, which resulted in a greater number of seeds set per visit. Various pollinator types have also been reported to visit the fragrant flowers of Clarkia breweri, with diurnal pollinators accounting for around 20% of the pollen transfer and nocturnal moth visits accounting for 80% of transfer (Raguso and Pichersky, 1995). The diurnal visitors were not described in this study, nor was seed set data. However, these and many studies have shown that plants may attract multiple taxonomic groups of insects to the flowers, but there is likely variability in the effectiveness of the pollinator as reflected in the number of seeds set. In many cases, the relative abundance of the pollinators is also a factor, as more effective, less abundant

PAGE 33

21 pollinators may actually generate similar numbers of seeds to more frequent, but less effective visitors (Wolff et al., 2003; Slauson, 2000; Young, 2002). Another constraint to the pollinator syndrome theory is the evolutionary history of the plant. The genus Datura illustrates this concept. Based on floral characteristics, Datura stramonium is a classic example of a nocturnal moth pollinated flower as it is long, tubular, and heavily scented at night (Grant, 1983). However, studies specifically addressing pollination biology of this plant showed that it was predominately self pollinating as well as visited by bees (Motten and Antonovics, 1992; Motten and Stone, 2000). The more generalist view takes the classic pollinator type tendencies based on pollinator and floral characteristics into account, but also considers other factors such as insect behavior and populations, environmental influences, as well as evolutionary history. Waser et al. (1996) suggests that plant-pollinator interactions should be considered as a web of interactions rather than specialized interactions only and that all flower visitors should be taken into account in plant-pollinator interaction studies. Floral Fragrance and Insects Insects use fragrance cues for localization and selection of potential food sources, nesting sites, and mates (reviewed in Knudsen et al., 1993). While plants produce hundreds of volatile compounds (e.g., Knudsen et al., 1993), not all of them have been found to be perceived by insects and elicit a behavioral response (Fraser et al., 2003). One of the ways that chemical detectability and responses in insects has been studied is by electroantennographic detection (Roelofs and Comeau, 1971). In general, insect antennas are the primary sites of olfactory sensing. Insect antennae are made up of thousands of differentially, finely tuned olfactory receptor neurons that selectively respond to odors, and in total produce an antennal response known as the

PAGE 34

22 electroantennogram (EAG) (Park et al., 2002). Measurements of insect EAGs have shown that the detectability and level of sensitivity to different chemicals is variable among species and between sexes of the same species (Fraser et al., 2003). For example, male and female Manduca sexta moths have similar antennal responses to many plant derived volatiles, but in general female moths are more sensitive to varying concentrations of the volatiles (Fraser et al., 2003). Fraser et al. (2003), Shields and Hildebrand (2001), and others show that insect antennae are not excited by all plant-derived volatiles and that the level of the excitatory responses is dose responsive to the volatile. While these studies demonstrate that an insect can detect odors, it does not indicate behavioral relevance. Studies of insect behavior to odors, such as proboscis extension responses to floral volatiles (Honda et al., 1998), together with EAGs are informative for determining if an odor elicits excitatory feeding responses. Insects have demonstrated the capability to perform associative learning in foraging activities and the ability to discriminate differences in floral odors. These abilities presumably improve foraging efficiency (Williams, 1998). In the bee community, flowers previously visited are avoided due to chemical footprints left on the flower after a visit by a bee (Goulsen et al., 2000). The duration of the repellency was variable depending on the evaporation of the footprint and has been reported to reflect the nectar secretion rate of the plant. These studies show there is an inverse relationship between repellency duration and nectar secretion rate (Stout and Goulsen, 2002). In a conditioning study with the hawkmoth Manduca sexta, an excitatory feeding response to different odors could be conditioned with sucrose solution reinforcement (Daly et al., 2001). The moths were presented with an odor followed by the sucrose conditioning

PAGE 35

23 reinforcement. After conditioning, the odor that was followed by sucrose elicited an excitatory response while an unconditioned odor did not. The response was independent of the odor used, with exception to methyl jasmonate, which did not excite the moths even with sucrose reinforcement. Separate sets of moths were used for each odor, so the ability of the moths to learn a new odor after having learned a different odor was not addressed. The emission of volatiles from flowers is primarily associated with attracting pollinators and the temporal regulation of floral fragrance sometimes reflects this. While it is controversial whether the majority of plants are generalist or specialized with respect to pollinators and pollination syndrome (Waser et al., 1996), there are reports of high fragrance emission at times when pollinator activities are high and cases where high fragrance emission does not correlate with pollinator activity. An example of the former has been observed in the wind and insect pollinated Trimnia moorei flowers. The flowers of this species emit stronger fragrance in the morning, temporally coinciding with visits that result in pollen transfer from both flies and bees (Bernhardt et al., 2003). In Nicotiana, there have been reports of hummingbirds, hawkmoths, and bees that pollinate some of the species in this genus (summarized in Raguso et al., 2003). Based on the pollinator syndrome hypothesis, the hummingbird pollinated Nicotiana forgetiana and N. langsdorffii would be predicted to not be heavily scented, as hummingbird pollinated flowers are not generally associated with a fragrance (Van Riper, 1960). However, the flowers did produce scent, which was primarily emitted at night. Since many of the other species in this genus are hawkmoth pollinated, it is thought that this is a result of its genetic background (Raguso et al., 2003).

PAGE 36

24 Fragrance Composition Floral fragrance is a ubiquitous, diverse, and dynamic characteristic in the plant kingdom. It is comprised of a mixture of small molecular weight (100-290D) volatile compounds (Knudsen et al., 1993; Dudareva and Pichersky 2000). They are derived from multiple pathways including fatty acid derivatives (e.g., methyl jasmonate), monoterpenoids (e.g., linalool), sesquiterpenoids (e.g., caryophyllene), and benzenoids (e.g., methyl salicylate). Floral fragrance is a diverse characteristic, as there have been hundreds of individual volatile components identified from flowers of many different genera (Knudsen et al., 1993). The variations in floral fragrance are compounded by the specific blends of each of the volatile components in quantity, identity, and combination. In many cases, floral fragrance varies throughout the life of the flower as a response to the developmental stage, time of day, and pollination status (Altenburger et al., 1990; Jakobsen et al., 1994; Schiestl et al., 1997; Helsper et al., 1998; Dudareva et al., 2000; Kolosova et al., 2001). This is a dynamic that is regulated developmentally, temporally, and spatially and in response to various stimuli. Fragrance Biochemistry Brief History Identification of volatile components has largely been the focus of much of the work to date on floral fragrance, while there has been less progress on the volatile associated enzyme biochemistry and gene cloning. One of the first enzymatic analyses of floral volatile production was in the annual Clarkia breweri, a plant native to California and the only identified species in the Clarkia genus that emits a strong fragrance (Raguso and Pichersky, 1995; Pichersky et al., 1995). The fragrance of C. breweri flowers is largely dominated by linalool, a terpenoid derivative, with lower levels of linalool oxides and benzenoid derivatives contributing to total fragrance (Raguso and Pichersky, 1995).

PAGE 37

25 The linalool synthase (LIS) enzyme was purified by various chromatographic steps and shown to catalyze the formation of linalool from geranyl pyrophosphate in the presence of a divalent metal cofactor (Pichersky et al., 1995). By N-terminal sequencing of LIS, the corresponding cDNA was cloned from a Clarkia cDNA library (Dudareva et al., 1996). Using this approach and homology based searches of cDNA libraries, other floral volatile biosynthetic genes have been identified including an iso-eugenol methyltransferase (IEMT; Wang et al., 1997), salicylic acid carboxyl methyltransferase (SAMT; Ross et al., 1999), benzoic acid carboxyl methyltranferase (BAMT; Murfitt et al., 2000), benzyl alcohol benzoyl transferase (BEBT; DAuria et al., 2002) and benzoic acid: salicylic acid carboxyl methyltransferase (BSMT; Negre et al., 2003). These studies have shown in vitro enzyme activity data to demonstrate catalytic specificity and molecular biological studies on the developmental and spatial gene regulation patterns. Fragrance Biochemistry Ado-Met Dependent Methyltransferases Volatile methyl esters are among some of the most abundant floral volatiles. To date, enzymes responsible for the synthesis of methyl esters have been cloned and characterized in vitro in Clarkia breweri (Ross et al., 1999), Antirrinhum majus cv. Maryland True Pink (Murfitt et al., 2000; Negre et al., 2002), Arabidopsis thaliana (Seo et al., 2001), Stephonitis floribunda (Pott et al., 2002), and Petunia x hybrida cv. Mitchell Dipoid (Negre et al., 2003). In general, these enzymes use S-adenosyl-L-methionine (SAM or Ado-Met) as a methyl donor to methylate the carboxylic acid moieties of substrate methyl acceptors. The salicylic acid (SA) and benzoic acid (BA) carboxyl methyltransferases are enzymes that catalyze the methylation of the carboxylic acid moiety of SA and BA to methyl salicylate and methyl benzoate, respectively. In the case of the Clarkia SAMT,

PAGE 38

26 Antirrhinum SAMT, and Petunia BSMT, the preferred substrate in vitro is salicylic acid, but they will also accept benzoic acid as a substrate (Ross et al., 1999; Negre et al. 2002, Negre et al., 2003). In contrast, the Antirrinhum BAMT will only accept benzoic acid out of the chemically similar substrates tested (Murfitt et al., 2000). To date, this is the only cloned floral volatile associated carboxyl methyltransferase that exhibits strict substrate specificity. Crystallization of the Clarkia SAMT has shown that differences in substrate preference are largely due to steric constraints brought about by differences in the amino acids that make up the active site of these carboxyl methyltransferases (Zubieta et al., 2003). In sequence comparison of SAMT orthologs with the Clarkia SAMT, it was found that residues responsible for SAM binding and orientation of the substrate and cofactor for methyl transfer are largely conserved (Zubieta et al., 2003). This study further showed that through selected substitutions of amino acids in the active site, the substrate specificity of the Clarkia SAMT could be altered so that jasmonic acid, a comparatively large substrate to salicylic acid, could be methylated. In other methyltranferases, the amino acids in the substrate binding site determine the specificity of the enzyme. For example, when the amino acids that bind iso-eugenol from the Clarkia iso-eugenol methyltransferase (IEMT) were changed to the amino acids that bind caffeic acid in the Clarkia caffeic acid O-methyltransferase (COMT), the specificity of the enzyme was changed so that the hybrid IEMT protein accepted caffeic acid, but not iso-eugenol (Wang and Pichersky, 1998; Wang and Pichersky, 1999). This was demonstrated for a COMT hybrid enzyme as well, which accepted iso-eugenol but not caffeic acid (Wang and Pichersky, 1999).

PAGE 39

27 Results of these in vitro assays do not completely reflect or predict what is synthesized in vivo by each of these gene products. In vivo product formation not only depends on enzyme specificity and abundance, but relative levels of substrates accessible to the enzyme. For example, in Clarkia breweri methyl benzoate has not been detected, but these flowers do emit methyl salicylate (Raguso and Pichersky, 1995). In petunia cv. MD, the flowers do not emit methyl salicylate consistently, but do emit large amounts of methyl benzoate (Verdonk et al., 2003; Chapter 4, in dissertation). Additionally in Antirrinhum, the BAMT is thought to primarily contribute to methyl benzoate emission, as SAMT is not expressed at very high levels (Negre et al., 2002). Transgenic plants with altered expression of these genes and measuring expression levels will be helpful in understanding volatile methyl ester production in these systems. A wide variety of reactions use SAM as a cofactor for methylation and as a substrate in metabolism. This includes nucleic acids, proteins, cell wall constituents, synthesis of ethylene and polyamines, and various secondary metabolites (reviewed in Moffatt and Weredtilnyk, 2001). SAM is the most widely used methyl donor in cellular methylation reactions as it is 10 3 fold more reactive than other methyl donors such as folate and betaine (Cantoni, 1975; Fauman et al., 1999). Due to the number of cellular methylation reactions that use SAM as a cofactor and reactions that use SAM as a substrate, the cellular demand for SAM is high (Moffat and Weretilnyk, 2001). Cells maintain adequate supplies of SAM through the methionine salvage cycle, which recycles de-methylated SAM, S-adenosyl-homocysteine (SAH), to homocysteine and adenosine. Most SAM dependent methyltransferases are inhibited by SAH (Mann and Mudd, 1963) and thus removal of SAH into the recycling pathway is critical in maintaining

PAGE 40

28 methylation reactions in the cell. Homocysteine and adenosine are further metabolized through the SAM/methionine recycling pathway for the regeneration of methionine and subsequent synthesis of SAM. Fragrance Regulation In general, many studies have shown that volatile biosynthetic genes are spatially, developmentally, and temporally regulated and activity and emission patterns correspond. Studies of floral volatile regulation have shown that gene expression, protein activity, and volatile emission patterns are all highly correlated (Pichersky et al., 1994; Dudareva et al., 1996, 2000; DAuria et al., 2002). In Clarkia, linalool synthase (LIS) and benzoyl-coenzyme A: benzyl alcohol benzoyl transferase (BEBT) are expressed in all floral organs, with highest mRNA levels of both occurring in the stigmas (Dudareva et al., 1996; DAuria et al., 2002). Since emission of the volatiles synthesized by LIS and BEBT spatially corresponds with mRNA expression and protein activity (Pichersky et al., 1994; Dudareva et al., 1996; DAuria et al., 2002), it is presumed that volatile synthesis is actively taking place at the site of emission. In snapdragon, the conical cells of the inner epidermal layer of the petals are presumed to be the site of methyl benzoate synthesis based on immunolocalization studies of the BAMT protein (Kolosova et al., 2001). The methylation reaction is proposed to take place in the cytoplasm, as this is where BAMT was detected at the subcellular level (Kolosova et al., 2001). Many of the identified volatile biosynthetic genes and corresponding volatiles have similar developmental patterns of expression. LIS and BEBT mRNA expression and protein activity is detectable in the early bud stage and peaks at anthesis for the following two days (Dudareva et al., 1996; DAuria et al., 2002). In snapdragon, BAMT expression, protein activity, and emission of methyl benzoate increases after flower opening and

PAGE 41

29 continues to increase with development (Dudareva et al., 2000) to a stage where pollination would result in the highest number of seeds set and pollinator visits are highest (Jones et al., 1998). After this stage, expression, activity, and methyl benzoate emission decrease in non-pollinated flowers (Dudareva et al., 2000). Fragrance Emission in Petunia Much of the work to date on fragrance regulation and biochemistry has concentrated on Clarkia breweri and Antirrhinum majus (Dudareva and Pichersky, 2000). Recently, Petunia hybrida cv Mitchell has been used as a model for studying fragrance regulation. Thus far, studies have shown that fragrance emission in petunia is rhythmic, with emission primarily occurring during the evening and night from the petals (Verdonk et al., 2003; Kolosova et al., 2001). The largest component of the fragrance emission profile is methyl benzoate (Verdonk et al., 2003; Kolosova et al., 2001), which peaks around midnight (Kolosova et al., 2001). The precursor to methyl benzoate, benzoic acid, is also rhythmic with maximal levels at night (Kolosova et al., 2001). Recently, the genes encoding for the enzymes that catalyze the formation of methyl benzoate were cloned in our lab (BSMT1, Accession #AAO45012 and BSMT2, Accession #AAO45013) and shown in vitro to catalyze the synthesis of methyl salicylate and methyl benzoate (Negre et al., 2003). Genetic Engineering for Improved Flower Fragrance Cloning and characterization of genes involved with floral fragrance production has presented a novel opportunity for genetic modification of this trait. This is an interesting prospect because many flower-breeding programs have been focused on plant morphology, disease resistance, and characteristics other than fragrance. As a result, floral fragrance has been diluted or is lacking in many plants. Floral fragrance

PAGE 42

30 engineering was first attempted in petunia with the Clarkia breweri linalool synthase gene (Lucker et al., 2001). Six independent transformed lines were obtained and two exhibited 3:1 segregation ratios and expression of the LIS transgene in the T 1 While there was LIS activity in these plants, there was little to no linalool detected. Further experiments showed that linalool was being glucosylated to linalyl--D-glucoside. The authors speculate that conjugation of linalool was a detoxification mechanism because terpenoids have been reported to be detrimental to biological structures. The same gene was transformed into carnation also with the goal of improving the fragrance (Lavy et al., 2002). Linalool was detected in 15 independent transformed plants, with linalool comprising up to 6% of the total volatiles and linalool oxides up to 4% of the total volatiles produced from the flowers. The linalool producing carnation flowers were used in scent panels to test if humans could detect a difference in the fragrance (Lavy et al., 2002). However, the overall fragrance of the flowers was not detectably different by human olfaction. This is clearly an important aspect in improving floral fragrance for commercial value through genetic engineering. Additionally, engineering plants for changes in the volatile profile will also be interesting for ecological studies since insects presumably use fragrance for flower localization. These plants will be useful in answering questions amount the importance of specific components of floral fragrance in pollinator attraction. Petunia The geographic origin of Petunia is South America and it was established as a genus in 1803 by Jusseau (Sink, 1984). It is a member of the Solanaceae family, which is made up of many horticulturally and agronomically important plants including tomato, pepper, potato, eggplant, and tobacco. Many of the modern Petunia hybrida cultivars

PAGE 43

31 have been bred from crosses between Petunia axillaris and Petunia integrefolia (Ando et al., 1999). While there are many Petunia hybrida cultivars in commercial breeding programs, it has also been used for genetic studies in many research laboratories. Anther culture was used with different backcrosses of P. hybrida and many different petunia lines to generate haploid petunia plants (Mitchell et al., 1980). The Petunia x hybrida cv Mitchell is one of these haploid (n=7), anther-derived plants generated from a selected plant from a backcross of Petunia axillaris X Petunia hybrida cv. Rose du Ciel (Mitchell et al., 1980). In tissue culturing of the haploid petunia Mitchell, diploidy was observed to spontaneously arise and polyploidy was induced by treatment with colchicine (Griesbach and Kamo, 1996). The petunia Mitchell Diploid (MD) line was selected from one of these diploid lines. Petunia x hybrida MD is a useful plant for scientific studies for many reasons. It is self-compatible allowing for the maintenance of highly inbred lines. It is derived from the doubling of a haploid plant and therefore does not exhibit any genetic variation in self pollinated progeny. The floral organs are relatively large and the plants are floriforus, allowing for easy manipulation and access to large amounts of tissue for study. The life cycle is relatively short and hundreds of seeds can be produced from a single pollination. The flowers exhibit a predictable senescence response, which is mediated by ethylene (Wilkinson et al., 1997). The flowers are heavily scented, allowing for studies on floral fragrance (Verdonk et al., 2003). Additionally, there is an established transformation protocol that can be used for analysis of transgenic plants (Jorgenson et al., 1996). This plant is used regularly in our lab for transformations and as a model system for studying floral biology.

PAGE 44

32 Research Objectives The purpose of these experiments was to establish genomics tools for Petunia for studying the changes in gene expression induced during ethylene-mediated floral senescence. This was achieved by making floral cDNA libraries, cDNA microarrays, and using the microarrays for identifying ethylene upand down-regulated genes. This is of interest because the changes that occur during floral senescence are not completely understood and an understanding of some of the changes that are occurring at the level of gene expression would allow for a better understanding of this process. From this initial research, ethylene down-regulated gene expression of benzoic acid: salicylic acid carboxyl methyltransferases (PhBSMT1 and PhBSMT2) was observed. The function of PhBSMTs was demonstrated with transgenic plants engineered for reduced expression of PhBSMT1 and PhBSMT2 by RNA interference. In detail expression analysis of PhBSMT1 and PhBSMT2 was examined in all floral organs spatially and temporally in response to ethylene application and pollination. Emission of methyl benzoate and other major floral volatiles was examined after pollination and ethylene treatment. A role for ethylene in regulation of floral volatiles is discussed.

PAGE 45

33 CH3SOHONH3 CH3S+OHONH3Adenine-ribose +NH3OOH CH2CH2 methylations,polyamines, protein synthesisACC oxidase O2CO2 + HCNACC synthase SCH3Adenosine ATPPPi+ PiSAM synthetaseL-methionineS-AdoMetMTAACCethylene CH3SOHONH3 CH3S+OHONH3Adenine-ribose +NH3OOH CH2CH2 methylations,polyamines, protein synthesisACC oxidase O2CO2 + HCNACC synthase SCH3Adenosine ATPPPi+ PiSAM synthetaseL-methionineS-AdoMetMTAACCethylene CH3SOHONH3 CH3S+OHONH3Adenine-ribose +NH3OOH CH2CH2 methylations,polyamines, protein synthesisACC oxidase O2CO2 + HCNACC synthase SCH3Adenosine ATPPPi+ PiSAM synthetaseL-methionineS-AdoMetMTAACCethylene Figure 1-1. Ethylene biosynthesis in plants. Ethylene is synthesized in two steps from S-AdoMet. S-AdoMet is also used as a cofactor for various methylations and protein synthesis, as well as a substrate for ethylene and polyamine biosynthesis.

PAGE 46

34 HKHKRRRR membraneEthylene receptordimercomplexETR1, ERS1, ETR2, EIN4, & ERS2 CTR Phosphorylation in absence of ethylene P EIN2 nucleusEIN3 ERF1 Ethylene regulatedGene trasncription HKHKRRRR membraneEthylene receptordimercomplexETR1, ERS1, ETR2, EIN4, & ERS2 CTR Phosphorylation in absence of ethylene P EIN2 nucleusEIN3 ERF1 Ethylene regulatedGene trasncription Figure 1-2. Proposed model of ethylene signal transduction in Arabidopsis thaliana (adapted from Wang et al., 2002). When ethylene binds to the receptor dimer complex, signaling through CTR is repressed and leading to an ethylene response. The receptors are comprised of a gene family (ETR1, ERS1, ETR2, EIN4, and ERS2 in Arabidopsis). ERS1 and ERS2 do not contain response regulator domains. HK: histidine kinase; RR: response regulator. See text for discussion of signaling components.

PAGE 47

CHAPTER 3 MICROARRAY ANALYSIS OF ETHYLENE-INDUCED FLORAL SENESCENCE IN PETUNIA Introduction Flowers are responsible for sexual reproduction of the planets most diverse group of seed plants. Angiosperms originated 120 million years ago and have evolved into thousands of distinctive flowering plant species that are represented today (Gorelick, 2001). Modern day flora have been shaped through time, selective pressures, and development of interactions with plants and animals that serve as pollinators and seed dispersers. As a result, there are diverse and complex means that plants have adapted in flower development, morphology, and regulation for fulfilling one of the key processes in the life of a plant. One extreme example of floral complexity can be observed in flowers of the parasitic Rafflesia plants. Rafflesia spp. produces red flowers up to three feet in diameter, that emit CO 2 through thermoregulation and sulfuric volatiles that smell like rotting meat to attract pollinating flies (Patino et al., 2002). In many plants, once a flower is successfully pollinated, flowers senesce and/or abscise the floral structures no longer required for growth and development of the fruit and seeds. These processes also can occur with aging, but in many plants pollination accelerates the progression of floral organ senescence and abscission (ONeill, 1993). Senescence and abscission are fundamentally different processes by definition, but with respect to post-pollination processes, both result in the termination of floral organs (van Doorn, 1997). This programmed event likely confers benefits to plants, as these processes are observed in diverse genera. Perhaps through eliminating the attractive parts 35

PAGE 48

36 of the flower, pollinators are directed to non-pollinated flowers, allowing for more efficient pollen dispersal. Additionally, these processes may allow for more efficient use of nutrients and energy for the development of the fruit after a pollination event. In many plants the phytohormones ethylene and auxin are key regulators of senescence and abscission. Ethylene either promotes floral organ abscission and senescence or has no effect on senescence of flowers with strict developmental floral termination programs (van Doorn, 2002). In contrast, auxin has both antagonistic and synergistic effects with ethylene in promoting senescence and abscission in ethylene sensitive species (Taylor and Whitelaw, 2001). Pollination and ethylene-induced senescence and abscission processes have been observed to take place in multiple families. van Doorn (1997) reported ethylene sensitivity in hundreds of representative species from 11 families, many of which have horticultural value. Processes involved with floral organ senescence and abscission are of interest for understanding a vital stage in plant development as well as for the potential of finding ways to prevent or slow down this process for horticultural purposes. A definitive role for ethylene in the regulation of floral senescence has been shown in plants genetically engineered for reduced ethylene synthesis and insensitivity (Savin et al., 1995; Wilkinson et al., 1997; Tieman et al., 2001; Shaw et al., 2002). In petunias expressing the Arabidopsis etr1-1 allele, senescence is delayed in response to exogenous ethylene treatments, pollination, and natural senescence (Wilkinson et al., 1997; Gubrium et al., 2000). While these plants show that ethylene signaling is essential for eliciting senescence, there remain some unanswered questions: What are the underlying molecular changes that occur during senescence? What genes are affected during

PAGE 49

37 Petunia floral senescence? How do these gene expression changes correlate with physiological changes in Petunia? Genomic and microarray technologies provide the opportunity for large-scale gene expression studies (Schena et al., 1995). These types of experiments are useful for screening thousands of genes and identifying differentially regulated candidate genes that may be involved with or affected during a particular physiological process. Candidate genes are identified by comparison of two sets of mRNAs, collected from a reference control tissue and an experimental tissue, making fluorescently labeled cDNA probes from these mRNAs, and examining which have higher levels of a particular message by hybridization to a microarray (reviewed in Lemieux et al., 1998). The hybridized microarray is scanned with a laser and camera to capture the fluorescent image, and computer analysis is performed to identify differentially regulated genes by a comparison of the intensity of the two colored fluorescent probes. The processes of microarray fabrication are shown in figure 3-1 and experiments conducted in this research are shown in figure 3-2. The focus of this study was to examine broad changes in gene expression during ethylene-induced floral senescence. The process of generating the necessary genomics tools for identifying differentially regulated genes is described in this study (Fig. 3-1). For this project, an EST (Expressed Sequence Tag) database was created from clones sequenced from three petunia floral cDNA libraries. These ESTs were subject to bioinformatic analysis for construction of a non-redundant set of clones and functional categorization. Microarrays were constructed and experiments were performed for identifying putative ethylene differentially regulated genes in petunia flowers. These

PAGE 50

38 differentially regulated clones would be used in future physiological studies of processes in the flowers that are affected by ethylene and senescence. Results and Discussion Establishment of a Petunia EST Collection An EST collection from MD petunia was established from three randomly sequenced cDNA libraries made from flowers in a series of developmental stages, ethylene-treated, and post-pollination. The goal in making these three flower libraries was to obtain a broad representation of genes expressed throughout the life of a petunia flower. Random DNA sequencing was performed from the 5 end, generating sequence information for 2603 clones from the floral development library, 2989 clones from the ethylene-treated library and 960 clones from the post-pollinated library (Table 3-1). More clones from these libraries were sequenced; however, no or low quality sequence data were obtained for approximately 6%, 8%, and 13% of total clones sequenced from the developmental, ethylene, and post-pollination libraries, respectively. These numbers are in agreement with other small-scale EST projects with regard to the number of no data clones (Gang et al., 2001). Losses of 6-13% are likely due to sequencing errors or sequencing of plasmids with no inserts. DNA sequencing was continued until close to 50% redundancy was reached for the developmental and ethylene-treated libraries (Table 3-1). At this point, it was estimated that more than two clones would have to be sequenced in order to gain a unique sequence. The post-pollination library was constructed at a later date and sequencing of this library is still in progress. Sequence data for each of these libraries are online and can be found at http://helix.biotech.ufl.edu with permission. BLASTx homology data for the libraries are accessible by password online at http://genomics3.biotech.ufl.edu:8080/bq/blastquest.jsp

PAGE 51

39 EST Analysis A non-redundant (nr) set of clones was assembled in preparation for microarray construction. This set of clones was made in order to allow the maximal number of different cDNAs to be assayed on the microarrays and have a replicate spot on the same array. These data were informative for approximation of how many genes are represented in the library databases and on the microarrays. Random sequencing of cDNA libraries results in some redundancy because intracellular mRNA levels of particular transcripts are often very abundant. As a result, these cDNAs are sequenced multiple times from independent cloning of the multiple transcripts. The nr set consisted of 1370 clones chosen from contigs and 1350 clones representing singlets from the combined developmental and ethylene-treated libraries, resulting in a two fold reduction in redundancy of the clones between the two libraries. All 960 clones sequenced from the post-pollination library were not subjected to this analysis and were included for microarray spotting, as sequencing from this library was in progress at the time the nr set was made. In total, 3,394 clones were amplified by PCR from the three libraries for construction of microarrays. This represents an estimated 3040 expressed genes with some redundancy due to overlap between the pollination library and the non-redundant set taken from the developmental and ethylene libraries. Later analysis of the redundancy of this library showed that it was redundant for approximately 20.7% of the clones. Additionally, seven percent of the total number of clones picked for PCR did not correctly amplify and were subsequently excluded from the set of cDNAs for microarray spotting. Each of the clones in the EST collection was assigned a putative function based on translated sequence similarity with proteins in the NCBI protein databases. While these

PAGE 52

40 data do not absolutely define the function of a gene, they do assist in database mining and as discussed later, aid in picking interesting clones for study from an EST collection. These data also enabled us to assign the clones into putative functional categories (Fig. 3-3), thus describing the range of genes that were represented on the petunia cDNA microarrays. The functional groups will also be useful in future studies if there is a need for assaying a large set of genes involved with a particular process. For example, defense was a category used in grouping the ESTs into functional groups. This group of putative defense-related clones could be used for construction of a microarray for experiments to study how putative defense-related gene expression changes during a stress response. The two largest categories for all of the libraries were unknown and metabolism. The overall percent breakdown of ESTs in each of the libraries was similar with the largest difference in the protein synthesis category, with the greatest number represented in the developmental library (9% in developmental, 6% ethylene-treated, 3 % post-pollinated). A higher representation of protein synthesis cDNAs in the developmental stage library is likely because the library represents tissues that are actively growing from a young bud stage to anthesis, which are stages of active, rapid growth a likely time of greater protein synthesis. The categorization of such a large number of clones as unknown is likely an overestimation as this is based on sequence data from average read lengths of approximately 550 bases. Sequence reads of this length are average for many EST projects (Rounsley et al., 1996; Lange et al., 2000; Gang et al., 2001). Many of the cloned inserts are longer than the average sequence read and may exhibit similarity to a previously identified gene if more sequence information is obtained. This may also be

PAGE 53

41 the case for all of the clones that did not show significant homology (expect value > 1.0 x 10-5) with any proteins from the NCBI database. Searches with longer sequence reads may indicate similarity to proteins in the database. Additionally, if the cloned insert is short and sequence information is from the 3 untranslated region of a clone, it is unlikely to have significant similarities with an ortholog with known function than if it was compared with sequence from the coding region. This could possibly contribute to error in assigning putative functions, since this is a region of high sequence divergence even among gene family members. The numbers of clones that were not functionally characterized from each library, presumably for one of the aforementioned reasons, are in Table 3-2. Categorization of singlets and contigs separately did not drastically change the results of the groupings. The majority of contigs from each of the libraries contained sequences from 2 clones (>69% for each library). There were a few contigs from each of the libraries containing greater than ten clones per contig (~3%, developmental; ~9% ethylene-treated; ~1% post-pollination). These representative abundant clones are listed in Table 3-3. Microarray Expression Analysis of Ethylene-Treated Flowers Microarrays were used to identify differentially regulated genes in flowers during ethylene-induced senescence in order to study physiological processes affected by these treatments. Using the non-redundant set of cDNAs and all clones available from the PP library, the cDNAs were spotted onto glass slides as outlined in Fig. 3-1. As a primary screen for all differentially regulated genes in floral organs, whole flowers were used for microarray analysis (procedure outlined in Fig. 3-2). The microarray experiments undertaken are listed in table 4 and results from these experiments with detailed categorizations of the putative differentially regulated genes listed in appendix 1. These

PAGE 54

42 experiments yielded more genes being ethylene up-regulated than down-regulated (Table 3-4; 7.4% total up-regulated and 0.6% down-regulated) based on a cy3:cy5 ratio > () 2. At this ratio, expression of the representative gene is estimated to be expressed by a difference of two-fold between the two probe sets. A subset of these putative differentially regulated genes were verified by sequencing to check the identity of the cDNA assigned to the array spot of interest. All of the cDNAs checked were identical with the clone assigned to each spot indicating there were no major organizational errors during microarray construction. RNA gel blots were performed with a subset of the clones that were putatively differentially regulated (cy3:cy5 ratio > () 2). This was done to check the results of the microarrays and as a secondary screen for identification of genes specifically differentially regulated by ethylene in petals (Table 3-5). These results show that only a subset of the cDNAs identified as differentially regulated on the microarrays corresponded with RNA gel blot data (Table. 3-5) as shown in blot numbers 3-5a through 3-5l. Many of the putative differentially regulated genes identified by microarray analysis were not differentially regulated in petals (Table 3-5) as shown in blot numbers 3-5m through 3-5ab. There are multiple possibilities that could account for this apparent discrepancy. First, it is possible that many of the putative differentially regulated genes identified from the microarrays were primarily regulated in stigma+styles or ovaries and not the petals. Second, it is also possible that using a heterogeneous tissue like whole flowers as the source for the probes caused a dilution of differentially regulated mRNAs. In support of the latter, many of the differentially regulated genes which corresponded with RNA gel blot data, had high expression levels, as inferred by the intensity of the

PAGE 55

43 expression (Table 3-5 blot numbers 3-5a, 3-5b, 3-5d, 3-5e, 3-5f, 3-5g, 3-5j, and 3-5k) and abundance clones aligning with these cDNAs in the library database (data not shown). Third, it is possible that some of the genes have low expression levels and are therefore difficult to detect by hybridization on a total RNA gel blot. Perhaps a more sensitive detection technique would help to resolve this issue. It is also possible that plant growth conditions were slightly different between times of collection of tissue for microarrays and collection of tissue for RNA gel blots. These reasons could together account for the differences in expression patterns from the microarrays and RNA gel blots. Regardless of the reason for the differences in the data, some clones did exhibit differential regulation in the petals during ethylene treatment and these were considered for future work in studying ethylene-regulated petal senescence. Upon a detailed functional classification (Appendix), the majority of the putative differentially regulated genes fell into three categories based on similarity: nucleotide binding and transcription/translation related, secondary metabolism, and pathogen/stress defense responses. Expression levels in the transcription and defense/stress response category were mostly ethylene up-regulated. One of the ways that plants respond to stress is through transcriptional regulation of genes encoding for proteins that assist in defense and stress responses (Xu et a., 1994; Lorenzo et al., 2003). Many of the genes up-regulated by ethylene have similarity to proteins involved with production of metabolites that have defense-related and therefore overlap with the defense category. For example, a cDNA similar to anthranilate N-hydroxycinnamoyl/benzoyltransferase from carnation (HCBT) was upregulated by ethylene (Table 3-5, blot number 3-5h). The carnation HCBT catalyzes the formation of the defense-related phytoalexin N

PAGE 56

44 benzoylanthranilate from anthranilate, a tryptophan precursor (Yang et al., 1997). In contrast, there were 11 genes with higher expression in the air treatment and 12 genes with higher expression in the ethylene treatments that fell into the category of secondary metabolism. While many of these may have critical roles in plant growth and development (like SAM synthetase, Table. 3-5, blot number 3-5i), they are classified here because they also are essential to the production of secondary metabolites. These changes in expression suggest that there is a change in the secondary metabolism of the flower when exposed to ethylene. Phenylalanine ammonia lyase (PAL) (Table 3-5, blot number 3-5j), a key enzyme in the control of phenylpropanoid metabolism, has higher expression in air at 16 hours. If gene expression is correlating with the level of protein, this suggests that the phenylpropanoid synthesis pathway is reduced in response to ethylene or senescence in petunia flowers. Of the interesting clones strongly down-regulated in petals in response to ethylene were genes exhibiting homology with salicylic acid: carboxyl methyltransferases (SAMT) (Table 3-5, blot number 3-5k). In vitro, SAMT catalyzes the methylation of salicylic acid and benzoic acid to form methyl salicylate and methyl benzoate (Ross et al., 1999). There were multiple spots on the array that corresponded to the petunia clone, as it was found abundantly in the post-pollination library. Sequencing of the full-length cDNAs showed that there were two distinct cDNAs in our EST collection and that the two exhibited high identity, except at the 3 ends of the clones. It is highly likely that both transcripts were detected simultaneously on the microarray and by RNA gel blot hybridization. These results were of interest because they indicated that ethylene could have a role in regulation of a floral volatile at the level of gene expression, which had not

PAGE 57

45 been shown before. These cDNAs were chosen for future studies to examine the role of ethylene in regulation of petunia floral fragrance (Negre et al., 2003; Underwood et al., submitted). This study examined changes in gene expression taking place during ethylene-induced senescence. Genetic tools were created including cDNA libraries and databases, as framework for this study and future molecular biological studies on petunia floral physiology. The cDNAs were used to construct microarrays for screening genes differentially regulated in response to ethylene. A small subset of the cDNAs was identified as differentially expressed in response to ethylene in whole flowers, with most of them being up-regulated by ethylene. Based on sequence similarity, many of these cDNAs may encode proteins with putative functions in transcription/translation, defense, and secondary metabolism. In constrast, there were a few cDNAs associated with secondary metabolism that were more abundant in the air treated flowers. This included two cDNAs with similarity to floral volatile carboxyl methyltransferases, which were chosen for the focus of future studies. In total, these results indicate that ethylene possibly up-regulates defense related processes, while shutting down processes no longer required, like fragrance emission, as petunia flowers senesce. Materials and Methods Plant Cultural Conditions Petunia x hybrida Mitchell Diploid (MD) was used for all experiments described in this research. Seeds were sown on Fafard #2 potting mix (Fafard Co.; Apopka, FL) in six-pack seed trays and placed in misthouse for mist every 30 minutes for 10 seconds. One tablespoon of vermiculite was placed on top of soil three days after seed sowing. Germinating seeds were left in mist until cotyledons had emerged and were visible. After

PAGE 58

46 germinating, seedlings were transferred into greenhouses set at day/night temperatures of 25C/18C. Plants were grown in flats for approximately four weeks until transplanting. Plants were transplanted into 1.2 L pots with Fafard #2 potting mix and fertilized every other watering with 150 mgL -1 Excel 15-5-15 (Scotts Company; Marysville, Ohio). Tissue Collections and RNA Extractions Floral tissues were collected from Petunia x hybrida MD for RNA extractions and subsequent cDNA library synthesis. There were three sets of whole flower tissue were collected for synthesis of three cDNA libraries: 1. developmental stage (from flowers collected on the same day at five stages of development beginning at early bud to anthesis); 2. ethylene-treated (collected after 2.5 LL -1 ethylene treatments for 30 minutes, 1 hour, 3 hours, 6 hours, and 12 hours); and 3. post-pollinated (collected 1, 2, 5, 10, 24, and 34 hours after pollination). For all collections, whole flowers were harvested at the indicated times, placed into 50 mL falcon tube, and immediately put into liquid nitrogen and stored separately at C for preservation until RNA extractions. Total RNA was extracted for the developmental stage, ethylene-treated, and post-pollination libraries (Wan and Wilkins, 1994 and Ciardi et al., 2000). Samples from each set of tissues were extracted separately for each respective timepoint. Following extraction and quantification, total RNA from each timepoint was combined in equal molar amounts resulting in three representative samples of developmental, ethylene-treated, and post-pollination RNAs for synthesis of each cDNA library. Poly (A)+ mRNA was isolated using a Poly (A)+ collection kit from Stratagene (LaJolla, CA). cDNA libraries were constructed from poly (A)+ mRNA using a uni-directional -ZAPII cDNA synthesis kit from Stratagene (LaJolla, CA). Mass excisions were conducted on

PAGE 59

47 unamplified library, plated on LB ampicillin media, and taken to the Interdisciplinary Center for Biotechnology Research Genomics Sequencing Core Laboratory (University of Florida; Gainesville, FL) for dye terminator capillary sequencing. Approximately 10% of bacterial cultures from plated mass excisions were picked randomly for sequencing to reduce the amount of redundant sequencing. All clones were organized as glycerol stocks in 96 well plates for long-term storage at -80C. The clones were numbered in this database according to library, plate number, and well number. For example, Petunia-PP-12-A01 is a clone from the petunia post-pollination library and corresponding glycerol stock is located in plate PP-12 in well A01. Bioinformatic Analysis One-pass sequence reads from the cloned unidirectional cDNAs in pBluescript were obtained using the T3 plasmid primer site, corresponding to the 5 end of each cDNA. The ESTs were organized into directories for separation of sequences into unique reads (singlets or singletons) or non-unique redundant reads (contigs). Assembly of the ESTs into groups was facilitated by the use of programs Phred and Phrap (University of Washington; Seattle, WA). After assembling the ESTs into singlets and contigs, Consed was used to view the contig alignments (University of Washington; Seattle, WA). EST redundancy was calculated by dividing the numbers of clones in contigs by the total number of clones sequenced. A non-redundant set of clones was generated by picking all singlet clones and one clone from each contig if the clones were in good alignment as viewed by contig sequence alignments in Consed. The criteria for chosing clones from contigs for the non-redundant set was based on the following sequence characterisitics: Sequence longer than 200 base pairs. If all clones aligned in contig, the EST with longest sequence read chosen.

PAGE 60

48 Clones aligned at ends of sequence reads (< 50 base pairs) both chosen. Clones not truly in alignment were all chosen. Functional Categorization of ESTs Contigs and singlets were categorized into functional groups based on homology with the NCBI protein databases (http://www.ncbi.nlm.nih.gov/Database/index.html). Functional group categories were in accordance with the categories in the Munich Information Center for Protein Sequences (MIPS; http://mips.gsf.de) and have been used by multiple groups for this purpose (Van der Hoevan et al., 2002; Guterman et al., 2002). The categories are as follows: metabolism; energy; cell cycle and DNA processing; transcription; protein synthesis; protein fate; cellular transport and transport mechanisms; cellular communications and signal transduction; cell rescue, defense, and virulence; regulation and interaction with cellular environment; cell fate; control of cellular organization; subcellular localization; protein activity regulation; protein with binding function or cofactor requirement; transport facilitation; classification not yet clear cut; and unknown. ESTs with expect values less than 1.0 x 10-5 were assigned into functional groups, as this was the threshold for homology set by Guterman et al. (2002). Microarray Fabrication Clones chosen for the non-redundant set were picked from the 96 well plate glycerol stocks into plates with wells containing 150 L Luria Broth with Ampicillin selection (50 mg/mL). Cultures were picked into 96-well plates, placed in 37C non-shaking incubator, and grown overnight. The following day, PCR was performed using the overnight-grown E. coli cultures to amplify the petunia cDNA inserts using T3 and T7 primers. Reactions were set up in 96 well plate format and inoculated with 2 L of

PAGE 61

49 bacterial culture, while the remaining culture was made into glycerol stocks for long-term storage of a non-redundant set of clones. PCR reactions were run under the following conditions: 95C for five minutes, followed by 35 cycles of 94C for one minute, 53C for one minute, and 72C for one minute, and finishing with 72C for seven minutes. After PCR was performed, products were analyzed by gel electrophoresis to verify amplification and absence of multiple bands, and the remainder of amplified stock was stored at C. Twelve percent of clones picked for culturing and PCR repeatedly did not grow, amplified more than one band, or did not amplify and were excluded from spotting onto the arrays. Products from the successful PCR reactions (22 L) were aliquoted into 384-well plates with four microliters of spotting solution (20X SSC and 20% sarkosyl). Plates containing PCR products and spotting solution were prepared for spotting and stored at 4C one day prior to use in the arrayer. Gold Seal glass slides (Corning, Toledo, OH) were used as the surface support for spotting microarrays. Slides were prepared and processed according to a modified procedure described by Eisen and Brown (1999). First, the slides were thoroughly cleaned in alkaline ethanol solution for 120 minutes by gently shaking at room temperature in a metal slide rack and glass chamber from Shandon Lipshaw (Pittsburgh, PA). Following the cleaning, slides were rinsed with approximately four liters of filter-sterilized dH 2 0 with minimal exposure to air and dust following this step. Next, the slides were coated with poly-L-lysine by transferring rinsed slides into a chamber containing a freshly prepared coating solution of 10% poly-L-lysine/10% PBS (Sigma-Aldrich Corp, St. Louis, MO). Slides were dried at room temperature overnight and then

PAGE 62

50 placed into a plain plastic slide box placed in a plastic container with Drie-Rite dessicant (Xenia, OH) until spotting the following day. cDNAs were spotted on to poly-L-lysine coated slides using an Affymetrix 418 Robotic four-pin Arrayer (Santa Clara, CA). cDNAs were spotted in an array pattern with 375 M distance between the centers of the spots, each spot was stamped twice in the same location to ensure deposition of each sample. Each cDNA sample was spotted in two locations on the slide so that all samples were replicated on each slide. After spotting, the slides were numbered and the arrayed area was etched into the slide with a diamond tip scribe (Fisher Scientific; Hampton, NH), then stored at room temperature in a plain slide box in a container with dessicant until processing and use. As the microarrays were needed, slides were processed (modified procedure from Eisen and Brown, 1999) by brief steam hydration, UV crosslinking, washing in 0.2% SDS for 10 minutes, rinsing with 2 L filtered dH 2 0, denaturing in boiling in water bath for 10 minutes, and dehydrating in 95% cold ethanol. All microarray experiments were performed within two months of spotting and processing was done as each experiment was performed. Probe Synthesis and Microarray Experimental Procedures The microarrays were used to screen for genes up and down-regulated in response to exogenous ethylene. All treatments were initiated the day after anthesis at 10 a.m. under sunny weather conditions to help eliminate developmental, temporal, and environmental variability. Flowers were treated with ethylene or air for 2 hours and 8.5 hours. The 16 hour sample was collected from flowers treated with ethylene for 8.5 hours to induce senescence, removed from ethylene treatment and then samples were collected 7.5 hours later. For ethylene and air treatments, flowers were excised and

PAGE 63

51 placed into 1.5 mL microfuge tubes containing 1.0 mL dH 2 0. Flowers were sealed in 37.85 L glass chambers and treated with 2-3 LL -1 ethylene. For air treatments, flowers were placed in the same conditions, but no ethylene was added to the chamber. Concentrations of exogenous ethylene were verified at the beginning and end of indicated treatment times using a Hewlett Packard Gas Chromatograph (Model 5890, Series II) equipped with a flame ionization detector and an alumina column. For tissue collection, flowers were collected at the indicated times, placed into 50 mL falcon tubes in liquid nitrogen, and stored at C until RNA extraction. Probes were prepared from total RNA extracted from the treated flowers by Phenol-Chloroform extractions followed by lithium chloride precipitations (Ciardi et al., 2000). After extractions, RNA was cleaned using a Qiagen RNeasy kit (Qiagen Inc; Valencia, CA) and following the manufacturers instructions. RNA was quantified by spectrophotometer readings and quality was checked by gel electrophoresis. Fluorescent probes were made using the Submicro EX Expression Array Detection Kit from Genisphere (Hatsfield, PA). Using this kit, cDNA probes were synthesized by selectively reverse transcribing mRNA from 50 g total RNA with an oligo poly-T primer tagged with a binding site for fluorescent dyes cyanine 3 (cy3) or cyanine 5 (cy5). Approximately one quarter of the newly generated cDNAs was aliquoted for each labeling reaction (approximately 12.5 g total RNA was used per dye per hybridization). The dye was tagged to the cDNA probes by incubation in the dark at 55C for one hour. Probes were hybridized to the microarray in the dark at 55C for at least 24 hours. The following day, slides were washed in the dark in three salt solutions (2X SSC, 2% SDS, 20 minutes at 55C; 2X SSC 20 minutes at room temperature; 0.2X SSC, 20 minutes at room temperature) and immediately scanned in an Affymetrix 428

PAGE 64

52 scanning microscope (Santa Clara, CA). The scanned images were analyzed for intensity readings using the Affymetrix Jaguar 1.0 software (Santa Clara, CA). Intensity ratios for the dyes (cy3 intensity/cy5 intensity) at each of the spots were used to find genes of interest. Each experiment was done in triplicate with three slides and only spots that showed consistent results as being up or down regulated by a cy3/cy5 ratio of at least ( + ) 2.0 for four out of the six spots were considered for further analysis. Verification of Microarray Data Results obtained from microarray experiments were checked by RNA gel blots. Total RNA was extracted from tissue (Ciardi et al., 2000), quantified, and quality was verified by gel electrophoresis on a 1% agarose, 1X TBE gel. Total RNA was separated on a denaturing formaldehyde gel and blotted as described (Kneissl and Deikman, 1996). Probes were prepared from PCR amplified cDNA inserts (PCR as described above) from the petunia cDNA libraries and labeled with 32 P dCTP using a Prime-It II Random Primer Labeling Kit from Stratagene (LaJolla, CA). Membranes were hybridized and washed as described (Deikman and Fischer, 1988).

PAGE 65

53 Figure 3-1. Flowchart of petunia EST project and expression studies. cDNA libraries were synthesized from three sets of Petunia MD floral tissues (1,2), randomly sequenced (3), analyzed for redundancy and putative functions (4), microarrays made from non-redundant cDNA set (5), and used for gene expression studies of senescing floral tissue (6). Interesting genes were verified for expression patterns by RNA gel blot analysis of petal tissues.

PAGE 66

54 Extract RNA AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA RT & Label AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA Hybridize probes to microarray AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAScan & Data Analysis Air treated Flower (A)C2H4treated Flower (B) Red ABExtract RNA AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA RT & Label AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA Hybridize probes to microarray AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAScan & Data Analysis Air treated Flower (A)C2H4treated Flower (B) Red AB Figure 3-2. Outline of microarray experimental procedures. Putative differentially expressed genes were identified according to these types of experiments during ethylene-induced senescence in the flowers. First, RNA was extracted from air treated or ethylene treated flowers, reverse transcribed to make cDNAs and for labeling with cy3 (green) or cy5 (red) fluorescent dyes. Dye labeled cDNAs are hybridized to microarray and microarrays are scanned. The intensity of the two colors is calculated, with ratios above two indicating a significant difference in message levels in the two tissues.

PAGE 67

55 18%5%2%4%6%5%2%7%8%3%31%3%2%1%0%2%1%2% 15%5%3%4%9%6%2%5%9%4%29%1%1%2%3%0%1%0% 17%5%1%6%3%6%0%6%9%1%3%33%2%0%1%0%2%3% Energy Transcription Protein Fate Cellular Communications & Signal Transduction Regulation & Interaction with Cellular Environment Control of Cellular Organization Protein Activity Regulation Transport Facilitation Unknown Metabolism Cell Cycle & DNA Processing Protein Synthesis Cellular Transport & Transport Mechanisms Cell Rescue, Defense, & Virulence Cell Fate Subcellular Localization Protein with Binding Function or Cofactor Requirement Classification not yet clear cut ACB 18%5%2%4%6%5%2%7%8%3%31%3%2%1%0%2%1%2% 15%5%3%4%9%6%2%5%9%4%29%1%1%2%3%0%1%0% 17%5%1%6%3%6%0%6%9%1%3%33%2%0%1%0%2%3% Energy Transcription Protein Fate Cellular Communications & Signal Transduction Regulation & Interaction with Cellular Environment Control of Cellular Organization Protein Activity Regulation Transport Facilitation Unknown Metabolism Cell Cycle & DNA Processing Protein Synthesis Cellular Transport & Transport Mechanisms Cell Rescue, Defense, & Virulence Cell Fate Subcellular Localization Protein with Binding Function or Cofactor Requirement Classification not yet clear cut 18%5%2%4%6%5%2%7%8%3%31%3%2%1%0%2%1%2% 15%5%3%4%9%6%2%5%9%4%29%1%1%2%3%0%1%0% 17%5%1%6%3%6%0%6%9%1%3%33%2%0%1%0%2%3% Energy Transcription Protein Fate Cellular Communications & Signal Transduction Regulation & Interaction with Cellular Environment Control of Cellular Organization Protein Activity Regulation Transport Facilitation Unknown Metabolism Cell Cycle & DNA Processing Protein Synthesis Cellular Transport & Transport Mechanisms Cell Rescue, Defense, & Virulence Cell Fate Subcellular Localization Protein with Binding Function or Cofactor Requirement Classification not yet clear cut Energy Transcription Protein Fate Cellular Communications & Signal Transduction Regulation & Interaction with Cellular Environment Control of Cellular Organization Protein Activity Regulation Transport Facilitation Unknown Metabolism Cell Cycle & DNA Processing Protein Synthesis Cellular Transport & Transport Mechanisms Cell Rescue, Defense, & Virulence Cell Fate Subcellular Localization Protein with Binding Function or Cofactor Requirement Classification not yet clear cut ACB Figure 3-3. Putative functional categories of ESTs. This represents cDNAs from the (A) developmental, (B) ethylene-treated, and (C) post-pollinated flower libraries. ESTs were categorized based on homology with proteins in NCBI protein databases if expect value was > 1.0 x 10 -5 Functional categories were assigned according to MIPS functional categories. Number of clones not included in this analysis are listed in table 2.

PAGE 68

56 Table 3-1. Sequence characteristics of petunia floral cDNA libraries. Percent redundancy was calculated by dividing the number of clones in contigs by the total number of clones with sequence data. Tissue Library name Clones sequenced Clones with data % redundancy Developmental flower DevA 2774 2603 45.50% Ethylene-treated flower C 2 H 4 3264 2989 45.00% Post-pollinated flower 9/2001 PP 960 835 N/A Post-pollinated flower present PP 1632 1396 29.60% Table 3-2. Number of clones from cDNA libraries not included in the functional analysis. These were excluded because there was not high homology with proteins in the NCBI protein database (expect value > 1.0 x10 -5 ) or there was no sequence information. Library High E value No Data Developmental 466 167 Ethylene-treated 399 181 Post-pollination 13 170 Table 3-3. Contigs from each of the cDNA libraries with the greatest number of clones. Library Most abundant clones # Clones Developmental putative oxidoreductase 11 ascorbate peroxidase 11 Polyubiquitin 12 Unknown 13 Elicitor inducible gene product 13 putative metallothionein-like protein 16 Ethylene-treated lipid transfer protein 10 putative metallothionein-like protein 14 ACC oxidase 16 polyubiquitin 17 isoflavone reductase-like protein 17 metallothionein-like protein type 2 23 Post-pollinated pectate lyase 7 SAM: SA carboxyl methyltransferase 8 isoflavone reductase-like protein 8 AMP-binding enzyme 8 S-adenosylmethionine synthetase 10

PAGE 69

57 Table 3-4. Number of cDNAs putatively differentially regulated by ethylene. These numbers summarize the expression results from the microarray experiments with probes derived from ethylene-treated vs. air-treated Petunia flowers. Treatment Ethylene up-regulated Ethylene downregulated 2 hours air vs 2 hours ethylene 159 6 8.5 hours air vs 8.5 hours ethylene 63 0 16 hours air vs 16 hours ethylene 30 14

PAGE 70

58 Table 3-5. Expression patterns in petunia corollas of differentially regulated cDNAs. These cDNAs were identified from the microarray screen for ethylene-regulated genes and had cy3 cy5 ratios > 2. RNA gel blots were made frm 20 g total RNA from corollas excised from ethylene treated flowers.

PAGE 71

59 Table 3-5. Continued.

PAGE 72

60 Table 3-5. Continued.

PAGE 73

CHAPTER 4 ETHYLENE-REGULATED FLORAL VOLATILE SYNTHESIS IN PETUNIA COROLLAS Introduction Flowers facilitate sexual reproduction in plants, and the diversity of strategies plants have evolved for pollination and seed set is evident in the plant kingdom. Flowers have complex morphological characteristics, providing visual and olfactory cues, as well as nectar rewards for insect and animal pollinators. Once pollination and fertilization are successfully achieved, it is unnecessary for the plant to maintain floral structures not involved with subsequent fruit and seed development. As a response to pollination and in some cases fertilization, plants have evolved various senescence and abscission programs to terminate floral structures that are no longer needed. Following pollination and fertilization, many changes take place in the transition to fruit and seed development. These changes include petal wilting and abscission, color changes, flower closure, and swelling of the ovary as fruit development is initiated (reviewed in ONeill, 1997). In many plant species, the phytohormone ethylene coordinates several of these processes (van Doorn, 1997) and is synthesized spatially and temporally after pollination promoting senescence of the petals. In Petunia x hybrida, within two to four hours after pollination, a short burst of ethylene is produced from the stigma and style (Hoekstra and Weges, 1986; Tang and Woodson, 1996; Jones et al., 2003), when pollen tubes have just started to germinate and grow into the stigma (Tang and Woodson, 1996). This is followed by sustained, autocatalytic ethylene production 61

PAGE 74

62 from the stigma+style and ovary, beginning approximately 12 hours and peaking at 24 hours after pollination. Ethylene production from the corolla is induced during between 24 and 36 hours after pollination (Jones et al., 2003). The latter phase of ethylene synthesis, which corresponds with the timing of fertilization, is thought to be responsible for corolla senescence. Evidence for this was shown when stylar tissue was treated with a competitive inhibitor of ethylene binding and the rate of corolla senescence was not significantly different compared with untreated, pollinated flowers (Hoekstra and Weges, 1986). Ethylene is also produced at approximately six days post anthesis in non-pollinated petunia flowers, with natural corolla senescence occurring after this developmental ethylene production begins (Whitehead et al., 1984). The role of ethylene in floral senescence in petunia was clearly demonstrated in plants engineered for heterologous expression of the Arabidopsis dominant mutant ethylene receptor, etr1-1 (Wilkinson et al., 1997). These ethylene-insensitive plants exhibit significantly delayed petal senescence after pollination or treatment with exogenous ethylene, as well as delayed developmental senescence (Wilkinson et al., 1997; Gubrium et al., 2000). Floral fragrance is composed of low molecular weight volatile organic compounds (voc) that together with other floral cues, are thought to stimulate pollinator activity. Floral VOCs are derived from multiple biosynthetic pathways in plant cells and include benzenoids, fatty acid derivatives, isoprenoids, and others (Knudsen et al., 1993b). The VOCs attract pollinators as the fragrance can indicate the presence of a food source or a site for nesting, with exceptions of biological mimicry where no pollinator reward is given (Altenburger and Matile, 1988). Many of the VOCs found in plants have been shown to be detectable by and stimulate antenna sensilla in the hawkmoth Manduca sexta

PAGE 75

63 (Shields and Hildebrand, 2001; Fraser et al., 2003). Additionally, many pollinators have the ability to discriminate differences in floral fragrance intensity and quality (e.g. Wright et al., 2002; Daly et al., 2001), and this ability has been shown in the hawkmoth, Manduca sexta, to be reinforced by sucrose availibility (Daly et al., 2001). In some plant species the availability of pollinator rewards and quality and intensity of the fragrance has been reported to reflect the pollination status (Burquez and Corbet, 1991; Schiestl et al., 2001). Intensity of floral fragrance has been shown to correlate with floral development, as fragrance is lower in young flowers, increases with flower age, and then gradually declines as non-pollinated flowers senesce or abscise from age (Dudareva et al., 2000). This pattern of emission through development correlates with pollen availability and in some cases, potential for the greatest number of seed set (Jones et al., 1998). Diverse floral volatile profiles have been characterized in many plant species (reviewed in Knudsen et al., 1993), but only a few of the enzymes responsible for catalyzing the synthesis of these volatiles have been characterized at the molecular level. Two ubiquitous floral volatile components are the benzenoid-type methyl esters, methyl salicylate (MeSA) and methyl benzoate (MeBA), and were reported in at least 47 and 34 genera, respectively (Knudsen et al., 1993). Synthesis of MeSA and MeBA is catalyzed by a family of carboxyl methyltransferases, and genes encoding these enzymes have been cloned from multiple plant species including Clarkia breweri, Antirrhinum majus, Stephonitis floribunda, and Petunia hybrida (Ross et al., 1999; Murfitt et al., 2000; Negre et al., 2002; Pott et al., 2002; Negre et al., in press). The general reaction these carboxyl methyltransferases catalyze is the methylation of the carboxylic acid moeity of salicylic acid (SA) or benzoic acid (BA) producing MeSA or MeBA using S-adenosine-L

PAGE 76

64 methionine (SAM) as a methyl donor. These and related carboxyl methyltransferases catalyze analogous reactions, but differ in substrate preference. For example, the Clarkia salicylic acid carboxyl methyltransferase (SAMT) and Petunia benzoic acid:salicylic acid carboxyl methyltransferase (BSMT) both have a lower K m for salicylic acid, but also methylate benzoic acid (Ross et al., 1998; Negre et al., in press). While the benzoic acid carboxyl methyltransferase (BAMT) from snapdragon has a lower K m for benzoic acid, BAMT also methylates salicylic acid (Murfitt et al., 2000). MeBA and MeSA have been proposed to have major roles in pollinator-attraction because of their abundance and regulation (Knudsen et al., 1993; Dudareva et al., 2000). Methyltransferase gene expression and corresponding volatile emission is high in flower petals of Clarkia breweri (Ross et al., 1999), Antirrhinum majus Maryland True Pink (Dudareva et al., 2000) and Petunia hybrida Mitchell Diploid (MD) (Verdonk et al., 2003; Negre et al., 2003) with RNA expression and volatile emission being much lower in other plant parts. In Maryland True Pink snapdragon, emission of floral volatiles is developmentally and temporally regulated. Peak MeBA emission occurs when flowers are most receptive to pollination (Dudareva et al., 2000) and MeBA emission is maximal in afternoon when bee pollinator activities are high (Kolosova et al., 2001). In contrast, the MD cultivar of petunia emits maximal MeBA at night (Kolosova et al., 2001), in association with attracting nocturnal moth pollinators (Knudsen and Tollsten, 1993). Temporal regulation of MeBA is most likely a result of substrate availability in both snapdragon and petunia. In snapdragon and petunia, BA substrate levels are rhythmic and correspond closely with MeBA emission patterns, while mRNA and protein activity exhibit less rhythmicity (Kolosova et al., 2001).

PAGE 77

65 The cloning of genes involved with volatile synthesis has made the concept of altering floral fragrance through genetic engineering feasible. This idea was first attempted by transfomation of the Clarkia breweri (S)-linalool synthase gene (CbLIS) into petunia, a plant that does not normally emit large amounts of linalool (Lucker et al., 2001). However, little emission of linalool was observed in transgenic lines because it was mostly conjugated as a non-volatile glucopyranoside (Lucker et al., 2001). Floral emission of linalool and linalool derivatives were detected by GC-MS when CbLIS was transformed into carnation (Lavy et al., 2002). Although these compounds comprised almost 10% of the volatiles emitted, there was no detectable change in scent as perceived by humans (Lavy et al., 2002). These studies are economically important because floral fragrance has not been a focus of many flower-breeding programs and as a result, fragrance is weak in many modern cultivars of commercially important floriculture plants. However, to engineer flowers for enhanced and novel fragrances, specific changes in the volatile profile will need to be detectable by humans. Additionally, how these specific changes in floral fragrance affect pollinator activity will be of interest, given the role in pollinator attraction. In a screen for ethylene-regulated genes in petunia flowers, we identified two cDNAs, PhBSMT1 and PhBSMT2, with similarity to salicylic acid carboxyl methyltransferases. Work presented here demonstrates the activity of the petunia BSMT in vivo is by RNA interference. Using wild type and transgenic ethylene-insensitive petunias, the pattern of PhBSMT1 and PhBSMT2 gene regulation in floral organs after pollination and ethylene treatment, measurements of substrate levels, and corresponding MeBA emission are shown. Emission of other major volatile components after ethylene

PAGE 78

66 treatments and pollination are also presented. This study demonstrates a role for ethylene sensitivity in regulation of floral volatile synthesis. The physiological implications of floral volatile regulation and in vivo BSMT function are discussed. Results RNAi PhBSMT Reduces Methyl Benzoate Emission and Changes Floral Fragrance in Petunia RNAi mediated post-transcriptional gene silencing of BSMT was employed to determine if the petunia BSMT1 and BSMT2 genes encode enzymes responsible for MeBA emission in vivo. Three primary transgenic lines out of 40 were selected having significantly reduced MeBA levels and BSMT expression. Two lines, BSMT-9 and BSMT-14, showed inheritance of the transgene and phenotype in the T 1 generation (Fig. 4-1) based on presence of NPTII, BSMT mRNA accumulation, and MeBA emission, while no phenotype was observed in the BSMT-33 T 1 transgenic progeny. In total, eight T 1 progeny were observed to have the phenotype and presence of transgene (seven from BSMT-9 and one from BSMT-14). Since many of the transgenic plants were not exhibiting a phenotype, it was inferred that the transgene was silenced in these plants. All lines with reduced BSMT mRNA levels exhibited lower MeBA emission than wild type (Fig. 4-1). NPTII positive plants that had apparently silenced, having wild type levels of MeBA, also had wild type levels of BSMT mRNA (Fig 4-1). These results together demonstrate that BSMT is responsible for MeBA synthesis in petunia. Phenotype-positive lines did not exhibit changes for other major volatiles (Fig. 4-2). A triangle test was used to determine if the fragrance of the plants with reduced MeBA (BSMT-9) was detectably different by human olfaction from the floral fragrance of wild type MD petunia. This test allowed us to assay the influence of one component,

PAGE 79

67 MeBA, of the volatile profile on the perception of the overall petunia fragrance, since other components of the floral fragrance remained the same (Fig. 4-2) and no new volatiles were detected. The human sensory panel was able to discriminate the differences in floral fragrance of the MeBA knockouts from MD wild type fragrance. In this panel, 48 of 60 participants chose the sample flower that was different from the other two given in the triangle test. The panelists were able to detect a significant difference between wild type fragrance and BSMT-9 fragrance at a probability of <0.1%. The BSMT9-9 T 1 progeny from this line was also detectably different at a probability of 0.1%, as 33 out of 60 panelists could detect a difference. Overall, the participants commented that the knockout flowers smelled less than wild type and many commented negatively on the floral fragrance of the BSMT knockouts. PhBSMT1 and PhBSMT2 Are Spatially and Temporally Regulated in Petunia Flowers Petunia flowers spatially and temporally produce ethylene in response to pollination (Jones et al., 2003) presumably in order to coordinate post-pollination changes in the individual floral organs (ONeill, 1997). Volatile production is also spatially regulated in the floral organs with the primary site of volatile emission being the corolla (Verdonk et al., 2003), specifically the petal limb (Fig. 4-3). Since these genes were ethylene down-regulated in whole flowers, expression levels in individual floral organs were measured to examine the spatial and temporal pattern of mRNA regulation in wild type MD and ethylene-insensitive 44568 after pollination and exogenous ethylene treatments. Expression of both BSMT1 and BSMT2 was highest in the petal limbs compared to the other floral organs (expression highest in petal limb>petal tube>ovary>stigma+style). BSMT1 comprised between 0.3 and 0.6% of total RNA and

PAGE 80

68 levels of BSMT2 mRNA between 0.6 and 0.8% of total RNA in petal limbs (Fig. 4-4). Beginning two hours after ethylene treatment and through the subsequent timecourse, the greatest decrease in both BSMT mRNAs was observed in the ovaries, petal limbs, and petal tubes in MD (Fig. 4-4). There was a decrease in mRNA levels in 44568 petal tube and ovary, but the magnitude of decrease was ultimately not as great as observed in MD. This decrease could possibly be due to some other type of regulation or effect, such as developmental regulation, flower excision, or a small amount of residual ethylene sensitivity in these tissues. These results indicate that ethylene down-regulates expression of BSMT1 and BSMT2 in petunia flowers. Down-regulation in response to ethylene is drastic and maintained in the corolla limb, the primary site of MeBA emission. These results suggest that pollination-induced ethylene production would down-regulate PhBSMT1 and PhBSMT2. Because ethylene production is spatially and temporally regulated after pollination (Jones et al., 2003), spatial expression of PhBSMT1 (Fig. 4-5) and PhBSMT2 (Fig. 4-6) in individual floral organs following pollination was examined. Down-regulation of both PhBSMTs was observed in MD stigma+styles beginning approximately 2 hours post pollination. At approximately 10 hours post pollination, expression of both PhBSMT1 and PhBSMT2 was reduced in the ovary, compared with corresponding expression in non-pollinated flowers. Down-regulation subsequently occurs at 24 hours post-pollination in the petal tubes and limbs, with some down-regulation occurring as early as 10 hours in the petal limb (PhBSMT1) and tube (PhBSMT2). Expression of both PhBSMTs in the ovary, petal tube, and petal limb of the ethylene-insensitive 44568 line was ultimately not different between pollinated and non

PAGE 81

69 pollinated control flowers. This pattern of down-regulation indicates that mobile ethylene is eliciting a response in surrounding organs proximal to the stigma+style. From these data it was concluded that pollination-induced ethylene elicits a decrease in both PhBSMT1 and PhBSMT2 mRNA levels temporally and spatially in the flowers. Substrate Regulation in Response to Pollination and Ethylene Treatments While regulation of MeBA emission by ethylene and pollination via mRNA levels may be a key point of regulation in petunia, it is likely that other factors, such as substrate availability may also regulate this process. In order to address the possibility of substrate level regulation of MeBA emission, BA, SA, and a likely substrate precursor, CA, levels were measured in MD and 44568 petals after ethylene treatment (Table 4-1) and pollination (Table 4-2). BA and CA were present in higher amounts in ethylene-treated corollas relative to air treated corollas in MD (12538 4619 ng gfw -1 and 4829 2042 ng gfw -1 respectively), but remained at similar levels in ethylene insensitive 44568 corollas (ethylene-treated 167 17 ng gfw -1 and air treated 244 1.6 ng gfw -1 ). At 36 hours after pollination, BA and CA were decreased in MD corollas (5895 2104 ng gfw -1 and 56 16.8 ng gfw -1 respectively) compared with non-pollinated samples (24214 2479 ng gfw -1 and 417 27.9 ng gfw -1 respectively) and remained unchanged in 44568. SA levels were largely unaffected by either the ethylene treatment or pollination in MD and 44568. Both BA and CA exhibited rhythmic regulation. Levels of both were higher in the night collected non-pollinated 36 hour control samples of MD and 44568 relative to the zero hours day collected sample.

PAGE 82

70 Volatile Emission Is Down-Regulated in Response to Exogenous Ethylene and Pollination MeBA emission was measured to determine if emission corresponded with the decreased benzoic acid levels (36 hours post-pollination) and decreased PhBSMT1 and PhBSMT2 mRNA levels (in the petals after 10 hours of ethylene treatment and 24 hours post-pollination). A reduction in MeBA emission was observed after 10 hours of ethylene treatment (Fig 4-7). Air-treated flowers emitted approximately 12 times as much MeBA as ethylene treated flowers. This reduction in MeBA emission was not observed with 44568 flowers after ethylene treatment, thus demonstrating a role for ethylene in regulation of this process. After pollination, MeBA emission decreased significantly in MD flowers beginning 24 hours after pollination (Fig. 4-7). Emission at this time was low in both genotypes, as this timepoint was collected during the day when emission was minimal. However, emission was decreased by 50% at 24 hours after pollination in MD compared to non-pollinated flowers. The timing of this reduction corresponds to when post-pollination ethylene production from the corolla begins and when corresponding PhBSMT mRNA levels have decreased compared with non-pollinated controls. Additionally, the timing also corresponds with the time of fertilization observed for petunia (Tang and Woodson, 1996). Based on these results, it was concluded that pollination-induced ethylene production regulates emission of MeBA in petunia. The data from the pollination timecourse also indicated that MeBA emission was rhythmic, shown in non-pollinated flowers. This is in agreement with observations by Kolosova et al. (2001) and Verdonk et al. (2003). In many cases, rhythmic biological patterns indicate the possibility of a circadian regulated process. In order to address this

PAGE 83

71 possibility, we collected floral volatiles through normal day-night cycles and then in continuous dark or light. When the plants were placed into complete darkness, emission of MeBA virtually ceased (Fig. 4-8) and PhBSMT gene expression did not continue to oscillate, as observed for the first two normal day-night cycles (Fig, 4-8b). When the plants were placed into continuous light, MeBA emission continued with no obvious oscillations from the first day through the remainder of the timecourse (Figure 4-9). However, gene expression continued rhythmicity through the first 24-hour period in constant light. Robust cycling was not observed the first day the plants were placed into the dark for either gene expression or MeBA emission. These data show an additional linkage between PhBSMT gene expression and MeBA emission; both are rhythmic with increased gene expression occurring approximately six hours prior to rises in emission. The production of other major floral volatile components in response to ethylene and pollination was also characterized. All the abundant floral volatiles were reduced by ethylene treatment (Fig. 4-10) and pollination (Fig. 4-11) in MD, but not in ethylene-insensitive 44568. Floral volatiles that decreased in response to both of these treatments included benzaldehyde, phenylacetaldehyde, benzyl alcohol, 2-phenylethanol, iso-eugenol, and benzyl benzoate. Iso-eugenol and benzyl alcohol were the least affected of all of these volatiles. These results show there is a coordinated down-regulation of floral volatile emission and it is dependent upon ethylene signaling. Discussion This study demonstrated in vivo that BSMT synthesizes methyl benzoate in petunia. The spatial and temporal regulation of PhBSMT gene expression, substrate levels, and MeBA emission after pollination and ethylene treatments were characterized. These results indicate that in Petunia, ethylene regulates floral fragrance at multiple

PAGE 84

72 levels that lead to an overall reduction in total floral fragrance mediated by the perception of ethylene. A primary role for ethylene in regulation of visible post-pollination changes has been established (Wilkinson et al., 1997). The impact of ethylene on these visual and olfactory cues on pollinator attractiveness is inferred to be negative, since insects are attracted by both visual cues and floral fragrance. Transgenic PhBSMT RNAi Plants Have Lower Methyl Benzoate Emission RNA interference was used to engineer petunia for decreased PhBSMT mRNA levels to test in vivo if BSMT synthesizes MeBA and to determine if this trait could be genetically engineered to be detectable by human olfaction. In transgenic lines, reduced PhBSMT mRNA levels resulted in greatly reduced levels of MeBA emission with the level of reduction of MeBA emission corresponding with the level of reduction of both PhBSMT1 and PhBSMT2 expression. These results link gene with function, indicating that in vivo both PhBSMT1 and PhBSMT2 catalyze the formation of MeBA in petunia flowers. More primary transgenic lines have been made using a different RNAi construct since pHannibal had low efficiency for knocking out this gene in petunia (three out of 15 pHannibal lines measured had reduced PhBSMT mRNA levels) and silencing was observed in the T 1 Many primary transgenic lines with the new construct were obtained and are expressing a reduced MeBA phenotype (R Dexter and D Clark, unpublished). The MeBA knockouts demonstrate that specific components of the floral volatile profile can be changed through genetic engineering. These plants will be useful in studying plant-insect interaction systems, as it is not known if specific components of floral fragrance are responsible for pollinator attraction. The fragrance of the MeBA knockout flowers was shown to be detectably different from wild type fragrance to humans. These are novel results demonstrating that flowers can be genetically modified

PAGE 85

73 for changed floral fragrance. The engineering of plants for enhanced scent characteristics has potential commercial value and possibly agronomic value, since fragrance is often lost in breeding for other traits. Novel and improved fragrances could feasibly be introduced into existing commercial cultivars through transformations with these and other scent genes. Human olfactory panels will help to answer questions of what fragrances, volatile combinations, and levels should be engineered to obtain flowers with pleasant fragrances. Ethylene Regulates PhBSMT Expression in Petunia Floral Organs PhBSMT1 and PhBSMT2 were both down-regulated quickly in response to exogenous ethylene and pollination in all MD floral organs, but not in ethylene insensitive 44568 flowers. The temporal and spatial down-regulation of both mRNAs in the floral organs follows the sequential pattern of post-pollination ethylene production observed in petunia (Tang and Woodson, 1996; Jones et al., 2003). Studies in petunia, Phalaenopsis spp., and carnation have shown that ethylene synthesis is temporally and spatially regulated in the flower and coordinates developmental changes such as petal senescence and ovule development (ONeill et al., 1993; Tang and Woodson, 1996; Bui and ONeill, 1998; Jones et al., 2003). Models for post-pollination regulation of ethylene production show that pollination induces production of ethylene first from the stigma+style, then from the ovary, followed by the petals. Here it is shown that both PhBSMT1 and PhBSMT2 are down-regulated in each floral organ, consistent with the pattern of spatial post-pollination ethylene production in petunia flowers observed by Tang and Woodson (1996) and Jones et al (2003). The data also suggest that the large amounts of ethylene produced from the stigma+style can regulate ethylene-sensitive processes in the corolla due to the close proximity of these organs. The corolla produces

PAGE 86

74 measurable ethylene beginning 24 hours after pollination (Jones et al., 2003) while down-regulation of PhBSMT2 begins as early as 10 hours in the MD petal tube, but not 44658. This implies mobile ethylene produced in the stigma+style regulates PhBSMT2 in the petal tube after pollination. There was some observed PhBSMT down-regulation after pollination in 44568 stigma+styles. It is likely that this is due to a small amount of ethylene sensitivity in this tissue as the constitutive 35S promoter driving expression of the etr1-1 allele does not confer high, equal expression levels of transgenes in all plant cells. Position effects of the transgene in the genome can cause expression levels of the transgene to differ in specific tissues (Holtorf et al., 1995; Van Leeuwen et al., 2000). In this case, position effects could result in certain cells or tissues having a small amount of ethylene sensitivity. However, the contribution of the stigma+style to total fragrance output is small and therefore changes in emission from this organ are likely to have little effect on the total fragrance output of the flower. Measurements of substrate levels after ethylene treatments and pollination indicated that ethylene does not immediately down-regulate substrate levels like it does PhBSMT gene expression levels. Free BA and CA levels increased in response to ethylene, while decreased after pollination in MD and remaining unchanged in 44568 for ethylene treatments and pollination. The increase in BA after ethylene treatment is likely due to substrate accumulation, as it is not being synthesized into MeBA (Fig. 4-7) due to decreased BSMT activity (Negre et al., 2003) and PhBSMT gene expression (Fig. 4-4). Increased levels of BA and CA after ethylene treatment and decreased levels after pollination in MD, but not 44568, indicate that there are likely multiple factors

PAGE 87

75 controlling the levels of BA and CA. Perhaps ethylene sensitivity is required, but there are also other factors such as senescence, pollination signals independent of ethylene that are required for substrate regulation. During pollination, these substrates may be used in other processes or be remobilized to other organs, resulting in the measurable decrease observed here. Additionally, ethylene treatments were conducted with excised tissue, which may cause differences if supply of these substrates involves a transport mechanism. In regard to SA, the levels of SA were relatively low compared with BA and CA. Since SA is a plant hormone and it is not actively being synthesized into MeSA during these treatments, low levels would be expected. Also, since SA is involved in eliciting cell death and plant defense responses, it is possible that maintenance of SA levels in petals during pollination may assist in promoting cell death or in defensive mechanisms against infection as the flower senesces around the developing fruit. These results together with those of Negre et al. (2003) thoroughly examine the synthesis and ethylene+pollination regulation of MeBA in petunia flowers. Here it is shown that MeBA emission is markedly reduced in response to ethylene (Fig. 4-7) and pollination (Fig. 4-7), in agreement with pollination data shown in Negre et al (2003). Down-regulation after pollination is controlled by ethylene through decreased mRNA levels in the corolla, possibly substrate levels in the corolla, and as observed by Negre et al. (2003) post-translationally. While post-translational regulation does appear to have a role in regulating MeBA (Negre et al., 2003), decreased mRNA levels are likely to have a major role since this would presumably limit enzyme abundance and disturb the substrate to enzyme ratio, and thus result in decreased product formation. Additionally, reducing the levels of mRNA by RNAi shows that mRNA abundance is a factor in how much

PAGE 88

76 MeBA is synthesized. As a whole, these results clearly demonstrate a role for ethylene in temporal and spatial regulation of MeBA at multiple levels. The nature of rhythmic MeBA emission was investigated in order to examine additional linkages between mRNA and MeBA emission and possible circadian rhythms. Rhythmic emission of MeBA has been demonstrated in multiple species including Petunia (Kolosova et al., 2001; Verdonk et al., 2003), Antirrinhum (Kolosova et al., 2001), and Stephanotis (Pott et al., 2002) and has been shown to be circadian in Antirrhinum. To demonstrate true circadian rhythmicity of a process, a robust rhythm must continue under constant environmental conditions (Jones and Mansfield, 1975). A robust rhythm of MeBA emission or mRNA expression was not observed when plants were placed into the dark. However, the maintenance of PhBSMT gene expression rhythmicity makes the interpretation less clear. It is possible that there was a shift in peak emission and gene expression when plants were placed into dark. More frequent sampling during the first 24-hour period in constant darkness and light might help with interpretation of these results. From these data it can be concluded that gene expression is rhythmic increasing approximately six hours ahead of high MeBA emission showing a linkage between PhBSMT mRNA expression and emission. Pollination and Ethylene Treatments Down-Regulate Floral Volatiles in Petunia Analysis of other major volatile components of petunia fragrance demonstrated that many of the floral volatiles are similarly down-regulated after pollination and in response to exogenous ethylene. Many of the volatile components that were significantly reduced have been shown to elicit excitation responses in antenna sensillas of the moth Manduca sexta (Shields and Hildebrand, 2001; Fraser et al., 2003). The volatile iso-eugenol was shown by Shields and Hildebrand (2001) to not elicit an excitatory response and this

PAGE 89

77 volatile component was not as reduced as were the other volatile components after pollination. Reduced emission after ethylene treatments in MD and continuous emission from 44568 flowers demonstrate that ethylene is likely to have key control of regulating most of the floral volatiles. All of the volatiles exhibited temporal down-regulation after pollination, analogous to the reduction in MeBA emission. This pattern of regulation corresponds to the second major phase of pollination induced ethylene production, when ethylene is simultaneously being produced from multiple floral organs. Ethylene treatments had a more pronounced effect on reducing floral volatile emission in MD compared with the pollinated flowers. The ethylene-treated flowers were treated in an enclosed system, which is conducive to eliciting a stronger ethylene response. Since air treated samples were not significantly altered compared with untreated controls, the effect is due to ethylene and not the enclosed treatment. Additionally, it is possible that constant exposure to ethylene and absence of pollination cues may contribute to a greater response. These results are of interest because they demonstrate that ethylene sensitivity controls total floral fragrance output in addition to controlling petal senescence. These two processes together may influence both visual and olfactory attractiveness of the flower to pollinators. The timing of floral volatile down-regulation has interesting ecological implications. In terms of pollinators that are primarily attracted by floral fragrance, there are multiple benefits to regulating floral volatiles after pollination. Some pollinators can distinguish differences in floral fragrance (Wright et al., 2002; Daly et al., 2001) and many of these volatile components elicit excitatory responses in a potential moth pollinator (Shields and Hildebrand, 2001). Some pollinator-behavior studies demonstrate

PAGE 90

78 higher visitation rates to flowers with greater scent levels (e.g. Wright et al., 2002). Results shown here indicate that petunia flowers may regulate fragrance intensity after pollination so that pollinated flowers are detectably different to insects. The timing of this process is important since reduction of fragrance emission too early after a pollination event may be wasteful if foreign pollen or a small amount of pollen is deposited. By reducing fragrance after a successful pollination event, more efficient distribution of pollen with less pollen spent on flowers that have already been pollinated could be achieved. It is possible that reducing floral fragrance is a type of defense mechanism, as these flowers may become less apparent to visiting insects. Fewer visits could result in less risk of perturbing the developing fruit, through introduction of parasites or pathogens by non-sterile pollinators. Additionally, if proportions of the volatiles emitted from the plant changes with pollinations, plant apparency (Feeny, 1976), or conspicuousness, may diminish and thus reduce pollinator visits if a number of pollinations were made. If the pollinators of petunia also use the plant as a site for laying eggs as in Nicotiana (Baldwin and Ohnmeiss, 1993), reduced apparency may decrease the probability of egg laying on the plant. These results show evidence of a role for ethylene in regulation of the major floral volatiles in petunia. In the case of MeBA, ethylene regulates mRNA expression levels in each of the floral organs in a spatial and temporal manner after pollination and this regulation corresponds with MeBA emission. Plants engineered for knockout of MeBA demonstrate two main points. First, BSMT synthesizes MeBA in petunia. Second, modification of floral scent through genetic engineering is possible in petunia and these changes can be used to alter the fragrance enough so that people perceive a difference in

PAGE 91

79 the floral fragrance. Scent panels will be key to engineering flowers for improved floral fragrance, as the MD wild type flowers were preferred over flowers with reduced MeBA. These plants should be useful tools for studying the effects of individual volatiles, changes in overall volatile blends, and the total effects of ethylene on floral attractiveness in pollinator behavior studies. Materials and Methods Plant Material In all experiments, Petunia x hybrida Mitchell Diploid (MD) was used as the wild type line and is also the genetic background of ethylene-insensitive 35S::etr1-1 line 44568 (Wilkinson et al., 1997). Plants were grown in air-conditioned glass greenhouses at 25C day/18C night. Plants were potted in Fafard 2B potting medium (Fafard Inc., Apopka, FL) in 1.2 L pots and fertilized at every irrigation with 150 mgL -1 Scotts Excel 15-5-15 (Scotts Co., Marysville, OH). cDNA Isolation Three cDNA libraries were constructed from petunia MD whole flowers collected at multiple developmental stages (from early bud to anthesis), ethylene-treated flowers (2.5 LL -1 ethylene treatments for 30 minutes, 1 hour, 3 hours, 6 hours, and 12 hours), and pollinated flowers (1, 2, 5, 10, 24, and 34 hours after pollination). Total RNA was extracted by a phenol:chloroform extraction method with lithium chloride precipitations as described in Ciardi et al. (2000). Messenger RNA was isolated using Oligotex mRNA purification (Qiagen Inc; Valencia, CA). cDNA libraries were constructed using a -ZAPII cDNA synthesis kit from Stratagene (LaJolla, CA). Approximately 6000 clones from these libraries were randomly sequenced. A minimally-redundant subset of these clones was used for microarray experiments to find ethylene regulated genes. From these

PAGE 92

80 microarray experiments, the salicylic acid carboxyl methyltransferase homologs were isolated as being candidate cDNAs that were down-regulated by ethylene (to be presented in another manuscript). Two full length cDNAs encoding for SAMT homologs were isolated from the cDNA libraries and were used in subsequent experiments. Northern analysis was used to verify down-regulation by ethylene (Negre et al., in press). Tissue Treatments and Collections All ethylene treatments and pollinations were initiated the day after anthesis at 10 a.m. under sunny weather conditions to help eliminate developmental and environmental variability. For ethylene and control-air treatments, flowers were excised and placed into 1.5 mL microfuge tubes containing 1.0 mL dH 2 0. Flowers were sealed in 37.85 L glass chambers and treated with 2-3 LL -1 ethylene. For air treatments, flowers were placed in the same conditions, but no ethylene was added and potassium permanganate (Fisher Scientific, Hampton, NH) was placed in the chambers. Concentrations of exogenous ethylene were verified at the beginning and end of indicated treatment times using a Gas Chromatograph (Hewlett Packard Model 5890, Series II; Palo Alto, CA) equipped with a flame ionization detector and an alumina column. In the pollinated flower collections, flowers were pollinated and remained on the plant until designated collection time. For every experiment all treated (ethylene or pollination) and control (air or non-pollinated) flowers were collected at the following times (with treatment times in parenthesis): 10:00 am (0 hours), 12:00 noon (2 hours), 8:00 pm (10 hours), 10:00 am (24 hours), 10:00 pm (36 hours), 10:00 am (48 hours) after ethylene treatment or pollination. Tissue was collected from plants placed into dark chambers at 25 + 3C for the constant dark circadian studies.

PAGE 93

81 Spatial and Temporal Analysis of mRNA Expression in Flowers Spatial and temporal mRNA accumulation was analyzed after ethylene treatments and pollination in petunia MD and ethylene-insensitive 44568. Expression was examined from individual floral organs including petal limbs, petal tubes, stigma+styles, and ovaries that were excised from ethylene-treated, air-treated, pollinated, and non-pollinated flowers. The day after anthesis, flowers were either collected for ethylene treatments or pollinated on the plant for the time courses described above. Harvested tissue was immediately frozen in liquid nitrogen and stored at C. Total RNA was extracted using an RNeasy Mini Plant RNA extraction kits with on-column DNase digestion performed during the extraction (Qiagen Inc., Valencia, CA). RNA was quantified by spectrophotometry and RNA quality was verified by gel electrophoresis. Real-Time RT-PCR was performed for quantification of PhBSMT mRNA transcripts from 100ng of total RNA using TaqMan One-Step RT-PCR reagents (Applied Biosystems; Foster City, CA). Reactions were conducted in 25 L volumes in 96 well optical reaction plates on an Gene Amp 5700 Sequence Detection System (Applied Biosystems; Foster City, CA). Primers and TaqMan probes were designed using Primer Express Software (Applied Biosystems; Foster City, CA). Specificity of each of the primer and probe sets was verified by performing PCR reactions with in vitro transcribed PhBSMT1 template with the primer and probe set specific to PhBSMT2 and vice-versa. In vitro transcribed RNA was synthesized using a MAXIscript In vitro Transcription Kit (Ambion, Austin, TX) according to manufacturers instructions. PhBSMT1 and PhBSMT2 were used as templates for in vitro transcription and the transcripts were collected on separate gels to prevent contamination. Primer and probe sequences used

PAGE 94

82 for individual detection of each gene corresponded to the 3 untranslated region of the cDNA and are as follows: PhBSMT1 forward primer, AAATGTCATCATCTCCTTGACCAA; PhBSMT1 reverse primer, CGGATCACTACTAAAATATTCGGGTTT; PhBSMT TaqMan probe, 6FAM-AAGGCACTCAATGTCTATTTTCGGTCGA-BHQ1; PhBSMT2 forward primer, TGTACCAATTCTCTATTGTTGTTTTGC; PhBSMT2 reverse primer, CTGAAAGGACCCCTAGTGTACAAGA; PhBSMT2 TaqMan probe, 6FAM-CTTCATAGGTGGTCGAGGTGCTAATTTATCTAGTC-BHQ1. TaqMan Real-Time PCR reactions were run under the following conditions: 48C for 30 minutes, 95C for 10 minutes, followed by 40 cycles of 95C for 15 seconds and 60C for 1 minute. Reactions were repeated twice with one set of RNAs and once with RNA collected from separate, duplicate tissue. PCR reactions of in vitro transcibed PhBSMT1 or PhBSMT2 standards were run in duplicate and in tandem with the sample RNAs to generate a standard curve from which the level of each PhBSMT mRNA in the samples was quantified. Generation of Transgenic PhBSMT RNAi Petunias RNAi constructs were made using the pHannibal RNAi cloning vector system (Wesley et al., 2001). The region cloned into the pHannibal vector includes the coding region from base 661 to base 1002 from PhBSMT1 in sense and antisense orientations flanking the intron segment. The RNAi chimeric gene was subsequently cloned into an Agrobacterium transformation vector containing the neomycin phosphotransferase II (NPTII) gene. This transformation vector was introduced into Agrobacterium tumefaciens and used for transforming leaf explants from five week-old MD seedlings grown in tissue culture according to the methods of Jorgenson et al (1996). Primary

PAGE 95

83 transformants were grown under greenhouse conditions described above, and transgenic plants were selected by PCR for presence of the neomycin phosphotransferase II gene (NPTII), reduced PhBSMT mRNA levels, and reduced MeBA emission. Three independent, transgenic lines were obtained: BSMT-9, BSMT-14, and BSMT-33. Flowers were self pollinated to produce T 1 progeny from these lines, which were analyzed for presence of transgene by PCR analysis. Sixteen T 1 progeny plants from each line were analyzed for phenotypes by measurement of MeBA emission and measurement of PhBSMT mRNA levels. Volatile Collection and Analysis Floral volatiles were collected from excised ethylene-treated, pollinated, and PhBSMT RNAi flowers. Volatiles from exogenous ethylene and pollination timecourses were treated or pollinated, then collected at the times indicated previously. An additional untreated control was included for the ethylene experiments to control for flower excision induced variability in air and ethylene treatments. For this control, flowers at the same developmental stages were collected fresh from the plants to compare with air treated control flowers. Three flowers were collected per treatment and each timepoint was repeated 3 times. Flowers from the PhBSMT RNAi screen were collected three times with 3-5 flowers per collection at 8:00 pm for the initial screen and once more with putative positive lines at midnight to verify that reduced MeBA emission was reduced when MeBA emission is maximal in MD wild type (Kolosova et al., 2001). Volatiles were collected for one hour according to collection protocol described by Schmelz et al (2001). Identification of each of the floral volatiles was verified by GC-MS (Schmelz, et al., 2001).

PAGE 96

84 Benzoic Acid & Salicylic Acid Extraction and Quantification Benzoic acid and salicylic acid were extracted and quantified by GC-MS (Schmelz et al., 2003). Petal tissue was excised from MD and 44568 whole flowers after 10 hrs treatment with 2-3 LL ethylene or 36 hrs after pollination and stored at C until extraction. Two replicate sets of tissues were used for quantification. Human Olfaction Panels Human sensory panels were used to determine if differences in fragrance of the reduced MeBA flowers and MD wild type flowers could be discriminated by human olfaction. A triangle test was performed with sixty human subjects, each randomly given a set of three unmarked flowers for sampling of the floral fragrances. The flower samples were prepared from freshly excised flowers at anthesis from MD and knockout line BSMT-9. Excised flowers were placed immediately into 5 mL water agarose blocks, then placed into 210 mL glass jars and and sealed with lids for approximately 120 minutes before testing. Each set of flowers consisted of two flowers of the same genotype and one of the other genotype. The test performed both with two controls and one knockout or two knockouts and one control and panelists were asked to judge which flower had a different fragrance. Additional descriptive comments were also solicited from the test subjects to determine if there were preferences in floral fragrance. The statistical significance of the correct number of judgements was determined as described (Lawless and Heyman, 1998).

PAGE 97

85 0204060801001209--99--209--69--229--49--319--2814-1514-1433-16MD control% of control MeBA emissio n 00.20.40.60.811.21.4% BSMT mRNA MeBA BSMT-1 BSMT-2 0204060801001209--99--209--69--229--49--319--2814-1514-1433-16MD control% of control MeBA emissio n 00.20.40.60.811.21.4% BSMT mRNA MeBA BSMT-1 BSMT-2 Figure 4-1. PhBSMT RNAi reduces MeBA emission and PhBSMT mRNA. Mean ( + SE; n=3) MeBA emission (blue solid bars, left axis) and mRNA levels (right axis) of PhBSMT1 (striped, light purple bars) and PhBSMT2 (dotted, dark purple bars) in T 1 lines of BSMT-9, BSMT-14, and BSMT-33. Silencing was observed in BSMT-33 (represented by 33-16). Line 14-14 was transgenic, but also silenced.

PAGE 98

86 Figure 4-2. PhBSMT RNAi reduces MeBA emission only. Mean ( + SE; n=3) emission of major volatiles in MD compared with most reduced MeBA T 1 line BSMT9-9.

PAGE 99

87 80120160 020 hl () l i b li b l SSO SR OS R A Benzldehyde B Benzyl Alcohol C Phenylacetaldehyde D Methyl Benzoate E 2-Phenylethanol F Iso-Eugenol G Benzyl Benzoate EFGDCBAEFGDCBAEFGDCBAEFGDCBAEFGDCBA WholeflowerNo limbLimb onlyNo CorollaORS LOv SS T R LOv SS T R LOv SS T R g g fw-1 hr-1 80120160 020 hl () l i b li b l SSO SR OS R A Benzldehyde B Benzyl Alcohol C Phenylacetaldehyde D Methyl Benzoate E 2-Phenylethanol F Iso-Eugenol G Benzyl Benzoate EFGDCBAEFGDCBAEFGDCBAEFGDCBAEFGDCBA WholeflowerNo limbLimb onlyNo CorollaORS LOv SS T R LOv SS T R LOv SS T R 80120160 020 hl () l i b li b l SSO SR OS R A Benzldehyde B Benzyl Alcohol C Phenylacetaldehyde D Methyl Benzoate E 2-Phenylethanol F Iso-Eugenol G Benzyl Benzoate EFGDCBAEFGDCBAEFGDCBAEFGDCBAEFGDCBA WholeflowerNo limbLimb onlyNo CorollaORS LOv SS T R LOv SS T R LOv SS T R g g fw-1 hr-1 Figure 4-3. Volatile emission patterns from MD floral organs. Volatiles were collected from whole flowers, flowers with petal limb excised, limbs only, flowers with excised corollas, and ovary+receptacle+sepals (ORS). Floral organs are explained in the picture to the right: L, petal limb; T, petal tube; S, stigma+style; O, ovary; R, receptacle.

PAGE 100

88 0.000.010.020.030.040.050.060.070210243648% BSMT mRN A Stigma+Style 0.000.020.040.060.080.100.120.140210243648% BSMT mRN A Ovary 0.000.200.400.600.801.001.201.400210243648% BSMT mRN A Petal Tube 0.000.200.400.600.801.001.201.400210243648% BSMT mRNA Petal Limb hours after ethylene hours after ethylene hours after ethylene hours after ethylene BSMT1 44568 BSMT2 44568 BSMT1 MD BSMT2 MD 0.000.010.020.030.040.050.060.070210243648% BSMT mRN A Stigma+Style 0.000.020.040.060.080.100.120.140210243648% BSMT mRN A Ovary 0.000.200.400.600.801.001.201.400210243648% BSMT mRN A Petal Tube 0.000.200.400.600.801.001.201.400210243648% BSMT mRNA Petal Limb hours after ethylene hours after ethylene hours after ethylene hours after ethylene BSMT1 44568 BSMT2 44568 BSMT1 MD BSMT2 MD Figure 4-4. PhBSMT mRNA expression after ethylene treatment. Flowers were treated with 2.5 ppm ethylene and then dissected for collection of individual floral organs. RNA measurements are represented as mean + SE. Treatment times are indicated on the x axis and % RNA on the y axis. Note differences in scale.

PAGE 101

89 0.0000.0020.0040.0060.0080.0100.0120.0140210243648% BSMT1 mRN A Stigma+Style 0.000.010.020.030.040.050.060.070210243648% BSMT1 mRNA Ovary 0.000.050.100.150.200.250.300210243648% BSMT1 mRN A Petal Tube 0.000.100.200.300.400.500.600210243648% BSMT1 mRN A Petal Limb MD N P MD Polln 44568 NP 44 568 P o lln hours after pollination hours after pollination hours after pollination hours after pollination 0.0000.0020.0040.0060.0080.0100.0120.0140210243648% BSMT1 mRN A Stigma+Style 0.000.010.020.030.040.050.060.070210243648% BSMT1 mRNA Ovary 0.000.050.100.150.200.250.300210243648% BSMT1 mRN A Petal Tube 0.000.100.200.300.400.500.600210243648% BSMT1 mRN A Petal Limb MD N P MD Polln 44568 NP 44 568 P o lln hours after pollination hours after pollination hours after pollination hours after pollination Figure 4-5. PhBSMT1 mRNA expression in MD and 44568 after pollination. Percent PhBSMT1 mRNA mean + SE is on the y axis and hours after pollination is on the x axis. Note differences in scale.

PAGE 102

90 0.000.010.020.030.040.050.060.070210243648% BSMT2 mRN A 0.000.020.040.060.080.100.120.140210243648% BSMT2 mRNA Ovary Stigma+Style 0.000.100.200.300.400.500.600210243648% BSMT2 mRN A Petal Tube 0.000.200.400.600.801.001.200210243648% BSMT2 mRN A Petal Limb MD N P M D P o lln 44568 NP 445 68 P o l l n hours after pollination hours after pollination hours after pollination hours after pollination 0.000.010.020.030.040.050.060.070210243648% BSMT2 mRN A 0.000.020.040.060.080.100.120.140210243648% BSMT2 mRNA Ovary Stigma+Style 0.000.100.200.300.400.500.600210243648% BSMT2 mRN A Petal Tube 0.000.200.400.600.801.001.200210243648% BSMT2 mRN A Petal Limb MD N P M D P o lln 44568 NP 445 68 P o l l n hours after pollination hours after pollination hours after pollination hours after pollination Figure 4-6. PhBSMT2 mRNA expression in MD and 44568 after pollination. Percent PhBSMT2 mRNA mean + SE is on the y axis and hours after pollination is on the x axis. Note differences in scale.

PAGE 103

91 020406080100 methyl benzoate MD NP MD x 44568 NP 44568 x 020406080100 g g fw-1 hr-1g g fw-1 hr-1hours after pollination 0210243648 NTAirC2H4NTAirC2H4MD44568 020406080100 methyl benzoate MD NP MD x 44568 NP 44568 x 020406080100 g g fw-1 hr-1g g fw-1 hr-1hours after pollination 0210243648 NTAirC2H4NTAirC2H4MD44568 AB 020406080100 methyl benzoate MD NP MD x 44568 NP 44568 x 020406080100 g g fw-1 hr-1g g fw-1 hr-1hours after pollination 0210243648 NTAirC2H4NTAirC2H4MD44568 020406080100 methyl benzoate MD NP MD x 44568 NP 44568 x 020406080100 g g fw-1 hr-1g g fw-1 hr-1hours after pollination 0210243648 NTAirC2H4NTAirC2H4MD44568 AB Figure 4-7. MeBA emission after ethylene treatment (A) and pollination (B). Mean ( + SE; n=3) after 10 hours of 2.5 ppm ethylene treatment of MD and 44568 whole flowers. In ethylene treatments (A) a not-treated (NT) control was measured to account for air and ethylene chamber treatment effects and flower excision for treatment. On x axis in 4B is hours after pollination.

PAGE 104

92 0.000.200.400.600.801.00% BSMT mRNA dk light light constant dark BSMT1 BSMT2 228ampm228ampm828pm8228ampm228am88pm28am Day 1Day 2Day 3Day 4Day 5D6 0.000.200.400.600.801.00% BSMT mRNA dk light light constant dark BSMT1 BSMT2 228ampm228ampm828pm8228ampm228am88pm28am Day 1Day 2Day 3Day 4Day 5D6 228ampm228ampm828pm8228ampm228am88pm28am Day 1Day 2Day 3Day 4Day 5D6 0102030405060 228ampm228ampm828pm8228ampm228am88pm28am dk constant dark light light Day 1Day 2Day 3Day 4Day 5D6 methyl benzoate 0102030405060 228ampm228ampm828pm8228ampm228am88pm28am dk constant dark light light Day 1Day 2Day 3Day 4Day 5D6 methyl benzoate A B g g fw-1 hr-1 0.000.200.400.600.801.00% BSMT mRNA dk light light constant dark BSMT1 BSMT2 228ampm228ampm828pm8228ampm228am88pm28am Day 1Day 2Day 3Day 4Day 5D6 0.000.200.400.600.801.00% BSMT mRNA dk light light constant dark BSMT1 BSMT2 228ampm228ampm828pm8228ampm228am88pm28am Day 1Day 2Day 3Day 4Day 5D6 228ampm228ampm828pm8228ampm228am88pm28am Day 1Day 2Day 3Day 4Day 5D6 0102030405060 228ampm228ampm828pm8228ampm228am88pm28am dk constant dark light light Day 1Day 2Day 3Day 4Day 5D6 methyl benzoate 0102030405060 228ampm228ampm828pm8228ampm228am88pm28am dk constant dark light light Day 1Day 2Day 3Day 4Day 5D6 methyl benzoate A B g g fw-1 hr-1 Figure 4-8. Rhythmic emission of MeBA and rhythmic expression of PhBSMTs. Flowers were collected from plants through two regular day-night cycles and then transfer to complete darkness. (A) mean ( + SE) mRNA expression of PhBSMT1 and PhBSMT2 and (B) Mean ( + SE) MeBA emission.

PAGE 105

93 0102030405060708090 1 00 methylbenzoate 2 am Day 1Day 2Day 3Day 42 pm2 am 2 pm2 am 2 pm2 am 2 pm 0.000.100.200.300.400.500.600.700.800.901.00 BSMT1 BSMT2 dk light light2 am Day 1Day 2Day 3Day 42 pm82 am 2 pm82 am 2 pm82 am 2 pm82 am 2 pm88888Day 5 % BSMT mRNAg g fw-1BA dk dark light dark light 0102030405060708090 1 00 methylbenzoate 2 am Day 1Day 2Day 3Day 42 pm2 am 2 pm2 am 2 pm2 am 2 pm 0.000.100.200.300.400.500.600.700.800.901.00 BSMT1 BSMT2 dk light light2 am Day 1Day 2Day 3Day 42 pm82 am 2 pm82 am 2 pm82 am 2 pm82 am 2 pm88888Day 5 % BSMT mRNAg g fw-1BA dk dark light dark light 0102030405060708090 1 00 methylbenzoate 2 am Day 1Day 2Day 3Day 42 pm2 am 2 pm2 am 2 pm2 am 2 pm 0.000.100.200.300.400.500.600.700.800.901.00 BSMT1 BSMT2 dk light light2 am Day 1Day 2Day 3Day 42 pm82 am 2 pm82 am 2 pm82 am 2 pm82 am 2 pm88888Day 5 % BSMT mRNAg g fw-1BA dk dark light dark light Figure 4-9. Rhythmic emission of MeBA and rhythmic expression of PhBSMTs. Flowers were collected from plants through two regular day-night cycles and then transfer to constant light. (A) mean ( + SE) mRNA expression of PhBSMT1 and PhBSMT2 and (B) Mean ( + SE) MeBA emission.

PAGE 106

94 0510152025 Benzaldehyde 00.20.40.60.811.21.41.6 Benzyl alcohol 0 2468 1 0 1 2 Benzyl Benzoate 0 2468 1 0 Iso-Eugenol 00.20.40.60.811.21.41.6 2-Phenylethanol gg 01234567 Phenylacetaldehyde NTAirC2H4NTAirC2H4MD44568 NTAirC2H4NTAirC2H4MD44568 NTAirC2H4NTAirC2H4MD44568 NTAirC2H4NTAirC2H4MD44568 NTAirC2H4NTAirC2H4MD44568 NTAirC2H4NTAirC2H4MD44568 g g fw-1 hr-1g g fw-1 hr-1g g fw-1 hr-1g g fw-1 hr-1g g fw-1 hr-1g g fw-1 hr-1 0510152025 Benzaldehyde 00.20.40.60.811.21.41.6 Benzyl alcohol 0 2468 1 0 1 2 Benzyl Benzoate 0 2468 1 0 Iso-Eugenol 00.20.40.60.811.21.41.6 2-Phenylethanol gg 01234567 Phenylacetaldehyde NTAirC2H4NTAirC2H4MD44568 NTAirC2H4NTAirC2H4MD44568 NTAirC2H4NTAirC2H4MD44568 NTAirC2H4NTAirC2H4MD44568 NTAirC2H4NTAirC2H4MD44568 NTAirC2H4NTAirC2H4MD44568 NTAirC2H4NTAirC2H4MD44568 NTAirC2H4NTAirC2H4MD44568 NTAirC2H4NTAirC2H4MD44568 NTAirC2H4NTAirC2H4MD44568 NTAirC2H4NTAirC2H4MD44568 NTAirC2H4NTAirC2H4MD44568 g g fw-1 hr-1g g fw-1 hr-1g g fw-1 hr-1g g fw-1 hr-1g g fw-1 hr-1g g fw-1 hr-1 Figure 4-10. Regulation of volatiles from MD and 44568 flowers in response to ethylene (mean + SE; n=3). Air treated control flowers were treated the same as ethylene treated flowers, except no ethylene was added to the chambers and KMnO 4 was included in the chamber. An additional non-treated (NT) control was used to identify changes induced by the treatment.

PAGE 107

95 44568 NP 44568 x MD NP MD x 024681012 Benzyl Benzoate 00.20.40.60.811.21.41.6 Benzyl Alcohol 05101520253035 Benzaldehyde 0 246810121416 Iso-Eugenol 00.511.522.5 2-Phenylethanol gg 0123456789 Phenylacetaldehyde hours after pollination 0210243648 hours after pollination 0210243648 hours after pollination 0210243648 hours after pollination 0210243648 hours after pollination 0210243648 hours after pollination 0210243648 g g fw-1 hr-1g g fw-1 hr-1g g fw-1 hr-1g g fw-1 hr-1g g fw-1 hr-1g g fw-1 hr-1 44568 NP 44568 x MD NP MD x 024681012 Benzyl Benzoate 00.20.40.60.811.21.41.6 Benzyl Alcohol 05101520253035 Benzaldehyde 0 246810121416 Iso-Eugenol 00.511.522.5 2-Phenylethanol gg 0123456789 Phenylacetaldehyde hours after pollination 0210243648 hours after pollination 0210243648 hours after pollination 0210243648 hours after pollination 0210243648 hours after pollination 0210243648 hours after pollination 0210243648 g g fw-1 hr-1g g fw-1 hr-1g g fw-1 hr-1g g fw-1 hr-1g g fw-1 hr-1g g fw-1 hr-1 Figure 4-11. Regulation of volatiles from MD and 44568 flowers in response to pollination (x) and non-pollination (NP) (mean + SE; n=3).

PAGE 108

96 Table 4-1. Benzoic acid, salicylic acid, and cinnamic acid levels after ethylene treatment. Mean + SE in ngg fw -1 in petunia MD corollas after 10 hrs 2-3 LL -1 ethylene treatment or air treatment of whole flowers. Increased benzoic acid observed for BA between zero and 10 hour treatment is likely due to rhythmic regulation of BA, also observed by Kolosova et al (2001). 0 hrs10 hrs Air10 hrs Ethylen e Benzoic acid2577 + 7784829 + 204212538 + 4619MDSalicylic acid117 + 23.7103 + 23.784 + 5.2Cinnamic acid43 + 13.670 + 21.3189 + 74.6Benzoic acid2111 + 64718251 + 155817637 + 167844568Salicylic acid100 + 13.3106 + 8.5124 + 15.9Cinnamic acid33 + 2.8244 + 1.6167 + 17.0TreatmentGenotype Tab1e 4-2. Benzoic acid, salicylic acid, and cinnamic acid after pollination. Mean + SE in ngg fw -1 in petunia MD corollas 36 hrs after pollination (Pol) or non-pollinated (NP) whole flowers. Increased benzoic acid 36 hours after pollination is due to rhythmic regulation, also observed by Kolosova et al (2001). Time0 hrs 36 hrs NP36 hrs PolBenzoic acid3965 + 44224214 + 24795895 + 2104MDSalicylic acid122 + 13.7191 + 6.0 105 + 17.2Cinnamic acid43 + 1.9417 + 27.956 + 16.8Benzoic acid5311 + 32325313 + 22220541 + 135144568Salicylic acid191 + 29.9233 + 14.9265 + 47.0Cinnamic acid27 + 4.3491 + 10.5483 + 46.9Genotype

PAGE 109

CHAPTER 5 GENERAL DISCUSSION AND CONCLUSIONS The results of these studies show that ethylene induces changes in gene expression and regulates floral fragrance in petunia flowers. Work in chapter two was performed with the intent of finding genes that are responsive to ethylene in questioning what processes are affected by ethylene during floral senescence (Chapter 2). A broad high-throughput approach, via microarrays, was taken for expression profiling of genes expressed in petunia flowers. While this process was initially labor intensive, the efficiency and high throughput capability that microarray experiments offered made this approach attractive for identifying differentially regulated genes. In using this methodology, thousands of expressed genes from petunia were simultaneously screened and the expression patterns of a small set of cDNAs were identified to be affected by ethylene. From these data, PhBSMT1 and PhBSMT2 were chosen for further study. A role for ethylene in regulation of the synthesis of methyl benzoate or synthesis of any other volatile in petunia had not been shown prior to this study. The second part of this project examined the in vivo function of BSMT, investigated the mRNA expression patterns in the flower and addressed the effects of ethylene on other floral volatile components (Chapter 3). Building Fundamental Tools for Genomic Studies Petunia is used in many research laboratories for molecular and genetic studies, but it is not as widely used as Arabidopsis thaliana. Thus, the genetic resources for Petunia 97

PAGE 110

98 are more limited. As a starting point, the genomics tools required for gene expression studies in floral tissues were constructed and implemented in this work. First, cDNA libraries were created from RNAs extracted from flowers undergoing different physiological and developmental processes. While sequencing of these libraries was not absolutely required for expression studies, there are multiple long-term advantages that outweigh the initial costs. First, the sequences can be compiled into a searchable database allowing rapid identification of genes most valuable for subsequent analysis. Such decisions can be aided by knowing the identity of many differentially expressed genes in a potentially long list of upand down-regulated genes. Additionally, these types of collections and databases facilitate collaborative interactions between laboratories and institutions, as there are thousands of cDNAs readily available for study. Microarray Analysis of Ethylene Regulated Genes The microarray experiments indicated that there were over 200 genes differentially regulated in response to ethylene in whole flowers (Chapter 2). However, these results were not consistent with expression analysis via RNA gel blot analysis. With these inconsistent results, one might question two things microarray quality and experimental design. The first consideration, microarray quality, can be addressed by a comparison with another experiment that was done with the same set of arrays. In the other study, the purpose was to identify genes specifically expressed in the petals. Probes generated from RNA of petals were used in comparison with probes generated from RNA of various plant organs (excluding the petals). The microarray expression results highly corresponded with expression data from northern blots. Therefore, it is inferred that the quality of the arrays was acceptable for finding differentially regulated genes. The other consideration is the experimental design. The experiments presented in chapter 2

PAGE 111

99 involved the hybridization of probes generated from RNA extracted from whole flowers. In using whole flowers, multiple organs were combined into one sample after treatment. Pooling of all of these RNAs together from multiple floral organs and comparing the RNAs from two pooled sets of tissues possibly introduced substantial error into the experiments. It is likely that each of these organs exhibits differential sensitivity to ethylene, which may even be compounded by small differences in the stage of the tissue at time of collection. Additionally, it is possible that the air-treated control samples were producing small amounts of ethylene, as has been observed for air controls in carnation (Jones, 2002). Ethylene produced by the controls would likely deplete the number of differentially regulated clones identified in the two experiments. All of these factors together likely complicated the results of these experiments. If these studies were to be done again, more specific tissues should be used as the RNA source for the probes to help reduce variability. Regardless of the limitations in these experiments, a number of cDNAs were identified to be differentially regulated in response to ethylene and two of these were chosen for in depth study (Chapter 3). Ethylene Regulates Floral Fragrance in Petunia The down regulation of the two PhBSMT genes was the first indication that ethylene was regulating synthesis of floral volatiles (Chapter 3). They were identified as ethylene down-regulated in whole flowers and in corollas (Chapter 2). The activity of the genes was confirmed in vitro (Negre et al., in press) and in vivo (Underwood et al., submitted) demonstrating that the BSMTs catalyze the synthesis of MeBA. The expression of these genes was sensitive to ethylene, both being down regulated with exogenous ethylene treatments and during pollination in wild type MD flowers. These results corresponded with decreased MeBA emission levels after ethylene treatment and

PAGE 112

100 pollination in MD. The emission of six other major volatiles was also decreased in response to ethylene application and pollination. These effects were not observed in the ethylene insensitive line. Thus, ethylene affects not only the floral display of petunia flowers, but also the fragrance. These findings are important from an ecological perspective and provoke many questions for future studies. Since many of the volatiles are reduced after pollination, it will be of interest to determine how these changes affect pollinator attraction to individual flowers. It is possible that plants regulated floral volatile emission to redirect pollinators to flowers that have not been pollinated. This might be beneficial for maximizing pollen flow to receptive flowers. In Nicotiana, a Manduca hawkmoth species is attracted to, and will pollinate the flowers but also lays eggs on the plant (Baldwin and Ohnmeiss, 1993). This behavior has also been observed in Manduca sexta with petunia. The Manduca spp. larvae forage on these plants upon development. Perhaps decreasing fragrance emission after pollination is a mechanism that has evolved as a strategy to decrease the apparency of the plant to pollinators that lay eggs. If there are different pollinators, the changes may be perceived by some pollinators, but not others. In addition to changes in floral fragrance, nectar is also an important component of attracting pollinators. The effects of pollination on nectar secretion in petunia would tie in another aspect to these studies. In Nicotiana, there is evidence that nectar has a role in defense against microbial attack in addition to a well-known role as a pollinator reward (Thornburg et al., 2003). It has been reported that nectar is reabsorbed after pollinations (Burquez and Corbet, 1991; Pettersson and Knudsen, 2001). Since nectar rewards are

PAGE 113

101 likely learned in association with fragrance (Daly et al., 2001), it is of interest to determine the affect of pollination on nectar secretion in petunia. Plants engineered for reduced MeBA emission had a significantly different fragrance compared to wild type. Human olfaction panels were important for determining that the genetic change had a significant effect on the fragrance. With the characterization of fragrance genes, there is opportunity to improve the fragrance of flowers. While there are limitations in doing this, due to conjugation of the product (Lucker et al., 2001) and availability of substrates for the reactions, many floral components are common to diverse plant genera (Knudsen et al., 1993) and could feasibly be introduced, enhanced, or otherwise modified in heterologous plants. Human olfaction panels will be helpful in determining what fragrances are pleasing to people.

PAGE 114

APPENDIX SUPPLEMENTAL DATA FOR CHAPTER 3 This appendix contains table A-1 with data of functional categorization of cDNAs identified as putative ethylene up or down regulated by microarray analysis (Table A-1) (Chapter 3). cDNAs were selected on the basis of having cy3:cy5 ratios above 2 or below -2 for four or more spots (out of six total from three arrays). Clones that have been analyzed by RNA gel blots are in bold. 102

PAGE 115

Table A-1. Functional categorization of cDNAs identified from microarray analysis. These cDNAs were identified as having a cy3 cy5 ratio > 2 for four out of six spots on the microarrays. 103 Nucleotide and Transcription related (nucleotide binding proteins, transcription factors, complex)Clone #BLAST homologyScoree valueHigherAve RatioSDER2-60Petunia-PP5-B06JAB Jun Activation Binding Protein4240E 8.5 hrs4.203.67ER1-92Petunia-DevA-10-F07.gribosomal S29 -like protein1163.00E-25E 2hrs4.164.18ER2-50Petunia-PP12-C04MADS-box transcription factor FBP23 2720E 8.5 hrs3.330.88ER1-5Petunia-4-F08.gputative replication factor A80.91.00E-14E 2hrs3.311.53ER2-10Petunia-DevA-15-G10.gNucleoid DNA binding like protein1401.00E-47E-16 hrs3.221.39Petunia-DevA-15-G10.gnucleoid DNA-binding like protein1406.00E-47E 2hrs2.140.15ER2-11Petunia-DevA-16-B02.gmRNA binding protein precursor 3930E 8.5 hrs3.181.83ER1-70Petunia-C2H4-5-D07.gflowering protein CONSTANS, putative 1107.18E-24E 8.5 hrs3.080.90ER2-3Petunia-DevA-13-D07.gnucleoid DNA-binding-like protein 2145.00E-59E 2hrs2.810.60ER2-39Petunia-DevA8-A12.gRNA Binding Protein 45 3322.00E-90E 2hrs2.811.44ER2-12Petunia-DevA-16-G04.gRNA helicase -like protein1263.00E-34E 2hrs2.810.53ER1-65Petunia-C2H4-4-A10.gribosomal protein L11-like1304.00E-31E 2hrs2.780.91ER1-3Petunia-4-C07.ghistone H3, putative2154.00E-55E 2hrs2.620.40ER1-90Petunia-DevA-10-A09.g60S RIBOSOMAL PROTEIN L31 1622.00E-39E 2hrs2.610.45ER1-94Petunia-DevA-11R-C05.g60S ribosomal33.10.63E-16 hrs2.450.42ER1-51Petunia-C2H4-24-H07.gsplicing factor SC35 2020E 8.5 hrs2.420.46ER1-46Petunia-C2H4-22-D07.g60S ACIDIC RIBOSOMAL PROTEIN P01101.00E-23E 2hrs2.400.61ER2-65Petunia-PP7-B01ribosomal protein S20 2210E 8.5 hrs2.350.17ER1-60Petunia-C2H4-2-B08.gputative zinc-binding protein91.31.00E-17E 2hrs2.350.19ER1-54Petunia-C2H4-27-B06.gputative 60S ribosomal protein L181773.00E-44E 2hrs2.280.47Petunia-C2H4-27-B06.gputative 60S ribosomal protein L18 1773.00E-44E 8.5 hrs2.230.29ER2-54Petunia-PP3-A11AP2 domain containing protein 76.33.00E-13E 2hrs2.230.17ER1-72Petunia-C2H4-5-F11.gprobable transcription factor SF3 3060E 8.5 hrs2.200.34ER2-41Petunia-DevA-9-B06.gEukaryotic peptide chain release factor subunit 1 (ERF1)3420E 8.5 hrs2.180.17ER1-93Petunia-DevA-10-G09.gsimilar to AP2 domain transcription factor95.93.00E-19E-16 hrs2.160.11RespirationClone #BLAST homologyScoree valueHigherAve RatioSDER1-52Petunia-C2H4-25-H10.gORF270/2 81.65.00E-32E 2hrs2.110.10ER2-61Petunia-PP-5-G06.gADP,ATP carrier protein, mitochondrial precursor 379e-104E 2 hrs2.320.39

PAGE 116

Table A-1. Continued. Nucleotide DegradationClone #BLAST homologyScoree valueHigherAve RatioSDER2-53Petunia-PP1-G05.gribonuclease T2 377e-104E 2hrs5.654.03Protein DegradationClone #BLAST homologyScoree valueHigherAve RatioSDER2-14Petunia-DevA-17R-F04.gcysteine proteinase3306.00E-90E-16 hrs2.730.70Petunia-DevA-17R-F04.gcysteine proteinase3307.00E-90E 2hrs2.300.44ER2-22Petunia-C2H4-28-F01.gcysteine protease 2723.00E-72E 2hrs2.550.07ER1-35Petunia-DevA-20-H01.gcysteine proteinase1912.00E-48E 2hrs2.500.71ER1-56Petunia-C2H4-18-E08.gcysteine protease 2279.00E-59E 8.5 hrs2.640.43Petunia-C2H4-18-E08.gcysteine protease2279.00E-59E 2hrs2.500.37Secondary MetabolismClone #BLAST homologyScoree valueHigherAve RatioSDER2-15Petunia-DevA-17R-G04.gMajor allergen Pru av 1 (Pru a 1).951.60E-40E 8.5 hrs4.673.67ER1-59Petunia-C2H4-2-A12.gS-adenosylmethionine decarboxylase proenzyme2755.00E-77E 2hrs4.272.69Petunia-C2H4-2-A12.gS-ADENOSYLMETHIONINE DECARBOXYLASE PROENZY M 2755.00E-77E 8.5 hrs2.280.24ER1-47Petunia-C2H4-22-G01.gputative cinnamoyl CoA reductase1373.00E-44E 8.5 hrs3.180.50Petunia-C2H4-22-G01.gputative cinnamoyl CoA reductase 1373.00E-44E 2hrs2.430.59ER2-52Petunia-PP1-F02.gO-methyltransferase35.40.83E 2hrs2.730.61ER2-46Petunia-PP10-H11.gSalicylic acid carboxyl methyltransferase3231.00E-85A 16 hrs2.700.42ER1-41Petunia-C2H4-20-F03.gsimilarity to a dimethylaniline monooxygenase (N-oxide410.0044E 8.5 hrs2.680.63ER2-51Petunia-PP12-H03.gphenyl ammonium lyase555e-155A 16 hrs2.680.48ER2-31Petunia-DevA-28-B10.ganthranilate N-hydroxycinnamoyl/benzoyltransferase-like p 1054.87E-22E 8.5 hrs2.680.63ER1-36Petunia-C2H4-19-B04.gphenylcoumaran benzylic ether reductase3284.00E-89A 16 hrs2.640.67ER1-39Petunia-C2H4-1-F10.gputative peptidyl-prolyl cis-trans isomerase [Arabidopsis624.60E-09E 8.5 hrs2.600.73ER1-34Petunia-C2H4-17-G06.gallergenic isoflavone reductase like2683.00E-82A 16 hrs2.580.64ER2-23Petunia-DevA-22-H11.gputative N-acetylglucosaminyltransferase 3610E 8.5 hrs2.440.46ER1-45Petunia-C2H4-22-B09.gputative flavonol synthase-like protein1352.00E-31E 2hrs2.430.46ER2-63Petunia-PP6-D05.gphenyl ammonium lyase7410.00E+00A 16 hrs2.400.32ER1-31Petunia-C2H4-17-A09.gallergenic isoflavone reductase like2736.00E-73 A 16 hrs2.350.41ER2-4Petunia-DevA-13-H03.g3-hydroxyisobutyryl-coenzyme A hydrolase3740E 8.5 hrs2.350.47 104

PAGE 117

Table A-1. Continued. 105 Secondary MetabolismClone #BLAST homologyScoree valueHigherAve RatioSDER1-33Petunia-C2H4-17-D12.gphenylcoumaran benzylic ether reductase2339.00E-61A 16 hrs2.320.31ER2-67Petunia-PP8-A04.gSalicylic acid carboxyl methyltransferase2648.00E-68A 16 hrs2.320.40ER2-47Petunia-PP11-B07.gS-adenosyl methionine synthetase7890 A 16 hrs2.320.25ER1-43Petunia-C2H4-22-B04.gS-adenosyl methionine synthetase430e-120 A 16 hrs2.280.36ER2-59Petunia-PP4-G07S-adenosyl-L-methionine:salicylic acid carboxyl methyltransfe r 3250E 8.5 hrs2.230.40ER1-50Petunia-C2H4-24-F02.gallergenic isoflavone reductase like2652.00E-70A 16 hrs2.130.13Pathogen/Defense/StressClone #BLAST homologyScoree valueHigherAve RatioSDER2-36Petunia-DevA-7-A11.gsenescence-associated protein (SAG29)48.11.00E-04E 8.5 hrs6.041.99ER1-11Petunia-C2H4-10D-E10.ghypersensitive-induced response protein2644.00E-74E 2hrs5.474.25ER1-4Petunia-4-D05.gsenescence-associated protein (SAG29)486.00E-05E 2hrs3.471.03Petunia-4-D05.gsenescence-associated protein (SAG29)485.00E-05E 2 hrs3.471.03Petunia-4-D05.gsenescence-associated protein (SAG29)485.00E-05E 8.5 hrs4.531.40ER1-42Petunia-C2H4-22-A10.gMTN31418.00E-33E-16 hrs3.030.95Petunia-C2H4-22-A10.gMtN3 1418.00E-33E 2hrs3.591.70ER1-82Petunia-C2H4-7RR-G07.gAvr9/Cf-9 rapidly elicited protein 137 87.84.00E-17E 8.5 hrs3.481.16ER2-37Petunia-DevA-7-A11.gsenescence-associated protein (SAG29)48.11.00E-04E 2hrs3.250.70ER1-17Petunia-C2H4-14G50-A12.gputative leucine rich protein99.44.00E-20E-16 hrs3.200.92ER1-44Petunia-C2H4-22-B08.gwound-induced protein tomato66.23.00E-16E 2hrs3.061.41Petunia-C2H4-22-B08.gwound-induced protein tomato (fragment).66.23.00E-16E 8.5 hrs2.250.17ER2-20Petunia-DevA-18RR-F09.gproteinase inhibitor type II precursor NGPI-232.73.9E 2hrs3.060.58ER2-24Petunia-DevA-24-A05.gglutathione S-transferase/peroxidase1874.00E-47E 2hrs2.931.19ER1-89Petunia-C2H4-9D-B04.gHR7474.00E-09E 2hrs2.921.53ER1-80Petunia-C2H4-7RR-C12.gsymbiosis-related protein1893.00E-47E 2hrs2.840.76ER1-69Petunia-C2H4-5-C07.gputative disease resistance protein1924.00E-48E 2hrs2.780.54ER2-30Petunia-DevA-27-F04.gosmotin like precursor tomato435e-121E-16 hrs2.750.45ER1-55Petunia-C2H4-28-B08.gHEAT SHOCK COGNATE 70 KD PROTEIN 21008.00E-21E 2hrs2.700.42ER2-25Petunia-DevA-25-A03.gperoxidase 11801.00E-44E 2hrs2.630.73ER1-85Petunia-C2H4-8RR-D02.gGlycine-rich protein 2.511.79E-06E 8.5 hrs2.580.43ER1-96Petunia-DevA-11R-F05.ghypersensitive-induced response protein2546.00E-67E 2hrs2.500.57

PAGE 118

Table A-1. Continued. 106 Pathogen/Defense/StressClone #BLAST homologyScoree valueHigherAve RatioSDER1-24Petunia-C2H4-15-D03.gHR7 76.63.00E-13E 2hrs2.500.46ER2-6Petunia-DevA-14R-D08.gputative acyl CoA synthetase 3465.00E-95E-16 hrs2.480.46ER2-33Petunia-DevA-29-G10.gprobable cysteine proteinase inhibito r 1602.00E-38E-16 hrs2.400.16ER2-44Petunia-PP10-F08Elicitor inducible gene product Nt-SubE80 1715.00E-42E 2hrs2.380.43ER1-10Petunia-C2H4-10D-E08.guncoupling protein (clone StUCP7)1302.00E-29E 2hrs2.310.36ER1-63Petunia-C2H4-2-D11.gproline-rich protein V-beta 1 precursor38.10.1E 2hrs2.300.26Petunia-DevA-6-A06.gprobable legumin B350.39E 2hrs2.300.16ER1-91Petunia-DevA-10-E04.gcontains similarity to MTN3 protein1804.00E-45E-16 hrs2.300.29ER2-49Petunia-PP11-C08putative peroxidase82.44.00E-15E 2hrs2.170.14ER2-56Petunia-PP3-D09L-ascorbate oxidase precursor, pollen specific2780.00E+00E-16 hrs2.160.43ER1-30Petunia-C2H4-16-G09.gapoptosis-linked gene 430.87.9E 2hrs2.030.06Flower Development RelatedClone #BLAST homologyScoree valueHigherAve RatioSDER1-2Petunia-3-H03.gfil1 protein1108.00E-24E 2hrs2.600.66ER2-38Petunia-DevA-7-G12.gPOLLEN-SPECIFIC PROTEIN NTP303 PRECURSOR3314.00E-90E 2hrs2.270.18Petunia-DevA-7-G12.gPollen specific protein NTP3033314.00E-90E-16 hrs2.250.26Tissue Differentiation RelatedClone #BLAST homologyScoree valueHigherAve RatioSDER1-81Petunia-C2H4-7RR-E09.gYABBY2 1271.00E-28E 2hrs2.650.07Fruit Ripening RelatedClone #BLAST homologyScoree valueHigherAve RatioSDER2-18Petunia-DevA-18RR-A08.gputative beta-galactosidase441e-123E 2hrs3.581.36Senesence/Ripening RelatedClone #BLAST homologyScoree valueHigherAve RatioSDER1-76Petunia-C2H4-7-B08.gpectin methylesterase32.73E 2hrs5.934.95ER1-4Petunia-4-D05.gsenescence-associated protein (SAG29)48.16.00E-05E 2hrs3.471.03ER2-36Petunia-DevA-7-A11.gsenescence-associated protein (SAG29)48.11.00E-04E 2hrs3.250.70

PAGE 119

Table A-1. Continued. Senesence/Ripening RelatedClone #BLAST homologyScoree valueHigherAve RatioSDER2-58Petunia-PP4-C09pectinesterase, putative1363.00E-31E 2hrs2.430.53ER2-45Petunia-PP10-G12pectin methyl-esterase like protein53.52.00E-09E 2hrs2.280.13ER2-69Petunia-PP9-D07pectin methyl-esterase 75.13.00E-13E 2hrs2.120.08ER2-9Petunia-DevA-14R-H04.gcopper homeostasis facto r 1371.00E-31E 2hrs2.050.06ER1-25Petunia-C2H4-15-F02.gPectin methylesterase like protein2308.00E-60E 2 hrsEthylene RelatedClone #BLAST homologyScoree valueHigherAve RatioSDER2-43Petunia-PP10-A03ACC OXIDASE 12180E 8.5 hrs4.780.88Petunia-PP6-G05.g ACC OXIDASE 13840E 8.5 hrs3.821.17ER2-64Petunia-PP6-G05ACC OXIDASE 13840E 8.5 hrs2.560.48ER1-61Petunia-C2H4-2-B12.gethylene-response protein ETR1 homolog1494.00E-35E 2hrs3.181.27Petunia-C2H4-2-B12.gethylene receptor homolog [Lycopersicon esculentum].1494.00E-35E 8.5 hrs2.632.63Petunia-C2H4-2-B12.gethylene-response protein ETR1 homolog1494.00E-35E 2hrs2.520.44ER1-7Petunia-C2H4-10D-D12.gethylene receptor ERS homolog 428e-119E 2hrs2.470.28ER1-79Petunia-C2H4-7-F07.gethylene-responsive protein1128.00E-25E 2hrs2.340.11ER1-26Petunia-C2H4-16-A10.gACC OXIDASE 41920E 8.5 hrs2.250.85Hormone RelatedClone #BLAST homologyScoree valueHigherAve RatioSDER1-38Petunia-C2H4-1-C10.ggibberellin 2-oxidase No13213.00E-87E 2hrs2.870.53ER2-42Petunia-DevA-9-F02.gNt-iaa4.3 deduced protein723.00E-12E 2hrs2.580.77ER2-66Petunia-PP7-F12indoleacetic acid (IAA)-inducible gene (IAA7)2432.00E-63A 2hrs2.550.27ER1-6Petunia-C2H4-10D-C11.gBrassinosteroid-regulated protein BRU1 precursor.3530E 8.5 hrs2.460.44ER1-27Petunia-C2H4-16-B04.gbrassinosteroid insensitive 1 gene3025.00E-84E 2hrs2.450.30SignalingClone #BLAST homologyScoree valueHigherAve RatioSDER2-6Petunia-DevA-14R-D08.gputative acyl CoA synthetase 3465.00E-95E-16 hrs2.480.46 107

PAGE 120

Table A-1. Continued. SignalingClone #BLAST homologyScoree valueHigherAve RatioSDER2-6Petunia-DevA-14R-D08.gputative acyl-CoA synthetase3469.00E-96E 2hrs2.460.22ER2-5Petunia-DevA-14R-B10.gTGF-beta receptor-interacting protein 12492.00E-65E 2hrs2.300.25ER2-32Petunia-DevA-29-C10.gputative casein kinase 73.98.00E-13E 2hrs21.4348.72Petunia-DevA-29-C10.gcasein kinase like protein6919.78E-13E 8.5 hrs3.431.04ER2-57Petunia-PP-4-C08.gRAB7C, GTP binding protein386e-106E 8.5 hrs5.654.78ER1-74Petunia-C2H4-6RR-A02.gputative GTP pyrophosphokinase [Arabidopsis thaliana].2784.00E-74E 8.5 hrs4.240.76ER1-15Petunia-C2H4-11-D06.gGTP-binding protein Rab11e2647.00E-70E 2hrs 3.752.01ER1-83Petunia-C2H4-7RR-G10.gcalmodulin 7 2986.00E-80E 2hrs3.461.83ER1-49Petunia-C2H4-24-D04.gcalmodulin2666.00E-71E 2 hrs3.221.55ER1-18Petunia-C2H4-14G50-A12.gputative leucine rich protein99.44.00E-20E-16 hrs3.200.92ER1-1Petunia-3-D03.gdiacylglycerol kinase433e-120E 2hrs2.750.49ER1-77Petunia-C2H4-7-C12.gMAP kinase58.95.00E-08E 2hrs2.740.49ER1-53Petunia-C2H4-26-F02.gprotein kinase homolog F24O1.132213.00E-58E 2hrs2.690.40ER2-8Petunia-DevA-14R-F09.gprotein phosphatase 2C 344e-120E 8.5 hrs2.530.38ER1-16Petunia-C2H4-11-E09.gEF-hand Calcium binding protein-like 1001.90E-21E 8.5 hrs2.400.43ER1-88Petunia-C2H4-9-C12.gsmall GTP-binding protein Sar1BNt3239.00E-88E 2hrs2.300.27ER1-20Petunia-C2H4-14G50-D12.gputative protein kinase1085.00E-23E 2hrs2.280.15ER1-48Petunia-C2H4-23-E08.greceptor-like protein kinase 37.40.078E 2hrs2.170.94ER1-62Petunia-C2H4-2-D03.gprotein kinase3514.00E-96E 2hrs2.130.13LipidsClone #BLAST homologyScoree valueHigherAve RatioSDER2-7Petunia-DevA-14R-F03.glipid-transfer protein-like protein1933.73E-16E 8.5 hrs3.250.99ER1-23Petunia-C2H4-15-C06.gheme A farnesyltransferase homolog F16B22.160.58.00E-16E 2hrs2.981.37Petunia-C2H4-15-C06.gheme A farnesyltransferase homolog F16B22.160.58.00E-16E 8.5 hrs2.400.32ER2-1Petunia-DevA-11R-G07.gfatty acid hydroxylase like protein 2377.23E-07E 8.5 hrs2.930.29ER1-13Petunia-C2H4-11-B02.gglyoxysomal beta-ketoacyl-thiolase2878.00E-75E 2hrs2.560.38ER1-21Petunia-C2H4-14G50-E10.gputative acyl-CoA synthetase 1492.10E-35E 8.5 hrs2.540.67ER2-40Petunia-DevA8-D09.gCER2, fatty acid elongase1683.00E-41E 2hrs2.200.26ER1-32Petunia-C2H4-17-B01.gAcyl-CoA binding protein1278.00E-29E 2hrs2.160.15 108

PAGE 121

Table A-1. Continued. RespirationClone #BLAST homologyScoree valueHigherAve RatioSDER1-29Petunia-C2H4-16-E12.gUbiquinol-cytochromeC reductase Iron sulfur subunit1385e-106E 8.5 hrs6.588.09ER2-2Petunia-DevA-12-D05.gcitrate synthase -like protein1855.00E-46E 2hrs3.001.10ER2-13Petunia-DevA-17R-F03.gpredicted NADH dehydrogenase2552.00E-67E 2hrs2.720.89Petunia-DevA-17R-F03.gNADH-ubiquinone oxidoreductase 24 kDa subunit, mitoch o 2450E 8.5 hrs2.450.17Respiration continuedClone #BLAST homologyScoree valueHigherAve RatioSDER1-40Petunia-C2H4-1-H04.gcytochrome P4501944.00E-50E 2hrs2.700.83ER2-26Petunia-DevA-25-B02.gmitochondrial formate dehydrogenase precursor385e-106E 2hrs2.700.49ER1-75Petunia-C2H4-6RR-D05.gmitochondrial processing peptidase1447.00E-34E 2hrs2.510.34ER1-86Petunia-C2H4-8RR-G05.gprobable H+ transporting ATP synthase beta chain1711.00E-48E-16 hrs2.410.38ChaperonesClone #BLAST homologyScoree valueHigherAve RatioSDER1-14Petunia-C2H4-11-C02.g14-3-3 family protein2891.00E-77E 2hrs4.492.13ER2-29Petunia-DevA-27-C08.g14-3-3 family protein405e-112E 2hrs3.120.97ER2-19Petunia-DevA-18RR-E11.gCHAPERONIN CPN60-2, MITOCHONDRIAL 1454.13E-34E 8.5 hrs2.950.31Structural-CellClone #BLAST homologyScoree valueHigherAve RatioSDER1-78Petunia-C2H4-7-D06.gactin depolymerizing factor 2547.00E-67E 2hrs3.783.06ER2-21Petunia-DevA-19R-C07.gtubulin beta-2 chain3422.00E-93E-16 hrs3.091.58ER1-12Petunia-C2H4-10-G05.gextensin precursor38.50.07E 2hrs2.630.75ER1-8Petunia-C2H4-10D-E01.gLYTB-like protein precursor392e-108E-16 hrs2.350.35Petunia-C2H4-10D-E01.gLYTB-like protein precursor392e-108E 2hrs2.570.80Petunia-C2H4-2-D11.gproline-rich protein V-beta 1 precursor38.10.1E 2hrs2.300.26ER1-57Petunia-C2H4-29-H04.gXyloglucan endotransglycolyase1853.00E-46E-16 hrs2.100.00PhotosynthesisClone #BLAST homologyScoree valueHigherAve RatioSDER1-73Petunia-C2H4-5-G05.gRIBULOSE BISPHOSPHATE CARBOXYLASE SMALL CHAI N 1557.00E-60E 2hrs3.010.65ER2-68Petunia-PP8-B07Photosystem II 10 kDa polypeptideE 8.5 hrs2.130.15 109

PAGE 122

Table A-1. Continued. 110 Carbohydrate MetabolismClone #BLAST homologyScoree valueHigherAve RatioSDER1-67Petunia-C2H4-4-B05.gsuccinate dehydrogenase flavoprotein alpha subunit (em b 3432.00E-95E 8.5 hrs5.072.87ER1-37Petunia-C2H4-19-C01.gFructokinase 1398.00E-55E 2hrs2.640.65ER2-17Petunia-DevA-18RR-A03.gglyceraldehyde-3-phosphate dehydrogenase1415.00E-33E 2hrs2.570.49ER1-64Petunia-C2H4-32-H12.gfructose-1,6-bisphosphate aldolase1091.00E-23E 2hrs2.400.61ER1-58Petunia-C2H4-2-A04.gSOLUBLE GLYCOGEN [STARCH] SYNTHASE2232.00E-57E 2hrs2.370.27Petunia-C2H4-2-A04.gsoluble starch synthetase2232.00E-57E-16 hrs2.370.42ER1-95Petunia-DevA-11R-F01.gbeta-xylosidase70.53.00E-12E 2hrs2.230.17One Carbon MetabolismClone #BLAST homologyScoree valueHigherAve RatioSDER1-87Petunia-C2H4-8RR-H02.ghydroxymethyltransferase 279e-101E 2hrs2.490.65ER2-48Petunia-PP11-C05hydroxymethyltransferase412e-114E 2hrs2.080.10ER2-62Petunia-PP6-A03hydroxymethyltransferase 383e-105E 2hrs2.070.10Amino Acid MetabolismClone #BLAST homologyScoree valueHigherAve RatioSDER1-84Petunia-C2H4-7RR-H11.gDIHYDRODIPICOLINATE SYNTHASE2544.00E-79E 2hrs3.843.78ER1-19Petunia-C2H4-14G50-D09.gaspartate transaminase35.40.34E 2hrs2.150.13TransportClone #BLAST homologyScoree valueHigherAve RatioSDER1-68Petunia-C2H4-5-A03.gsulfate transporte r 792.00E-14E 2hrs3.250.88ER1-9Petunia-C2H4-10D-E02.gsynaptic glycoprotein SC2-like protein398e-110E 2hrs2.920.92ER1-22Petunia-C2H4-14G50-F09.gwater channel protein 83.23.00E-15E 2hrs2.861.12Petunia-C2H4-14G50-F09.gwater channel protein 83.23.00E-15E 8.5 hrs2.360.11ER1-71Petunia-C2H4-5-E01.gputative sugar transporter226e-110E 8.5 hrs2.850.40ER1-66Petunia-C2H4-4-A11.gsugar transporter like protein1841.00E-74E 2hrs2.740.99ER2-34Petunia-DevA5-E11.gphosphate/phosphoenolpyruvate translocator 2720E 8.5 hrs2.330.29ER2-16Petunia-DevA-18-G02.gpredicted vesicle-associated membrane protein 7C379.70E-02E 2hrs2.270.37

PAGE 123

Table A-1. Continued. 111 TransportClone #BLAST homologyScoree valueHigherAve RatioSDER2-55Petunia-PP3-B10.gHuman secretory carrier membrane protein604.00E-06A 16 hrs2.270.31ER2-9Petunia-DevA-14R-H04.gcopper homeostasis facto r 1371.00E-31E 2hrs2.050.06ER1-28Petunia-C2H4-16-D04.gputative membrane trafficking factor1283.00E-29E-16 hrs2.040.05TransposonsClone #BLAST homologyScoree valueHigherAve RatioSDER2-28Petunia-DevA-27-B06.gputative non-LTR retroelement reverse transcriptase45.86.00E-06E 2hrs3.201.41ER2-27Petunia-DevA-26-C10.gputative transposon protein50.86.00E-09E 2hrs2.600.95Unknown/HypotheticalClone #BLAST homologyScoree valueHigherAve RatioSDPetunia-C2H4-19-D07.gCG13846 gene product40.40.019E 2hrs2.640.56Petunia-DevA-19R-F12.gExpressed protein 62.81.00E-09E-16 hrs3.741.38Petunia-DevA-19R-F12.ghypothetical protein 62.81.00E-09E 2hrs3.021.43Petunia-3-A10.ghypothetical protein2877.00E-77E-16 hrs2.610.80Petunia-3-A10.gAt2g16090/F7H1.11 2877.00E-77E 2hrs3.601.89Petunia-DevA-20-A05.gevidence:NAS~putative~unclassifiable 32.71.80E+00E-16 hrs9.9011.37Petunia-DevA-20-A05.gputative32.72.2E 2hrs5.698.88Petunia-3-D08.gno similaritiesE-16 hrs3.201.29Petunia-DevA-19R-A10.gunknown protein2414.00E-63E-16 hrs2.580.38Unknown/HypotheticalClone #BLAST homologyScoree valueHigherAve RatioSDPetunia-C2H4-30-G05.gunknown protein 1544.00E-37E-16 hrs2.100.20Petunia-C2H4-18-F12.gputative protein627.00E-09E-16 hrs2.880.85Petunia-C2H4-3-B06.gputative protein 1962.00E-49E-16 hrs2.751.26Petunia-PP5-G09.gUnknown protein 99.42.00E-20A 16 hrs2.751.26Petunia-4-D07.gunknown protein 1213.00E-27E-16 hrs2.430.66Petunia-PP1-G12.ghypothetical protein3379.00E-92E 2hrs2.630.36Petunia-DevA-27-E01.ghypothetical protein F12E4.360 70.56.00E-12A 2hrs2.932.04Petunia-DevA-11R-D02.gunknown protein85.92.00E-16E 2hrs2.650.47Petunia-C2H4-15-H04.gConserved hypothetical protein 322.20E+00E 2hrs3.100.94

PAGE 124

Table A-1. Continued. Unknown/HypotheticalClone #BLAST homologyScoree valueHigherAve RatioSDPetunia-C2H4-7RR-H04.gunknown protein90.11.00E-17E 2hrs3.131.36Petunia--G02.gexpressed protein 96.35.00E-20E 2hrs2.680.64Petunia-DevA5-B08.gUnknown Protein90.12.00E-17E 2hrs2.550.56Petunia-DevA5-B08.gunknown protein90.12.00E-95E 8.5 hrs5.002.10Petunia-DevA-10-B05.ghypothetical protein 43.10.003E 2hrs2.430.25Petunia-DevA-9-G08.ghypothetical protein31.24.4E 2hrs3.792.40Petunia-DevA-10-F08.gputative protein 1822.00E-45E 2hrs3.210.99Petunia-C2H4-19-G11.gunknown78.63.00E-14E 2hrs2.690.53Petunia-C2H4-28-H09.gunknown protein86.32.00E-24E 2hrs2.550.21Petunia-C2H4-2-B10.gputative protein2223.00E-57E 2hrs3.281.82Petunia-PP1-F09.ghypothetical protein 33.90.8E 2hrs2.630.81Petunia-DevA-7-D03.ghypothetical protein1424.00E-33E 2hrs2.370.62Petunia-DevA-10-E01.gunknown protein 1253.00E-28E 2hrs2.240.19Petunia--H07.gunknown protein90.52.00E-17E 2hrs2.871.13Petunia-DevA-16-D05.ghypothetical protein F28P10.5064.72.00E-18E 2hrs2.940.61Petunia-C2H4-1-G05.gunknown protein2333.00E-66E 2hrs2.510.42Petunia-C2H4-27-E08.gHypothetical protein79.72.00E-29E 2hrs2.900.94Petunia-DevA-15-C12.gputative protein1022.00E-21E 2hrs2.430.29Petunia-C2H4-6RR-H07.gunknown protein2148.00E-55E 2hrs2.610.84Petunia-C2H4-7RR-E02.gunknown protein2809.00E-75E 2hrs2.600.78Petunia-C2H4-6RR-G07.gno hitsE 2hrs2.710.94Petunia-DevA-15-A03.ghypothetical protein SPAP8A3.0231.61.5E 2hrs3.271.05Petunia-C2H4-29-F06.ghypothetical protein 35.40.17E 2hrs2.450.60Petunia-C2H4-32-A05.ghypothetical protein SPBC21.0433.14A 2hrs3.520.68Petunia-C2H4-32-B05.gunknown protein 323.2E 2hrs2.550.47Petunia-C2H4-32-B05.glow quality sequenceE-16 hrs2.670.73Petunia-C2H4-28-B02.gno hitsE 2hrs3.601.20Petunia-PP9-B09unknown protein1032.00E-21E 2hrs2.150.14Petunia-PP9-C10hypothetical protein 1525.00E-36E 2hrs2.230.38Petunia-PP11-B08unknown protein 92.83.00E-18E 2hrs2.250.28Petunia-C2H4-17-C10.gno hitsE 2hrs3.300.47 112

PAGE 125

Table A-1. Continued. Unknown/HypotheticalClone #BLAST homologyScoree valueHigherAve RatioSDPetunia-C2H4-16-C04.ghypothetical protein 1 common tobacco transposable elemen t 52.84.00E-06E 2hrs3.050.71Petunia-PP6-B12similar to AMP-binding protein1914.00E-50A 2hrs3.020.97Petunia-C2H4-7RR-G07.gputative protein 87.84.00E-17E 2hrs2.991.02Petunia-PP11-F02CG2839 gene product 47.83.00E-04A 2hrs3.701.27Petunia-C2H4-11-A07.gexpressed protein1187.00E-26E 2hrs3.170.81Petunia-PP-3-D08.gexpressed protein97.87.00E-20E 2 hrs2.850.79Petunia-DevA-15-E11.gputative protein35.89.30E-02E 8.5 hrs4.554.44Petunia-C2H4-8RR-A05.gunknown2230E 8.5 hrs4.353.10Petunia-DevA-6-H08.ghypothetical protein Arabidopsis thaliana1514.00E-36E 8.5 hrs3.952.60Clones with No/Low Quality Sequence DataClone #BLAST homologyScoree valueHigherAve RatioSDER2-35Petunia-DevA-6-A03.gno dataE-16 hrs2.961.51Petunia-DevA-15-B01.glow quality sequenceE-16 hrs3.000.75Petunia-C2H4-31-E04.gno hits returnedE-16 hrs3.821.72Petunia-C2H4-18-H02.glow quality sequenceA 16 hrs2.560.44Petunia-C2H4-32-B08.glow quality sequenceE-16 hrs2.760.70Petunia-C2H4-32-B08.gT24P13.832.30.92E 2hrs2.800.45Petunia-DevA-13-E12.glow quality sequenceE-16 hrs2.490.47Petunia-PP9-A02no sequence dataE 2hrs2.170.14Petunia-PP12-F08no sequence dataE 2hrs2.130.05Petunia-PP12-C09no sequence dataE 2hrs2.080.10Petunia-PP11-C12no sequence dataE 2hrs2.360.45Petunia-C2H4-10-C08.gmaturase K 30.45.40E+00E 2hrs4.332.47Petunia-PP5-A01.gDanio rerio mRNA for hypothetical protein566.00E-05A 16 hrs2.800.57Petunia-C2H4-7-A06.gmostly low quality sequenceA 2 hrs2.450.50Petunia-DevA-24-A02.glow quality sequenceA 2hrs2.540.33Petunia-PP-9-B04.gno sequence dataA 2 hrs2.350.88Petunia-PP9-H04no sequence dataA 2 hrs2.660.70Petunia-PP9-H04E 8.5 hrs5.375.23Petunia-PP4-H01low quality sequenceE 8.5 hrs4.101.14 113

PAGE 126

Table A-1. Continued. Clones with No/Low Quality Sequence DataClone #BLAST homologyScoree valueHigherAve RatioSDPetunia--C11.gE 8.5 hrs2.250.31Petunia-C2H4-10D-F01.gE 8.5 hrs2.380.28Petunia-C2H4-11-D02.gE 8.5 hrs4.350.55Petunia-C2H4-11-H01.gE 8.5 hrs2.300.34Petunia-C2H4-14G50-B03.gE 8.5 hrs2.540.78Petunia-C2H4-15-G12.gE 8.5 hrs4.424.52Petunia-C2H4-22-E04.gE 8.5 hrs2.430.33Petunia-C2H4-23-A06.gE 8.5 hrs2.540.18Petunia-C2H4-28-C12.gE 8.5 hrs3.550.86Clones with No/Low Quality Sequence DataClone #BLAST homologyScoree valueHigherAve RatioSDPetunia-C2H4-2-C03.gE 8.5 hrs2.350.34Petunia-C2H4-32-E07.gE 8.5 hrs3.180.53Petunia-C2H4-3-D12.gE 8.5 hrs2.280.21Petunia-C2H4-5-E06.gE 8.5 hrs3.021.08Petunia-C2H4-6-B11.gE 8.5 hrs2.230.26Petunia-C2H4-7-A05.gE 8.5 hrs2.881.19Petunia-C2H4-7RR-F05.gE 8.5 hrs2.580.44Petunia-C2H4-9-B01.gE 8.5 hrs2.881.43Petunia-DevA-11R-C08.gE 8.5 hrs2.740.79Petunia-DevA-13-C04.gE 8.5 hrs2.951.07Petunia-DevA-15-C03.gE 8.5 hrs2.800.79Petunia-DevA-15-C03.gchitin synthase34.30.99E 2hrs3.150.92Petunia-DevA-15-E11.gE 8.5 hrs4.554.44Petunia-DevA-17R-B08.gE 8.5 hrs2.280.25Petunia-DevA-18RR-F09.gE 8.5 hrs2.950.63Petunia-DevA-19R-A04.gE 8.5 hrs2.300.36Petunia-DevA-19R-F06.gE 8.5 hrs2.530.47Petunia-DevA-20-C08.gE 8.5 hrs2.600.56Petunia-DevA-23-E05.gE 8.5 hrs3.601.50Petunia-DevA-23-E05.gputative retro element polyprotein35.42.00E-01E-16 hrs3.431.74 114

PAGE 127

Table A-1. Continued. Clones with No/Low Quality Sequence DataClone #BLAST homologyScoree valueHigherAve RatioSDPetunia-DevA-25-A11.gE 8.5 hrs2.680.96Petunia-DevA-25-E11.gE 8.5 hrs3.150.96Petunia-DevA-26-E04.gE 8.5 hrs2.600.88Petunia-DevA-9-D01.gE 8.5 hrs2.360.13Petunia-DevA-9-F06.gE 8.5 hrs2.180.10Petunia-PP11-C02E 8.5 hrs2.560.64Petunia-PP11-E08E 8.5 hrs2.950.70Petunia-PP11-E08FlgL protein 32.35.2E 2hrs2.550.48Petunia-PP2-C07.gE 8.5 hrs2.630.56Petunia-PP2-C09.gE 8.5 hrs2.400.33Petunia-PP2-F01.gE 8.5 hrs10.8219.11Petunia-PP5-A11E 8.5 hrs3.100.85Petunia-PP6-F09E 8.5 hrs3.231.40Petunia-PP7-A10E 8.5 hrs2.370.23Petunia-PP7-D04E 8.5 hrs2.440.61Petunia-C2H4-3-C05.gE 2hrs3.151.12Petunia-DevA-14R-A04.gE 2 hrs3.241.42Petunia-DevA-27-C11.gE 2 hrs3.732.02Petunia-DevA-27-D08.gE 2 hrs2.290.43Petunia-C2H4-15-G01.gE 2 hrs2.710.64 115

PAGE 128

LIST OF REFERENCES Abeles FB, Morgan PW, Saltveit ME (1992) Ethylene in Plant Biology Ed 2. Academic Press, San Diego, CA Acaster MA, Kende H (1983) Properties and partial purification of 1-aminocyclopropane-1-carboxylate synthase. Plant Physiol 72: 139-145 Adams DO, Yang S-F (1979) Ethylene biosynthesis: Identification of 1-aminocyclopropane-1-carboxylic acid as an intermediate in the conversion of methionine to ethylene. Proc Natl Acad Sci USA 76: 170-174 Alonso JM, Hirayama T, Roman G, Nourizadeh S, Ecker JR (1999) EIN2, a bifunctional transducer of ethylene and stress responses in Arabidopsis. Science 284: 2148-2152 Altenburger R, Matile P (1988) Circadian rhythmicity of fragrance emission in flowers of Hoya carnosa R. Br. Planta 174: 248-252 Altenburger R, Matile P (1990) Further observations on rhythmic emission of fragrance in flowers. Planta 180: 194-197 Ando T, Saito N, Tatsuzawa F, Kakefuda T, Yamakage K, Ohtani E, Koshi-ishi M, Matsusake Y, Kokubun H, Watanabe H, Tsukamoto T, Ueda Y, Hashimoto G, Marchesi E, Asakura K, Hara R, Seki H (1999) Floral anthocyanins in wild taxa of Petunia (Solanaceae). Biochem Syst Ecol 27: 623-650 Baldwin IT, Ohnmeiss TE (1993) Alkaloidal responses to damage in Nicotiana native to North America. J Chem Ecol 19: 1143-1153 Bielskei RL, Reid MS (1992) Physiological-changes accompanying senescence in the emphemeral daylily flower. Plant Physiol 98: 1042-1049 Bernhardt P (1999) The Roses Kiss, A Natural History of Flowers. Island Press, Washington, D.C. Bernhardt P, Sage T, Weston P, Azuma H, Lam M, Thien LB, Bruhl J (2003) The pollination of Trimenia moorei (Trimeniaceae): Floral volatiles, insect/wind pollen vectors and stigmatic self-incompatibility in a basal angiosperm. Ann Bot-London 92: 445-458 116

PAGE 129

117 Berrocal-Lobo M, Molina A, Solano R (2002) Constitutive expression of the ETHYLENE-RESPONSE-FACTOR1 in Arabidopsis confers resistance to several necrotrophic fungi. Plant J 29: 23-32 Bleecker AB, Estelle MA, Somerville C, Kende H (1988) Insensitivity to ethylene conferred by a dominant mutation in Arabidopsis thaliana. Science 241: 1086-1089 Borochov A, Woodson WR (1989) Physiology and biochemistry of flower petal senescence. Hort Rev 11: 15-43 Bovy AG, Angenent GC, Dons HJM, van Altvorst AC (1999) Heterologous expression of the Arabidopsis etr1-1 allele inhibits the senescence of carnation flowers. Mol Breeding 5: 301-308 Buchanan-Wollaston V (1994) Isolation of cDNA clones for genes that are expressed during leaf senescence in Brassica napus identification of a gene encoding a sensescence-specific metallothionein-like protein. Plant Physiol 105: 839-846 Bui AQ, O'Neill SD (1998) Three 1-aminocyclopropane-1-carboxylate synthase genes regulated by primary and secondary pollination signals in orchid flowers. Plant Physiol 116: 419-428 Buquez A, SA Corbet (1991) Do flowers resorb nectar? Funct Ecol 5: 369-379 Cameron AC, Reid MS (1981) The use of the silver thiosulfate anionic complex as a foliar spray .2. prevention of shattering in potted geraniums. HortScience 16: 405-405 Cantoni GL (1975) Biological methylation: selected aspects. Ann Rev Biochem 44: 435-451 Chae HS, Faure F, Kieber JJ (2003) The eto1, eto2, and eto3 mutations and cytokinin treatment increase ethylene biosynthesis in Arabidopsis by increasing the stability of ACS protein. Plant Cell 15: 545-559 Chao QM, Rothenberg M, Solano R, Roman G, Terzaghi W, Ecker JR (1997) Activation of the ethylene gas response pathway in Arabidopsis by the nuclear protein ETHYLENE-INSENSITIVE3 and related proteins. Cell 89: 1133-1144 Chang C, Kwok SF, Bleecker AB, Meyerowitz EM (1993) Arabidopsis ethylene-response gene ETR1 similarity of product to 2-component regulators. Science 262: 539-544 Ciardi J, Tieman DM, Lund ST, Jones JB, Stall RE, Klee HJ(2000) Response to Xanthomonas campestris pv. vesicatoria in tomato involves regulation of ethylene receptor gene expression. Plant Physiol 123: 81-92

PAGE 130

118 Clark DG, Gubrium EK, Barrett JE, Nell TA, Klee HJ (1999) Root formation in ethylene-insensitive plants. Plant Physiol 121:53-59 Coruzzi G, Last R (2000) Amino Acids. In B Buchanan, W Gruissem, R Jones, eds, Biochemistry and Molecular Biology of Plants, Ed 1, American Society of Plant Physiologists, Rockville, MD Culley TM, Weller SG, Sakai AK (2002) Evolution of wind pollination in angiosperms. Trends Ecol Evol 17: 491-491 Daly KC, Durtschi ML, Smith BH (2001) Olfactory-based discrimination learning in the moth, Manduca sexta. J Insect Physiol 47: 375-384 D'Auria JC, Chen F, Pichersky E (2002) Characterization of an acyltransferase capable of synthesizing benzylbenzoate and other volatile esters in flowers and damaged leaves of Clarkia breweri. Plant Physiol 130: 466-476 Dudareva N, Cseke L, Blanc VM, Pichersky E (1996) Evolution of floral scent in Clarkia: Novel patterns of S-linalool synthase gene expression in the C breweri flower. Plant Cell 8: 1137-1148 Dudareva N, Pichersky E (2000) Biochemical and molecular genetic aspects of floral scents. Plant Physiol 122: 627-633 Dudareva N, Murfitt LM, Mann CG, Gorenstein N, Kolosova N, Kish CM, Bonham C, Wood K (2000) Developmental Regulation of Methyl Benzoate Biosynthesis and Emission in Snapdragon Flowers. Plant Cell 12: 949-961 Eisen MB, Brown PO (1999) DNA arrays for analysis of gene expression. Methods Enzymol 303: 179-205 Euler M, Baldwin IT (1996) The chemistry of defense and apparency in the corollas of Nicotiana attenuata. Oecologia 107: 102-112 Fauman EB, Blumenthal RM, Cheng X (1999) Structure and evolution of AdoMet-dependent methyltransferases. In X Cheng, RM Blumenthal, eds, S-adenosylmethionine-Dependent Methyltransferases: Structures and Functions, World Scientific Publishing, Singapore, pp1-38 Feeny PP (1976) Plant apparency and chemical defense. In JW Walace, RL Mansell, eds, Biochemical interaction between plants and insects, Rec Adv Phytochem 10: 1-40 Fraser AM, Mechaber WL, Hildebrand JG (2003) Electroantennographic and behavioral responses of the sphinx moth Manduca sexta to host plant headspace volatiles. J Chem Ecol 29: 1813-1833

PAGE 131

119 Gang DR, Wang J, Dudareva N, Nam KH, Simon JE, Lewinsohn E, Pichersky E (2001) An investigation of the storage and biosynthesis of phenylpropenes in sweet basil. Plant Physiol 125: 539-555 Gilissen LJW, Hoekstra FA (1984) Pollination-induced corolla wilting in Petunia hybrida rapid transfer through through the style of a wilting-inducing substance. Plant Physiol 75: 496-498 Goulson D, Stout JC, Langley JG, Hughes WOH (2000) The identity and function of scent marks deposited by foraging bumblebees. J Chem Ecol 26: 2897-2911 Grant V (1983) The systematic and geographical distribution of hawkmoth flowers in the temperate North American flora. Bot Gaz 144: 439-449 Griesbach RJ, Kamo KK (1996) The effect of induced polyploidy on the flavonols of Petunia Mitchell. Phytochemistry 42: 361-363 Gubrium EK, Clevenger DJ, Clark DG, Barrett JE, Nell TA (2000) Reproduction and horticultural performance of transgenic ethylene-insensitive petunias. J Amer Soc Hort Sci 125: 277-281 Guterman I, Shalit M, Menda N, Piestun D, Dafny-Yelin M, Shalev G, Bar E, Davydov O, Ovadis M, Emanuel M, Wang J, Adam Z, Pichersky E, Lewinsohn E, Zamir D, Vainstein A, Weiss D (2002) Rose scent: Genomics approach to discovering novel floral fragrance-related genes. Plant Cell 14: 2325-2338 Guzman P, Ecker JR (1990) Exploiting the triple response of Arabidopsis to identify ethylene-related mutants. Plant Cell 2: 513-523 Hamilton AJ, GW Lycett, D Grierson (1990) Antisense gene that inhibits synthesis of the hormone ethylene in transgenic plants. Nature 346: 284-287 Hamilton AJ, Bouzayen M, Grierson D (1991) Identification of a tomato gene for the ethylene-forming-enzyme by expression in yeast. Proc Nat Acad Sci USA 88: 7434-7437 Helsper JPFG, Davies JA, HJ Bouwmeester, Krol AF, van Kampen MH (1998) Circadian rhythmicity in emission of volatile compounds by flowers of Rosa hybrida L. cv. Honesty. Planta 207: 88-95 Hoekstra FA, van Roekel GC (1986). Effects of interspecific pollination on stylar ethylene production and flower longevity in Petunia hybrida. Acta Hort 181:105-112 Hoekstra FA, Weges R (1986) Lack of control by early pistillated ethylene of the accelerated wilting of Petuniahybrida flowers. Plant Physiol 80: 403-408

PAGE 132

120 Holden MJ, Marty JA, Singh-Cundy A (2003) Pollination-induced ethylene promotes the early phase of pollen tube growth in Petunia inflata. J Plant Physiol 160: 261-269 Holtorf S, Apel K, Bohlmann H (1995) Comparison of different constitutive and inducible promoters for the overexpression of transgenes in Arabidopsis thaliana. Plant Mol Biol 29: 637-646 Honda K, Omura H, Hayashi N (1998) Identification of floral volatiles from Ligustrum japonicum that stimulate flower visiting by cabbage butterfly. J Chem Ecol 24: 2167-2180 Hua J, Meyerowitz EM (1998) Ethylene responses are negatively regulated by a receptor gene family in Arabidopsis thaliana. Cell 94: 261-271 Ichimura K, Suto K (1998) Role of ethylene in acceleration of flower senescence by filament wounding in Portulaca hybrid. Physiol Plant 104: 603-607 Jakobsen HB, Olsen CE (1994) Influence of climatic factors on emission of flower volatiles in situ. Planta 192: 365-371 Jerardo A (2003) Floriculture and Nursery Crops Outlook Sept 17, 2003. United States Department of Agriculture Jones KN, Reithel JS, Irwin RE (1998) A trade-off between the frequency and duration of bumblebee visits to flowers. Oecologia 117: 161-168 Jones MB, Mansfield TA (1975) Circadian rhythms in plants. Sci Prog Oxf 62: 103-125 Jones ML (2002) Ethylene responsiveness in carnation styles is associated with stigma receptivity. Sex Plant Reprod 15: 107-112 Jones ML, Woodson WR (1997) Pollination-induced ethylene in carnation Role of stylar ethylene in corolla senescence. Plant Physiol 115: 205-212 Jones ML, Langston BJ, Johnson F (2003) Pollination-induced senescence of ethylene sensitive and insensitive petunias. In M Vendrell, H Klee, JC Pech, F Romojaro, eds, Biology and Biotechnology of the Plant Hormone Ethylene III, IOS Press, Amsterdam, The Netherlands, pp 324-326 Jorgensen RA, Cluster PD, English J, Que QD, Napoli CA (1996) Chalcone synthase cosuppression phenotypes in petunia flowers: Comparison of sense vs antisense constructs and single-copy vs complex T-DNA sequences. Plant Mol Biol 31: 957-973 Kevan P, Giurfa M, Chittka L (1996) Why are there so many and so few white flowers? Trends Plant Sci 1: 280-284

PAGE 133

121 Kieber JJ, Rothenburg M, Roman G, Feldmann KA, Ecker JR (1993) CTR1,a negative regulator of the ethylene response pathway in Arabidopsis, encodes a member of the raf family of protein kinases. Cell 72: 427-441 Klee H (2002) Control of ethylene-mediated processes in tomato at the level of receptors. J Exp Bot 53: 2057-2063 Kneissl ML, Diekman J (1996) The tomato E8 gene influences ethylene biosynthesis in fruit but not in flowers. Plant Physiol 112: 537-547 Knight LE, Rose RC, Crocker W (1910) Effects of various gases and vapors upon etiolated seedlings of sweet pea. Science 31: 635-636 Knoester M, van Loon LC, van den Heuvel J, Hennig J, Bol JF, Linthorst HJM (1998) Ethylene-insensitive tobacco lacks nonhost resistance against soil-borne fungi. Proc Natl Acad Sci USA 95: 1933-1937 Knudsen JT, Tollsten L (1993) Trends in floral scent chemistry in pollination syndromes: floral scent composition in moth pollinated taxa. Bot J Linn Soc 113: 263-284 Knudsen JT, Tollsten L, Bergstrom LG (1993) Review article number 76 Floral scents a checklist of volatile compounds isolated by head-space techniques. Phytochemistry 33: 253-280 Kolosova N, Gorenstein N, Kish CM, Dudareva N (2001) Regulation of circadian methyl benzoate emission in diurnally and nocturnally emitting plants. Plant Cell 13: 2333-2347 Lanahan MB, Yen HC, Giovannoni JJ, Klee HJ (1994) The never ripe mutation blocks ethylene perception in tomato. Plant Cell 6: 521-530. Lange BM, Wildung MR, Stauber EJ, Sanchez C, Pouchnik D, Croteau R (2000) Probing essential oil biosynthesis and secretion by functional evaluation of expressed sequence tags from mint glandular trichomes. Proc Natl Acad Sci USA 97: 2934-2939 Lavy M, Zuker A, Lewinsohn E, Larkov O, Ravid U, Vainstein A, Weiss D (2002) Linalool and linalool oxide production in transgenic carnation flowers expressing the Clarkia breweri linalool synthase gene. Mol Breeding 9: 103-111 Lawless HT, Heymann H (1998) Sensory Evaluation of Food Principles and Practices. Chapman and Hall Publishers; New York, New York, pp 116-139 Lawton KA, Raghothama KG, Goldsrough PB, Woodson WR (1990) Regulation of senescence-related gene-expression in carnation flower petals by ethylene. Plant Physiol 93: 1370-1375

PAGE 134

122 Lohman KN, Gan SS, John MC, Amasino RM (1994) Molecular analysis of natural leaf senescence in Arabidopsis-thaliana. Physiol Plant 92: 322-328 Lorenzo O, Piqueras R, Sanchez-Serrano JJ, Solano R (2003) ETHYLENE RESPONSE FACTOR1 integrates signals from ethylene and jasmonate pathways in plant defense. Plant Cell 15:165-178 Lucker J, Bouwmeester HJ, Schwab W, Blaas J, van der Plas LHW, Verhoeven HA (2001) Expression of Clarkia S-linalool synthase in transgenic Petunia plants results in the accumulation of S-linalyl-beta-D-glucopyranoside. Plant J 27: 315-324 Lemieux B, Aharoni A, Schena M (1998) Overview of DNA chip technology. Mol Breeding 4: 277-289 Mann JD, Mudd SH (1963) Alkaloids and plant metabolism. IV. The tyramine methylpherase of barley roots. J Biol Chem 238: 381-385 McKeon TA, Fernandez-Maculet JC, Yang S-F (1995) Biosynthesis and metabolism of ethylene. In Plant Hormones Physiology, Biochemistry, and Molecular Biology, ed, PJ Davies, Kluwer Academic Publishers Dordrecht, The Netherlands, pp 118-139 Mitchell AZ, Hanson MR, Skvirsky RC, Ausubel FM (1980) Anther culture of Petunia genotypes with high-frequency of callus, root, or plantlet formation. Z Pflanzenphysiol 100: 131-146 Moffatt BA, Weretilnyk EA (2001) Sustaining S-adenosyl-L-methionine-dependent methyltransferase activity in plant cells. Physiol Plant 113: 435-442 Motten AF, Antonovics J (1992) Determinants of outcrossing rate in a predominantly self-fertilizing weed, Datura stramonium (Solanaceae). Amer J Bot 79: 419 427 Motten AF, Stone JL (2000) Heritability of stigma position and the effect of stigmaanther separation on outcrossing in a predominantly self-fertilizing weed, Datura stramonium (Solanaceae). Amer J Bot 87:339-347 Murfitt LM, Kolosova N, Mann CJ, Dudareva N (2000) Purification and characterization of S-adenosyl-L-methionine: Benzoic acid carboxyl methyltransferase, the enzyme responsible for biosynthesis of the volatile ester methyl benzoate in flowers of Antirrhinum majus. Arch Biochem Biophys 382: 145-151 Negre F, Kolosova N, Knoll J, Kish CM, Dudareva N (2002) Novel S-adenosyl-L-methionine : salicylic acid carboxyl methyltransferase, an enzyme responsible for biosynthesis of methyl salicylate and methyl benzoate, is not involved in floral scent production in snapdragon flowers. Arch Biochem Biophys 406: 261-270

PAGE 135

123 Negre F, Kish CM, Boatright J, Underwood BA, Shibuya K, Wagner C, Clark DG, and Dudareva N (in press) Regulation of methylbenzoate emission after pollination in snapdragon and petunia flowers. Plant Cell. ONeill SD (1997) Pollination regulation of flower development. Ann Rev Plant Physiol Plant Mol Biol 48: 547-574 ONeill SD, Nadeau JA, Zhang XS, Bui AQ, Halevy AH (1993) Interorgan regulation of ethylene biosynthetic genes by pollination. Plant Cell 5: 419-432 Panavas T, Pikula A, Reid PD, Rubinstein B, Walker EL (1999) Identification of senescence-associated genes from daylily petals. Plant Mol Biol 40: 237-248 Park KC, Ochieng SA, Zhu JW, Baker TC (2002) Odor discrimination using insect electroantennogram responses from an insect antennal array. Chem Senses 27: 343-352 Patino S, Aalto T, Edwards AA, Grace J (2002) Is Rafflesia an endothermic flower? New Phytol 154: 429-437 Pichersky E, Lewinsohn E, Croteau R (1995) Purification and characterization of S-linalool synthase an enzyme involved in the production of floral scent in Clarkia breweri. Arch Biochem Biophys 316: 803-807 Peiser G, Yang SF (1998) Evidence for 1-(malonylamino) cyclopropane-1-carboxylic acid being the major conjugate of aminocyclopropane-1-carboxylic acid in tomato fruit. Plant Physiol 116: 1527-1532 Pettersson S, Knudsen JT (2001) Floral scent and nectar production in Parkia biglobosa Jacq. (Leguminosae : Mimosoideae). Bot J Linn Soc 135: 97-106 Porat R, Nadeau JA, Kirby JA, Sutter EG, ONeill SD (1998) Characterization of the primary pollen signal in the postpollination syndrome of Phalaenopsis flowers. J Plant Growth Reg 24: 109-117 Pott MB, Pichersky E, Piechulla B (2002) Evening specific oscillations of scent emission, SAMT enzyme activity, and SAMT mRNA in flowers of Stephanotis floribunda. J Plant Physiol 159: 925-934 Procter M, Yeo P, Lack A (1996) The natural history of pollination. Harper Collins, London Quirino BF, Normanly J, Amasino RM (1999) Diverse range of gene activity during Arabidopsis thaliana leaf senescence includes pathogen-independent induction of defense-related genes. Plant Mol Biol 40: 267-278

PAGE 136

124 Raguso RA, Pichersky E (1995) Floral volatiles from Clarkia breweri and C concinna (Onagraceae) recent evolution of floral scent and moth pollination. Plant Syst Evol 194: 55-67 Raguso RA, Levin RA, Foose SE, Holmberg MW, McDade LA (2003) Fragrance chemistry, nocturnal rhythms and pollination "syndromes" in Nicotiana. Phytochemistry 63: 265-284 Reid MS, Fujino DW, Hoffman NE, Whitehead CS (1984) 1-aminocyclopropane-1-carboxylic acid (ACC). The transmitted stimulus in pollinated flowers. J Plant Growth Regul 3: 189-196. Riechmann JL, Meyerowitz EM (1998) The AP2/EREBP family of plant transcription factors. Biol Chem 379: 633-646 Riov J, Yang S-F (1982) Autoinhibition of ethylene production in citrus peel disks suppression of 1-aminocyclopropane-1-carboxylic acid synthesis. Plant Physiol 69: 687-690 Rodriguez FI, Esch JJ, Hall AE, Binder BM, Schaller GE, Bleecker AB (1999) A copper cofactor for the ethylene receptor ETR1 from Arabidopsis. Science 283: 996-998 Roelofs W, Comeau A (1971) Sex-pheromone perception: electroantennogram responses of the red-banded leaf roller moth. J Insect Physiol 17: 1969-1982 Roman G, Lubarsky B, Kieber JJ, Rothenberg M, Ecker JR (1995) Genetic-analysis of ethylene signal-transduction in Arabidopsis-thaliana novel mutant loci integrated into a stress-response pathway. Genetics 139: 1393-1409 Ross JR, Nam KH, D'Auria JC, Pichersky E (1999) S-adenosyl-L-methionine : salicylic acid carboxyl methyltransferase, an enzyme involved in floral scent production and plant defense, represents a new class of plant methyltransferases. Arch Biochem Biophys 367: 9-16 Rubinstein B (2000) Regulation of cell death in flower petals. Plant Mol Biol 44: 303-318 Sato T, Theologis A (1989) Cloning the mRNA encoding 1-aminocyclopropane-1-carboxylate synthase, the key enzyme for ethylene biosynthesis in plants. Proc Natl Acad Sci USA 86: 6621-6625 Savin KW, Baudinette SC, Graham MW, Michael MZ, Nugent GD, Chandler SF, Cornish EC (1995) Antisense ACC oxidase RNA delays carnation petal senescence. HortScience 30: 970-972 Schaller GE, Bleecker AB (1995) Ethylene-binding sites generated in yeast expressing the Arabidopsis ETR1 gene. Science 270: 1809-1811

PAGE 137

125 Schaller GE, Ladd AN, Lanahan MB, Spanbauer JM, Bleecker AB (1995) The ethylene response mediator ETR1 from Arabidopsis forms a disulfide-linked dimer. J Biol Chem 270: 12526-12530 Schena M, Shalon D, Davis RW, Brown PO (1995) Quantitative monitoring of gene expression patterns with a complementary DNA microarray. Science 270: 467-470 Schenk PM, Kazan K, Wilson I, Anderson JP, Richmond T, Somerville SC, Manners JM (2000) Coordinated plant defense responses in Arabidopsis revealed by microarray analysis. Proc Natl Acad Sci USA 97: 11655-11660 Schiestl FP, Ayasse M, HF Paulus, Erdmann D, Francke W (1997) Variation of floral scent emission and post-pollination changes in individual flowers of Ophrys sphegodes subsp. sphegodes. J Chem Ecol 23: 2881-2895 Schiestl FP, Ayasse M (2001) Post-pollination emission of a repellent compound in a sexually deceptive orchid: a new mechanism for maximising reproductive success? Oecologia 126: 531-534 Schmelz EA, Alborn HT, Tumlinson JH (2001) The influence of intact-plant and excised-leaf bioassay designs on volicitinand jasmonic acid-induced sesquiterpene volatile release in Zea mays. Planta 214: 171-179 Schmelz EA, Engelberth J, Alborn HT, ODonnell P, Sammons M, Toshima H, Tumlinson JH (2003) Simultaneous analysis of phytohormones, phytotoxins, and volatile organic compounds in plants. Proc Natl Acad Sci USA 100: 10522-10557 Seo HS, Song JT, Cheong JJ, Lee YH, Lee YW, Hwang I, Lee JS, Choi YD (2001) Jasmonic acid carboxyl methyltransferase: A key enzyme for jasmonate-regulated plant responses. Proc Natl Acad Sci USA 98: 4788-4793 Serek M, Tamari G, Sisler EC, Borochov A (1995) Inhibition of ethylene-induced senescence symptoms by 1-methylcyclopropane, a new inhibitor of ethylene action. Physiol Plant 94: 229-232 Shaw JF, Chen HH, Tsai MF, Kuo CI, Huang LC (2002) Extended flower longevity of Petunia hybrida plants transformed with boers, a mutated ERS gene of Brassica oleracea. Mol Breeding 9: 211-216 Shields VDC, Hildebrand JG (2001) Responses of a population of antennal olfactory receptor cells in the female moth Manduca sexta to plant-associated volatile organic compounds. J Comp Physiol A 186: 1135-1151 Singh A, Evensen KB, Kao TH (1992) Ethylene synthesis and floral senescence folling compatible and incompatible pollinations in Petunia-inflata. Plant Physiol 99: 38-45 Sink KC (1984) Petunia. Springer Verlag, Berlin, Germany

PAGE 138

126 Sisler E, Pian A (1973) Effect of ethylene and cyclic olefins on tobacco leaves. Tob Sci 17: 68-72 Sisler E, Serek M (1997) Inhibitors of ethylene responses in plants at the receptor level: Recent developments. Physiol Plant 100:577-582 Slauson LA (2000) Pollination biology of two chiropterophilous agaves in Arizona. Amer J Bot 87: 825-836 Solano R, Stepanova A, Chao Q, Ecker JR (1998) Nuclear events in ethylene signaling: a transcriptional cascade mediated by ETHYLENE-INSENSITIVE3 and ETHYLENE-RESPONSE-FACTOR1. Gene Dev 12: 3703-3714 Spanu P, Reinhardt D Boller T (1991) Analysis and cloning of the ethylene-forming-enzyme from tomato by functional expression of its messenger-RNA in Xenopus laevis oocytes. EMBO J 10: 2007-2013 Spellman PT, Sherlock G, Zhang MQ, Iyer VR, Anders K, Eisen MB, Brown PO, Botstein D, Futcher B (1998) Comprehensive identification of cell cycle-regulated genes of the yeast Saccharomyces cerevisiae by microarray hybridization. Mol Biol Cell 9: 3273-3297 Stead AD (1992) Pollination-induced flower senescence a review. J Plant Growth Regul 11: 13-20 Stout JC, Goulson D (2002) The influence of nectar secretion rates on the responses of bumblebees (Bombus spp.) to previously visited flowers. Behav Ecol Sociobiol 52: 239-246 Tang XY, Woodson WR (1996) Temporal and spatial expression of 1-aminocyclopropane-1-carboxylate oxidase mRNA following pollination of immature and mature Petunia flowers. Plant Physiol 112: 503-511 Tang XY, Gomes AMTR, Bhatia A, Woodson WR (1994) Pistil-specific and ethylene-regulated expression of 1-aminocyclopropane-1-carboxylate oxidase genes in Petunia flowers. Plant Cell 6: 1227-1239 Taylor JE, Whitelaw CA (2001) Signals in abscission. New Phytol 151: 323-339 Thornburg RW, Carter C, Powell A, Mittler R, Rizhsky L, Horner HT (2003) A major function of the tobacco floral nectary is defense against microbial attack. Plant Syst Evol 238: 211-218 Tieman DV, Taylor MG, Ciardi JA, Klee HJ (2000) The tomato ethylene receptors NR and LeETR4 are negative regulators of ethylene response and exhibit functional compensation within a multigene family. Proc Natl Acad Sci USA 97: 5663-5668

PAGE 139

127 Tieman DM, Ciardi JA, Taylor MG, Klee HJ (2001) Members of the tomato LeEIL (EIN3-like) gene family are functionally redundant and regulate ethylene responses throughout plant development. Plant J 26: 47-58 Underwood BA, Shibuya K, Tieman DM, Loucas HM, Dexter RJ, Simkin AJ, Sims CA, Schmelz EA, Klee HJ, Clark DG (2003) Ethylene-regulated floral volatile synthesis in Petunia corollas. submitted. Van der Hoeven R, Ronning C, Giovannoni J, Martin G, Tanksley S (2002) Deductions about the number, organization, and evolution of genes in the tomato genome based on analysis of a large expressed sequence tag collection and selective genomic sequencing. Plant Cell 14: 1441-1456 van Doorn WG (1997) Effects of pollination on floral attraction and longevity. J Exp Bot 48: 1615-1622 van Doorn WG (2001) Categories of petal senescence and abscission: A re-evaluation. Ann Bot London 87: 447-456 van Doorn WG (2002) Does ethylene treatment mimic the effects of pollination on floral lifespan and attractiveness? Ann Bot London 89: 375-383 Van Leeuwen W, Hagenfoorn MJM, Ruttink T, Van Poecke R, Van Der Plas LHW (2000) Patterns of green fluorescent protein expression in transgenic plants. Plant Mol Biol Rep 18: 143-144 Van Riper W (1960) Does a hummingbird find its way to nectar through its sense of smell? Sci Amer 202: 157-166 Verdonk JC, de Vos CHR, Verhoeven HA, Haring MA, van Tunen AJ, Schuurink RC (2003) Regulation of floral scent production in petunia revealed by targeted metabolomics. Phytochemistry 62: 997-1008 Vogel JP, Woeste KW, Theologis A, Kieber JJ (1998) Recessive and dominant mutations in the ethylene biosynthetic gene ACS5 of Arabidopsis confer cytokinin insensitivity and ethylene overproduction. Plant Physiol 119: 521-530 Vroege PW, Stelleman P (1990) Insecet and wind pollination in Salix repens L. and Salix caprea L. Israel J Bot 39: 125-132 Wan CY, Wilkins TA (1994) A modified hot borate method significantly enhances the yield of high-quality RNA from cotton (Gossypium hirsutum L.). Anal Biochem 223: 7-12 Wang JH, Dudareva N, Bhakta S, Raguso RA, Pichersky E (1997) Floral scent production in Clarkia breweri (Onagraceae) .2. Localization and developmental modulation of the enzyme S-adenosyl-L-methionine:(iso)eugenol O-methyltransferase and phenylpropanoid emission. Plant Physiol 114: 213-221

PAGE 140

128 Wang JH, Pichersky E (1998) Characterization of S-adenosyl-L-methionine: (iso)eugenol O-methyltransferase involved in floral scent production in Clarkia breweri. Arch Biochem Biophys 349: 153-160 Wang JH, Pichersky E (1999) Identification of specific residues involved in substrate discrimination in two plant O-methyltransferases. Arch Biochem Biophys 368: 172-180 Wang KL-C, Li H, Ecker JR (2002) Ethylene biosynthesis and signaling networks. Plant Cell S131-S151 Waser NM, Chittka L, Price MV, Williams NM, Ollerton J (1996) Generalization in pollination systems, and why it matters. Ecology 77: 1043-1060 Wesley SV, Helliwell CA, Smith NA, Wang MB, Rouse DT, Liu Q, Gooding PS, Singh SP, Abbott D, Stoutjesdijk PA, Robinson SP, Gleave AP, Green AG, Waterhouse PM (2001) Construct design for efficient, effective and high-throughput gene silencing in plants. Plant J 27 (6): 581-590 Whitehead CS, Halevy AH, Reid MS (1984) Roles of ethylene and 1-Aminocyclopropane-1-carboxylic acid in pollination and wound-induced senescence of Petunia-hybrida flowers. Physiol Plant 61: 643-648 Whitelaw CA, Lyssenko NN, Chen L, Zhou D, Mattoo AK, Tucker ML (2002) Delayed Abscission and Shorter Internodes Correlate with a Reduction in the Ethylene Receptor LeETR1 Transcript in Transgenic Tomato. Plant Physiol. 128: 978-987 Wilkinson JQ, Lanahan MB, Yen HC, Giovannoni JJ, Klee HJ (1995) An ethylene-inducible component of signal-transduction encoded by never-ripe. Science 270: 1807-1809 Wilkinson JQ, Lanahan MB, Clark DG, Bleecker AB, Chang C, Meyerowitz EM, Klee HJ (1997) A dominant mutant receptor from Arabidopsis confers ethylene insensitivity in hererologous plants. Nat Biotech 15: 444-447 Williams CS (1998) The identity of the previous visitor influences flower rejection by nectar collecting bees. Animal Behavior 56: 673-681 Wolff D, Braun M, Liede S (2003) Nocturnal versus diurnal pollination success in Isertia laevis (Rubiaceae): A sphingophilous plant visited by hummingbirds. Plant Biol 5: 71-78 Woltering EJ, van Doorn WJ (1988) Role of ethylene in senescence of petals morphological and taxonomical relationships. J Exp Bot 39: 1605-1616 Woodson WR (1987) Changes in protein and mRNA populations during carnation petal senescence. Physiol Plant 71: 495-502

PAGE 141

129 Woodson WR, Park KY, Drory A, Larsen PB, Wang H (1992) Expression of ethylene biosynthetic pathway transcripts in senescing carnation flowers. Plant Physiol 99: 526-532 Woodson WR, Brandt AS, Itzhaki H, Maxon JM, Park KY, Wang H (1993) Regulation and function of flower senescence-related genes. Acta Hort 336: 41-46 Wright GA, Skinner BD, Smith BH (2002) Ability of Honeybee, Apis mellifera, to detect and discriminate odors of varieties of canola (Brassica rapa and Brassica napus) and snapdragon flowers (Antirrhinum majus). J Chem Ecol 28: 721-740 Xu Y, Chang PF, Liu D, Narasimhan ML, Raghothama ML, Hasegawa KG, Bressan RA (1994) Plant defense genes are synergistically induced by ethylene and methyl jasmonate. Plant Cell 6: 1077-1085 Yang S-F (1987) Regulation of biosynthesis and action of ethylene. Acta Hort 201: 53-59 Yang Q, Reinhard K, Schiltz A, Matern U (1997) Characterization and heterologous expression of hydroxycinnamoyl/benzoyl-CoA: anthranilate N-hydroxycinnamoyl/benzoyltransferase from elicited cultures of carnation, Dianthus caryophyllus L. Plant Mol Biol 35: 777-789 Young HJ (2002) Diurnal and nocturnal pollination of Silene alba (Caryophyllaceae). Amer J Bot 89: 433-440 Yu YB, Adams DO, Yang S-F (1979) 1-aminocyclopropane-1-carboxylate synthase, a key enzyme in ethylene biosynthesis. Arch Biochem Biophys198: 280-286 Zhang XS, ONeill SD (1993) Ovary and gametophyte development are coordinately regulated by auxin and ethylene following pollination. Plant Cell 5: 403-418 Zubieta C, Ross JR, Koscheski P, Yang Y, Pichersky E, Noel JP (2003) Structural basis for substrate recognition in the salicylic acid carboxyl methyltransferase family. Plant Cell 15: 1704-1716

PAGE 142

BIOGRAPHICAL SKETCH Beverly Ann Underwood was born in South Miami, Florida, on November 4, 1977. She was raised in the Redland area by her father and mother, Ronnie and Linda Johnson, with her two older sisters, Susan and Diana. In their life there, her family had a tropical fruit grove and enjoyed many aspects of living in a sub-tropical environment including snorkeling and diving for lobster. After graduating from high school, she came to Gainesville to attend the University of Florida. During her undergraduate years, she enjoyed working at the University of Florida Tropical Research and Education Center during the summers and taking botany classes during the school year. She graduated with a Bachelor of Science in botany and minor in plant molecular and cellular biology in August 1999. Seeking a further education, she entered the Plant Molecular and Cellular Biology Ph.D. program with Dr. David G. Clark as her advisor. Under the guidance of Dr. Clark, she studied gene regulation and floral fragrance regulation in Petunia x hybrida flowers in response to ethylene. During her graduate career, she met her husband, Stuart, a recent Ph.D. graduate of the University of Florida Microbiology Department. They were married June 29, 2002, in West Jefferson, North Carolina, and look forward to a long and happy life together. 130


Permanent Link: http://ufdc.ufl.edu/UFE0002394/00001

Material Information

Title: Effects of Ethylene on Floral Fragrance of Petunia x hybrida 'Mitchell Diploid'
Physical Description: Mixed Material
Copyright Date: 2008

Record Information

Source Institution: University of Florida
Holding Location: University of Florida
Rights Management: All rights reserved by the source institution and holding location.
System ID: UFE0002394:00001

Permanent Link: http://ufdc.ufl.edu/UFE0002394/00001

Material Information

Title: Effects of Ethylene on Floral Fragrance of Petunia x hybrida 'Mitchell Diploid'
Physical Description: Mixed Material
Copyright Date: 2008

Record Information

Source Institution: University of Florida
Holding Location: University of Florida
Rights Management: All rights reserved by the source institution and holding location.
System ID: UFE0002394:00001


This item has the following downloads:


Full Text












EFFECTS OF ETHYLENE ON FLORAL FRAGRANCE OF
PETUNIA XHYBRIDA 'MITCHELL DIPLOID'
















By

BEVERLY ANN UNDERWOOD


A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY

UNIVERSITY OF FLORIDA


2003

































Copyright 2003

by

Beverly Ann Underwood


































This document is dedicated to my husband, Stuart, my family, and my faithful feline
Pookey.















ACKNOWLEDGMENTS

I would like to thank my advisor and mentor, Dr. David G. Clark, for his insight,

guidance, and patience through my program. His positive attitude and excitement for

science and life have been inspirational. I truly appreciate all that he has done for me.

I extend deep appreciation to Dr. Harry Klee in his continued guidance, patience,

and generosity through my graduate program. I also would like to thank Dr. Karen Koch,

Dr. Don McCarty, and Dr. Jim Barrett. Without their guidance, time, and generosity this

work would not have been possible. Your influence will have lasting effects on my life

and my future in science.

Special thanks to Dr. Kenichi Shibuya for his friendship, lessons, humor, making

wonderful cDNA libraries, and showing me together with Holly that 40 RNA gel blot

hybridizations are no problem. I thank Holly Loucas for her friendship, all of her help

through the years with all aspects of the lab, plant transformations, and interpretations. I

thank Kris Barry and Jason Jandrew for their friendship, kindness, and sharing and help

with plants. I thank Jenny Davis for her kindness, sharing and help with plants, and leave

my best wishes to her in her research with the many petunia sequences, wherever that

path may go. I thank Rick Dexter for all of his help with plants, scent panels, lab work,

outright sarcasm, and friendship.

I especially want to thank Dr. Denise Tieman. I will never forget her

thoughtfulness, generosity, and wisdom. She really had a huge impact on me at many

levels and I will never forget her true kindness. Also, I wish to thank Dr. Joseph Ciardi









for teaching me from the beginning, showing me how to clone, and to have a lot of fun

with all of it. I thank all of the past and present members of Dr. Clark's and Dr. Klee's

laboratories and to the Floriculture group for their friendship, perspectives, and help with

my research. I truly appreciate and am honored to have worked with all of you.

I thank my husband Stuart, my Mom, my Dad, and sisters for all of their support,

unconditional love, and patience with me when I needed it. I especially thank my faithful

cat Pookey for her complete understanding, companionship, and company next to the

computer at all hours.
















TABLE OF CONTENTS
page

A C K N O W L E D G M E N T S ................................................................................................. iv

LIST OF TABLES ......... ....... ................................................ ix

LIST OF FIGURES ............................... ... ...... ... ................. .x

ABSTRACT ........ .............. ............. ...... ...................... xi

CHAPTER

1 IN TRODU CTION ................................................. ...... .................

2 LITER A TU R E REV IEW ............................................................. ....................... 4

E th y le n e ................................................................................. 4
Ethylene B iosynthesis.................................................................... 4
R regulation of Ethylene B iosynthesis..................................... .................................... 5
Ethylene Perception and Signal Transduction................................... .....................7
F loral S en escen ce .................................... ................................ ..... ........... 1 1
Pollination-Regulation of Ethylene Biosynthesis.....................................................12
Prevention of Ethylene Effects ........................................................ ............. 15
Gene Expression During Senescence ........................................ ...... ............... 18
Flow ers and Pollination .................. ........................... .... .. ... .. ........ .... 19
Floral Fragrance and Insects ......................................................... .............. 21
Fragrance C om position........................................................ ........................... 24
Fragrance Biochemistry Brief History................................ ...... ............... 24
Fragrance Biochemistry Ado-Met Dependent Methyltransferases .......................25
F ragrance R regulation ......... ............................................................... .... .... .... .. 28
Fragrance Emission in Petunia ......................................................................... 29
Genetic Engineering for Improved Flower Fragrance..............................................29
P e tu n ia ............................................................................... 3 0
R research O bjectives.......... ................................................................. ......... ....... 32

3 MICROARRAY ANALYSIS OF ETHYLENE-INDUCED FLORAL
SENESCENCE IN PETUNIA ............................................................................. 35

In tro d u ctio n .......................................................................................3 5
R results and D discussion ............................... .. .................................... ............. 38
Establishment of a Petunia EST Collection................................................38









E S T A n aly sis ....................... ........... ... ...... ....................... .............. 3 9
Microarray Expression Analysis of Ethylene-Treated Flowers ..........................41
M materials and M methods ....................................................................... ..................4 5
Plant Cultural Conditions ............................................................................... 45
Tissue Collections and RNA Extractions.........................................................46
B ioinform atic A nalysis............................................................. ............... 47
Functional Categorization of ESTs .......................................... ............... 48
M icroarray Fabrication ............................... ....... ......... ........................... 48
Probe Synthesis and Microarray Experimental Procedures .............................50
V erification of M icroarray D ata................................... .................................... 52

4 ETHYLENE-REGULATED FLORAL VOLATILE SYNTHESIS IN PETUNIA
C O R O L L A S ...............................................................................................................6 1

In tro d u ctio n ...................................... ................................................ 6 1
R esu lts ................................. ................... .. ...... .... .. ....... ....... ...............66
RNAi PhBSMTReduces Methyl Benzoate Emission and Changes Floral
Fragrance in Petunia...... ................................ ....... ...............66
PhBSMT1 and PhBSMT2 Are Spatially and Temporally Regulated in Petunia
F low ers................ ................... ........ .. ....... ... ... ...............6 7
Substrate Regulation in Response to Pollination and Ethylene Treatments........69
Volatile Emission Is Down-Regulated in Response to Exogenous Ethylene
an d P o llin atio n ................................................. ................ 7 0
D discussion ...................................... ......... .... ..... ........ ......................71
Transgenic PhBSMTRNAi Plants Have Lower Methyl Benzoate Emission .....72
Ethylene Regulates PhBSMT Expression in Petunia Floral Organs .................73
Pollination and Ethylene Treatments Down-Regulate Floral Volatiles in
P e tu n ia ...................................... ............ ................ ................ 7 6
M materials and M methods ....................................................................... ..................79
P lant M material ......................................................................79
cD N A Isolation ................................................... ................ 7 9
Tissue Treatm ents and Collections ........................ ...................................... 80
Spatial and Temporal Analysis of mRNA Expression in Flowers ...................... 81
Generation of Transgenic PhBSMTRNAi Petunias........................................... 82
V olatile Collection and A nalysis...................... ..... ......... ............... ... 83
Benzoic Acid & Salicylic Acid Extraction and Quantification...........................84
H um an O lfaction Panels............................................. ............................. 84

5 GENERAL DISCUSSION AND CONCLUSIONS ..............................................97

Building Fundamental Tools for Genomic Studies .......................................... 97
Microarray Analysis of Ethylene Regulated Genes.................................................98
Ethylene Regulates Floral Fragrance in Petunia .....................................................99









APPENDIX SUPPLEMENTAL DATA FOR CHAPTER 3.....................................102

L IST O F R E F E R E N C E S ........................................................................ .................... 116

B IO G R A PH IC A L SK E T C H .................................................................. ....................130



















































viii
















LIST OF TABLES


Table page

3-1 Sequence characteristics of petunia floral cDNA libraries. ..............................56

3-2 Number of clones from cDNA libraries not included in the functional analysis.....56

3-3 Contigs from each of the cDNA libraries with the greatest number of clones. .......56

3-4 Number of cDNAs putatively differentially regulated by ethylene .......................57

3-5 Expression patterns in petunia corollas of differentially regulated cDNAs............58

4-1 Benzoic acid, salicylic acid, and cinnamic acid levels after ethylene treatment......96

4-2 Benzoic acid, salicylic acid, and cinnamic acid after pollination ..........................96

A-1 Functional categorization of cDNAs identified from microarray analysis ..........103
















LIST OF FIGURES


Figure p

1-1 Ethylene biosynthesis in plants ........................................... ......................... 33

1-2 Proposed model of ethylene signal transduction in Arabidopsis thaliana ..............34

3-1 Flowchart of petunia EST project and expression studies .................................53

3-2 Outline of microarray experimental procedures........................................... 54

3-3 Putative functional categories of ESTs ......................................... ...........................55

4-1 PhBSMTRNAi reduces MeBA emission and PhBSMTmRNA............................85

4-2 PhBSM TRNAi reduces M eBA emission only .............. ............ .....................86

4-3 Volatile emission patterns from MD floral organs........................................87

4-4 PhBSMTmRNA expression after ethylene treatment............................................88

4-5 PhBSMT] mRNA expression in MD and 44568 after pollination........................... 89

4-6 PhBSMT2 mRNA expression in MD and 44568 after pollination.........................90

4-7 MeBA emission after ethylene treatment (A) and pollination (B).........................91

4-8 Rhythmic emission of MeBA and rhythmic expression of PhBSMTs...................92

4-9 Rhythmic emission of MeBA and rhythmic expression of PhBSMTs...................93

4-10 Regulation of volatiles from MD and 44568 flowers in response to ethylene.........94

4-11 Regulation of volatiles from MD and 44568 flowers in response to pollination
(x) and non-pollination (N P) ........................................................ ............. 95















Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy

EFFECTS OF ETHYLENE ON FLORAL FRAGRANCE OF
PETUNIA XHYBRIDA 'MITCHELL DIPLOID'



By

Beverly Ann Underwood

December 2003

Chair: David G. Clark
Major Department: Plant Molecular and Cellular Biology

Ethylene is involved with regulating many plant processes including stress

responses, fruit ripening, and flower senescence. In these studies, the effects of ethylene

on gene expression in Petunia x hybrida 'Mitchell Diploid' (MD) flowers were examined

using Petunia cDNA microarrays. The process of developing tools for genomic studies

through microarrays is described and is followed by the characterization of two genes

isolated from the microarray experiments.

Two cDNAs from Petunia hybrida encoding benzoic acid:salicylic acid carboxyl

methyltransferases (PhBSMT1 and PhBSMT2) were identified as ethylene down-

regulated in petunia flowers. Literature reports show that in vitro these and orthologous

carboxyl methyltransferases catalyze the synthesis of methyl benzoate, a ubiquitous floral

scent volatile in flowering plants. Expression of the genes inplanta was reduced by

RNA interference and these plants emitted methyl benzoate less than 10% of wild type









levels, thus demonstrating BSMT function. Human olfaction panels with flowers of these

plants demonstrated that the floral fragrance was significantly changed relative to wild

type. Wild type and transgenic ethylene-insensitive petunias, 35S::etrl-1, were used to

examine ethylene regulation of methyl benzoate emission. The expression of both

PhBSMT] and PhBSMT2 was down-regulated by exogenous ethylene and by pollination-

induced ethylene in the floral organs of wild type, but not 35S::etrl-]. Ethylene

treatments also reduced emission of other major floral volatiles in wild type, but not

emission from 35S::etrl-] flowers. These results demonstrate a unique role for ethylene

in regulation of floral fragrance in petunia.














CHAPTER 1
INTRODUCTION

Fragrant and colorful flowers have long attracted the attention of people and

animals. In addition to providing an aesthetically pleasing display for humans, they are a

food source for many living creatures. Flowers evolved as an efficient means to facilitate

sexual reproduction and have evolved into thousands of different forms, using a

fascinating diversity of strategies to coordinate pollination and subsequent fertilization.

Floral diversity in the plant kingdom is reflected by a continuum of species that vary in

fragrance, floral color, gross morphology, and phenology. Plants invest energy in the

production of chemicals that make flowers visually attractive and also chemicals that are

perceived by olfaction to attract pollinators. Together these characteristics can increase

the attractiveness of the plant to pollinators flying in the area, luring them to the flower

and thus increasing the chance for a pollination event. One extreme example of a

pollinator attraction strategy is observed in many species of orchids. These flowers have

heavily reinforced floral organs with thick, waxy cuticles that preserve the flower for

many weeks, sometimes months, until the flowers are pollinated (O'Neill et al., 1993;

Bernhardt, 1999). Flowers of one species of orchid, Grammatophyllum multiflorum,

have been recorded to live as long as nine months (Bernhardt, 1999). While most plants

do not exhibit the extreme flower longevity observed in some orchid species, there are

many plants that exhibit senescence and abscission of organs involved with pollinator

attraction in response to pollination. In many plant species the senescence and abscission

responses are mediated by the phytohormone ethylene (Stead, 1992).









Broad taxonomic studies on plants that exhibit senescence have shown that diverse

genera exhibit senescence in response to pollination (Woltering and van Doom, 1988;

van Doom, 2001; 2002). In these plants, pollination often accelerates senescence through

the production and subsequent perception of ethylene in the flower (O'Neill, 1997). In

response to this ethylene and other pollination signals, the flower is initiated into post-

pollination development and transition into fruit and seed development.

The purpose of these studies was to investigate processes that are affected by

ethylene in petunia flowers. An understanding of these processes is important for

understanding plant physiology and identification of ways to circumvent or prevent the

effects of ethylene for commercial purposes. This research specifically addressed the

effect of ethylene on gene expression in petunia flowers. The process of generating the

necessary genetic tools for addressing the effects of ethylene on gene expression is

described and data from these experiments are presented. This is followed by

characterization of expression of two genes that were ethylene down regulated in the

flowers. These two genes were similar to salicylic acid carboxyl methytransferases that

have been reported in vitro to catalyze the synthesis of methyl salicylate and methyl

benzoate, two ubiquitous floral volatiles. The two petunia genes were named benzoic

acid: salicylic acid carboxyl methyltransferase (PhBSMT] and PhBSMT2) based on in

vivo and in vitro data presented here and in Negre et al. (2003). The ethylene down-

regulation ofPhBSMT] and PhBSMT2 was verified by examination of expression in

flowers of transgenic ethylene insensitive petunias generated by Wilkinson et al (1997)

that constitutively express an Arabidopsis thaliana etrl-1 ethylene receptor mutant allele

(35S::etrl-1). Corresponding emission patterns of methyl benzoate and other major









floral volatiles emitted by petunia flowers in response to ethylene and pollination are

presented.














CHAPTER 2
LITERATURE REVIEW

Ethylene

Ethylene is a simple, two-carbon, gaseous plant hormone that is involved in many

aspects of plant development. It has roles in regulating seedling growth, vascular

differentiation, cell elongation, root development, stress responses, fruit ripening,

abscission, and senescence (Abeles et al., 1992; Wang et al., 2002). Since ethylene is

involved in such a diverse array of plant processes in many plant species, it has been of

considerable interest for biological studies. One of the classical ethylene responses used

to identify and study many components of ethylene biosynthesis and signaling is the

seedling triple response. The seedling triple response is the development of a short, thick

hypocotyl with a pronounced apical hook that develops in the presence of ethylene

(Knight et al., 1910). Using the triple response as a genetic screen, mutants in ethylene

responses, biosynthesis, and signal transduction have been identified (reviewed in Wang

et al., 2002). These mutants largely group into two categories: lack of an ethylene

response in the presence of ethylene (ethylene insensitive etr, ein2, ein3) and ethylene

response in the absence of ethylene (eto, ctr). Studies on many plant species demonstrate

that ethylene biosynthesis, perception, and signaling is conserved and highly regulated to

mediate appropriate ethylene responses in the plant life cycle.

Ethylene Biosynthesis

Ethylene is synthesized in plants in a two-step reaction from S-adenosyl-L-

methionine (SAM). SAM is synthesized from methionine and ATP in the methionine-









recycling pathway by SAM synthetase (Adams and Yang, 1979). While SAM is a

precursor to ethylene, SAM is also a widely used substrate and cofactor for other

reactions in plant cells, such as protein synthesis, polyamine biosynthesis, and ubiquitous

methylation reactions (Coruzzi and Last, 2000). The first committed step to ethylene

biosynthesis is synthesis of 1-aminocycloproprane-l-carboxylic acid (ACC) and side

product 5'-methylthioadenosine (MTA) from SAM, which is catalyzed by ACC synthase

(ACS) in the presence of a pyridoxal phosphate cofactor (Yu et al., 1979; Sato and

Theologis, 1989). The second and final step of ethylene biosynthesis is catalyzed by

ACC oxidase (ACO), which oxidizes ACC to ethylene (C2H4) and side products of

carbon dioxide (CO2) and hydrogen cyanide (HCN) (Hamilton et al., 1991; Spanu et al.,

1991). The side products of each of these reactions are further metabolized for sulfur

recycling (MTA) and prevention of cellular toxicity (HCN) (reviewed in Wang et al.,

2002). Genetic evidence for ACO activity was first shown with tomato plants expressing

an antisense ACO transgene (Hamilton et al., 1990). These plants exhibited decreased

ethylene biosynthesis resulting in delayed fruit ripening (Hamilton et al., 1990).

Regulation of Ethylene Biosynthesis

Ethylene synthesis is induced in plants in response to various stresses and

developmental processes (Yang, 1987). For example, during the onset of ripening of

climacteric fruit there is a rapid increase in ethylene synthesis, which coordinates many

of the ripening-associated processes in these fruits (reviewed in Abeles et al., 1992).

Ethylene synthesis is both positively and negatively regulated in plants. It is generally

controlled through regulation of ACS mRNA levels, enzyme activity, and protein

degradation, as well as through conjugation of ACC and control of ACO activity (Wang

et al., 2002).









Since the initial cloning of ACS from zucchini (Sato and Theologis, 1989) and ACO

from tomato (Hamilton et al., 1990), multiple ACS and ACO genes have been identified,

studied in various plant species, and demonstrated to be inducible by various factors.

Expression levels of both ACS and ACO are generally correlated with ethylene

production levels (Acaster and Kende, 1991), and expression of subsets of ACS and ACO

genes have been shown to be ethylene inducible, depending on the tissue and

developmental stage (e.g., Woodson et al., 1992; Bui and O'Neill, 1998; reviewed in

McKeon et al., 1995). The ethylene inducible nature of some ACS and ACO genes

illustrates the primary way autocatalytic ethylene synthesis can be induced. Through

inducible expression of ethylene biosynthetic genes, ethylene triggers synthesis of itself.

However, ethylene synthesis can also be auto-inhibitory. One mechanism for auto-

inhibition of ethylene synthesis takes place through inhibition of ACC synthase activity,

which restricts ACC supply (Riov and Yang, 1982; Abeles et al., 1992). In addition to

ACS and ACO being ethylene-inducible, some ACS and ACO genes are reported to be

auxin inducible (e.g., Bui and O'Neill, 1998). The differential nature of ACS and ACO

gene family regulation likely aids in integration of multiple signals for mediation of

appropriate physiological responses at specific times and stages of plant development.

Additional post-translational control of ethylene biosynthesis has been

demonstrated with the ethylene overproducing mutants (eto mutants) and studies of ACC

regulation. Three eto mutants have been identified to overproduce ethylene compared

with wild type, except in the presence of ethylene synthesis inhibitors (Guzman and

Ecker, 1990; Kieber et al., 1993). Two of the eto mutations identified are disrupted in a

key regulatory domain in carboxyl terminal end of two ACS isoforms (eto2 -ACS5, eto3









-ACS9), resulting in increased ACS stability (Vogel et al., 1998; Chae et al., 2003). This

is thought to account for the increased ethylene synthesis phenotype observed in eto2 and

eto3 mutant plants, thus implicating post-translational control of ACS in regulation of

ethylene synthesis. Whether or not all of the ACS proteins are regulated post-

translationally in this manner remains to be examined. Ethylene production is also

controlled through regulation of ACC levels. Conjugation of ACC to 1-

malonylaminocyclopropane-1-carboxylic acid (MACC) by ACC N-malonyltransferase

regulates levels ofACC. (Yang et al., 1987; Peiser and Yang, 1998). This activity

creates a sink for excess ACC. This activity also provides an additional level of auto-

inhibitory ethylene regulation, as ethylene can increase levels of malonyltransferase

(Abeles et al., 1992), and no role for malonylated ACC has been established (McKeon et

al., 1995). In summary, there is differential control ofACS and ACO transcript

abundance, protein levels, activity, and ACC levels, and these are all potential points of

ethylene synthesis regulation in plants.

Ethylene Perception and Signal Transduction

The genetic components of the ethylene-signaling network have largely been

isolated through mutant screens using the seedling triple response in Arabidopsis

(reviewed in Wang et al., 2002). The signal transduction mutants fall into two broad

categories based on their phenotype: reduced or absence of a response to ethylene

(ethylene insensitive phenotype, etr, ers, ein2, ein3) or ethylene responses in the absence

of ethylene (constitutive triple response, ctr). Epistasis analysis of these mutants has

been used to create a general map of the components in ethylene response signaling

networks (Roman et al., 1995).









Ethylene is perceived through a membrane-bound receptor protein dimer, ETR

(ethylene response) (Chang et al., 1993; Schaller and Bleecker, 1995). The ethylene

receptor was identified through a triple response mutant screen (Bleecker et al., 1988)

and subsequent cloning of ETR and mutant alleles by chromosome walking (Chang et al.,

1993). ETR was shown to be an ethylene receptor genetically, by conferring ethylene

insensitivity in plants transformed with a mutant allele (Chang et al., 1993) and

biochemically by generating ethylene binding sites in yeast transformed with ETR

(Schaller and Bleecker, 1995). Five confirmed ethylene receptors have been identified in

Arabidopsis thaliana (AtETR1, AtERS1, AtETR2, AtEIN4, and AtERS2) (reviewed in

Wang et al., 2002) and six in Lycopersicum esculentum, five of which have been

confirmed to bind ethylene (LeETR1, LeETR2, NR, LeETR4, LeETR5, and LeETR6)

(reviewed in Klee, 2002). Each of the receptor proteins exhibits similarity to bacterial

two-component regulators. Bacterial two-component regulators are known signal

transducers that consist of a sensor and a response regulator, which function together to

regulate cellular responses to environmental stimuli. The amino terminus of ETR, or the

sensor, has hydrophobic stretches that are responsible for membrane localization

(Schaller et al., 1995), dimerization through covalent disulfide bonds (Schaller et al.,

1995), copper co-factor binding (Rodriguez et al., 1999), and ethylene binding (Schaller

and Bleecker, 1995). The carboxyl terminus is divergent among the receptors, as some

have a conserved histidine kinase domain, while others do not, and in some receptors a

response regulator domain is present. The role of each of these domains in ethylene

signaling is not yet clear.









Single mutant lines for each of the ethylene receptors in Arabidopsis have all

exhibited dominant ethylene insensitivity (reviewed in Wang et al., 2002). One of the

mutant alleles, etrl-1, has one missense mutation in the amino terminus, which abolishes

ethylene binding (Schaller and Bleecker, 1995). The etrl-1 lesion is a Cys65 to Tyr

missense mutation (Chang et al., 1993), which disturbs coordination of a Cu2+ cation that

is required for ethylene binding (Rodriguez et al., 1999). Plants expressing mutant alleles

of the ethylene receptors are disrupted in many responses throughout the life cycle of the

plant, including seedling growth, fruit ripening, delayed flower and leaf senescence, and

inhibition of adventitious root formation (Lanahan et al., 1994; Wilkinson et al., 1995,

1997; Clark et al., 1999; Gubrium et al., 2000). The dominant nature of the mutation

indicates that the receptor negatively regulates ethylene responses. The signal

transduction model states that in wild type plants in the absence of ethylene, the ethylene

receptor is signaling, repressing an ethylene response. In the presence of ethylene,

signaling is inactivated and repression of ethylene responses is relieved. In plants

expressing a mutant etr allele, ethylene insensitivity is observed because the mutant

receptors continue to signal, thus repressing an ethylene response even in the presence of

ethylene. Experiments with receptor loss-of-function mutants in Arabidopsis and

antisense lines in tomato have validated this proposed model of receptor action for both

plant species (Hua and Meyerowitz, 1998; Tieman et al., 2000). In Arabidopsis, single

receptor loss-of-function lines do not show a phenotype, while triple and quadruple loss-

of-function mutants exhibit ethylene hypersensitivity (Hua and Meyerowitz, 1998). In

cases of single and double loss-of-function mutants, functional redundancy between the

receptors likely compensates for the loss of the absent receptors, while triple and









quadruple loss-of-function mutants are no longer able to compensate. Due to decreased

levels of the receptors in these plants, signaling through the pathway is presumably

reduced to a level that does not repress ethylene responses.

Research in tomato uncovered some interesting aspects of receptor regulation.

Antisense lines for the NR tomato ethylene receptor exhibit no change in ethylene

sensitivity, while antisense LeETR4 lines exhibit increased ethylene sensitivity. When

receptor mRNA levels were measured in the two lines, it was discovered that LeETR4

expression increased in the NR antisense line compensating for decreased NR expression

and thus wild type levels of ethylene sensitivity were maintained in the antisense NR

plants (Tieman et al., 2000). These results indicate that ethylene sensitivity is regulated

by receptor levels, thus allowing for fine control of an ethylene response.

The ethylene receptor signals to downstream components presumably through a

phosphorylation cascade to repress ethylene responses. The next component that has

been placed in the pathway is CTR (constitutive triple response), a Raf-like protein-

kinase and negative regulator of ethylene responses (Kieber et al., 1993). Plants

homozygous for a ctr null mutation exhibit a constitutive ethylene response phenotype

even in the presence of ethylene biosynthesis and perception inhibitors (Kieber et al.,

1993). Analogous signal cascades in animals and yeast indicate that signaling from CTR

may be propagated via a MAP-kinase cascade (Wang et al., 2002) to EIN2 (ethylene

insensitive) (Roman et al., 1995). EIN2 mutants are recessive and loss-of-function

mutants are completely insensitive to ethylene, indicating EIN2 is required for ethylene

responses and that it positively regulates ethylene responses (Alonso et al., 1999).

Genetic screens for ethylene insensitivity have also led to the discovery of nuclear factors









that have been shown to regulate ethylene responsive gene expression. EIN3 (Ethylene

Insensitive 3) and related EILs (Ethylene Insensitive 3 Like) encode for nuclear localized

transcription factors that regulate ethylene responsive genes (Chao et al., 1997; Solano et

al., 1998). One of the EIN3 positively regulated genes identified is ethylene response

factor (ERF1), which encodes a transcription factor that binds to GCC consensus

ethylene response elements (ERE) in the promoters of ethylene responsive genes (Solano

et al., 1998). Subsequent studies with ERF1 showed that it is induced in response to

infection by necrotrophic fungi via ethylene and jasmonic acid signaling. ERF 1 in turn

regulates expression of a subset of pathogen responsive defense genes (Berrocal-Lobo et

al., 2002; Lorenzo et al., 2003). However, ERF1 does not regulate all ethylene

responsive genes. For example, in Arabidopsis the ethylene sensitive HOOKLESS1 gene

is not induced by ERF1 even though the promoter contains an ERE (Solano et al., 1998).

It is likely that there are ERFs regulating subsets of ethylene responsive genes, as the

ERFs are a part of a large gene family in Arabidopsis (Riechmann and Meyerowitz,

1998).

Floral Senescence

Floral senescence is a genetically regulated process in plants that represents the

terminal phase of flower development. Since this process affects the visual display of the

flower, there have been significant efforts to understand how this process takes place for

the goal of increasing flower longevity for commercial purposes. In many plant species,

this process is developmentally programmed, but is accelerated by pollination induced

ethylene and exposure to ethylene (van Doom, 1997). Additionally, stress and wounding

can cause ethylene synthesis in the flower and therefore induce floral senescence

(Ichimura, 1998; reviewed in Rubenstien, 2000). However, in some plants ethylene does









not accelerate floral senescence and developmentally programmed proteolytic events in

the flower are predominately responsible for hastening senescence (Beileski and Reid,

1992; Stead, 1992).

Pollination-Regulation of Ethylene Biosynthesis

One of the first detectable responses to pollination in many flowers is the

production of ethylene (O'Neill, 1997; Tang and Woodson, 1996). The role of

pollination-induced ethylene in mediating post-pollination changes including flower

senescence and abscission has been implicated in species from many families of plants

(van Doom, 1997). Studies of ethylene biosynthetic gene expression and corresponding

ethylene emission have demonstrated directly that endogenous ethylene synthesis

increases following pollination (O'Neill, 1997; Bui and O'Neill, 1998). Pollination

signals and ethylene have a role in coordinating morphological changes in the flower that

include stigma closure, ovary maturation, ovule differentiation, and perianth senescence

(Zhang and O'Neill, 1993; O'Neill et al., 1993). There are changes in gene expression as

well as post-translational events that take place prior to senescence and are important for

mediating these processes (Lawton et al., 1990; Abeles et al., 1992). These pollination

induced programmed events are thought to have evolved in some plant genera because

after pollination and fertilization the purpose of the petals has been fulfilled. Therefore,

maintenance of a floral structure that is attractive to pollinators is no longer necessary and

the flower must switch to a phase of fruit and seed development (Rubenstein, 2000).

Phalaenopsis, carnation, and petunia flowers all senesce in response to ethylene

and exhibit underlying similarities in post-pollination development. In each of these

flowers, ethylene synthesis is stimulated shortly after pollination through induction of

ACC synthase (ACS) and ACC oxidase (ACO) expression. Both of these enzymes are









encoded by multigene families in these plants and are temporally and spatially regulated

in the flowers (O'Neill et al., 1993; Bui and O'Neill, 1998; Tang et al., 1994; Jones and

Woodson, 1999). There are differences in these flowers in the temporal pattern of

ethylene production that reflect the spatial patterns of ACS and ACO gene expression.

The identity of the initiating signal from the pollination event is still obscure. There is

evidence for multiple phytohormones auxinn, ethylene) and hormone precursors (ACC) in

initiation of the post-pollination syndrome (O'Neill, 1997; Porat et al., 1998). It is

possible that the initiating signal is unique to taxonomic groups of plants.

Bui and O'Neill studied post-pollination responses in Phalaenopsis and proposed a

model for inter-organ regulation of the ethylene biosynthetic genes, corresponding

enzyme activity, and subsequent morphological and developmental changes (Bui and

O'Neill, 1998). The Phalaenopsis pollination model is an excellent model system for

study because morphological changes in the ovary and perianth are triggered by

pollination and not aging. Therefore, there is a clear distinction in changes induced by

pollination (O'Neill et al., 1993). First, pollination initiates expression of ACS2 in the

stigma, which together with basal levels of ACO activity induces ethylene production

(O'Neill et al., 1993; Bui and O'Neill, 1998). Upon the initial synthesis of ethylene,

autocatalytic ethylene synthesis initiated as expression of the ethylene responsive ACS1

gene is induced. Pollination also induces expression of ACS3 in the ovary. However,

very low levels of ACO activity and ethylene emission have been detected from the ovary

(O'Neill et al., 1993), so it is thought that ACC produced here is mostly transported to the

perianth and labellum (Bui and O'Neill, 1998). In the petal organs, ACC together with

diffusible ethylene induces ACS1 and ACO1 expression in the labellum and ACO1









expression in the perianth. This stimulates autocatalytic ethylene production from these

tissues and signals for senescence. The primary signal inducing ACS2 andACS3 in

Phalaenopsis is thought to possibly be auxin (Bui and O'Neill, 1998). Evidence for this

was demonstrated with auxin treatments of the stigmas, which induced expression of

ACS2 in the stigma and ACS3 in the ovary (Bui and O'Neill, 1998). Auxin can also

induce ovary growth and differentiation of the ovules in the absence of pollination

(Zhang and O'Neill, 1993). In summary, the induction of ACS, ACO, and corresponding

ethylene is spatially and temporally regulated in Phalaenopsis by pollination, auxin, and

ethylene, coordinating post-pollination events in each of the floral organs (Bui and

O'Neill, 1998). This type of inter-organ regulation of ethylene-sensitive biosynthetic

genes demonstrates that ethylene sensitivity is also required for eliciting ethylene

responses. However, studies of ethylene receptor regulation in Phalaenopsis have not

been published to date.

The post-pollination senescence response has also been extensively studied in

Petunia. In contrast to Phalaenopsis flowers, petunia flowers exhibit developmental

senescence as well as pollination-accelerated and ethylene-induced senescence

(Whitehead et al., 1984; Wilkinson et al., 1997; Gubrium et al., 2000). Additionally,

pollination-induced ethylene synthesis in Petunia is likely not induced by auxin

(Hoekstra and Van Roekel, 1986). After a compatible pollination, germination of the

pollen begins as early as one hour post-pollination (PP) (Tang and Woodson, 1996). The

stigma produces a large amount of ethylene beginning 2-4 hours PP (Tang and Woodson,

1996; Wilkinson et al., 1997; Jones et al., 2003). Ethylene produced at this time does not

hasten petal senescence (Hoekstra and Weges, 1986) and is thought to promote pollen









tube growth in Petunia inflata (Holden et al., 2003). After pollen tubes have begun to

grow through the style, another burst of ethylene is produced beginning at 12 hours PP

from the stigma+style and ovary. Ethylene is produced from these organs continuously

thereafter, peaking around 24 hours PP (Jones et al., 2003), corresponding with the time

of fertilization (Tang and Woodson, 1996). These later phases of ethylene production are

thought to be responsible for inducing corolla senescence for two reasons. First,

treatment of P. hybrida stigmas with an ethylene action inhibitor, 2,4-norbornadiene,

prevented ethylene synthesis from the stigma, but did not delay senescence of the corolla

(Hoekstra and Weges, 1986). Next, incompatible pollinations in P. inflata do not result

in a second peak of ethylene production or senescence (Singh et al., 1992). While the

early burst of ethylene does not appear to be responsible for petal senescence, it is clear

that there is a wilt-inducing substance that is being transmitted by the stigma+style within

the first five hours after pollination (Gilissen and Hoekstra, 1984). Experimental

evidence for this was shown when the stigma+styles were removed from the flower at

five hours post-pollination and petal senescence was still initiated. Experiments with

radiolabeled ACC are suggestive of ACC as the wilt-inducing substance (Reid et al.,

1984). It is possible that translocated ACC and diffusible ethylene produced in the

stigma+style and ovary are factors in inducing petal senescence in Petunia.

Prevention of Ethylene Effects

Due to the detrimental effect of ethylene on flower longevity and display, there

have been many efforts to find ways to avoid ethylene effects to increase the longevity of

flowers. Increasing flower longevity has tremendous commercial value since annual

floriculture crop sales total billions of dollars worldwide. Estimated sales of petunia

plants in the United States alone in 2002 were approximately 140 million dollars









(Jerardo, 2003). Means of improving floral longevity in ethylene responsive flowers

include chemical applications which inhibit endogenous ethylene production or block

ethylene perception and genetically engineering flowers for decreased ethylene

biosynthesis or perception. Treatments with ethylene biosynthetic inhibitors such as

aminoethoxyacetic acid (AOA) and aminoethoxyvinylglycine (AVG) reduce endogenous

ethylene production (Serek et al., 1995). However, these treatments do not prevent

effects from exogenous ethylene sources and are thus commercially limited. Approaches

that block the perception of ethylene have been used to successfully diminish the effects

of exogenous and endogenously produced ethylene. Silver thiosulfate (STS) has been

used for this purpose for many years (Cameron and Reid, 1982) and works by binding to

the ethylene receptors and likely displacing the Cu2+ cation that is required for ethylene

sensitivity, and thus reducing ethylene binding and subsequent responses (Rodriguez,

1999). Compounds that have similar chemical structures to ethylene and competitively

bind to the receptor have also been used for blocking ethylene responses. This includes

chemicals such as 2,4-norbomadiene (2,4-NBD) (Sisler and Pian, 1973) and 1-

methylcyclopropene (1-MCP) (Serek et al., 1994). The latter is a more recent

development and has been shown to be advantageous to other receptor binding

compounds because it is effective at low concentrations and is non-toxic, whereas 2,4-

NBD is highly noxious and STS is a heavy metal and potential ground water contaminant

(Sisler and Serek, 1997). While these compounds are effective in blocking existing

ethylene receptors, these inhibitors do not block ethylene receptors that are synthesized

after the blocking treatments.









Insensitivity to ethylene has also been generated by genetically engineering plants

for reduced ethylene biosynthesis or altered ethylene signaling. This approach is

advantageous in that no chemical treatments are required and the effects are long lasting.

Decreased ethylene biosynthesis has been engineered in carnation through constitutive

antisense expression of an ACC oxidase gene (Savin et al., 1995). While these plants

synthesize reduced levels of ethylene compared with wild type, the plants were still

sensitive to exogenous sources of ethylene. Engineering of ethylene insensitivity through

alteration of ethylene signaling has been very successful in increasing flower longevity.

While extended floral longevity isn't directly valuable in tomato, plants engineered for

ethylene insensitivity at multiple points in the signaling pathway displayed extended

flower life in addition to altered fruit ripening (Wilkinson et al., 1997; Tieman et al.,

2001; Whitelaw et al., 2002). These results are of interest because they can be applied to

ethylene-sensitive floriculture crops. Extended floral longevity through altering ethylene

signaling was first demonstrated in tomato, petunia, and tobacco in plants constitutively

expressing the Arabidopsis etrl-1 allele (Wilkinson et al., 1997). Similar results were

also observed when petunia was transformed with a Brassica oleracea mutant ERS gene

(Shaw et al., 2002). In plants expressing heterologous mutant forms of ethylene receptors,

flower longevity was increased, but there were negative side effects caused by

constitutive ethylene insensitivity (Knoester et al., 1998; Shaw et al., 2002). These plants

were significantly reduced in their ability to form adventitious roots (Clark et al., 1999)

and were more susceptible to disease and death (Knoester et al., 1998; Shaw et al., 2002).

Through flower specific expression of etrl-1, extended floral life was observed in

carnation (Bovy et al., 1999). However, horticultural performance characteristics were









not reported for these plants. In tomato plants engineered for reduced EIL (EIN3-like)

expression, delayed floral senescence and abscission was observed (Tieman et al., 2001).

Since the ethylene signaling pathway seems to be highly conserved in multiple plant

species, it is likely that engineering of these points in the pathway will be useful in

generating longer lasting flowers. Perhaps future engineering of ethylene insensitivity

through tissue specific expression will help to alleviate some of these unfavorable side

effects. Horticultural performance studies with these plants will ultimately be required to

help answer this question and provide longer lasting flowers to the commercial

floriculture markets.

Gene Expression During Senescence

A number of studies have shown that senescence processes involve changes in gene

expression (Buchanan-Wollaston, 1994; Panavas et al., 1999, Quirino et al., 1999).

Studies with inhibitors of transcription and translation have demonstrated that synthesis

of new RNAs and proteins are required for senescence (Abeles et al., 1992; Borochov

and Woodson, 1989; Woodson, 1987; 1993). Through these studies, some differentially

regulated genes have been found including cysteine proteases, ribosomal proteins,

ethylene biosynthetic genes, nucleases, genes with putative roles in defense, and also

genes with unknown function (Lohman et al., 1994; Quirino et al., 1999; Panavas et al.,

1999). Many of these studies have focused on leaves for identification of senescence-

associated genes (SAGs) and a few of these have examined expression of the SAGs in

other senescing organs, like the flowers (Quirino et al. 1999). Quirino et al. (1999)

found that many of the genes differentially regulated during leaf senescence were also

similarly regulated during flower senescence and in response to pathogen attack. These

results present the possibility that there are underlying mechanisms controlling and/or









facilitating the senescence process at the level of gene expression. These results also

demonstrate that changes in gene expression are likely important for senescence

processes affected by ethylene.

Flowers and Pollination

Plants have evolved flowers for facilitating sexual reproduction. There is a

tremendous diversity of floral forms and strategies that plants use to facilitate pollination

and subsequent fertilization. In general, biotic (insect or animal), abiotic (wind), and

ambophily (both biotic and abiotic) means of pollination are utilized in flowering plants.

In the case of biotic pollination, flowers often have a vivid display of petals and color,

secrete nectar, have sticky pollen, and often emit a fragrance (Procter et al., 1996). In

contrast, wind pollinated flowers are often less complex and produce large amounts of

powdery pollen from flowers situated on the outside of the plant (Culley et al., 2002).

Plants capable of both types of mechanisms may be able to take advantage of conditions

that may favor one pollination mechanism over the other. An example of this has been

observed in Salix spp., where wind pollination was observed to be favored in more open

areas and insect pollination was favored in sheltered, forested areas (Vroege and

Stelleman, 1990).

There are two major views regarding the degree of specialization of biotic

pollination in plants. Historically, biotic pollination has been discussed as being a highly

specialized relationship that is the result of co-evolution and that there is an evolutionary

trend towards specialization of pollinators and plants. In this regard, many plant species

have been characterized as having a particular "pollination syndrome" based on a suite of

floral characteristics including morphology, chemical characteristics (fragrance and

nectar composition, color), and phenological characteristics (flower opening, temporal









regulation of fragrance) that together attract a particular class or order of pollinators (e.g.,

beetles or birds; bees or moths). For example, nocturnal moth-pollinated flowers are

generally typified as having white, tubular shaped flowers that are heavily scented at

night (Knudsen and Tollsten, 1993), while bee-pollinated flowers are usually patterned

with UV absorbing chemicals demarcating the nectar and pollen location to pollinating

insects and have a weak fragrance (Kevan et al., 1996).

More recently, there have been reports that challenge the "pollination syndrome"

dogma with the hypothesis that pollinator systems are more generalized than specialized.

One clear, recently published example of this is a study with Isertia laevis (Rubiaceae),

which exhibits floral characteristics typical of moth pollination (Wolff et al., 2003).

However, both hummingbird species and sphinx moths were observed to visit the

flowers, with hummingbird visits being more frequent than the sphingid moths.

Hummingbird visits produced a greater number of fruit set than the sphinx moth, but the

sphinx moths deposited more pollen than the hummingbirds, which resulted in a greater

number of seeds set per visit. Various pollinator types have also been reported to visit

the fragrant flowers of Clarkia breweri, with diurnal pollinators accounting for around

20% of the pollen transfer and nocturnal moth visits accounting for 80% of transfer

(Raguso and Pichersky, 1995). The diurnal visitors were not described in this study, nor

was seed set data. However, these and many studies have shown that plants may attract

multiple taxonomic groups of insects to the flowers, but there is likely variability in the

effectiveness of the pollinator as reflected in the number of seeds set. In many cases, the

relative abundance of the pollinators is also a factor, as more effective, less abundant









pollinators may actually generate similar numbers of seeds to more frequent, but less

effective visitors (Wolff et al., 2003; Slauson, 2000; Young, 2002).

Another constraint to the pollinator syndrome theory is the evolutionary history of

the plant. The genus Datura illustrates this concept. Based on floral characteristics,

Datura stramonium is a classic example of a nocturnal moth pollinated flower as it is

long, tubular, and heavily scented at night (Grant, 1983). However, studies specifically

addressing pollination biology of this plant showed that it was predominately self

pollinating as well as visited by bees (Motten and Antonovics, 1992; Motten and Stone,

2000). The more generalist view takes the classic pollinator type tendencies based on

pollinator and floral characteristics into account, but also considers other factors such as

insect behavior and populations, environmental influences, as well as evolutionary

history. Waser et al. (1996) suggests that plant-pollinator interactions should be

considered as a web of interactions rather than specialized interactions only and that all

flower visitors should be taken into account in plant-pollinator interaction studies.

Floral Fragrance and Insects

Insects use fragrance cues for localization and selection of potential food sources,

nesting sites, and mates (reviewed in Knudsen et al., 1993). While plants produce

hundreds of volatile compounds (e.g., Knudsen et al., 1993), not all of them have been

found to be perceived by insects and elicit a behavioral response (Fraser et al., 2003).

One of the ways that chemical detectability and responses in insects has been studied is

by electroantennographic detection (Roelofs and Comeau, 1971). In general, insect

antennas are the primary sites of olfactory sensing. Insect antennae are made up of

thousands of differentially, finely tuned olfactory receptor neurons that selectively

respond to odors, and in total produce an antennal response known as the









electroantennogram (EAG) (Park et al., 2002). Measurements of insect EAGs have

shown that the detectability and level of sensitivity to different chemicals is variable

among species and between sexes of the same species (Fraser et al., 2003). For example,

male and female Manduca sexta moths have similar antennal responses to many plant

derived volatiles, but in general female moths are more sensitive to varying

concentrations of the volatiles (Fraser et al., 2003). Fraser et al. (2003), Shields and

Hildebrand (2001), and others show that insect antennae are not excited by all plant-

derived volatiles and that the level of the excitatory responses is dose responsive to the

volatile. While these studies demonstrate that an insect can detect odors, it does not

indicate behavioral relevance. Studies of insect behavior to odors, such as proboscis

extension responses to floral volatiles (Honda et al., 1998), together with EAGs are

informative for determining if an odor elicits excitatory feeding responses.

Insects have demonstrated the capability to perform associative learning in foraging

activities and the ability to discriminate differences in floral odors. These abilities

presumably improve foraging efficiency (Williams, 1998). In the bee community,

flowers previously visited are avoided due to chemical footprints left on the flower after a

visit by a bee (Goulsen et al., 2000). The duration of the repellency was variable

depending on the evaporation of the footprint and has been reported to reflect the nectar

secretion rate of the plant. These studies show there is an inverse relationship between

repellency duration and nectar secretion rate (Stout and Goulsen, 2002). In a

conditioning study with the hawkmoth Manduca sexta, an excitatory feeding response to

different odors could be conditioned with sucrose solution reinforcement (Daly et al.,

2001). The moths were presented with an odor followed by the sucrose conditioning









reinforcement. After conditioning, the odor that was followed by sucrose elicited an

excitatory response while an unconditioned odor did not. The response was independent

of the odor used, with exception to methyl jasmonate, which did not excite the moths

even with sucrose reinforcement. Separate sets of moths were used for each odor, so the

ability of the moths to learn a new odor after having learned a different odor was not

addressed.

The emission of volatiles from flowers is primarily associated with attracting

pollinators and the temporal regulation of floral fragrance sometimes reflects this. While

it is controversial whether the majority of plants are generalist or specialized with respect

to pollinators and pollination syndrome (Waser et al., 1996), there are reports of high

fragrance emission at times when pollinator activities are high and cases where high

fragrance emission does not correlate with pollinator activity. An example of the former

has been observed in the wind and insect pollinated Trimnia moorei flowers. The flowers

of this species emit stronger fragrance in the morning, temporally coinciding with visits

that result in pollen transfer from both flies and bees (Bernhardt et al., 2003). In

Nicotiana, there have been reports of hummingbirds, hawkmoths, and bees that pollinate

some of the species in this genus (summarized in Raguso et al., 2003). Based on the

pollinator syndrome hypothesis, the hummingbird pollinated Nicotianaforgetiana andN.

langsdorffii would be predicted to not be heavily scented, as hummingbird pollinated

flowers are not generally associated with a fragrance (Van Riper, 1960). However, the

flowers did produce scent, which was primarily emitted at night. Since many of the other

species in this genus are hawkmoth pollinated, it is thought that this is a result of its

genetic background (Raguso et al., 2003).









Fragrance Composition

Floral fragrance is a ubiquitous, diverse, and dynamic characteristic in the plant

kingdom. It is comprised of a mixture of small molecular weight (100-290D) volatile

compounds (Knudsen et al., 1993; Dudareva and Pichersky 2000). They are derived

from multiple pathways including fatty acid derivatives (e.g., methyl jasmonate),

monoterpenoids (e.g., linalool), sesquiterpenoids (e.g., caryophyllene), and benzenoids

(e.g., methyl salicylate). Floral fragrance is a diverse characteristic, as there have been

hundreds of individual volatile components identified from flowers of many different

genera (Knudsen et al., 1993). The variations in floral fragrance are compounded by the

specific blends of each of the volatile components in quantity, identity, and combination.

In many cases, floral fragrance varies throughout the life of the flower as a response to

the developmental stage, time of day, and pollination status (Altenburger et al., 1990;

Jakobsen et al., 1994; Schiestl et al., 1997; Helsper et al., 1998; Dudareva et al., 2000;

Kolosova et al., 2001). This is a dynamic that is regulated developmentally, temporally,

and spatially and in response to various stimuli.

Fragrance Biochemistry Brief History

Identification of volatile components has largely been the focus of much of the

work to date on floral fragrance, while there has been less progress on the volatile

associated enzyme biochemistry and gene cloning. One of the first enzymatic analyses of

floral volatile production was in the annual Clarkia breweri, a plant native to California

and the only identified species in the Clarkia genus that emits a strong fragrance (Raguso

and Pichersky, 1995; Pichersky et al., 1995). The fragrance of C. breweri flowers is

largely dominated by linalool, a terpenoid derivative, with lower levels of linalool oxides

and benzenoid derivatives contributing to total fragrance (Raguso and Pichersky, 1995).









The linalool synthase (LIS) enzyme was purified by various chromatographic steps and

shown to catalyze the formation of linalool from geranyl pyrophosphate in the presence

of a divalent metal cofactor (Pichersky et al., 1995). By N-terminal sequencing of LIS,

the corresponding cDNA was cloned from a Clarkia cDNA library (Dudareva et al.,

1996). Using this approach and homology based searches of cDNA libraries, other floral

volatile biosynthetic genes have been identified including an iso-eugenol

methyltransferase (IEMT; Wang et al., 1997), salicylic acid carboxyl methyltransferase

(SAMT; Ross et al., 1999), benzoic acid carboxyl methyltranferase (BAMT; Murfitt et

al., 2000), benzyl alcohol benzoyl transferase (BEBT; D'Auria et al., 2002) and benzoic

acid: salicylic acid carboxyl methyltransferase (BSMT; Negre et al., 2003). These

studies have shown in vitro enzyme activity data to demonstrate catalytic specificity and

molecular biological studies on the developmental and spatial gene regulation patterns.

Fragrance Biochemistry Ado-Met Dependent Methyltransferases

Volatile methyl esters are among some of the most abundant floral volatiles. To

date, enzymes responsible for the synthesis of methyl esters have been cloned and

characterized in vitro in Clarkia breweri (Ross et al., 1999), Antirrinhum majus cv.

'Maryland True Pink' (Murfitt et al., 2000; Negre et al., 2002), Arabidopsis thaliana (Seo

et al., 2001), Stephonitisfloribunda (Pott et al., 2002), and Petunia x hybrida cv. Mitchell

Dipoid (Negre et al., 2003). In general, these enzymes use S-adenosyl-L-methionine

(SAM or Ado-Met) as a methyl donor to methylate the carboxylic acid moieties of

substrate methyl acceptors.

The salicylic acid (SA) and benzoic acid (BA) carboxyl methyltransferases are

enzymes that catalyze the methylation of the carboxylic acid moiety of SA and BA to

methyl salicylate and methyl benzoate, respectively. In the case of the Clarkia SAMT,









Antirrhinum SAMT, and Petunia BSMT, the preferred substrate in vitro is salicylic acid,

but they will also accept benzoic acid as a substrate (Ross et al., 1999; Negre et al. 2002,

Negre et al., 2003). In contrast, the Antirrinhum BAMT will only accept benzoic acid out

of the chemically similar substrates tested (Murfitt et al., 2000). To date, this is the only

cloned floral volatile associated carboxyl methyltransferase that exhibits strict substrate

specificity.

Crystallization of the Clarkia SAMT has shown that differences in substrate

preference are largely due to steric constraints brought about by differences in the amino

acids that make up the active site of these carboxyl methyltransferases (Zubieta et al.,

2003). In sequence comparison of SAMT orthologs with the Clarkia SAMT, it was

found that residues responsible for SAM binding and orientation of the substrate and

cofactor for methyl transfer are largely conserved (Zubieta et al., 2003). This study

further showed that through selected substitutions of amino acids in the active site, the

substrate specificity of the Clarkia SAMT could be altered so that jasmonic acid, a

comparatively large substrate to salicylic acid, could be methylated. In other

methyltranferases, the amino acids in the substrate binding site determine the specificity

of the enzyme. For example, when the amino acids that bind iso-eugenol from the

Clarkia iso-eugenol methyltransferase (IEMT) were changed to the amino acids that bind

caffeic acid in the Clarkia caffeic acid O-methyltransferase (COMT), the specificity of

the enzyme was changed so that the hybrid IEMT protein accepted caffeic acid, but not

iso-eugenol (Wang and Pichersky, 1998; Wang and Pichersky, 1999). This was

demonstrated for a COMT hybrid enzyme as well, which accepted iso-eugenol but not

caffeic acid (Wang and Pichersky, 1999).









Results of these in vitro assays do not completely reflect or predict what is

synthesized in vivo by each of these gene products. In vivo product formation not only

depends on enzyme specificity and abundance, but relative levels of substrates accessible

to the enzyme. For example, in Clarkia breweri methyl benzoate has not been detected,

but these flowers do emit methyl salicylate (Raguso and Pichersky, 1995). In petunia cv.

'MD', the flowers do not emit methyl salicylate consistently, but do emit large amounts

of methyl benzoate (Verdonk et al., 2003; Chapter 4, in dissertation). Additionally in

Antirrinhum, the BAMT is thought to primarily contribute to methyl benzoate emission,

as SAMTis not expressed at very high levels (Negre et al., 2002). Transgenic plants with

altered expression of these genes and measuring expression levels will be helpful in

understanding volatile methyl ester production in these systems.

A wide variety of reactions use SAM as a cofactor for methylation and as a

substrate in metabolism. This includes nucleic acids, proteins, cell wall constituents,

synthesis of ethylene and polyamines, and various secondary metabolites (reviewed in

Moffatt and Weredtilnyk, 2001). SAM is the most widely used methyl donor in cellular

methylation reactions as it is 103 fold more reactive than other methyl donors such as

folate and betaine (Cantoni, 1975; Fauman et al., 1999). Due to the number of cellular

methylation reactions that use SAM as a cofactor and reactions that use SAM as a

substrate, the cellular demand for SAM is high (Moffat and Weretilnyk, 2001). Cells

maintain adequate supplies of SAM through the methionine salvage cycle, which recycles

de-methylated SAM, S-adenosyl-homocysteine (SAH), to homocysteine and adenosine.

Most SAM dependent methyltransferases are inhibited by SAH (Mann and Mudd, 1963)

and thus removal of SAH into the recycling pathway is critical in maintaining









methylation reactions in the cell. Homocysteine and adenosine are further metabolized

through the SAM/methionine recycling pathway for the regeneration of methionine and

subsequent synthesis of SAM.

Fragrance Regulation

In general, many studies have shown that volatile biosynthetic genes are spatially,

developmentally, and temporally regulated and activity and emission patterns correspond.

Studies of floral volatile regulation have shown that gene expression, protein activity, and

volatile emission patterns are all highly correlated (Pichersky et al., 1994; Dudareva et

al., 1996, 2000; D'Auria et al., 2002). In Clarkia, linalool synthase (LIS) and benzoyl-

coenzyme A: benzyl alcohol benzoyl transferase (BEBT) are expressed in all floral

organs, with highest mRNA levels of both occurring in the stigmas (Dudareva et al.,

1996; D'Auria et al., 2002). Since emission of the volatiles synthesized by LIS and

BEBT spatially corresponds with mRNA expression and protein activity (Pichersky et al.,

1994; Dudareva et al., 1996; D'Auria et al., 2002), it is presumed that volatile synthesis is

actively taking place at the site of emission. In snapdragon, the conical cells of the inner

epidermal layer of the petals are presumed to be the site of methyl benzoate synthesis

based on immunolocalization studies of the BAMT protein (Kolosova et al., 2001). The

methylation reaction is proposed to take place in the cytoplasm, as this is where BAMT

was detected at the subcellular level (Kolosova et al., 2001).

Many of the identified volatile biosynthetic genes and corresponding volatiles have

similar developmental patterns of expression. LIS and BEBTmRNA expression and

protein activity is detectable in the early bud stage and peaks at anthesis for the following

two days (Dudareva et al., 1996; D'Auria et al., 2002). In snapdragon, BAMT expression,

protein activity, and emission of methyl benzoate increases after flower opening and









continues to increase with development (Dudareva et al., 2000) to a stage where

pollination would result in the highest number of seeds set and pollinator visits are

highest (Jones et al., 1998). After this stage, expression, activity, and methyl benzoate

emission decrease in non-pollinated flowers (Dudareva et al., 2000).

Fragrance Emission in Petunia

Much of the work to date on fragrance regulation and biochemistry has

concentrated on Clarkia breweri and Antirrhinum majus (Dudareva and Pichersky, 2000).

Recently, Petunia hybrida cv Mitchell has been used as a model for studying fragrance

regulation. Thus far, studies have shown that fragrance emission in petunia is rhythmic,

with emission primarily occurring during the evening and night from the petals (Verdonk

et al., 2003; Kolosova et al., 2001). The largest component of the fragrance emission

profile is methyl benzoate (Verdonk et al., 2003; Kolosova et al., 2001), which peaks

around midnight (Kolosova et al., 2001). The precursor to methyl benzoate, benzoic

acid, is also rhythmic with maximal levels at night (Kolosova et al., 2001). Recently, the

genes encoding for the enzymes that catalyze the formation of methyl benzoate were

cloned in our lab (BSMT1, Accession #AA045012 and BSMT2, Accession #AA045013)

and shown in vitro to catalyze the synthesis of methyl salicylate and methyl benzoate

(Negre et al., 2003).

Genetic Engineering for Improved Flower Fragrance

Cloning and characterization of genes involved with floral fragrance production has

presented a novel opportunity for genetic modification of this trait. This is an interesting

prospect because many flower-breeding programs have been focused on plant

morphology, disease resistance, and characteristics other than fragrance. As a result,

floral fragrance has been diluted or is lacking in many plants. Floral fragrance









engineering was first attempted in petunia with the Clarkia breweri linalool synthase

gene (Lucker et al., 2001). Six independent transformed lines were obtained and two

exhibited 3:1 segregation ratios and expression of the LIS transgene in the T1. While

there was LIS activity in these plants, there was little to no linalool detected. Further

experiments showed that linalool was being glucosylated to linalyl-P-D-glucoside. The

authors speculate that conjugation of linalool was a detoxification mechanism because

terpenoids have been reported to be detrimental to biological structures. The same gene

was transformed into carnation also with the goal of improving the fragrance (Lavy et al.,

2002). Linalool was detected in 15 independent transformed plants, with linalool

comprising up to 6% of the total volatiles and linalool oxides up to 4% of the total

volatiles produced from the flowers. The linalool producing carnation flowers were used

in scent panels to test if humans could detect a difference in the fragrance (Lavy et al.,

2002). However, the overall fragrance of the flowers was not detectably different by

human olfaction. This is clearly an important aspect in improving floral fragrance for

commercial value through genetic engineering. Additionally, engineering plants for

changes in the volatile profile will also be interesting for ecological studies since insects

presumably use fragrance for flower localization. These plants will be useful in

answering questions amount the importance of specific components of floral fragrance in

pollinator attraction.

Petunia

The geographic origin of Petunia is South America and it was established as a

genus in 1803 by Jusseau (Sink, 1984). It is a member of the Solanaceae family, which

is made up of many horticulturally and agronomically important plants including tomato,

pepper, potato, eggplant, and tobacco. Many of the modem Petunia hybrida cultivars









have been bred from crosses between Petunia axillaris and Petunia integrefolia (Ando et

al., 1999). While there are many Petunia hybrida cultivars in commercial breeding

programs, it has also been used for genetic studies in many research laboratories. Anther

culture was used with different backcrosses of P. hybrida and many different petunia

lines to generate haploid petunia plants (Mitchell et al., 1980). The Petunia x hybrida cv

Mitchell is one of these haploid (n=7), anther-derived plants generated from a selected

plant from a backcross of Petunia axillaris X Petunia hybrida cv. 'Rose du Ciel'

(Mitchell et al., 1980). In tissue culturing of the haploid petunia 'Mitchell', diploidy was

observed to spontaneously arise and polyploidy was induced by treatment with colchicine

(Griesbach and Kamo, 1996). The petunia 'Mitchell Diploid' (MD) line was selected

from one of these diploid lines.

Petunia x hybrida 'MD' is a useful plant for scientific studies for many reasons. It

is self-compatible allowing for the maintenance of highly inbred lines. It is derived from

the doubling of a haploid plant and therefore does not exhibit any genetic variation in self

pollinated progeny. The floral organs are relatively large and the plants are floriforus,

allowing for easy manipulation and access to large amounts of tissue for study. The life

cycle is relatively short and hundreds of seeds can be produced from a single pollination.

The flowers exhibit a predictable senescence response, which is mediated by ethylene

(Wilkinson et al., 1997). The flowers are heavily scented, allowing for studies on floral

fragrance (Verdonk et al., 2003). Additionally, there is an established transformation

protocol that can be used for analysis oftransgenic plants (Jorgenson et al., 1996). This

plant is used regularly in our lab for transformations and as a model system for studying

floral biology.









Research Objectives

The purpose of these experiments was to establish genomics tools for Petunia for

studying the changes in gene expression induced during ethylene-mediated floral

senescence. This was achieved by making floral cDNA libraries, cDNA microarrays, and

using the microarrays for identifying ethylene up- and down-regulated genes. This is of

interest because the changes that occur during floral senescence are not completely

understood and an understanding of some of the changes that are occurring at the level of

gene expression would allow for a better understanding of this process. From this initial

research, ethylene down-regulated gene expression of benzoic acid: salicylic acid

carboxyl methyltransferases (PhBSMT1 and PhBSMT2) was observed. The function of

PhBSMTs was demonstrated with transgenic plants engineered for reduced expression of

PhBSMT1 and PhBSMT2 by RNA interference. In detail expression analysis of

PhBSMT1 and PhBSMT2 was examined in all floral organs spatially and temporally in

response to ethylene application and pollination. Emission of methyl benzoate and other

major floral volatiles was examined after pollination and ethylene treatment. A role for

ethylene in regulation of floral volatiles is discussed.









































C02+


Figure 1-1. Ethylene biosynthesis in plants. Ethylene is synthesized in two steps from
S-AdoMet. S-AdoMet is also used as a cofactor for various methylations and
protein synthesis, as well as a substrate for ethylene and polyamine
biosynthesis.


0


H3CS OH

L-methionine NH3











membrane

Ethylene receptor
dimer complex

S^ ETR1, ERS1, ETR2,
w EIN4, & ERS2




CTR p Phosphorylation in
C absence of ethylene



nucleus

EIN2EI


ERF1
e Ethylene regulated
Gene trasncription




Figure 1-2. Proposed model of ethylene signal transduction in Arabidopsis thaliana
(adapted from Wang et al., 2002). When ethylene binds to the receptor dimer
complex, signaling through CTR is repressed and leading to an ethylene
response. The receptors are comprised of a gene family (ETR1, ERS1, ETR2,
EIN4, and ERS2 in Arabidopsis). ERS 1 and ERS2 do not contain response
regulator domains. HK: histidine kinase; RR: response regulator. See text for
discussion of signaling components.













CHAPTER 3
MICROARRAY ANALYSIS OF ETHYLENE-INDUCED FLORAL SENESCENCE IN
PETUNIA

Introduction

Flowers are responsible for sexual reproduction of the planet's most diverse group

of seed plants. Angiosperms originated 120 million years ago and have evolved into

thousands of distinctive flowering plant species that are represented today (Gorelick,

2001). Modern day flora have been shaped through time, selective pressures, and

development of interactions with plants and animals that serve as pollinators and seed

dispersers. As a result, there are diverse and complex means that plants have adapted in

flower development, morphology, and regulation for fulfilling one of the key processes in

the life of a plant. One extreme example of floral complexity can be observed in flowers

of the parasitic Rafflesia plants. Rafflesia spp. produces red flowers up to three feet in

diameter, that emit CO2 through thermoregulation and sulfuric volatiles that smell like

"rotting meat" to attract pollinating flies (Patino et al., 2002).

In many plants, once a flower is successfully pollinated, flowers senesce and/or

abscise the floral structures no longer required for growth and development of the fruit

and seeds. These processes also can occur with aging, but in many plants pollination

accelerates the progression of floral organ senescence and abscission (O'Neill, 1993).

Senescence and abscission are fundamentally different processes by definition, but with

respect to post-pollination processes, both result in the termination of floral organs (van

Doom, 1997). This programmed event likely confers benefits to plants, as these

processes are observed in diverse genera. Perhaps through eliminating the attractive parts









of the flower, pollinators are directed to non-pollinated flowers, allowing for more

efficient pollen dispersal. Additionally, these processes may allow for more efficient use

of nutrients and energy for the development of the fruit after a pollination event.

In many plants the phytohormones ethylene and auxin are key regulators of

senescence and abscission. Ethylene either promotes floral organ abscission and

senescence or has no effect on senescence of flowers with strict developmental floral

termination programs (van Doom, 2002). In contrast, auxin has both antagonistic and

synergistic effects with ethylene in promoting senescence and abscission in ethylene

sensitive species (Taylor and Whitelaw, 2001). Pollination and ethylene-induced

senescence and abscission processes have been observed to take place in multiple

families, van Doom (1997) reported ethylene sensitivity in hundreds of representative

species from 11 families, many of which have horticultural value. Processes involved

with floral organ senescence and abscission are of interest for understanding a vital stage

in plant development as well as for the potential of finding ways to prevent or slow down

this process for horticultural purposes.

A definitive role for ethylene in the regulation of floral senescence has been shown

in plants genetically engineered for reduced ethylene synthesis and insensitivity (Savin et

al., 1995; Wilkinson et al., 1997; Tieman et al., 2001; Shaw et al., 2002). In petunias

expressing the Arabidopsis etrl-1 allele, senescence is delayed in response to exogenous

ethylene treatments, pollination, and natural senescence (Wilkinson et al., 1997; Gubrium

et al., 2000). While these plants show that ethylene signaling is essential for eliciting

senescence, there remain some unanswered questions: What are the underlying

molecular changes that occur during senescence? What genes are affected during









Petunia floral senescence? How do these gene expression changes correlate with

physiological changes in Petunia?

Genomic and microarray technologies provide the opportunity for large-scale gene

expression studies (Schena et al., 1995). These types of experiments are useful for

screening thousands of genes and identifying differentially regulated candidate genes that

may be involved with or affected during a particular physiological process. Candidate

genes are identified by comparison of two sets of mRNAs, collected from a reference

control tissue and an experimental tissue, making fluorescently labeled cDNA probes

from these mRNAs, and examining which have higher levels of a particular message by

hybridization to a microarray (reviewed in Lemieux et al., 1998). The hybridized

microarray is scanned with a laser and camera to capture the fluorescent image, and

computer analysis is performed to identify differentially regulated genes by a comparison

of the intensity of the two colored fluorescent probes. The processes of microarray

fabrication are shown in figure 3-1 and experiments conducted in this research are shown

in figure 3-2.

The focus of this study was to examine broad changes in gene expression during

ethylene-induced floral senescence. The process of generating the necessary genomics

tools for identifying differentially regulated genes is described in this study (Fig. 3-1).

For this project, an EST (Expressed Sequence Tag) database was created from clones

sequenced from three petunia floral cDNA libraries. These ESTs were subject to

bioinformatic analysis for construction of a non-redundant set of clones and functional

categorization. Microarrays were constructed and experiments were performed for

identifying putative ethylene differentially regulated genes in petunia flowers. These









differentially regulated clones would be used in future physiological studies of processes

in the flowers that are affected by ethylene and senescence.

Results and Discussion

Establishment of a Petunia EST Collection

An EST collection from 'MD' petunia was established from three randomly

sequenced cDNA libraries made from flowers in a series of developmental stages,

ethylene-treated, and post-pollination. The goal in making these three flower libraries

was to obtain a broad representation of genes expressed throughout the life of a petunia

flower. Random DNA sequencing was performed from the 5' end, generating sequence

information for 2603 clones from the floral development library, 2989 clones from the

ethylene-treated library and 960 clones from the post-pollinated library (Table 3-1).

More clones from these libraries were sequenced; however, no or low quality sequence

data were obtained for approximately 6%, 8%, and 13% of total clones sequenced from

the developmental, ethylene, and post-pollination libraries, respectively. These numbers

are in agreement with other small-scale EST projects with regard to the number of no

data clones (Gang et al., 2001). Losses of 6-13% are likely due to sequencing errors or

sequencing of plasmids with no inserts. DNA sequencing was continued until close to

50% redundancy was reached for the developmental and ethylene-treated libraries (Table

3-1). At this point, it was estimated that more than two clones would have to be

sequenced in order to gain a unique sequence. The post-pollination library was

constructed at a later date and sequencing of this library is still in progress. Sequence

data for each of these libraries are online and can be found at http://helix.biotech.ufl.edu

with permission. BLASTx homology data for the libraries are accessible by password

online at http://genomics3.biotech.ufl.edu:8080/bq/blastquest.i sp.









EST Analysis

A non-redundant (nr) set of clones was assembled in preparation for microarray

construction. This set of clones was made in order to allow the maximal number of

different cDNAs to be assayed on the microarrays and have a replicate spot on the same

array. These data were informative for approximation of how many genes are

represented in the library databases and on the microarrays. Random sequencing of

cDNA libraries results in some redundancy because intracellular mRNA levels of

particular transcripts are often very abundant. As a result, these cDNAs are sequenced

multiple times from independent cloning of the multiple transcripts. The nr set consisted

of 1370 clones chosen from contigs and 1350 clones representing singlets from the

combined developmental and ethylene-treated libraries, resulting in a two fold reduction

in redundancy of the clones between the two libraries. All 960 clones sequenced from

the post-pollination library were not subjected to this analysis and were included for

microarray spotting, as sequencing from this library was in progress at the time the nr set

was made. In total, 3,394 clones were amplified by PCR from the three libraries for

construction of microarrays. This represents an estimated 3040 expressed genes with

some redundancy due to overlap between the pollination library and the non-redundant

set taken from the developmental and ethylene libraries. Later analysis of the redundancy

of this library showed that it was redundant for approximately 20.7% of the clones.

Additionally, seven percent of the total number of clones picked for PCR did not

correctly amplify and were subsequently excluded from the set of cDNAs for microarray

spotting.

Each of the clones in the EST collection was assigned a putative function based on

translated sequence similarity with proteins in the NCBI protein databases. While these









data do not absolutely define the function of a gene, they do assist in database mining and

as discussed later, aid in picking interesting clones for study from an EST collection.

These data also enabled us to assign the clones into putative functional categories (Fig. 3-

3), thus describing the range of genes that were represented on the petunia cDNA

microarrays. The functional groups will also be useful in future studies if there is a need

for assaying a large set of genes involved with a particular process. For example,

defense was a category used in grouping the ESTs into functional groups. This group of

putative defense-related clones could be used for construction of a microarray for

experiments to study how putative defense-related gene expression changes during a

stress response.

The two largest categories for all of the libraries were unknown and metabolism.

The overall percent breakdown of ESTs in each of the libraries was similar with the

largest difference in the protein synthesis category, with the greatest number represented

in the developmental library (9% in developmental, 6% ethylene-treated, 3 % post-

pollinated). A higher representation of protein synthesis cDNAs in the developmental

stage library is likely because the library represents tissues that are actively growing from

a young bud stage to anthesis, which are stages of active, rapid growth a likely time of

greater protein synthesis. The categorization of such a large number of clones as

unknown is likely an overestimation as this is based on sequence data from average read

lengths of approximately 550 bases. Sequence reads of this length are average for many

EST projects (Rounsley et al., 1996; Lange et al., 2000; Gang et al., 2001). Many of the

cloned inserts are longer than the average sequence read and may exhibit similarity to a

previously identified gene if more sequence information is obtained. This may also be









the case for all of the clones that did not show significant homology (expect value > 1.0 x

10-5) with any proteins from the NCBI database. Searches with longer sequence reads

may indicate similarity to proteins in the database. Additionally, if the cloned insert is

short and sequence information is from the 3' untranslated region of a clone, it is unlikely

to have significant similarities with an ortholog with known function than if it was

compared with sequence from the coding region. This could possibly contribute to error

in assigning putative functions, since this is a region of high sequence divergence even

among gene family members. The numbers of clones that were not functionally

characterized from each library, presumably for one of the aforementioned reasons, are in

Table 3-2. Categorization of singlets and contigs separately did not drastically change

the results of the groupings. The majority of contigs from each of the libraries contained

sequences from 2 clones (>69% for each library). There were a few contigs from each of

the libraries containing greater than ten clones per contig (-3%, developmental; -9%

ethylene-treated; -1% post-pollination). These representative abundant clones are listed

in Table 3-3.

Microarray Expression Analysis of Ethylene-Treated Flowers

Microarrays were used to identify differentially regulated genes in flowers during

ethylene-induced senescence in order to study physiological processes affected by these

treatments. Using the non-redundant set of cDNAs and all clones available from the PP

library, the cDNAs were spotted onto glass slides as outlined in Fig. 3-1. As a primary

screen for all differentially regulated genes in floral organs, whole flowers were used for

microarray analysis (procedure outlined in Fig. 3-2). The microarray experiments

undertaken are listed in table 4 and results from these experiments with detailed

categorizations of the putative differentially regulated genes listed in appendix 1. These









experiments yielded more genes being ethylene up-regulated than down-regulated (Table

3-4; 7.4% total up-regulated and 0.6% down-regulated) based on a cy3:cy5 ratio > () 2.

At this ratio, expression of the representative gene is estimated to be expressed by a

difference of two-fold between the two probe sets. A subset of these putative

differentially regulated genes were verified by sequencing to check the identity of the

cDNA assigned to the array spot of interest. All of the cDNAs checked were identical

with the clone assigned to each spot indicating there were no major organizational errors

during microarray construction.

RNA gel blots were performed with a subset of the clones that were putatively

differentially regulated (cy3:cy5 ratio > () 2). This was done to check the results of the

microarrays and as a secondary screen for identification of genes specifically

differentially regulated by ethylene in petals (Table 3-5). These results show that only a

subset of the cDNAs identified as differentially regulated on the microarrays

corresponded with RNA gel blot data (Table. 3-5) as shown in blot numbers 3-5a through

3-51. Many of the putative differentially regulated genes identified by microarray

analysis were not differentially regulated in petals (Table 3-5) as shown in blot numbers

3-5m through 3-5ab. There are multiple possibilities that could account for this apparent

discrepancy. First, it is possible that many of the putative differentially regulated genes

identified from the microarrays were primarily regulated in stigma+styles or ovaries and

not the petals. Second, it is also possible that using a heterogeneous tissue like whole

flowers as the source for the probes caused a dilution of differentially regulated mRNAs.

In support of the latter, many of the differentially regulated genes which corresponded

with RNA gel blot data, had high expression levels, as inferred by the intensity of the









expression (Table 3-5 blot numbers 3-5a, 3-5b, 3-5d, 3-5e, 3-5f, 3-5g, 3-5j, and 3-5k) and

abundance clones aligning with these cDNAs in the library database (data not shown).

Third, it is possible that some of the genes have low expression levels and are therefore

difficult to detect by hybridization on a total RNA gel blot. Perhaps a more sensitive

detection technique would help to resolve this issue. It is also possible that plant growth

conditions were slightly different between times of collection of tissue for microarrays

and collection of tissue for RNA gel blots. These reasons could together account for the

differences in expression patterns from the microarrays and RNA gel blots. Regardless

of the reason for the differences in the data, some clones did exhibit differential

regulation in the petals during ethylene treatment and these were considered for future

work in studying ethylene-regulated petal senescence.

Upon a detailed functional classification (Appendix), the majority of the putative

differentially regulated genes fell into three categories based on similarity: nucleotide

binding and transcription/translation related, secondary metabolism, and pathogen/stress

defense responses. Expression levels in the transcription and defense/stress response

category were mostly ethylene up-regulated. One of the ways that plants respond to

stress is through transcriptional regulation of genes encoding for proteins that assist in

defense and stress responses (Xu et a., 1994; Lorenzo et al., 2003). Many of the genes

up-regulated by ethylene have similarity to proteins involved with production of

metabolites that have defense-related and therefore overlap with the defense category.

For example, a cDNA similar to anthranilate N-hydroxycinnamoyl/benzoyltransferase

from carnation (HCBT) was upregulated by ethylene (Table 3-5, blot number 3-5h). The

carnation HCBT catalyzes the formation of the defense-related phytoalexin N-









benzoylanthranilate from anthranilate, a tryptophan precursor (Yang et al., 1997). In

contrast, there were 11 genes with higher expression in the air treatment and 12 genes

with higher expression in the ethylene treatments that fell into the category of secondary

metabolism. While many of these may have critical roles in plant growth and

development (like SAM synthetase, Table. 3-5, blot number 3-5i), they are classified here

because they also are essential to the production of secondary metabolites. These

changes in expression suggest that there is a change in the secondary metabolism of the

flower when exposed to ethylene. Phenylalanine ammonia lyase (PAL) (Table 3-5, blot

number 3-5j), a key enzyme in the control of phenylpropanoid metabolism, has higher

expression in air at 16 hours. If gene expression is correlating with the level of protein,

this suggests that the phenylpropanoid synthesis pathway is reduced in response to

ethylene or senescence in petunia flowers.

Of the interesting clones strongly down-regulated in petals in response to ethylene

were genes exhibiting homology with salicylic acid: carboxyl methyltransferases

(SAMT) (Table 3-5, blot number 3-5k). In vitro, SAMT catalyzes the methylation of

salicylic acid and benzoic acid to form methyl salicylate and methyl benzoate (Ross et al.,

1999). There were multiple spots on the array that corresponded to the petunia clone, as

it was found abundantly in the post-pollination library. Sequencing of the full-length

cDNAs showed that there were two distinct cDNAs in our EST collection and that the

two exhibited high identity, except at the 3' ends of the clones. It is highly likely that

both transcripts were detected simultaneously on the microarray and by RNA gel blot

hybridization. These results were of interest because they indicated that ethylene could

have a role in regulation of a floral volatile at the level of gene expression, which had not









been shown before. These cDNAs were chosen for future studies to examine the role of

ethylene in regulation of petunia floral fragrance (Negre et al., 2003; Underwood et al.,

submitted).

This study examined changes in gene expression taking place during ethylene-

induced senescence. Genetic tools were created including cDNA libraries and databases,

as framework for this study and future molecular biological studies on petunia floral

physiology. The cDNAs were used to construct microarrays for screening genes

differentially regulated in response to ethylene. A small subset of the cDNAs was

identified as differentially expressed in response to ethylene in whole flowers, with most

of them being up-regulated by ethylene. Based on sequence similarity, many of these

cDNAs may encode proteins with putative functions in transcription/translation, defense,

and secondary metabolism. In contrast, there were a few cDNAs associated with

secondary metabolism that were more abundant in the air treated flowers. This included

two cDNAs with similarity to floral volatile carboxyl methyltransferases, which were

chosen for the focus of future studies. In total, these results indicate that ethylene

possibly up-regulates defense related processes, while shutting down processes no longer

required, like fragrance emission, as petunia flowers senesce.

Materials and Methods

Plant Cultural Conditions

Petunia x hybrida 'Mitchell Diploid' (MD) was used for all experiments described

in this research. Seeds were sown on Fafard #2 potting mix (Fafard Co.; Apopka, FL) in

six-pack seed trays and placed in misthouse for mist every 30 minutes for 10 seconds.

One tablespoon of vermiculite was placed on top of soil three days after seed sowing.

Germinating seeds were left in mist until cotyledons had emerged and were visible. After









germinating, seedlings were transferred into greenhouses set at day/night temperatures of

25C/18C. Plants were grown in flats for approximately four weeks until transplanting.

Plants were transplanted into 1.2 L pots with Fafard #2 potting mix and fertilized every

other watering with 150 mgL-1 Excel 15-5-15 (Scotts Company; Marysville, Ohio).

Tissue Collections and RNA Extractions

Floral tissues were collected from Petunia x hybrida 'MD' for RNA extractions

and subsequent cDNA library synthesis. There were three sets of whole flower tissue

were collected for synthesis of three cDNA libraries: 1. developmental stage (from

flowers collected on the same day at five stages of development beginning at early bud to

anthesis); 2. ethylene-treated (collected after 2.5 [L'L-1 ethylene treatments for 30

minutes, 1 hour, 3 hours, 6 hours, and 12 hours); and 3. post-pollinated (collected 1, 2, 5,

10, 24, and 34 hours after pollination). For all collections, whole flowers were harvested

at the indicated times, placed into 50 mL falcon tube, and immediately put into liquid

nitrogen and stored separately at -800C for preservation until RNA extractions.

Total RNA was extracted for the developmental stage, ethylene-treated, and post-

pollination libraries (Wan and Wilkins, 1994 and Ciardi et al., 2000). Samples from each

set of tissues were extracted separately for each respective timepoint. Following

extraction and quantification, total RNA from each timepoint was combined in equal

molar amounts resulting in three representative samples of developmental, ethylene-

treated, and post-pollination RNAs for synthesis of each cDNA library. Poly (A)+

mRNA was isolated using a Poly (A)+ collection kit from Stratagene (LaJolla, CA).

cDNA libraries were constructed from poly (A)+ mRNA using a uni-directional X-ZAPII

cDNA synthesis kit from Stratagene (LaJolla, CA). Mass excisions were conducted on









unamplified library, plated on LB ampicillin media, and taken to the Interdisciplinary

Center for Biotechnology Research Genomics Sequencing Core Laboratory (University

of Florida; Gainesville, FL) for dye terminator capillary sequencing. Approximately 10%

of bacterial cultures from plated mass excisions were picked randomly for sequencing to

reduce the amount of redundant sequencing. All clones were organized as glycerol

stocks in 96 well plates for long-term storage at -800C. The clones were numbered in this

database according to library, plate number, and well number. For example, Petunia-PP-

12-A01 is a clone from the petunia post-pollination library and corresponding glycerol

stock is located in plate PP-12 in well A01.

Bioinformatic Analysis

One-pass sequence reads from the cloned unidirectional cDNAs in pBluescript

were obtained using the T3 plasmid primer site, corresponding to the 5' end of each

cDNA. The ESTs were organized into directories for separation of sequences into unique

reads (singlets or singletons) or non-unique redundant reads (contigs). Assembly of the

ESTs into groups was facilitated by the use of programs Phred and Phrap (University of

Washington; Seattle, WA). After assembling the ESTs into singlets and contigs, Consed

was used to view the contig alignments (University of Washington; Seattle, WA). EST

redundancy was calculated by dividing the numbers of clones in contigs by the total

number of clones sequenced. A non-redundant set of clones was generated by picking all

singlet clones and one clone from each contig if the clones were in good alignment as

viewed by contig sequence alignments in Consed. The criteria for choosing clones from

contigs for the non-redundant set was based on the following sequence characteristics:

* Sequence longer than 200 base pairs.
* If all clones aligned in contig, the EST with longest sequence read chosen.









* Clones aligned at ends of sequence reads (< 50 base pairs) both chosen.
* Clones not truly in alignment were all chosen.


Functional Categorization of ESTs

Contigs and singlets were categorized into functional groups based on homology

with the NCBI protein databases (http://www.ncbi.nlm.nih.gov/Database/index.html).

Functional group categories were in accordance with the categories in the Munich

Information Center for Protein Sequences (MIPS; http://mips.gsf.de) and have been used

by multiple groups for this purpose (Van der Hoevan et al., 2002; Guterman et al., 2002).

The categories are as follows: metabolism; energy; cell cycle and DNA processing;

transcription; protein synthesis; protein fate; cellular transport and transport mechanisms;

cellular communications and signal transduction; cell rescue, defense, and virulence;

regulation and interaction with cellular environment; cell fate; control of cellular

organization; subcellular localization; protein activity regulation; protein with binding

function or cofactor requirement; transport facilitation; classification not yet clear cut;

and unknown. ESTs with expect values less than 1.0 x 10-5 were assigned into

functional groups, as this was the threshold for homology set by Guterman et al. (2002).

Microarray Fabrication

Clones chosen for the non-redundant set were picked from the 96 well plate

glycerol stocks into plates with wells containing 150 [tL Luria Broth with Ampicillin

selection (50 mg/mL). Cultures were picked into 96-well plates, placed in 370C non-

shaking incubator, and grown overnight. The following day, PCR was performed using

the overnight-grown E. coli cultures to amplify the petunia cDNA inserts using T3 and

T7 primers. Reactions were set up in 96 well plate format and inoculated with 2 p.L of









bacterial culture, while the remaining culture was made into glycerol stocks for long-term

storage of a non-redundant set of clones. PCR reactions were run under the following

conditions: 95C for five minutes, followed by 35 cycles of 940C for one minute, 53C

for one minute, and 720C for one minute, and finishing with 720C for seven minutes.

After PCR was performed, products were analyzed by gel electrophoresis to verify

amplification and absence of multiple bands, and the remainder of amplified stock was

stored at -200C. Twelve percent of clones picked for culturing and PCR repeatedly did

not grow, amplified more than one band, or did not amplify and were excluded from

spotting onto the arrays. Products from the successful PCR reactions (22 [tL) were

aliquoted into 384-well plates with four microliters of spotting solution (20X SSC and

20% sarkosyl). Plates containing PCR products and spotting solution were prepared for

spotting and stored at 40C one day prior to use in the arrayer.

Gold Seal glass slides (Coming, Toledo, OH) were used as the surface support for

spotting microarrays. Slides were prepared and processed according to a modified

procedure described by Eisen and Brown (1999). First, the slides were thoroughly

cleaned in alkaline ethanol solution for 120 minutes by gently shaking at room

temperature in a metal slide rack and glass chamber from Shandon Lipshaw (Pittsburgh,

PA). Following the cleaning, slides were rinsed with approximately four liters of filter-

sterilized dH20 with minimal exposure to air and dust following this step. Next, the

slides were coated with poly-L-lysine by transferring rinsed slides into a chamber

containing a freshly prepared coating solution of 10% poly-L-lysine/10% PBS (Sigma-

Aldrich Corp, St. Louis, MO). Slides were dried at room temperature overnight and then









placed into a plain plastic slide box placed in a plastic container with Drie-Rite

dessicant (Xenia, OH) until spotting the following day.

cDNAs were spotted on to poly-L-lysine coated slides using an Affymetrix 418

Robotic four-pin Arrayer (Santa Clara, CA). cDNAs were spotted in an array pattern

with 375 [tM distance between the centers of the spots, each spot was stamped twice in

the same location to ensure deposition of each sample. Each cDNA sample was spotted

in two locations on the slide so that all samples were replicated on each slide. After

spotting, the slides were numbered and the arrayed area was etched into the slide with a

diamond tip scribe (Fisher Scientific; Hampton, NH), then stored at room temperature in

a plain slide box in a container with dessicant until processing and use. As the

microarrays were needed, slides were processed (modified procedure from Eisen and

Brown, 1999) by brief steam hydration, UV crosslinking, washing in 0.2% SDS for 10

minutes, rinsing with 2 L filtered dH20, denaturing in boiling in water bath for 10

minutes, and dehydrating in 95% cold ethanol. All microarray experiments were

performed within two months of spotting and processing was done as each experiment

was performed.

Probe Synthesis and Microarray Experimental Procedures

The microarrays were used to screen for genes up and down-regulated in response

to exogenous ethylene. All treatments were initiated the day after anthesis at 10 a.m.

under sunny weather conditions to help eliminate developmental, temporal, and

environmental variability. Flowers were treated with ethylene or air for 2 hours and 8.5

hours. The 16 hour sample was collected from flowers treated with ethylene for 8.5

hours to induce senescence, removed from ethylene treatment and then samples were

collected 7.5 hours later. For ethylene and air treatments, flowers were excised and









placed into 1.5 mL microfuge tubes containing 1.0 mL dH20. Flowers were sealed in

37.85 L glass chambers and treated with 2-3 tLL-L1 ethylene. For air treatments, flowers

were placed in the same conditions, but no ethylene was added to the chamber.

Concentrations of exogenous ethylene were verified at the beginning and end of indicated

treatment times using a Hewlett Packard Gas Chromatograph (Model 5890, Series II)

equipped with a flame ionization detector and an alumina column. For tissue collection,

flowers were collected at the indicated times, placed into 50 mL falcon tubes in liquid

nitrogen, and stored at -80C until RNA extraction. Probes were prepared from total

RNA extracted from the treated flowers by Phenol-Chloroform extractions followed by

lithium chloride precipitations (Ciardi et al., 2000). After extractions, RNA was cleaned

using a Qiagen RNeasy kit (Qiagen Inc; Valencia, CA) and following the manufacturers

instructions. RNA was quantified by spectrophotometer readings and quality was

checked by gel electrophoresis. Fluorescent probes were made using the Submicro EX

Expression Array Detection Kit from Genisphere (Hatsfield, PA). Using this kit, cDNA

probes were synthesized by selectively reverse transcribing mRNA from 50 [tg total RNA

with an oligo poly-T primer tagged with a binding site for fluorescent dyes cyanine 3

(cy3) or cyanine 5 (cy5). Approximately one quarter of the newly generated cDNAs was

aliquoted for each labeling reaction (approximately 12.5 tg total RNA was used per dye

per hybridization). The dye was tagged to the cDNA probes by incubation in the dark at

55C for one hour. Probes were hybridized to the microarray in the dark at 55C for at

least 24 hours. The following day, slides were washed in the dark in three salt solutions

(2X SSC, 2% SDS, 20 minutes at 55C; 2X SSC 20 minutes at room temperature; 0.2X

SSC, 20 minutes at room temperature) and immediately scanned in an Affymetrix 428









scanning microscope (Santa Clara, CA). The scanned images were analyzed for intensity

readings using the Affymetrix Jaguar 1.0 software (Santa Clara, CA). Intensity ratios for

the dyes (cy3 intensity/cy5 intensity) at each of the spots were used to find genes of

interest. Each experiment was done in triplicate with three slides and only spots that

showed consistent results as being up or down regulated by a cy3/cy5 ratio of at least (+)

2.0 for four out of the six spots were considered for further analysis.

Verification of Microarray Data

Results obtained from microarray experiments were checked by RNA gel blots.

Total RNA was extracted from tissue (Ciardi et al., 2000), quantified, and quality was

verified by gel electrophoresis on a 1% agarose, IX TBE gel. Total RNA was separated

on a denaturing formaldehyde gel and blotted as described (Kneissl and Deikman, 1996).

Probes were prepared from PCR amplified cDNA inserts (PCR as described above) from

the petunia cDNA libraries and labeled with 32P dCTP using a Prime-It II Random Primer

Labeling Kit from Stratagene (LaJolla, CA). Membranes were hybridized and washed as

described (Deikman and Fischer, 1988).









1) Extract RNA 2) Synthesize cDNA libraries
44 Petunia cDNAs
inserted in plasmids
in E cofi
AAAA i)

<9/Zz-)


6) Array experiments
& data analysis


5) Array Fabrication

PCR + Printing
******
::::::::


3) Random Sequencing


II


4) Bioinformatics


12,


I


7) Verify data by
northern blot


On going


- 8) Pick interesting
clones for further
studv


aown up
Figure 3-1. Flowchart of petunia EST project and expression studies. cDNA libraries
were synthesized from three sets of Petunia 'MD' floral tissues (1,2),
randomly sequenced (3), analyzed for redundancy and putative functions (4),
microarrays made from non-redundant cDNA set (5), and used for gene
expression studies of senescing floral tissue (6). Interesting genes were
verified for expression patterns by RNA gel blot analysis of petal tissues.









Air treated
Flower (A)


S_ AAAAAAAA
' AAAAAAAA

A AA AAAAAA
AAAAAAAA


1111


C2H4 treated
Flower (B)


Extract RNA ASAAAAAAAA


~ S AAAAAAAAA
AAAA
AAAAAAAA


RT & Label


A A=B
A>B


Yellow
Green


Scan & Data Analysis
Figure 3-2. Outline of microarray experimental procedures. Putative differentially
expressed genes were identified according to these types of experiments
during ethylene-induced senescence in the flowers. First, RNA was extracted
from air treated or ethylene treated flowers, reverse transcribed to make
cDNAs and for labeling with cy3 (green) or cy5 (red) fluorescent dyes. Dye
labeled cDNAs are hybridized to microarray and microarrays are scanned.
The intensity of the two colors is calculated, with ratios above two indicating
a significant difference in message levels in the two tissues.


Hybridize probes to microarray



-, M 0.Red

















5%
3%
2%
4%
4%

4% 9% 3 6%
9% 3%
3% 3%-

0% 2% 0 7
S 9% 5% 2%5
1%
1%- 1% 2%

S17% Metabolism

Protein Synthesis
33% 6Cellular Transport & Transport Mechanisms
5% *Cell Rescue, Defense, & Virulence
nCell Fate
1% Subcellular Localization
elmo [n]CelliCat iond ost-pcle l cuint
MProtein with Binding Function or Cofactor Requirement
6% Eclassification not yet clear cut
iEnergy
3% OTranscription
EProtein Fate
3% 6% oCellular Communications & Signal Transduction
3%b MRegulation & Interaction with Cellular Environment
2%6 0% OControl of Cellular Organization
13/
0 /1 9% M Protein Activity Regulation
1% 0%2% 6% mTransport Facilitation
iUnknow n

Figure 3-3. Putative functional categories of ESTs. This represents cDNAs from the (A)
developmental, (B) ethylene-treated, and (C) post-pollinated flower libraries.
ESTs were categorized based on homology with proteins in NCBI protein
databases if expect value was > 1.0 x 10-5. Functional categories were
assigned according to MIPS functional categories. Number of clones not
included in this analysis are listed in table 2.









Table 3-1. Sequence characteristics of petunia floral cDNA libraries. Percent
redundancy was calculated by dividing the number of clones in contigs by the
total number of clones with sequence data.
Tissue Library Clones Clones with %
name sequenced data redundancy
Developmental
Developmental DevA 2774 2603 45.50%
flower
Ethylene-treated C2H4
CfH4 3264 2989 45.00%
flower
Post-pollinated 960 835 N/A
flower 9/2001
Post-pollinated 1632 1396 29.60
flower present

Table 3-2. Number of clones from cDNA libraries not included in the functional
analysis. These were excluded because there was not high homology with
proteins in the NCBI protein database (expect value> 1.0 x105) or there was
no sequence information.
Library High E value No Data
Developmental 466 167
Ethylene-treated 399 181
Post-pollination 13 170

Table 3-3. Contigs from each of the cDNA libraries with the greatest number of clones.
Library Most abundant clones # Clones
Developmental putative oxidoreductase 11
ascorbate peroxidase 11
Polyubiquitin 12
Unknown 13
Elicitor inducible gene product 13
putative metallothionein-like protein 16

Ethylene-treated lipid transfer protein 10
putative metallothionein-like protein 14
ACC oxidase 16
polyubiquitin 17
isoflavone reductase-like protein 17
metallothionein-like protein type 2 23

Post-pollinated pectate lyase 7
SAM: SA carboxyl methyltransferase 8
isoflavone reductase-like protein 8
AMP-binding enzyme 8
S-adenosylmethionine synthetase 10






57


Table 3-4. Number of cDNAs putatively differentially regulated by ethylene. These
numbers summarize the expression results from the microarray experiments
with probes derived from ethylene-treated vs. air-treated Petunia flowers.
Ethylene down-
Treatment Ethylene up-regulated tlee d
regulated
2 hours air vs 2 hours
159 6
ethylene
8.5 hours air vs 8.5 hours
63 0
ethylene
16 hours air vs 16 hours
ethylene___










Table 3-5. Expression patterns in petunia corollas of differentially regulated cDNAs.
These cDNAs were identified from the microarray screen for ethylene-
regulated genes and had cy3 cy5 ratios >+2. RNA gel blots were made frm 20
jtg total RNA from corollas excised from ethylene treated flowers.
Blot Clone # Homology Microarray Petal hrs C2H4
Ratio Hiaher 0 12 24

3.47+103 2E
3-5a Petunia-4-D05 senescence-associated protein 29
4.53+1.4 85E



3-5b Petunia-C2H4-17-AO9 allergenic isoflavone reductase like 2.35+ 0.4 16 A




3-5c Petunia-C2H4-22-G01 putative cinnamoyl CoA reductase 3 18 + 0 5 8 5 E



2.5+ 0.7 2E El.
3-5d Petunia-C2H4-18-E08 cysteine protease p w :
264+04 85E


427+2.7 2 E
3-5e Petunia-C2H4-2-A12 SAM decarboxylase
228+0.23 16E



3-5f Petunla-C2H4-5-A03 sulfate transporter 3.25 + 0 88 2 E




3-5g Petunia-C2H4-5-C07 putative disease resistance protein 2.78 + 0 54 2 E




3-5h Petunla-DevA-28-B10 g anthranilate N-hydroxycinnamoyll 26 + 062 85E 'iil,
benzoyltransferase-like protein



3-5i Petunia-PP11-B07 g S-adenosyl methionine synthetase 232 + 0.25 16 E




3-5j Petunia-PP6-D05 g phenyl ammonium lyase 24 + 032 16E E .











Table 3-5. Continued.
Microarray Petal Blot
Blot Clone# Homology Ratio Higher 0 12 24

2.23 + 0.4 8.5 E .
3-5k Petunla-PP8-A04.g Salicylic acid 8
carboxyl methyltransferase 2 32+ 0.4 16 E



3-51 Petunia-C2H4-10D-EO6 uncoupling protein 2.31 + 0.36 2 E *




3-5m Petunia-C2H4-10D-E1C hypersensitive-induced response 547+425 2E
protein



3-5n Petunia-C2H4-15-D03 hairy roots 7 2.5 +0.46 2E



2.72+ 0 89 2 E
3-50 Petunia-DevA-17R-F03 predicted NADH dehydrogenase !!
2.45+0 17 85E



3-5p Petunia-C2H4-16-GO9 apoptosis-linked gene 4 2.03 + 006 2E i lE -




3-5q Petunia-C2H4-1-C10 lGbl:-erein 2-oxidase No1 2.87 + 0.5 2E E




3-5r Petunia-C2H4-22-B08 wound-induced protein-tomato 3.06 + 1 41 2E



2.37 +0.27 2E
3-5s Petunia-C2H4-2-A04 soluble starch synthetase
2.37+042 16E



3-5t Petunia-C2H4-4-B05 succinate dehydrogenase 5 07 + 2.9 8 5 E
flavoprotein alpha subunit











Table 3-5. Continued.


Microarray


Petal Blot


Blot Clone# Homology Ratio Higher 0 12 24


3-5u Petunla-DevA-13-H03 3-hydroxyisobutyryl-coenzyme A 2.35 + 0.47 8.5 E
hydrolase



3-5v Petunia-DevA-14R-D08 putative acyl CoA synthetase 2 48 + 0.46 16 E




3-5w Petunia-DevA-14R-F03 lipid-transfer protein-like protein 3.25 + 0 99 8 5




3-5x Petunia-DevA-14R-H04 copper homeostasis factor 2.05 + 0 06 2 E



214+14 2E
3-5y Petunia-DevA-15-G10 Nucleoid DNA binding like protein
322+14 16E


23+044 2E
3-5z Petunla-DevA-17R-F04 cysteine proteinase
273+07 16E



3-5aa Petunla-DevA-22-H11 g putative N-acetylglucosaminyl- 2.44+ 046 8 5 E
transferase



3-5ab Petunia-DevA-29-G10.c probable cysteine proteinase 24 + 0 16 16 E
inhibitor














CHAPTER 4
ETHYLENE-REGULATED FLORAL VOLATILE SYNTHESIS IN PETUNIA
COROLLAS

Introduction

Flowers facilitate sexual reproduction in plants, and the diversity of strategies

plants have evolved for pollination and seed set is evident in the plant kingdom. Flowers

have complex morphological characteristics, providing visual and olfactory cues, as well

as nectar rewards for insect and animal pollinators. Once pollination and fertilization are

successfully achieved, it is unnecessary for the plant to maintain floral structures not

involved with subsequent fruit and seed development. As a response to pollination and in

some cases fertilization, plants have evolved various senescence and abscission programs

to terminate floral structures that are no longer needed.

Following pollination and fertilization, many changes take place in the transition to

fruit and seed development. These changes include petal wilting and abscission, color

changes, flower closure, and swelling of the ovary as fruit development is initiated

(reviewed in O'Neill, 1997). In many plant species, the phytohormone ethylene

coordinates several of these processes (van Doom, 1997) and is synthesized spatially and

temporally after pollination promoting senescence of the petals. In Petunia x hybrida,

within two to four hours after pollination, a short burst of ethylene is produced from the

stigma and style (Hoekstra and Weges, 1986; Tang and Woodson, 1996; Jones et al.,

2003), when pollen tubes have just started to germinate and grow into the stigma (Tang

and Woodson, 1996). This is followed by sustained, autocatalytic ethylene production









from the stigma+style and ovary, beginning approximately 12 hours and peaking at 24

hours after pollination. Ethylene production from the corolla is induced during between

24 and 36 hours after pollination (Jones et al., 2003). The latter phase of ethylene

synthesis, which corresponds with the timing of fertilization, is thought to be responsible

for corolla senescence. Evidence for this was shown when stylar tissue was treated with

a competitive inhibitor of ethylene binding and the rate of corolla senescence was not

significantly different compared with untreated, pollinated flowers (Hoekstra and Weges,

1986). Ethylene is also produced at approximately six days post anthesis in non-

pollinated petunia flowers, with natural corolla senescence occurring after this

developmental ethylene production begins (Whitehead et al., 1984). The role of ethylene

in floral senescence in petunia was clearly demonstrated in plants engineered for

heterologous expression of the Arabidopsis dominant mutant ethylene receptor, etrl-1

(Wilkinson et al., 1997). These ethylene-insensitive plants exhibit significantly delayed

petal senescence after pollination or treatment with exogenous ethylene, as well as

delayed developmental senescence (Wilkinson et al., 1997; Gubrium et al., 2000).

Floral fragrance is composed of low molecular weight volatile organic compounds

(voc) that together with other floral cues, are thought to stimulate pollinator activity.

Floral VOCs are derived from multiple biosynthetic pathways in plant cells and include

benzenoids, fatty acid derivatives, isoprenoids, and others (Knudsen et al., 1993b). The

VOCs attract pollinators as the fragrance can indicate the presence of a food source or a

site for nesting, with exceptions of biological mimicry where no pollinator reward is

given (Altenburger and Matile, 1988). Many of the VOCs found in plants have been

shown to be detectable by and stimulate antenna sensilla in the hawkmoth Manduca sexta









(Shields and Hildebrand, 2001; Fraser et al., 2003). Additionally, many pollinators have

the ability to discriminate differences in floral fragrance intensity and quality (e.g. Wright

et al., 2002; Daly et al., 2001), and this ability has been shown in the hawkmoth,

Manduca sexta, to be reinforced by sucrose availability (Daly et al., 2001). In some plant

species the availability of pollinator rewards and quality and intensity of the fragrance

has been reported to reflect the pollination status (Burquez and Corbet, 1991; Schiestl et

al., 2001). Intensity of floral fragrance has been shown to correlate with floral

development, as fragrance is lower in young flowers, increases with flower age, and then

gradually declines as non-pollinated flowers senesce or abscise from age (Dudareva et al.,

2000). This pattern of emission through development correlates with pollen availability

and in some cases, potential for the greatest number of seed set (Jones et al., 1998).

Diverse floral volatile profiles have been characterized in many plant species

(reviewed in Knudsen et al., 1993), but only a few of the enzymes responsible for

catalyzing the synthesis of these volatiles have been characterized at the molecular level.

Two ubiquitous floral volatile components are the benzenoid-type methyl esters, methyl

salicylate (MeSA) and methyl benzoate (MeBA), and were reported in at least 47 and 34

genera, respectively (Knudsen et al., 1993). Synthesis of MeSA and MeBA is catalyzed

by a family of carboxyl methyltransferases, and genes encoding these enzymes have been

cloned from multiple plant species including Clarkia breweri, Antirrhinum majus,

Stephonitisfloribunda, and Petunia hybrida (Ross et al., 1999; Murfitt et al., 2000; Negre

et al., 2002; Pott et al., 2002; Negre et al., in press). The general reaction these carboxyl

methyltransferases catalyze is the methylation of the carboxylic acid moeity of salicylic

acid (SA) or benzoic acid (BA) producing MeSA or MeBA using S-adenosine-L-









methionine (SAM) as a methyl donor. These and related carboxyl methyltransferases

catalyze analogous reactions, but differ in substrate preference. For example, the Clarkia

salicylic acid carboxyl methyltransferase (SAMT) and Petunia benzoic acid:salicylic acid

carboxyl methyltransferase (BSMT) both have a lower Km for salicylic acid, but also

methylate benzoic acid (Ross et al., 1998; Negre et al., in press). While the benzoic acid

carboxyl methyltransferase (BAMT) from snapdragon has a lower Km for benzoic acid,

BAMT also methylates salicylic acid (Murfitt et al., 2000).

MeBA and MeSA have been proposed to have major roles in pollinator-attraction

because of their abundance and regulation (Knudsen et al., 1993; Dudareva et al., 2000).

Methyltransferase gene expression and corresponding volatile emission is high in flower

petals of Clarkia breweri (Ross et al., 1999), Antirrhinum majus 'Maryland True Pink'

(Dudareva et al., 2000) and Petunia hybrida 'Mitchell Diploid' (MD) (Verdonk et al.,

2003; Negre et al., 2003) with RNA expression and volatile emission being much lower

in other plant parts. In 'Maryland True Pink' snapdragon, emission of floral volatiles is

developmentally and temporally regulated. Peak MeBA emission occurs when flowers

are most receptive to pollination (Dudareva et al., 2000) and MeBA emission is maximal

in afternoon when bee pollinator activities are high (Kolosova et al., 2001). In contrast,

the MD cultivar of petunia emits maximal MeBA at night (Kolosova et al., 2001), in

association with attracting nocturnal moth pollinators (Knudsen and Tollsten, 1993).

Temporal regulation of MeBA is most likely a result of substrate availability in both

snapdragon and petunia. In snapdragon and petunia, BA substrate levels are rhythmic

and correspond closely with MeBA emission patterns, while mRNA and protein activity

exhibit less rhythmicity (Kolosova et al., 2001).









The cloning of genes involved with volatile synthesis has made the concept of

altering floral fragrance through genetic engineering feasible. This idea was first

attempted by transformation of the Clarkia brewer (S)-linalool synthase gene (CbLIS)

into petunia, a plant that does not normally emit large amounts of linalool (Lucker et al.,

2001). However, little emission of linalool was observed in transgenic lines because it

was mostly conjugated as a non-volatile glucopyranoside (Lucker et al., 2001). Floral

emission of linalool and linalool derivatives were detected by GC-MS when CbLIS was

transformed into carnation (Lavy et al., 2002). Although these compounds comprised

almost 10% of the volatiles emitted, there was no detectable change in scent as perceived

by humans (Lavy et al., 2002). These studies are economically important because floral

fragrance has not been a focus of many flower-breeding programs and as a result,

fragrance is weak in many modern cultivars of commercially important floriculture

plants. However, to engineer flowers for enhanced and novel fragrances, specific

changes in the volatile profile will need to be detectable by humans. Additionally, how

these specific changes in floral fragrance affect pollinator activity will be of interest,

given the role in pollinator attraction.

In a screen for ethylene-regulated genes in petunia flowers, we identified two

cDNAs, PhBSMT] and PhBSMT2, with similarity to salicylic acid carboxyl

methyltransferases. Work presented here demonstrates the activity of the petunia BSMT

in vivo is by RNA interference. Using wild type and transgenic ethylene-insensitive

petunias, the pattern of PhBSMT] and PhBSMT2 gene regulation in floral organs after

pollination and ethylene treatment, measurements of substrate levels, and corresponding

MeBA emission are shown. Emission of other major volatile components after ethylene









treatments and pollination are also presented. This study demonstrates a role for

ethylene sensitivity in regulation of floral volatile synthesis. The physiological

implications of floral volatile regulation and in vivo BSMT function are discussed.

Results

RNAi PhBSMT Reduces Methyl Benzoate Emission and Changes Floral Fragrance
in Petunia

RNAi mediated post-transcriptional gene silencing of BSMT was employed to

determine if the petunia BSMT1 and BSMT2 genes encode enzymes responsible for

MeBA emission in vivo. Three primary transgenic lines out of 40 were selected having

significantly reduced MeBA levels and BSMT expression. Two lines, BSMT-9 and

BSMT-14, showed inheritance of the transgene and phenotype in the T1 generation (Fig.

4-1) based on presence of NPTII, BSMTmRNA accumulation, and MeBA emission,

while no phenotype was observed in the BSMT-33 T1 transgenic progeny. In total, eight

Ti progeny were observed to have the phenotype and presence of transgene (seven from

BSMT-9 and one from BSMT-14). Since many of the transgenic plants were not

exhibiting a phenotype, it was inferred that the transgene was silenced in these plants.

All lines with reduced BSMTmRNA levels exhibited lower MeBA emission than wild

type (Fig. 4-1). NPTII positive plants that had apparently silenced, having wild type

levels of MeBA, also had wild type levels of BSMTmRNA (Fig 4-1). These results

together demonstrate that BSMT is responsible for MeBA synthesis in petunia.

Phenotype-positive lines did not exhibit changes for other major volatiles (Fig. 4-2).

A triangle test was used to determine if the fragrance of the plants with reduced

MeBA (BSMT-9) was detectably different by human olfaction from the floral fragrance

of wild type MD petunia. This test allowed us to assay the influence of one component,









MeBA, of the volatile profile on the perception of the overall petunia fragrance, since

other components of the floral fragrance remained the same (Fig. 4-2) and no new

volatiles were detected. The human sensory panel was able to discriminate the

differences in floral fragrance of the MeBA knockouts from MD wild type fragrance. In

this panel, 48 of 60 participants chose the sample flower that was different from the other

two given in the triangle test. The panelists were able to detect a significant difference

between wild type fragrance and BSMT-9 fragrance at a probability of <0.1%. The

BSMT9-9 T1 progeny from this line was also detectably different at a probability of

0.1%, as 33 out of 60 panelists could detect a difference. Overall, the participants

commented that the knockout flowers smelled "less" than wild type and many

commented negatively on the floral fragrance of the BSMT knockouts.

PhBSMT1 and PhBSMT2 Are Spatially and Temporally Regulated in Petunia
Flowers

Petunia flowers spatially and temporally produce ethylene in response to

pollination (Jones et al., 2003) presumably in order to coordinate post-pollination

changes in the individual floral organs (O'Neill, 1997). Volatile production is also

spatially regulated in the floral organs with the primary site of volatile emission being the

corolla (Verdonk et al., 2003), specifically the petal limb (Fig. 4-3). Since these genes

were ethylene down-regulated in whole flowers, expression levels in individual floral

organs were measured to examine the spatial and temporal pattern of mRNA regulation

in wild type MD and ethylene-insensitive 44568 after pollination and exogenous ethylene

treatments. Expression of both BSMT1 and BSMT2 was highest in the petal limbs

compared to the other floral organs (expression highest in petal limb>petal

tube>ovary>stigma+style). BSMT1 comprised between 0.3 and 0.6% of total RNA and









levels ofBSMT2 mRNA between 0.6 and 0.8% of total RNA in petal limbs (Fig. 4-4).

Beginning two hours after ethylene treatment and through the subsequent timecourse, the

greatest decrease in both BSMTmRNAs was observed in the ovaries, petal limbs, and

petal tubes in MD (Fig. 4-4). There was a decrease in mRNA levels in 44568 petal tube

and ovary, but the magnitude of decrease was ultimately not as great as observed in MD.

This decrease could possibly be due to some other type of regulation or effect, such as

developmental regulation, flower excision, or a small amount of residual ethylene

sensitivity in these tissues. These results indicate that ethylene down-regulates

expression ofBSMT] and BSMT2 in petunia flowers. Down-regulation in response to

ethylene is drastic and maintained in the corolla limb, the primary site of MeBA

emission.

These results suggest that pollination-induced ethylene production would down-

regulate PhBSMT] and PhBSMT2. Because ethylene production is spatially and

temporally regulated after pollination (Jones et al., 2003), spatial expression of PhBSMT]

(Fig. 4-5) and PhBSMT2 (Fig. 4-6) in individual floral organs following pollination was

examined. Down-regulation of both PhBSMTs was observed in MD stigma+styles

beginning approximately 2 hours post pollination. At approximately 10 hours post

pollination, expression of both PhBSMT] and PhBSMT2 was reduced in the ovary,

compared with corresponding expression in non-pollinated flowers. Down-regulation

subsequently occurs at 24 hours post-pollination in the petal tubes and limbs, with some

down-regulation occurring as early as 10 hours in the petal limb (PhBSMT1) and tube

(PhBSMT2). Expression of both PhBSMTs in the ovary, petal tube, and petal limb of the

ethylene-insensitive 44568 line was ultimately not different between pollinated and non-









pollinated control flowers. This pattern of down-regulation indicates that mobile

ethylene is eliciting a response in surrounding organs proximal to the stigma+style. From

these data it was concluded that pollination-induced ethylene elicits a decrease in both

PhBSMT] and PhBSMT2 mRNA levels temporally and spatially in the flowers.

Substrate Regulation in Response to Pollination and Ethylene Treatments

While regulation of MeBA emission by ethylene and pollination via mRNA levels

may be a key point of regulation in petunia, it is likely that other factors, such as substrate

availability may also regulate this process. In order to address the possibility of substrate

level regulation of MeBA emission, BA, SA, and a likely substrate precursor, CA, levels

were measured in MD and 44568 petals after ethylene treatment (Table 4-1) and

pollination (Table 4-2). BA and CA were present in higher amounts in ethylene-treated

corollas relative to air treated corollas in MD (12538 4619 ng gfw-1 and 4829 + 2042

ng gfw-1, respectively), but remained at similar levels in ethylene insensitive 44568

corollas (ethylene-treated 167 17 ng gfw-1 and air treated 244 1.6 ng gfw-1). At 36

hours after pollination, BA and CA were decreased in MD corollas (5895 2104 ng gfw-

1 and 56 16.8 ng gfw-1, respectively) compared with non-pollinated samples (24214

2479 ng gfw-1 and 417 27.9 ng gfw-1, respectively) and remained unchanged in 44568.

SA levels were largely unaffected by either the ethylene treatment or pollination in MD

and 44568. Both BA and CA exhibited rhythmic regulation. Levels of both were higher

in the night collected non-pollinated 36 hour control samples of MD and 44568 relative

to the zero hours day collected sample.









Volatile Emission Is Down-Regulated in Response to Exogenous Ethylene and
Pollination

MeBA emission was measured to determine if emission corresponded with the

decreased benzoic acid levels (36 hours post-pollination) and decreased PhBSMT] and

PhBSMT2 mRNA levels (in the petals after 10 hours of ethylene treatment and 24 hours

post-pollination). A reduction in MeBA emission was observed after 10 hours of

ethylene treatment (Fig 4-7). Air-treated flowers emitted approximately 12 times as

much MeBA as ethylene treated flowers. This reduction in MeBA emission was not

observed with 44568 flowers after ethylene treatment, thus demonstrating a role for

ethylene in regulation of this process. After pollination, MeBA emission decreased

significantly in MD flowers beginning 24 hours after pollination (Fig. 4-7). Emission at

this time was low in both genotypes, as this timepoint was collected during the day when

emission was minimal. However, emission was decreased by 50% at 24 hours after

pollination in MD compared to non-pollinated flowers. The timing of this reduction

corresponds to when post-pollination ethylene production from the corolla begins and

when corresponding PhBSMT mRNA levels have decreased compared with non-

pollinated controls. Additionally, the timing also corresponds with the time of

fertilization observed for petunia (Tang and Woodson, 1996). Based on these results, it

was concluded that pollination-induced ethylene production regulates emission of MeBA

in petunia.

The data from the pollination timecourse also indicated that MeBA emission was

rhythmic, shown in non-pollinated flowers. This is in agreement with observations by

Kolosova et al. (2001) and Verdonk et al. (2003). In many cases, rhythmic biological

patterns indicate the possibility of a circadian regulated process. In order to address this









possibility, we collected floral volatiles through normal day-night cycles and then in

continuous dark or light. When the plants were placed into complete darkness, emission

of MeBA virtually ceased (Fig. 4-8) and PhBSMT gene expression did not continue to

oscillate, as observed for the first two normal day-night cycles (Fig, 4-8b). When the

plants were placed into continuous light, MeBA emission continued with no obvious

oscillations from the first day through the remainder of the timecourse (Figure 4-9).

However, gene expression continued rhythmicity through the first 24-hour period in

constant light. Robust cycling was not observed the first day the plants were placed into

the dark for either gene expression or MeBA emission. These data show an additional

linkage between PhBSMT gene expression and MeBA emission; both are rhythmic with

increased gene expression occurring approximately six hours prior to rises in emission.

The production of other major floral volatile components in response to ethylene

and pollination was also characterized. All the abundant floral volatiles were reduced by

ethylene treatment (Fig. 4-10) and pollination (Fig. 4-11) in MD, but not in ethylene-

insensitive 44568. Floral volatiles that decreased in response to both of these treatments

included benzaldehyde, phenylacetaldehyde, benzyl alcohol, 2-phenylethanol, iso-

eugenol, and benzyl benzoate. Iso-eugenol and benzyl alcohol were the least affected of

all of these volatiles. These results show there is a coordinated down-regulation of floral

volatile emission and it is dependent upon ethylene signaling.

Discussion

This study demonstrated in vivo that BSMT synthesizes methyl benzoate in

petunia. The spatial and temporal regulation of PhBSMTgene expression, substrate

levels, and MeBA emission after pollination and ethylene treatments were characterized.

These results indicate that in Petunia, ethylene regulates floral fragrance at multiple









levels that lead to an overall reduction in total floral fragrance mediated by the perception

of ethylene. A primary role for ethylene in regulation of visible post-pollination changes

has been established (Wilkinson et al., 1997). The impact of ethylene on these visual and

olfactory cues on pollinator attractiveness is inferred to be negative, since insects are

attracted by both visual cues and floral fragrance.

Transgenic PhBSMT RNAi Plants Have Lower Methyl Benzoate Emission

RNA interference was used to engineer petunia for decreased PhBSMTmRNA

levels to test in vivo if BSMT synthesizes MeBA and to determine if this trait could be

genetically engineered to be detectable by human olfaction. In transgenic lines, reduced

PhBSMTmRNA levels resulted in greatly reduced levels of MeBA emission with the

level of reduction of MeBA emission corresponding with the level of reduction of both

PhBSMT] and PhBSMT2 expression. These results link gene with function, indicating

that in vivo both PhBSMT] and PhBSMT2 catalyze the formation of MeBA in petunia

flowers. More primary transgenic lines have been made using a different RNAi construct

since pHannibal had low efficiency for knocking out this gene in petunia (three out of 15

pHannibal lines measured had reduced PhBSMTmRNA levels) and silencing was

observed in the T1. Many primary transgenic lines with the new construct were obtained

and are expressing a reduced MeBA phenotype (R Dexter and D Clark, unpublished).

The MeBA knockouts demonstrate that specific components of the floral volatile

profile can be changed through genetic engineering. These plants will be useful in

studying plant-insect interaction systems, as it is not known if specific components of

floral fragrance are responsible for pollinator attraction. The fragrance of the MeBA

knockout flowers was shown to be detectably different from wild type fragrance to

humans. These are novel results demonstrating that flowers can be genetically modified









for changed floral fragrance. The engineering of plants for enhanced scent characteristics

has potential commercial value and possibly agronomic value, since fragrance is often

lost in breeding for other traits. Novel and improved fragrances could feasibly be

introduced into existing commercial cultivars through transformations with these and

other scent genes. Human olfactory panels will help to answer questions of what

fragrances, volatile combinations, and levels should be engineered to obtain flowers with

pleasant fragrances.

Ethylene Regulates PhBSMT Expression in Petunia Floral Organs

PhBSMT] and PhBSMT2 were both down-regulated quickly in response to

exogenous ethylene and pollination in all MD floral organs, but not in ethylene

insensitive 44568 flowers. The temporal and spatial down-regulation of both mRNAs in

the floral organs follows the sequential pattern of post-pollination ethylene production

observed in petunia (Tang and Woodson, 1996; Jones et al., 2003). Studies in petunia,

Phalaenopsis spp., and carnation have shown that ethylene synthesis is temporally and

spatially regulated in the flower and coordinates developmental changes such as petal

senescence and ovule development (O'Neill et al., 1993; Tang and Woodson, 1996; Bui

and O'Neill, 1998; Jones et al., 2003). Models for post-pollination regulation of ethylene

production show that pollination induces production of ethylene first from the

stigma+style, then from the ovary, followed by the petals. Here it is shown that both

PhBSMT] and PhBSMT2 are down-regulated in each floral organ, consistent with the

pattern of spatial post-pollination ethylene production in petunia flowers observed by

Tang and Woodson (1996) and Jones et al (2003). The data also suggest that the large

amounts of ethylene produced from the stigma+style can regulate ethylene-sensitive

processes in the corolla due to the close proximity of these organs. The corolla produces









measurable ethylene beginning 24 hours after pollination (Jones et al., 2003) while down-

regulation of PhBSMT2 begins as early as 10 hours in the MD petal tube, but not 44658.

This implies mobile ethylene produced in the stigma+style regulates PhBSMT2 in the

petal tube after pollination.

There was some observed PhBSMT down-regulation after pollination in 44568

stigma+styles. It is likely that this is due to a small amount of ethylene sensitivity in this

tissue as the constitutive 35S promoter driving expression of the etrl-1 allele does not

confer high, equal expression levels oftransgenes in all plant cells. Position effects of

the transgene in the genome can cause expression levels of the transgene to differ in

specific tissues (Holtorf et al., 1995; Van Leeuwen et al., 2000). In this case, position

effects could result in certain cells or tissues having a small amount of ethylene

sensitivity. However, the contribution of the stigma+style to total fragrance output is

small and therefore changes in emission from this organ are likely to have little effect on

the total fragrance output of the flower.

Measurements of substrate levels after ethylene treatments and pollination

indicated that ethylene does not immediately down-regulate substrate levels like it does

PhBSMT gene expression levels. Free BA and CA levels increased in response to

ethylene, while decreased after pollination in MD and remaining unchanged in 44568 for

ethylene treatments and pollination. The increase in BA after ethylene treatment is likely

due to substrate accumulation, as it is not being synthesized into MeBA (Fig. 4-7) due to

decreased BSMT activity (Negre et al., 2003) and PhBSMT gene expression (Fig. 4-4).

Increased levels of BA and CA after ethylene treatment and decreased levels after

pollination in MD, but not 44568, indicate that there are likely multiple factors









controlling the levels of BA and CA. Perhaps ethylene sensitivity is required, but there

are also other factors such as senescence, pollination signals independent of ethylene that

are required for substrate regulation. During pollination, these substrates may be used in

other processes or be remobilized to other organs, resulting in the measurable decrease

observed here. Additionally, ethylene treatments were conducted with excised tissue,

which may cause differences if supply of these substrates involves a transport

mechanism. In regard to SA, the levels of SA were relatively low compared with BA and

CA. Since SA is a plant hormone and it is not actively being synthesized into MeSA

during these treatments, low levels would be expected. Also, since SA is involved in

eliciting cell death and plant defense responses, it is possible that maintenance of SA

levels in petals during pollination may assist in promoting cell death or in defensive

mechanisms against infection as the flower senesces around the developing fruit.

These results together with those ofNegre et al. (2003) thoroughly examine the

synthesis and ethylene+pollination regulation of MeBA in petunia flowers. Here it is

shown that MeBA emission is markedly reduced in response to ethylene (Fig. 4-7) and

pollination (Fig. 4-7), in agreement with pollination data shown in Negre et al (2003).

Down-regulation after pollination is controlled by ethylene through decreased mRNA

levels in the corolla, possibly substrate levels in the corolla, and as observed by Negre et

al. (2003) post-translationally. While post-translational regulation does appear to have a

role in regulating MeBA (Negre et al., 2003), decreased mRNA levels are likely to have a

major role since this would presumably limit enzyme abundance and disturb the substrate

to enzyme ratio, and thus result in decreased product formation. Additionally, reducing

the levels of mRNA by RNAi shows that mRNA abundance is a factor in how much









MeBA is synthesized. As a whole, these results clearly demonstrate a role for ethylene in

temporal and spatial regulation of MeBA at multiple levels.

The nature of rhythmic MeBA emission was investigated in order to examine

additional linkages between mRNA and MeBA emission and possible circadian rhythms.

Rhythmic emission of MeBA has been demonstrated in multiple species including

Petunia (Kolosova et al., 2001; Verdonk et al., 2003), Antirrinhum (Kolosova et al.,

2001), and Stephanotis (Pott et al., 2002) and has been shown to be circadian in

Antirrhinum. To demonstrate true circadian rhythmicity of a process, a robust rhythm

must continue under constant environmental conditions (Jones and Mansfield, 1975). A

robust rhythm of MeBA emission or mRNA expression was not observed when plants

were placed into the dark. However, the maintenance of PhBSMT gene expression

rhythmicity makes the interpretation less clear. It is possible that there was a shift in

peak emission and gene expression when plants were placed into dark. More frequent

sampling during the first 24-hour period in constant darkness and light might help with

interpretation of these results. From these data it can be concluded that gene expression

is rhythmic increasing approximately six hours ahead of high MeBA emission showing a

linkage between PhBSMTmRNA expression and emission.

Pollination and Ethylene Treatments Down-Regulate Floral Volatiles in Petunia

Analysis of other major volatile components of petunia fragrance demonstrated that

many of the floral volatiles are similarly down-regulated after pollination and in response

to exogenous ethylene. Many of the volatile components that were significantly reduced

have been shown to elicit excitation responses in antenna sensillas of the moth Manduca

sexta (Shields and Hildebrand, 2001; Fraser et al., 2003). The volatile iso-eugenol was

shown by Shields and Hildebrand (2001) to not elicit an excitatory response and this









volatile component was not as reduced as were the other volatile components after

pollination. Reduced emission after ethylene treatments in MD and continuous emission

from 44568 flowers demonstrate that ethylene is likely to have key control of regulating

most of the floral volatiles. All of the volatiles exhibited temporal down-regulation after

pollination, analogous to the reduction in MeBA emission. This pattern of regulation

corresponds to the second major phase of pollination induced ethylene production, when

ethylene is simultaneously being produced from multiple floral organs. Ethylene

treatments had a more pronounced effect on reducing floral volatile emission in MD

compared with the pollinated flowers. The ethylene-treated flowers were treated in an

enclosed system, which is conducive to eliciting a stronger ethylene response. Since air

treated samples were not significantly altered compared with untreated controls, the

effect is due to ethylene and not the enclosed treatment. Additionally, it is possible that

constant exposure to ethylene and absence of pollination cues may contribute to a greater

response. These results are of interest because they demonstrate that ethylene sensitivity

controls total floral fragrance output in addition to controlling petal senescence. These

two processes together may influence both visual and olfactory attractiveness of the

flower to pollinators.

The timing of floral volatile down-regulation has interesting ecological

implications. In terms of pollinators that are primarily attracted by floral fragrance, there

are multiple benefits to regulating floral volatiles after pollination. Some pollinators can

distinguish differences in floral fragrance (Wright et al., 2002; Daly et al., 2001) and

many of these volatile components elicit excitatory responses in a potential moth

pollinator (Shields and Hildebrand, 2001). Some pollinator-behavior studies demonstrate









higher visitation rates to flowers with greater scent levels (e.g. Wright et al., 2002).

Results shown here indicate that petunia flowers may regulate fragrance intensity after

pollination so that pollinated flowers are detectably different to insects. The timing of

this process is important since reduction of fragrance emission too early after a

pollination event may be wasteful if foreign pollen or a small amount of pollen is

deposited. By reducing fragrance after a successful pollination event, more efficient

distribution of pollen with less pollen spent on flowers that have already been pollinated

could be achieved. It is possible that reducing floral fragrance is a type of defense

mechanism, as these flowers may become less apparent to visiting insects. Fewer visits

could result in less risk of perturbing the developing fruit, through introduction of

parasites or pathogens by non-sterile pollinators. Additionally, if proportions of the

volatiles emitted from the plant changes with pollinations, plant apparency (Feeny, 1976),

or conspicuousness, may diminish and thus reduce pollinator visits if a number of

pollinations were made. If the pollinators of petunia also use the plant as a site for laying

eggs as in Nicotiana (Baldwin and Ohnmeiss, 1993), reduced apparency may decrease

the probability of egg laying on the plant.

These results show evidence of a role for ethylene in regulation of the major floral

volatiles in petunia. In the case of MeBA, ethylene regulates mRNA expression levels in

each of the floral organs in a spatial and temporal manner after pollination and this

regulation corresponds with MeBA emission. Plants engineered for knockout of MeBA

demonstrate two main points. First, BSMT synthesizes MeBA in petunia. Second,

modification of floral scent through genetic engineering is possible in petunia and these

changes can be used to alter the fragrance enough so that people perceive a difference in









the floral fragrance. Scent panels will be key to engineering flowers for improved floral

fragrance, as the MD wild type flowers were preferred over flowers with reduced MeBA.

These plants should be useful tools for studying the effects of individual volatiles,

changes in overall volatile blends, and the total effects of ethylene on floral attractiveness

in pollinator behavior studies.

Materials and Methods

Plant Material

In all experiments, Petunia x hybrida 'Mitchell Diploid' (MD) was used as the wild

type line and is also the genetic background of ethylene-insensitive 35S::etrl-] line

44568 (Wilkinson et al., 1997). Plants were grown in air-conditioned glass greenhouses

at 250C day/180C night. Plants were potted in Fafard 2B potting medium (Fafard Inc.,

Apopka, FL) in 1.2 L pots and fertilized at every irrigation with 150 mg-L1 Scott's Excel

15-5-15 (Scotts Co., Marysville, OH).

cDNA Isolation

Three cDNA libraries were constructed from petunia MD whole flowers collected

at multiple developmental stages (from early bud to anthesis), ethylene-treated flowers

(2.5 pL'L-1 ethylene treatments for 30 minutes, 1 hour, 3 hours, 6 hours, and 12 hours),

and pollinated flowers (1, 2, 5, 10, 24, and 34 hours after pollination). Total RNA was

extracted by a phenol:chloroform extraction method with lithium chloride precipitations

as described in Ciardi et al. (2000). Messenger RNA was isolated using Oligotex mRNA

purification (Qiagen Inc; Valencia, CA). cDNA libraries were constructed using a k-

ZAPII cDNA synthesis kit from Stratagene (LaJolla, CA). Approximately 6000 clones

from these libraries were randomly sequenced. A minimally-redundant subset of these

clones was used for microarray experiments to find ethylene regulated genes. From these









microarray experiments, the salicylic acid carboxyl methyltransferase homologs were

isolated as being candidate cDNAs that were down-regulated by ethylene (to be presented

in another manuscript). Two full length cDNAs encoding for SAMT homologs were

isolated from the cDNA libraries and were used in subsequent experiments. Northern

analysis was used to verify down-regulation by ethylene (Negre et al., in press).

Tissue Treatments and Collections

All ethylene treatments and pollinations were initiated the day after anthesis at 10

a.m. under sunny weather conditions to help eliminate developmental and environmental

variability. For ethylene and control-air treatments, flowers were excised and placed into

1.5 mL microfuge tubes containing 1.0 mL dH20. Flowers were sealed in 37.85 L glass

chambers and treated with 2-3 tL-L.1 ethylene. For air treatments, flowers were placed

in the same conditions, but no ethylene was added and potassium permanganate (Fisher

Scientific, Hampton, NH) was placed in the chambers. Concentrations of exogenous

ethylene were verified at the beginning and end of indicated treatment times using a Gas

Chromatograph (Hewlett Packard Model 5890, Series II; Palo Alto, CA) equipped with a

flame ionization detector and an alumina column. In the pollinated flower collections,

flowers were pollinated and remained on the plant until designated collection time. For

every experiment all treated (ethylene or pollination) and control (air or non-pollinated)

flowers were collected at the following times (with treatment times in parenthesis): 10:00

am (0 hours), 12:00 noon (2 hours), 8:00 pm (10 hours), 10:00 am (24 hours), 10:00 pm

(36 hours), 10:00 am (48 hours) after ethylene treatment or pollination. Tissue was

collected from plants placed into dark chambers at 25 + 30C for the constant dark

circadian studies.









Spatial and Temporal Analysis of mRNA Expression in Flowers

Spatial and temporal mRNA accumulation was analyzed after ethylene treatments

and pollination in petunia MD and ethylene-insensitive 44568. Expression was

examined from individual floral organs including petal limbs, petal tubes, stigma+styles,

and ovaries that were excised from ethylene-treated, air-treated, pollinated, and non-

pollinated flowers. The day after anthesis, flowers were either collected for ethylene

treatments or pollinated on the plant for the time courses described above. Harvested

tissue was immediately frozen in liquid nitrogen and stored at -800C. Total RNA was

extracted using an RNeasy Mini Plant RNA extraction kits with on-column DNase

digestion performed during the extraction (Qiagen Inc., Valencia, CA). RNA was

quantified by spectrophotometry and RNA quality was verified by gel electrophoresis.

Real-Time RT-PCR was performed for quantification ofPhBSMTmRNA transcripts

from 100ng of total RNA using TaqMan One-Step RT-PCR reagents (Applied

Biosystems; Foster City, CA). Reactions were conducted in 25 p.L volumes in 96 well

optical reaction plates on an Gene Amp 5700 Sequence Detection System (Applied

Biosystems; Foster City, CA). Primers and TaqMan probes were designed using Primer

Express Software (Applied Biosystems; Foster City, CA). Specificity of each of the

primer and probe sets was verified by performing PCR reactions with in vitro transcribed

PhBSMT] template with the primer and probe set specific to PhBSMT2 and vice-versa.

In vitro transcribed RNA was synthesized using a MAXIscript In vitro Transcription Kit

(Ambion, Austin, TX) according to manufacturer's instructions. PhBSMT] and

PhBSMT2 were used as templates for in vitro transcription and the transcripts were

collected on separate gels to prevent contamination. Primer and probe sequences used









for individual detection of each gene corresponded to the 3' untranslated region of the

cDNA and are as follows: PhBSMT] forward primer,

AAATGTCATCATCTCCTTGACCAA; PhBSMT] reverse primer,

CGGATCACTACTAAAATATTCGGGTTT; PhBSMT TaqMan probe, 6FAM-

AAGGCACTCAATGTCTATTTTCGGTCGA-BHQ 1; PhBSMT2 forward primer,

TGTACCAATTCTCTATTGTTGTTTTGC; PhBSMT2 reverse primer,

CTGAAAGGACCCCTAGTGTACAAGA; PhBSMT2 TaqMan probe, 6FAM-

CTTCATAGGTGGTCGAGGTGCTAATTTATCTAGTC-BHQ1. TaqMan Real-Time

PCR reactions were run under the following conditions: 480C for 30 minutes, 95C for 10

minutes, followed by 40 cycles of 95C for 15 seconds and 600C for 1 minute. Reactions

were repeated twice with one set of RNAs and once with RNA collected from separate,

duplicate tissue. PCR reactions of in vitro transcibed PhBSMT] or PhBSMT2 standards

were run in duplicate and in tandem with the sample RNAs to generate a standard curve

from which the level of each PhBSMTmRNA in the samples was quantified.

Generation of Transgenic PhBSMT RNAi Petunias

RNAi constructs were made using the pHannibal RNAi cloning vector system

(Wesley et al., 2001). The region cloned into the pHannibal vector includes the coding

region from base 661 to base 1002 from PhBSMT] in sense and antisense orientations

flanking the intron segment. The RNAi chimeric gene was subsequently cloned into an

Agrobacterium transformation vector containing the neomycin phosphotransferase II

(NPTII) gene. This transformation vector was introduced into Agrobacterium

tumefaciens and used for transforming leaf explants from five week-old MD seedlings

grown in tissue culture according to the methods of Jorgenson et al (1996). Primary









transformants were grown under greenhouse conditions described above, and transgenic

plants were selected by PCR for presence of the neomycin phosphotransferase II gene

(NPTII), reduced PhBSMTmRNA levels, and reduced MeBA emission. Three

independent, transgenic lines were obtained: BSMT-9, BSMT-14, and BSMT-33.

Flowers were self pollinated to produce T1 progeny from these lines, which were

analyzed for presence of transgene by PCR analysis. Sixteen T1 progeny plants from

each line were analyzed for phenotypes by measurement of MeBA emission and

measurement of PhBSMT mRNA levels.

Volatile Collection and Analysis

Floral volatiles were collected from excised ethylene-treated, pollinated, and

PhBSMTRNAi flowers. Volatiles from exogenous ethylene and pollination timecourses

were treated or pollinated, then collected at the times indicated previously. An additional

untreated control was included for the ethylene experiments to control for flower excision

induced variability in air and ethylene treatments. For this control, flowers at the same

developmental stages were collected fresh from the plants to compare with air treated

control flowers. Three flowers were collected per treatment and each timepoint was

repeated 3 times. Flowers from the PhBSMTRNAi screen were collected three times

with 3-5 flowers per collection at 8:00 pm for the initial screen and once more with

putative positive lines at midnight to verify that reduced MeBA emission was reduced

when MeBA emission is maximal in MD wild type (Kolosova et al., 2001). Volatiles

were collected for one hour according to collection protocol described by Schmelz et al

(2001). Identification of each of the floral volatiles was verified by GC-MS (Schmelz, et

al., 2001).









Benzoic Acid & Salicylic Acid Extraction and Quantification

Benzoic acid and salicylic acid were extracted and quantified by GC-MS (Schmelz

et al., 2003). Petal tissue was excised from MD and 44568 whole flowers after 10 hrs

treatment with 2-3 p.L-L ethylene or 36 hrs after pollination and stored at -80C until

extraction. Two replicate sets of tissues were used for quantification.

Human Olfaction Panels

Human sensory panels were used to determine if differences in fragrance of the

reduced MeBA flowers and MD wild type flowers could be discriminated by human

olfaction. A triangle test was performed with sixty human subjects, each randomly given

a set of three unmarked flowers for sampling of the floral fragrances. The flower samples

were prepared from freshly excised flowers at anthesis from MD and knockout line

BSMT-9. Excised flowers were placed immediately into 5 mL water agarose blocks, then

placed into 210 mL glass jars and and sealed with lids for approximately 120 minutes

before testing. Each set of flowers consisted of two flowers of the same genotype and

one of the other genotype. The test performed both with two controls and one knockout

or two knockouts and one control and panelists were asked to judge which flower had a

different fragrance. Additional descriptive comments were also solicited from the test

subjects to determine if there were preferences in floral fragrance. The statistical

significance of the correct number of judgements was determined as described (Lawless

and Heyman, 1998).









S120 1.4
.o E MeBA
.h 100 1.2
E DBSMT-1
S 80- BSIVT-2 m
S0.8
60 --
0.6 3
o
20
c -0.4
20 0.2
O0 -- -' 0





Figure 4-1. PhBSMTRNAi reduces MeBA emission and PhBSMT mRNA. Mean (+ SE;
n=3) MeBA emission (blue solid bars, left axis) and mRNA levels (right axis)
of PhBSMTJ (striped, light purple bars) and PhBSMT2 (dotted, dark purple
bars) in T1 lines ofBSMT-9, BSMT-14, and BSMT-33. Silencing was
observed in BSMT-33 (represented by 33-16). Line 14-14 was transgenic, but
also silenced.









100


mA Benzldehyde
60 oB Benzyl Alcohol
i m mC Phenylacetaldehyde
mD Methyl Benzoate
mE 2-Phenylethanol
II *F Iso-Eugenol
10 -nG Benzyl Benzoate





ABCDEFG ABCDEFG
MD BSMT9-9

Figure 4-2. PhBSMTRNAi reduces MeBA emission only. Mean (+ SE; n=3) emission of
major volatiles in MD compared with most reduced MeBA T1 line BSMT9-9.
















mA Benzldehyde
SB Benzyl Alcohol
* C Phenylacetaldehyde
SD Methyl Benzoate
SE 2-Phenylethanol
* F Iso-Eugenol
D G Benzyl Benzoate


nrL~


ABCDEFG ABCDEFG ABCDEFG ABCDEFG ABCDEFG


Whole
flower


No limb Limb only No Corolla


ORS


Figure 4-3. Volatile emission patterns from MD floral organs. Volatiles were collected
from whole flowers, flowers with petal limb excised, limbs only, flowers with
excised corollas, and ovary+receptacle+sepals (ORS). Floral organs are
explained in the picture to the right: L, petal limb; T, petal tube; S,
stigma+style; 0, ovary; R, receptacle.













Stigma+Style


0 2 10 24 36 48
hours after ethylene
Petal Tube


0 2 10 24 36 48
hours after ethylene


I




I I


0 2 10 24 36 48
hours after ethylene


2 10 24 36
hours after ethylene


BSMT1 MD BSMT1 44568
BSMT2 MD BSMT2 44568
Figure 4-4. PhBSMTmRNA expression after ethylene treatment. Flowers were treated
with 2.5 ppm ethylene and then dissected for collection of individual floral
organs. RNA measurements are represented as mean + SE. Treatment times
are indicated on the x axis and % RNA on the y axis. Note differences in
scale.


Ovary