Application of Proven Agronomic Biotechnologies to Ornamental Plants

In 1996, interdisciplinary scientists ushered in a new age of collaboration between molecular biologists and plant breeders with the introduction of herbicide- and insect-tolerant transgenic plants on a commercial agronomic scale. In the past 6–7 years, herbicide and insect tolerance traits have been utilised in many different agronomic and vegetable crops, and now make up a significant percentage of the acreage planted yearly in the United States for corn, soybeans and cotton. Although there have been no published reports on the production of herbicide- or insect-tolerant ornamental crops, research is currently being conducted on turfgrass, and a select number of significant floriculture crops to engineer glyphosate resistance. The transfer of glyphosate resistance into creeping bentgrass is an obvious example of how proven agronomic traits can be used to make weed control in municipal and highly managed turf environments more efficient. It is very likely that any number of horticultural crops could be engineered with herbicide resistance, but the trait will probably only be commercially viable in crops grown in the field or planted in the ground in large public areas, such as vegetable crops, turfgrass, bedding plants and nursery crops. The use of herbicide resistance as a selectable marker in tissue culture, and subsequently as a ‘stacked’ trait on top of other introduced traits may influence the appearance of this technology in future crops grown for ornamental purposes as well.

From a broad perspective, it appears that the future for application of biotechnology to ornamental crops is promising. Many technical advancements have been made that have provided ‘proof of concept’ of their commercial utility in ornamental crops, and this work will likely supply the next generation of flower breeders with many novel traits. As progress is made in sequencing new plant genomes and as functional genomic tools become more widely used in crops of lesser economic value, scientists working to apply biotechnology to ornamental crops will likely find themselves with more novel traits to work on than they have people in their labs. There is no question that the main technical limitations that exist for applying biotechnology to ornamental crops lie mainly in the area of development of transformation systems for the large number of plant species used in the industry. For this reason, it will be imperative for breeders to introgress commercially viable traits into breeding stocks, whether their goal is to reproduce their crops by sexual or asexual means.

The real potential of biotechnology in ornamental crops is probably not going to be determined by any major difficulties encountered in the technical realm—it appears that there will be many genes and promoters that will have commercial viability in the ornamental plant industry. The factors that will determine how the ornamental plant industry accepts and utilises biotechnology will be more influenced by economic and regulatory issues. It is obvious that in terms of regulatory hurdles, ornamental crops have a particular advantage over food crops because they are non-edible. Since consumers do not physically ingest these products, it is likely that much public relations benefit can be gained in terms of consumer perception of genetically engineered crops by providing them an ornamental plant that is extremely novel and desirable. The introduction of new and novel plants has been the basis of progress in the ornamental plant industry since its inception, so it should be no surprise if that happens.

It is also logical to think that since there may be fewer regulatory hurdles for ornamental plants, the cost of introducing a new biotechnological ornamental crop would be reduced. However, the current regulatory restrictions that have been developed around the introduction of genetically engineered food crops will have to be revisited when regulatory packages for new ornamental crops are submitted to federal agencies. While it may be feasible to submit regulatory packages for each individual cultivar of genetically engineered corn or soybean, it will be cost prohibitive for any company to acquire all of the required regulatory data necessary for the hundreds of individual engineered cultivars that could be developed and potentially released by any particular breeding company. Since an individual ornamental crop has much less commercial value than any individual agronomic crop and a much shorter market life expectancy, the costs of gaining regulatory approval from federal agencies will have far more impact on the economic decisions made by those parties interested in engineering these crops. With the current royalty structures and profit margins in the ornamental plant industry, combined with a comparatively low market volume, it would take a company several years to recur the initial cost of their biotechnology investment, even when they capture large percentages of a given market share. In addition, many cut flowers and vegetative propagules used in the ornamental plant industry are produced in Africa, Central and South America and Europe, and are imported into the United States and all throughout Europe and Asia. With the current global status of genetically engineered plants in most of these regions being more restrictive than in the United States, any segment of the industry depending on import or export of products outside the United States will have far more complications in the physical movement of genetically engineered ornamental plants. These types of scenarios ultimately restrict the application of plant biotechnology to ornamental crops far more than any limitations at the technical end.

Phytohormone Synthesis and Perception

Flower colour and scent of ornamental crops are important candidate target traits for genetic manipulation with biotechnology tools because they are the obvious reasons that these crops are grown. However, there has been much research attention given to the manipulation of phytohormone synthesis and perception in ornamental crops due to the wide range of physiological processes in which they are involved. Currently, several issues in crop production and post-harvest handling in the ornamental horticulture industry reflect a need to be able to understand physiological processes that are normally controlled by plant hormones. For example, significant effort and money is spent applying chemicals to plants to control the synthesis of gibberellic acid (GA) and subsequent plant height during crop production. In many potted flowering crops and most bedding plant crops, as many as three to five applications of growth-regulating compounds may be used during a crop cycle. This adds a significant amount of production cost for chemicals and labour to apply and handle them, while being a negative factor environmentally. In addition, many of the newly introduced ornamental species that have not been in commercial production for a long period of time are receiving particular attention in conventional breeding and selection for dwarf plants because their natural habit does not fit into the mass-market operation scheme requiring compact plants that fit into transport vessels. It is apparent that the manipulation of endogenous GA concentration in plants through genetic engineering has the potential to produce ornamental and flowering plants with a diverse array of reduced-height phenotypes.

Considerable research over the past two decades has been conducted to isolate GA biosynthesis mutants (for review, see Hedden and Proebsting, 1999) and to elucidate the GA biosynthetic pathway (for review, see Hedden and Kamiya, 1997; Hedden and Phillips, 2000). The initial cloning and characterisation of gibberellin 20-oxidase genes from Arabidopsis (Phillips et al., 1995) and spinach (Wu et al., 1996) provided some of the first practical molecular tools needed to develop plants with altered endogenous GA levels. In Arabidopsis thaliana, GA 20-oxidase catalyses consecutive steps late in GA biosynthesis, and is encoded by three genes with differential patterns of expression (Phillips et al., 1995; Coles et al., 1999). Over-expression of GA 20-oxidase in Arabidopsis leads to increased production of active GA (GA4) and subsequent stem elongation, while expression of antisense GA 20-oxidase leads to reduced levels of active GAs and reduced rates of stem elongation (Coles et al., 1999). Heterologous over-expression of a pumpkin GA 20-oxidase gene driven by constitutive promoters in both transgenic Solanum dulcamara (Curtis et al., 2000) and lettuce (Lactuca sativa ‘Vanguard’) (Niki et al., 2001) also resulted in the production of dwarf plants. In both cases, endogenous concentrations of active forms of GA, such as GA1 and GA4 were reduced in transgenic dwarf plants, while concentrations of various inactive forms of GA such as GA25 were increased, indicating a diversion of the normal pathway of GA biosynthesis towards the production of inactive forms of GA.

More recently, additional tools for use in engineering dwarf plants have been developed as a result of the cloning and characterisation of genes involved in the regulation of GA degradation. The first steps in the degradation of biologically active GA appear to involve hydroxylation reactions that are similar to the final steps of GA synthesis. Similarly to the GA3-oxidases, which hydroxylate the 3-position of GA precursors to form active GAs (Chiang et al., 1995; Talon et al., 1990), GA-2 oxidases hydroxylate the 2-position of active C19-GAs as a first step to render them inactive (Sakamoto et al., 2001; Martin et al., 1999; Thomas et al., 1999). Two additional GA2-oxidases (AtGAox7 and AtGAox8) have recently been isolated from Arabidopsis that hydroxylate C20-GA precursors, but not active C19GAs (Schomburg et al., 2003in press). When over-expressed in transgenic Arabidopsis and tobacco, these genes result in plants with decreased levels of active GAs and a range of corresponding dwarf phenotypes. Transgenic CaMV35S::AtGAox7 and AtGAox8 petunias have a range of dwarf phenotypes, suggesting that the control of endogenous GA levels may be applicable to important floriculture crops as well (Figure 45.1). It is interesting to note that the dwarf phenotype can be reversed by treatment of plants with exogenous GA (Figure 45.2). This is very important for the application of dwarfing technology to the ornamental plant industry, because producers of vegetative propagules may actually want to temporarily increase shoot elongation at various points during cutting production, then have the plants resume a dwarf phenotype after propagation.Breeder selection for the appropriate plant height among transgenic lines will be critical for applying these technologies to any given crop, and breeders should be able to select any particular plant height they desire. Traditional horticultural experiments similar to those used in any breeding programme with a focus directed towards plant height control should be relatively straightforward, and are well established in the industry.

It is likely that dwarfing technologies will find utility in many areas of ornamental plant production. There should be ample room in the industry for potted plants that fit in transit vessels easily, annual bedding plants that do not grow out of the confines of a small garden spot, and shrubs and trees that require less pruning. One extremely important application for dwarfing technology will likely be observed in plant species used in the turfgrass industry. Slower growing lawn grasses that require significantly less labour input for maintenance have been a dream of suburban homeowners and turfgrass maintenance experts for many years. This concept is being developed by commercial scientists at the Scotts Company, with a great deal of progress being made in producing dwarf creeping bentgrass (Agrostis sp.). In high-management areas, these ‘low mow’ grasses have a great deal of potential for saving money on labour costs and equipment maintenance. In addition, slow-growing grasses could help reduce lawn mower pollution and use of fossil fuels, and may use less water and fertilisers. It is likely that these ‘low-mow’ turfgrasses will be important for homeowners and municipal or roadside situations where there is a low amount of foot traffic. It is not likely that these grasses will be used to any great extent on golf courses or sports turf facilities. Due to the high amount of wear and tear that is experienced under these conditions,they may not grow fast enough to cover the damage inflicted by constant use.

Another aspect of hormonal control of plant morphology currently being pursued in ornamental crops concerns the engineering of plants with delayed leaf senescence. Senescence is the final developmental process in the lifecycle of a leaf through which macromolecules (e.g. chlorophyll, proteins, nucleic acids) of leaf cells are metabolised to basic components and transported to the growing shoot and reproductive organs of the plant (Matile et al., 1996; Noodén et al., 1997). Natural leaf senescence in many plants is characterised by lower leaf yellowing or chlorosis as nutrients and other components of the cells are degraded (especially chlorophyll). In horticultural terms, leaf chlorosis can cause a decreased aesthetic appearance of ornamental plants and thus a decrease in the salability of those plants.

One way to prevent leaf senescence is through the manipulation of cytokinins. Cytokinins are an important class of phytohormones that influence numerous aspects of plant growth and development, and have been shown to delay and, in some cases, reverse the leaf senescence process (Gan and Amasino, 1996). The limitation of the use of cytokinins in the prevention of senescence has most often been related to difficulties of controlling temporal and spatial delivery of cytokinins. Exogenously applied cytokinins may not enter the cells, nor be transported to the area needed, nor quickly become conjugated or metabolised to non-active forms (Klee and Lanahan, 1995; Gan and Amasino, 1996; Kaminek et al., 1997). Researchers have recently turned to the manipulation and engineering of endogenous cytokinin biosynthesis in order to solve these problems.

After many years of research on cytokinins, the mechanisms of biosynthesis and perception are just now being elucidated. One enzyme, isopentenyl transferase (IPT), has received much research attention because of its known involvement in cytokinin synthesis. IPT enzyme activity has been isolated from many organisms including plants (Nishinari and Syono, 1980; Blackwell and Horgan, 1994), Dictyostelium discoideum (Taya et al., 1978; Ihara et al., 1984), and Rhodococcus fasciens (Crespi et al., 1992). Although it is well known that there is IPT activity in plants, the genes that encode IPT have only recently been isolated and cloned from plants (Kakimoto, 2001; Takei et al., 2001). As a result, experiments focused on the manipulation of endogenous cytokinin synthesis in transgenic plants with these genes have been lacking. In contrast, a gene from Agrobacterium tumefaciens that encodes the IPT enzyme has been available for use in transgenic plant research for a number of years. IPT is encoded on the Ti (tumour inducing) plasmid of Agrobacterium tumefaciens (Akiyoshi et al., 1984; Barry et al., 1984). It catalyses the condensation of dimethylallylpyrophosphate (DMAPP) and 5′AMP to form isopentenyladenosine 5′-phostphate ([9R-5′P]iP), which is then quickly converted to different cytokinins. This is presumed to be the rate-limiting step in cytokinin biosynthesis since expression of this one gene can cause an over-production of cytokinins (Medford et al., 1989; Klee and Lanahan, 1995; Gan and Amasino, 1996). Attempts have been made to use this gene under the control of various promoters, for example, modified CaMV 35S (Faiss et al., 1997), Cu2+-inducible (McKenzie et al., 1998), or heat shock inducible (Smart et al., 1991) to stop leaf senescence. However, the results of such experiments were often complicated by abnormal growth patterns, or the possibility that the induction treatment (i.e. high temperature or CuSO4 treatment) was causing the observed phenotype (Smart et al., 1991; Buchanan-Wollaston, 1997).

In an attempt to overcome the problems associated with past work on transgenic plants overproducing cytokinins, Gan and Amasino (1995) developed a genetic construct that utilised the highly senescence-specific promoter from SAG 12 (PSAG-12) to drive IPT expression. This construct had three important features due to PSAG-12 specificity: temporal regulation, spatial regulation and quantitative regulation (Gan and Amasino, 1995, 1996). When leaf senescence is triggered, the transcription of IPT is activated by PSAG-12, leading to the production of functional IPT enzyme. The enzyme then catalyses cytokinin production, which in turn delays senescence. Without senescence signals, the PSAG-12 promoter attenuates IPT transcription and subsequent enzyme production, thus providing autoregulatory control of cytokinin synthesis. The PSAG-12-IPT construct was transformed into tobacco. The transformed plants displayed a normal growth habit except that senescence was inhibited and there was a significant increase in flower number, an increased biomass due to the presence of lower leaves, and increased seed yield (Gan and Amasino, 1995). Since then, transgenic PSAG-12-IPT lettuce plants (McCabe et al., 2001) Nicotiana alata (Schroeder et al., 2001) and petunias (Clark et al., in press) with similar delayed leaf senescence phenotypes have been produced (Figure 45.3). The PSAG-12 promoter has also been used to drive the expression of the KNOTTED-1 mutant gene from maize in transgenic tobacco plants to confer a delayed leaf senescence phenotype (Ori et al., 1999), thus extending the utility of delayed leaf senescence technologies. Transgenic PSAG-12:KNOTTED-1 and PSAG-13-IPT petunias both have delayed leaf senescence after nutrition stress induced at the onset of flowering. It is likely that horticultural performance studies will show that breeder selection of transgenic plants under a wide range of selection criteria will be essential for providing the growers with the methods required to produce these plants effectively on a commercial scale.
One hormone of particular interest in the ornamental plant industry is ethylene gas. Ethylene is involved in many physiological processes in plants including fruit ripening, petal senescence, abscission and seed germination. Floral senescence is of particular interest because quality loss of many important floriculture crops is known to occur due to ethylene gas in the post-harvest environment. Treatment of ethylene-sensitive flowers with chemical ethylene biosynthesis or perception inhibitors delays visual symptoms of corolla senescence (Jones and Woodson, 1997; Whitehead et al., 1984; Serek et al., 1995). Many economically important crops are grown for their floral display; therefore the control of ethylene synthesis and perception is thought to be the key to increasing display life and enhancing visual quality. Since ethylene is such a significant problem in both potted flowering crops and cut flowers there have been several attempts to produce chemical control methods for both ethylene synthesis and sensitivity. Effective control of ethylene synthesis using a chemical approach has been much less effective across the industry than the manipulation of ethylene sensitivity. Ethylene sensitivity has long been managed in floriculture crops through the use of silver thiosulfate (STS), which makes plant tissue insensitive to ethylene, but also has environmental downsides that appear to restrict its commercial use. Another chemical approach gaining popularity within the industry is the use of 1-MCP (1-methylcyclopropene), which is a compound that blocks the ethylene receptor protein and makes plant tissue insensitive to ethylene. Typically applied as a gas, 1-MCP provides a means by which large amounts of tissue can be treated for a short time. Although this compound has proved to be effective in some crops, there is some difficulty with the use of 1-MCP in crops that continue to produce new ethylene receptor proteins through development during post-harvest transit, thus limiting the residual effect of the compound (Cameron and Reid, 2001).

A great deal of research has been conducted to produce floriculture crops that synthesise less ethylene, or are insensitive to ethylene. Transformation with an antisense ACC oxidase has been shown to be effective in producing carnation (Savin et al., 1995) and torenia plants (Aida et al., 1998) that produce significantly less ethylene, and have delayed flower senescence. As has been the case with chemical approaches to control the effects of ethylene in ornamental crops, the method of control receiving the most research attention in terms of genetic engineering has been at the level of ethylene perception. Wilkinson et al. (1997) transformed petunia with a dominant mutant Arabidopsis ethylene receptor, etr1-1, under the control of a constitutive cauliflower mosaic virus 35S (CaMV35S) promoter to produce ethylene insensitivity throughout the whole plant (Figure 45.4). These petunias had an increase in natural and pollination-induced flower longevity compared to wild-type plants, but had physiological side effects that limit their commercial use (Wilkinson et al., 1997; Gubrium et al., 2000). Ethylene-insensitive petunias showed a significant reduction in adventitious root formation, and even exogenous treatments with auxin did not increase adventitious root formation to untreated wild-type controls (Clark et al., 1999). Since many horticultural species are often propagated through vegetative cuttings this characteristic would severely limit the commercial utility of ethylene-insensitive plants propagated by vegetative cuttings. After much experimental effort, it is clear that the key to manipulation of ethylene sensitivity will be the promoters being used to drive transcription of these proven transgenes. Effective temporal and spatial control of transgene expression will ultimately lead to plants that have longer lasting flowers with no negative side effects that would limit commercial production. Bovy et al. (1999) have supported this idea by producing transgenic ethylene-insensitive carnations by driving the etr1-1 transgene with the flower-specific transcriptional promoter fbp1 from petunia. Although they were able to produce plants with a delayed flower senescence phenotype, they did not report heritability of the trait or extensive investigation of other horticultural performance characteristics that would limit commercialisation of these plants. Future experiments directed towards discovering the appropriate promoter to drive transgenes conferring ethylene insensitivity will likely be the key to success for the development of long-lasting flowers in the future. It will be imperative to conduct a significant amount of field and greenhouse trailing of these plants in order to completely eliminate the possibility of negative side effects.

Applications of Plant Biotechnology to Ornamental Crops

The ornamental plant industry is technically diverse and is characterised by the use of literally hundreds of different plant species. In many cases, numerous different cultivars of the same species may be available, and plants may be reproduced through both sexual and vegetative propagation methods. Due to the large amount of genetic diversity used by this industry, there are often very complex issues that arise during crop production and post-harvest handling throughout the wholesale and retail markets. For example, crop production practices for potted poinsettia and chrysanthemum plants require manipulation of the photoperiod to initiate flowering, and the application of growth regulators to keep the plants compact in habit. This type of production scheme is vastly different from the production methods used for producing bedding plant species such as petunia, geranium and impatiens, but a typical greenhouse grower may be producing all these plants simultaneously. At the post-harvest and shipping end of the industry, cut flower species grown in the tropical regions of Central and South America may have optimal shipping and storage conditions that are vastly different from the optimal conditions needed for cut flower species grown in Northern Europe. However, all the flowers may be exported to a common port such as Miami, then repackaged and shipped together in the same transport vehicle under the same conditions to retailers in virtually any part of the United States. This type of scenario often results in poor product performance when it reaches the florist or is displayed on retail shelves.

In addition to the demands of complex production and marketing systems, the demand from consumers for a constant supply of new and interesting flowering plants with unique characteristics continues to increase. In the past 5–10 years, there has been a dramatic increase in the number of flowering plant species available to consumers worldwide, and this trend is continuing. As a result, there is very little information available on appropriate production and post-harvest schemes for many of these crops. With such a diverse industry it is logical to assume that the application of biotechnology to ornamental crops is going to be very difficult. Applying biotechnology concepts is quite abstract in ornamental plants for several reasons. First, scientists attempting to improve these crops are trying to hit a moving target. Turnover of new cultivars is constant, and there are very few instances of consistent use of inbred lines in commercial breeding programmes. This means that by the time a researcher has transformed a new trait into a particular cultivar and proved that the trait is of commercial interest, the original cultivar may actually have been replaced by a new ‘improved’ cultivar. As a result, it will be of utmost importance to get commercially viable trangenes into breeding stocks, whether the goal is to produce a crop that will be reproduced by seeds or by vegetative propagules. Second, calculating the value added by a trait is very esoteric in ornamental crop species. Calculating increases in corn yield may be relatively straightforward, but determining how much a new colour of pansy is worth, or whether a more fragrant rose is something that someone will buy and actually pay more for, is a bit more cumbersome. Since it is well known that there is significant cost associated with developing an approach for improvement of any plant species using biotechnology, it becomes much harder to determine the potential economic value of a new ornamental cultivar until after it has proved to be successful. Much of the profit to be made in the ornamental plant industry through the use of biotechnology will be made by either the breeding companies who invest in technology or by retailers who have the ability to market the value of a novel trait. Value will also be added for growers if the introduced traits significantly reduce production costs, and this impact should be particularly significant because this segment of the industry currently has lower profit margins than any other. This scenario is very common in many types of plant-based agriculture, but with ornamental plant germplasm being spread amongst so many breeding companies, no one company can justify the investment in biotechnology that has been seen for crops of agronomic scale without making sure that the technology is a financial risk worth taking. It is likely that a vertically aligned strategy capturing as much value as possible from breeder to producer to retailer will be the best approach to making the benefits of biotechnology outweigh the high costs of technology development.

The last major issue that makes the concept of applying biotechnology approaches to ornamental crops more difficult is probably the most important—there is a significant bottleneck in the number of ornamental plant species that have been genetically transformed to date. Although there have been published reports describing the genetic transformation of several of the major floriculture crops such as chrysanthemum, rose, carnation and petunia, other important crops such as poinsettia, hibiscus and impatiens have still not been reliably transformed. Logically, most of the crops that have been the focus of transformation efforts to date have been those with the most value economically, or plants that have been discovered to be easier to transform or culture in vitro. Even in crops that have received a fair amount of research attention for genetic transformation, a significant amount of difficulty has been encountered with developing transformation protocols that can be used successfully on all cultivars of a given plant species. In many cases, transformation of different cultivars or breeding lines has required the development of several different transformation protocols. The more aggressive transformation efforts in ornamental crops are being undertaken by private corporations, with the remainder of the work being taken care of (one species or cultivar at a time) by smaller academic tissue culture laboratories.

In many cases, biotechnology applications are proving to be very difficult, but several advances have been made with engineering a wide variety of genetic traits in floriculture crops and turfgrass. Significant gains have been made in cloning important genes that are proving to be involved with biological processes that scientists hope to manipulate in ornamental crops in the future. The purpose of the rest of this chapter is to provide a status report on the progress that has been made in applying biotechnology to plants grown for their ornamental characteristics, and to project where research efforts will be focused over the next few years.

New Traits—Old Concepts

Flower Colour

Flower colour is a key component that influences consumer choice among crops grown commercially for their ornamental characteristics. Flower colour has been the subject of a large amount of biochemical and applied genetics research for many years (Mol et al., 1998). Flower breeders have been supplying the markets with new colours using traditional breeding techniques and selection, and to a lesser extent mutation breeding. The main biological function of flower colour pigments is to attract pollinators. Various pollinators are attracted to a flower by particular colours, and the patterning of flower coloration makes flowers easy for pollinators to distinguish as they move about.

Although many species have a wide variety of possible flower colours, no single species contains all possible flower colours. Flower colours are produced in plants biochemically as betalains, carotenoids and flavonoids. Betalains are synthesised almost exclusively in the Caryophyllales, and are yellow to red nitrogen-containing compounds derived from tyrosine (Stafford, 1994). Because of their limited presence in the plant kingdom, very little attention has been given to this class of compounds for the purpose of genetically engineering flower colour. Carotenoids are responsible for yellow and orange colours such as those observed in sunflower, marigold and tomato flowers, and over 600 different carotenoid structures have been identified (Straub, 1987). Carotenoids function mainly in photosynthesis by assisting with light harvesting and by quenching singlet oxygen and triplet chlorophyll species that are derived from excessive light energy (Demmig-Adams and Adams, 2000). Carotenoid synthesis and storage occurs in plastids, but all of the biosynthetic genes isolated to date are nuclear-encoded.

Although there are many plant species in which yellow flower colour is not produced and cannot be introgressed through traditional genetics, very little progress has been made in terms of genetically engineering yellow flower colour in plants. The general biochemistry of carotenoid synthesis has been studied for over thirty years, but the carotenoid biosynthetic pathway genes have only been identified and characterised since the 1990s (reviewed by Hirschberg, 2001). There are many different genes involved in carotenoid synthesis, and many of these genes are represented in plants in small multigene families (Zhu et al., 2002; Moehs et al., 2001). Since carotenoids are involved in such a range of diverse biological functions, most of the research on engineering carotenoid synthesis in transgenic plants has focused on the alteration of nutritional characteristics in food plants. Interestingly, rice engineered to over-express the daffodil phytoene synthase gene for increased pro-vitamin A content had a characteristic ‘golden’ colour (Burkhardt et al., 1997). Although these plants may be valuable for helping solve worldwide vitamin A deficiency, they also demonstrate the ability to drive the expression of yellow colour in plants. It is likely that progress in engineering yellow flower colour through the manipulation of carotenoid synthesis will be aided by observations on food crops with altered carotenoid content. It is also possible that attempts to genetically manipulate enzymes of the carotenoid biosynthesis pathway for yellow flower colour may affect unintended physiological processes in the plants leading to undesirable horticultural characteristics. Thus, the choice of appropriate transcriptional promoters to drive the expression of carotenoid biosynthetic genes in specific plant tissues will be imperative in determining the success of engineering yellow flower colour in ornamental plants.

Flavonoids are a class of secondary metabolites that are responsible for pale yellow, red, purple and blue colours in flowers. The flavonoid biosynthetic pathway, in particular, the anthocyanin synthesis pathway, has been extensively reviewed (Heller and Forkman, 1994; Forkman, 1994; Holton and Cornish, 1995). Hundreds of anthocyanins have been purified and their chemical structures have been determined (Strack and Wray, 1994), and virtually all of the genes that encode anthocyanin biosynthetic enzymes have been isolated. Thus, the anthocyanin biosynthetic pathway has served as an excellent target for transgenic manipulation because of the extensive background studies of its chemistry and genetics.

There are six main anthocyanins in plant tissues: pelargonidin, cyanidin, peonidin, delphinidin, petunidin and malvidin. These compounds are usually modified by glycosylation and acylation to produce a broader range of anthocyanins. In addition to the variety of anthocyanin chemical structures, colour variation produced by anthocyanins can be influenced by vacuolar pH, quantities of co-pigments and metal ions, and cell shape (Tanaka et al., 1998; Mol et al., 1998). Increased vacuolar pH is known to correlate with the ‘blueing’ of flowers of many plant species as they age and senesce (Yoshida, et al., 1995), and it is likely that both genetic and environmental factors influence the control of vacuolar pH. Genetic research on petunia has defined seven genetic loci that produce flower blueing when mutated (Chuck et al., 1993; van Houwelingen et al., 1998). Although these mutations are known to lead to measurable increases in the pH of petal extracts, but not an alteration of the actual anthocyanin composition, the actual mechanism of cellular pH control is still unknown. The range of the possible colours produced by anthocyanins can also be affected by the presence or absence of metal ions and co-pigments. In particular, the degree of blueness in anthocyanin pigments can be greatly influenced by metal ions and co-pigments which form stacked complexes with anthocyanins, and change their light absorption spectra (Kondo et al., 1992; Brouillard and Dangles, 1994). Perceived colour is also influenced by the shape of cells that accumulate anthocyanins. Petal epidermal cells more flattened in shape produce fainter colours, while conically shaped epidermal cells produce a sheen that gives the colours a more ‘velvety’ appearance (Noda et al., 1994). Since the underlying molecular mechanisms for changes in cell shape are poorly understood, it may be a few years before a complete understanding can be had of how pH changes and co-pigmentation and cell shape work together to produce an almost infinite array of flower colours through anthocyanins. However, it is well established that many genes involved in anthocyanin synthesis are regulated at the transcriptional level, which suggests that different flower colours and pigmentation patterns in flowers must be largely controlled by the expression patterns of regulatory genes (Holton and Cornish, 1995).

Over the past 10–15 years, many researchers have been able to successfully alter flower colour by manipulating the expression of anthocyanin biosynthetic genes in transgenic plants. Since no species has the ability to produce all possible flower colours, most of the research conducted to date has focused on the modification of already existing anthocyanin production systems in plants, or on the introduction of new biosynthetic enzymes that are not normally found in a particular species to make novel colours. For example, rose and carnation do not produce purple/blue delphinidin derivatives because they lack flavonoid 3′5′-hydroxylase (F3′5′H) activity (Mol et al., 1998). Also, plants such as petunia do not normally produce orange pelargonidin derivatives because the petunia dihydroflavonol reductase enzyme does not use the required dihydrokaempferol precursor as a substrate (Meyer, 1987).

Much of the research, focused on manipulating anthocyanin levels in flowers, has centred around the enzymes chalcone synthase (CHS) and dihydroflavonol-4-reductase (DFR). The CHS enzyme is the first enzyme committed to flavonoid production and catalyses the formation of chalcones, which are the intermediates used in the synthesis of all flavonoids. The DFR enzyme reduces dihydroflavonols to leucoanthocyanidins, another early rate-limiting step in anthocyanin synthesis. Many attempts at producing white flowers by suppression of flavonoid synthesis through the suppression of CHS and DFR activities have been successful. This approach is useful in a practical sense because in many species used as an ornamental crop, it has proved to be rather difficult to produce purely white flowering plants through traditional breeding methods.

Suppression of CHS via antisense or co-suppression has been demonstrated to suppress anthocyanin formation in a wide variety of plant species such as petunia (van der Krol, 1988; Napoli et al., 1990), chrysanthemum (Courtney-Gutterson et al., 1994), gerbera (Elomaa et al., 1993) lisianthus (Deroles et al., 1998) and torenia (Aida et al., 2000a). Similar to CHS, suppression of anthocyanin synthesis has also been achieved through suppression of DFR activity, leading to reduced anthocyanin synthesis in petunia (van der Krol, 1990) and torenia (Aida et al., 2000b). With the suppression of both CHS and DFR, there have also been reports of the generation of novel flower colour patterns in addition to the reduction in anthocyanin in petunia (van der Krol, 1988, 1990), torenia (Aida et al., 2000a, 2000b) and lisianthus (Deroles et al., 1998; Bradley et al., 2000). Interestingly, a consensus developed as a result of suppression of CHS and DFR is that the consistency of new flower phenotypes in these transgenic plants can vary within and between individual transgenic lines (van der Krol et al., 1990; Meyer et al., 1992; van Blokland et al., 1993; Elomaa and Holton, 1994; Deroles et al., 1998). Since colour patterns of some transgenic CHS lines of lisianthus have proved to be more stably inherited than others (Bradley et al., 2000), it is likely that any transgenic approach to the alteration of flower colour or colour patterning will require a significant amount of breeder selection over successive generations to stabilise the phenotype.

Controlled over-expression of anthocyanin biosynthetic genes has resulted in the production of novel flower colours in transgenic plants. One particularly dramatic example has been illustrated with petunia plants engineered for orange coloration. Normally, petunia does not produce orange colour because its intrinsic DFR protein does not accept dihydrokaempferol as a substrate, so pelargonidin-based pigments are usually absent. Maize DFR, which has a different substrate specificity than the petunia DFR, is able to produce pelargonidin derivatives if dihydrokaempferol is present and available. By expressing the maize DFR gene in petunia, Oud et al. (1995) were able to produce flowers with a brick red colour resulting from the accumulation of pelargonidin-derived pigments. Although this colour was not considered commercially acceptable, hybrids based on F4 genetic lines derived from commercial germplasm were obtained that had a unique orange flower colour. By introgressing the maize DFR gene into various breeding lines, DFR expression was stabilised, thus establishing new colour profiles that were not already present in petunia. Once stabilised through successive generations of breeding, the maize DFR gene behaved normally, and was used to successfully develop F1 petunia varieties with orange flower colour with no deleterious side effects (Oud et al., 1995).

Perhaps the best-known example of the production of novel flower colours in plants has been the transgenic ‘blue’ carnations produced by Florigene Ltd. and Suntory Ltd., which have been marketed in the United States, Australia and Japan under the name ‘Moonshadow’. These transgenic carnations contain the petunia flavonoid 3′5′-hydroxylase gene (F3′5′H), which encodes a cytochrome (cyt) P450 enzyme that catalyses the 3′5′ hydroxylation of dihydroflavonols, the precursors of purple anthocyanins. The F3′5′H enzyme activity is normally absent in carnation (and roses), but when the petunia F3′5′H gene is expressed in carnation, the plants produce and accumulate delphinidin, making the flowers a violet-bluish colour. It is likely that further improvements will be possible in producing blue colour using F3′5′H combined with other enhancers of anthocyanin pigmentation. It is known that the addition of cyt b5 activity enhances cyt P450 hydroxylation reactions, and is required for full activity of F3′5′H (de Vetten et al., 1999). Since the substitution pattern of anthocyanin pigments is a main determining factor in flower colour, it is thought that the stimulation of F3′5′H activity by the cyt b5 protein will be useful in enhancing blue flower colour (de Vetten et al., 1999).

Although much research has been conducted to date on anthocyanin biosynthesis and manipulation in plants, many issues related to the mechanisms of flower colouration are still poorly understood. To date, success has been achieved in the manipulation of flower colour through engineering anthocyanin biosynthetic genes and proteins. It is encouraging that most of the transgenic plants resulting from these efforts have been produced through driving transgene expression with constitutive promoters, and few deleterious side effects have been observed (Tanaka et al., 1998). However, the controlled manipulation of co-pigmentation and vacuolar pH have still not been addressed to a great extent. Once these factors are understood, the production of blue colour in flowers may become a reality.