Flavonoid gene expression and metabolite profiling during fruit development in highbush blueberry (Vaccinium corymbosum L.)

(1)

blueberry (Vaccinium corymbosum L.)

by Michael Zifkin

BSc, St. Francis Xavier University, 2007 A Thesis Submitted in Partial Fulfillment of the Requirements for the Degree of

MASTER OF SCIENCE in the Department of Biology

 Michael Zifkin, 2010 University of Victoria

(2)

Supervisory Committee

Flavonoid gene expression and metabolite profiling during fruit development in highbush blueberry (Vaccinium corymbosum L.)

by Michael Zifkin

BSc, St. Francis Xavier University, 2007

Supervisory Committee

Dr. C. Peter Constabel, Department of Biology Supervisor

Dr. Jürgen Ehlting, Department of Biology Departmental Member

Dr. Robert L. Chow, Department of Biology Departmental Member

(3)

Abstract

Supervisory Committee

Dr. C. Peter Constabel, Department of Biology Supervisor

Dr. Jürgen Ehlting, Department of Biology Departmental Member

Dr. Robert L. Chow, Department of Biology Departmental Member

Highbush blueberry (Vaccinium corymbosum L.) has one of the highest antioxidant capacities and flavonoid concentrations of any fruit or vegetable, and regular

consumption of blueberries has been connected to a wide range of health benefits. A diversity of flavonoids (flavonols, anthocyanins, proanthocyanidins) are likely responsible for many of the health benefits, and these compounds also significantly contribute to the organoleptic properties of ripe blueberries. Despite the potential importance of these flavonoids in diet, there has been little investigation into the molecular genetics of blueberry flavonoid biosynthesis. Therefore, I developed a real-time quantitative PCR protocol to monitor expression of flavonoid genes throughout development and ripening. Following evaluation of five reference genes, expression profiling of biosynthetic genes revealed that flavonoid synthesis is tightly controlled at the transcriptional level in a biphasic developmental pattern. These results are discussed in relation to flavonoid metabolite accumulation profiles, which were produced as part of a collaboration. Finally, in conjunction with a second group of collaborating scientists, some promising preliminary evidence is provided suggesting that the hormone abscisic acid might have a role in regulating ripening initiation in blueberry.

(4)

List of Tables

Table 1-1. Summary of proteins known to be directly involved in flavonoid biosynthesis. ... 17 Table 2-1. Gene-specific primers used for real time qPCR designed using Vector NTI Advance 9. ... 37 Table 2-2. Blueberry fruit EST library statistics. ... 44 Table 2-3. Top 40 highly represented unigenes in the two blueberry fruit EST libraries. 46 Table 2-4. Blueberry unigenes used for qPCR analysis. ... 49 Table 2-5. Reference gene statistics, stability values and rankings. ... 57 Table 2-6. Summary of blueberry PA subunit composition following acid hydrolysis and phloroglucinol derivatization. ... 62

(7)

List of Figures

Figure 1-1. The most commonly occurring anthocyanidin, flavonol and flavan-3-ol

structures. ... 3

Figure 1-2. Simplified flavonoid biosynthesis pathway leading to flavonols, anthocyanins and PAs. ... 13

Figure 1-3. Simplified pathway of abscisic acid (ABA) metabolism in plants. ... 22

Figure 2-1. Blueberry developmental staging system... 42

Figure 2-2. Phylogeny of the flavonoid R2R3 MYB transcription factor family. ... 50

Figure 2-3. Phylogeny of the two related flavonoid hydroxylase families. ... 51

Figure 2-4. Phylogeny of the DFR and ANR protein families. ... 52

Figure 2-5. Phylogeny of a subset of the flavonoid-O-glycosyltransferase family. ... 53

Figure 2-6. Evaluation of reference genes for qPCR. ... 55

Figure 2-7. Proanthocyanidin gene expression throughout blueberry development. ... 59

Figure 2-8. Proanthocyanidin metabolite accumulation throughout blueberry development. ... 61

Figure 2-9. Expression of genes involved in flavonoid hydroxylation and flavonol and anthocyanin synthesis. ... 65

Figure 2-10. Seasonal flavonol and anthocyanidin aglycone accumulation profiles. ... 66

Figure 2-11. Localization of PAs and PA gene expression in blueberry fruit. ... 68

Figure 2-12. ABA biosynthesis and metabolism in developing blueberry fruit. ... 71

Figure 3-1. The reaction kinetics of product amplification during a qPCR run. ... 83

Figure 3-2. Analysis of RNA integrity. ... 87

Figure 3-3. Amplification from no reverse transcriptase (NRT) and no template (NTC) control wells. ... 90

Figure 3-4. Amplicon dissociation analysis. ... 96

Figure 3-5. Serial template dilution analysis of dynamic range and efficiency. ... 98

(8)

List of Supplementary Tables (Appendix B)

Supplementary Table 2-1. Putative ripening-associated ESTs in blueberry fruit EST libraries. ... 152

(9)

List of Supplementary Figures (Appendix B)

Supplementary Figure 2-1. Gene ontology (GO) categorization of blueberry unigenes. 155 Supplementary Figure 2-2. Flavonol localization in blueberry fruit cross-sections. ... 156 Supplementary Figure 2-3. Additional hormone quantification. ... 157

(10)

Acknowledgments

I would especially like to thank Pat Kerfoot for allowing me to freely sample fruit from his picturesque organic blueberry orchard (Sweet Briar Farm, Saanich, BC) throughout the duration of the project.

I am grateful for the work of our collaborators on this project: Alena Jin and Dr. Jocelyn Ozga at the University of Alberta, and Irina Zaharia and Dr. Suzanne Abrams at the Plant Biotechnology Institute in Saskatoon. Their hard work and expertise helped make the project a more complete story.

I would like to thank my supervisor Dr. Peter Constabel for allowing me to explore and work on a really interesting topic, and I would like to thank my committee members Dr. Jürgen Ehlting and Dr. Bob Chow for their assistance, attention and encouragement.

Finally, thank you to those past and present members of the lab that were always there to lend a hand and share and discuss their scientific interests and ideas.

(11)

1. Chapter One: General Introduction

1.1 Introduction to blueberry fruit

From an agricultural and human nutrition perspective, fruits are among the most important plant tissues. Fruits are limited to flowering plants, the angiosperms. Most common in nature are the dry fruits (e.g. cereal grains, nuts, legumes, Arabidopsis siliques), but fleshy fruits (e.g. berries, pome fruits) make up a significant contribution to the human diet as well. Fleshy fruits are most noticeably characterized by a sweet, juicy and soft flesh, and attractively colored skin. These characteristics likely evolved from dry fruit and function to attract and reward fruit-consuming animals (Seymour et al., 2008).

Blueberries (Vaccinium sp., section Cyanococcus of the Ericaceae family) are the second highest grossing fleshy fruit crop in Canada (behind tomatoes), trailing only the USA in total blueberry sales and production (United Nations, 2008). In Eastern Canada, the majority of blueberry production is from managed ‘wild’ lowbush plants (Vaccinium angustifolium Ait.), while cultivated autotetraploid highbush blueberries (Vaccinium corymbosum L.) are the preferred blueberry fruits in Western Canada (Hancock et al., 2008). Highbush blueberries have been cultivated since the early 1900s and tend to be larger than lowbush blueberries, which are more astringent (Hancock, 2008). The development and ripening of fleshy fruit such as blueberry is a complex process that is regulated coordinately with seed maturation. At key points in the growing season, specific biochemical and physiological activities occur, involving cell division and expansion, cell wall and tissue softening, metabolism of organic acids, and accumulation of soluble sugars, flavor compounds, anthocyanins and other flavonoids (Coombe, 1976; Brady, 1987; Klee, 2010).

Fruits are important components of a healthy, balanced diet, which is exemplified by the fact that an inverse correlation is often observed between regular fruit

consumption and incidence of cardiovascular disease, stroke and some types of cancer (Rimm, 2002; Dauchet et al., 2006; He et al., 2006). The health benefits of fruit are often attributed to the presence of various vitamins and minerals (Van Duyn and Pivonka, 2000). However, a specific group of ‘secondary’ metabolites that are known as

(12)

flavonoids have been implicated in the prevention of these diseases and as a consequence they have received considerable research focus (see e.g. Middleton et al., 2000; Santos-Buelga and Scalbert, 2000; Liu, 2003; Rasmussen et al., 2005; Zafra-Stone et al., 2007; Aron and Kennedy, 2008). Blueberries contain a substantial quantity and wide diversity of flavonoids compared to other fruits and vegetables (Gu et al., 2005; Wu et al., 2006), and this has been linked to a number of health benefits associated with blueberry

consumption (Beattie et al., 2005; Seeram, 2008). Despite this, very little is known about the genetics and biochemistry of flavonoid biosynthesis in blueberry.

In this introduction, I will first provide a general overview of flavonoids including structures, functions and potential health effects. In the second half, I will lead up to the objectives of my thesis by describing what is currently known about the molecular genetics of flavonoid synthesis and how flavonoid synthesis is regulated in coordination with fruit maturation in a selection of other species.

1.2 The diversity of flavonoid structures and functions

The three most common classes of flavonoids in fruit are the flavonols,

anthocyanins and proanthocyanidins (PAs, also referred to as condensed tannins). Each of these derives from a common structural skeleton known as a flavan, which consists of two C6 aromatic rings (called A ring and B-ring, respectively) linked by a C3 heterocyclic

ring (C-ring; Fig. 1-1A). Differences within each flavonoid class are due to hydroxylations and in some cases subsequent methoxylations, glycosylations and

acylations (Fig. 1-1). Variation in hydroxylation pattern most often occurs on the B-ring. Methoxylation can occur in both anthocyanins and flavonols, but is most often seen in anthocyanins. The level of B-ring modification in anthocyanins is generally proportional to their shade of blue (Tanaka et al., 2008). Glycosylation of anthocyanidin, flavonol and occasionally flavan-3-ol aglycones increases solubility and stability, while acylation with aromatic or aliphatic groups can also promote stability. The variety of specific acylated and glycosylated anthocyanins in a plant is species specific, but the majority of

anthocyanins found in nature derive from six common anthocyanidin structures (Fig. 1-1). Biosynthesis of flavonoids occurs in the cytosol, followed by storage in the central vacuole (Halbwirth, 2010).

(13)

Figure 1-1. The most commonly occurring anthocyanidin, flavonol and flavan-3-ol structures.

In the flavan-3-ol structures, the ligand on C3 can be either trans or cis to the aromatic B-ring on

C2 and gallic acid (GA) is occasionally present at R3. The trans flavan-3-ols can be collectively

referred to as catechins, and the cis flavan-3-ols as epicatechins. Anthocyanidin R1 R2 Pelargonidin H H Cyanidin OH H Peonidin OCH3 H Delphinidin OH OH Petunidin OCH3 OH

Malvidin OCH3 OCH3

Flavonol R1 R2 Kaempferol H H Quercetin OH H Isorhamnetin OCH3 H Myricetin OH OH Larycitrin OH OCH3

Syringetin OCH3 OCH3

Flavan-3-ol 2,3 stereochemistry R1 R2 R3

Afzelechin trans H H H

Catechin trans OH H H

Gallocatechin trans OH OH H

Epiafzelechin cis H H H

Epicatechin (EC) cis OH H H

EC-gallate cis OH H GA

Epigallocatechin (EGC) cis OH OH H

EGC-gallate cis OH OH GA A B C C A B

(14)

The pH of the vacuole and availability of some heavy metals can also modestly affect anthocyanin color (Tanaka et al., 2008).

Also in the vacuole, flavan-3-ols can be polymerized into PA oligomers as short as two units up to polymers around 100 units. The mechanism of polymerization is unknown, but PAs in most species are typically enriched in flavan-3-ols of the 2,3-cis stereochemistry, which are often collectively referred to as epicatechins (Fig. 1-1).

While technically not integral for the survival of a plant, it is now known that many ‘secondary’ metabolites, including flavonoids, do significantly improve fitness (Dixon and Paiva, 1995; Shirley, 1996; Winkel-Shirley, 2002; Taylor and Grotewold, 2005; Agati and Tattini, 2010). PAs are typically found in high concentrations in leaves of many plants as well as the skin and seeds of immature fruit. They are notable for their astringent and bitter organoleptic properties, suggesting a protective function against premature or unwanted feeding, although this has not been conclusively established (Levin et al., 1976; Herrera et al., 1982; Wrangham and Waterman, 1983; Kreuger and Potter, 1994; Ayres et al., 1997; Cipollini and Levey, 1997; Brossaud et al., 2001; Treutter, 2006). Further supporting a defensive role, they have significant in vitro antioxidant (Heim et al., 2002) and antimicrobial activity (Scalbert et al., 1991), and are biosynthetically induced in response to a number of biotic and abiotic stresses (Kouki and Manetas, 2002; Peters and Constabel, 2002; Paolocci et al., 2004; Miranda et al., 2007; Constabel and Mellway, 2009; Mellway et al., 2009; Akagi et al., 2010). Moreover, PAs are found in seed coats of most seed-producing plants. As darkly pigmented cross-linked structures, they provide a mechanical barrier against fungal and bacterial invasion. They also help maintain dormancy and prevent premature germination by inhibiting oxygen and water intake (Debeaujon et al., 2000).

Flavonols are the most ancient and ubiquitous of the flavonoids (Winkel-Shirley, 2002). They are usually found in highest concentrations in sun-exposed leaves and fruit skin, which spatially supports their role as protective chemicals against UV-B light (Li et al., 1993; Solovchenko and Schmitz-Eiberger, 2003; Agati and Tattini, 2010). In

addition, in flowers they can act as visual cues for pollinators that can see in the UV wavelength range. They can also be found in seed coats, where they assist PAs in pigmentation and maintaining dormancy (Auger et al., 2010).

(15)

Anthocyanins often accumulate in significant quantities in stressed and senescing leaves, but are perhaps most noticeable in flowers and fruit skin. In flowers, they function as pollinator cues, and in the skin of ripe fruit they function as red and blue pigments to attract animals. Animals that consume the fruit disperse seeds and aid reproductive potential (Wilson and Whelan, 1990; Allen et al., 2008). Therefore, flavonoids are important in ensuring Darwinian fitness in plants.

1.3 The potential health effects of blueberry flavonoids

Blueberries have attracted much attention due both to their superior antioxidant capacity compared to other fruits and vegetables (Prior et al., 1998), and their health benefits (Beattie et al., 2005; Seeram, 2008). Blueberry extracts have been implicated in slowing the development of cancer, cardiovascular disease, colitis, diabetes, liver

toxicity, immune deficiency, hypertension and neurodegenerative diseases in a variety of experimental and clinical experiments (Sweeney et al., 2002; Beattie et al., 2005; Lau et al., 2005; Martineau et al., 2006; Neto, 2007; Shaughnessy et al., 2008; 2009; Shukitt-Hale et al., 2008b; Seeram, 2008; 2010; Stull et al., 2010; Wang et al., 2010; Wu et al., 2010). While flavonoids alone have not been attributed to the positive effects in all of these studies, they are often suggested as a causal factor.

Blueberries produce a considerable quantity of flavonoids, more than most other common crops. For example, Gu et al. (2004) determined that amongst the most

commonly consumed fruits, blueberries ranked behind only cranberries and chokeberries in concentration of total PAs (cultivated highbush: 179.8 ± 50.8 mg/100 g-1 fwt; lowbush: 331.9 ± 14.0) and had one of the higher percentages of polymeric PAs (76 %). In

addition, blueberry ranked first in structural diversity and within the top five foods in concentration of anthocyanins (cultivated highbush: 386.6 ± 77.7 mg/100 g-1 fwt; lowbush: 486.5), over ten times the amount in grapes and apples (Wu et al., 2006). Blueberries also contain more flavonols than most fruits, including grapes, apples and raspberries (Justesen et al., 1998). The significant concentration of flavonoids in

blueberries results in high in vitro antioxidant activity (Prior et al., 1998; Ehlenfeldt et al., 2001; Sanchez-Moreno et al., 2003), which is most commonly measured using the

(16)

The most promising evidence generally supporting flavonoid-based health

benefits has come from epidemiological studies on healthy and at-risk populations, which have often found an inverse correlation between flavonoid intake (from whole foods) and relative risk of developing cardiovascular diseases (Hertog et al., 1993; 1995; Knekt et al., 1995; Geleijnse et al., 1999; 2002; Duffy et al., 2001; Hollman, 2001; Huxley and Neil, 2003; Beattie et al., 2005; Rasmussen et al., 2005; Hooper et al., 2008; Mursu et al., 2008; Hollman et al., 2010). Although no long-term epidemiological studies have been undertaken to determine the specific, individual effects of PA and anthocyanin

consumption on cardiovascular disease, short-term clinical experiments with grape seed, tea, cocoa and cranberry extracts (rich in flavan-3-ols and/or PAs) have shown a

reduction in low-density lipoprotein (LDL) oxidation and total serum LDL levels, as well as improvement in endothelial function (Arai et al., 2000; Heiss et al., 2005; Rasmussen et al., 2005; Schroeter et al., 2006; Baba et al., 2007; Balzer et al., 2008; Milbury et al., 2008; Monagas et al., 2009). Controlled in vitro experiments with flavonoid extracts from fruit, tea, wine and cocoa have generated similar results (Wilson et al., 1998; Middleton et al., 2000; Youdim et al., 2002; Beattie et al., 2005; Neto, 2007; Seeram, 2008).

A few epidemiological association studies have also revealed that flavonoids may prevent the development of certain cancers (Hertog et al., 1996; Peterson et al., 2003; Gates et al., 2007; Arts, 2008; Rossi et al., 2010). However, in general the evidence for flavonoid-based cancer prevention is not as convincing as the cardiovascular health research (Bobe et al., 2009), as most of the positive results have come from in vitro animal models and cell culture experiments (reviewed in Middleton et al., 2000; Santos-Buelga and Scalbert, 2000; Rasmussen et al., 2005; Zafra-Stone et al., 2007; Aron and Kennedy, 2008). A number of these experiments have used flavonoid-rich blueberry extracts (Beattie et al., 2005; Matchett et al., 2005; 2006; Yi et al., 2005; 2006; Schmidt et al., 2006; Seeram et al., 2006; Neto, 2007; Seeram, 2008; Bae et al., 2009; Gordillo et al., 2009; Stoner et al., 2010). The results of these studies are complicated by the fact that the authors often used flavonoid quantities well above predicted physiological

concentrations (Crozier et al., 2009).

Although the majority of flavonoid health research has been focused on the anticancer and cardioprotective properties of flavonoids, the most promising results on

(17)

blueberry flavonoids are related to their neuroprotective effects (Shukitt-Hale et al., 2008b; Seeram, 2008; Spencer et al., 2009). In general, it is thought that dietary antioxidant intake associated with regular fruit and vegetable consumption can slow progression of dementia and prevent stroke (Sweeney et al., 2002; Solfrizzi et al., 2003; ; Dauchet et al., 2005; Del Parigi et al., 2006; Barberger-Gateau et al., 2007; Spencer et al., 2009).

Long-term feeding trials in aging adult rats have revealed that flavonoid-rich blueberry extracts can improve cognitive function compared to controls, and are more effective than strawberry and spinach at enhancing motor performance (Joseph et al., 1999). Other experiments have confirmed this and also found that blueberry can significantly improve balance and coordination in aging rats (Youdim et al., 2000; Goyarzu et al., 2004). Furthermore, blueberry flavonoids can improve memory and promote neurogenesis in older humans and rats (Casadesus et al., 2004; Duffy et al., 2005; Ramirez et al., 2005; Barros et al., 2006; McGuire et al., 2006; Williams et al., 2008; Papandreou et al., 2009; Krikorian et al., 2010), and can even prevent cognitive retardation in irradiated juvenile rats (Shukitt-Hale et al., 2007). All of these effects have been attributed especially to anthocyanins and PAs (Spencer et al., 2009), although experimental evidence with purified blueberry flavonoids seems to be lacking.

Supporting the flavonoid-based hypothesis is the finding that small amounts of flavan-3-ols and anthocyanins are detectable in the brains of rats and pigs, and the concentration of anthocyanins seems to be proportional to cognitive effects (Andres-Lacueva et al., 2005; Talavera et al., 2005; Kalt et al., 2008; Faria et al., 2010; Milbury and Kalt, 2010). Despite these promising results, both short-term clinical and long-term intervention studies are needed to determine the extent to which specific blueberry flavonoids truly enhance memory and cognition and protect against age-related neurological decline in human subjects.

It should be noted that the evaluation of flavonoid health benefits is complicated by the issue of bioavailability. For flavonoids to truly exert effects on cancer,

cardiovascular health and the central nervous system, they likely have to be absorbed and metabolized. Early studies on the bioavailability of flavonoids often concluded that very small amounts were absorbed and that the compounds were quickly excreted (Manach et

(18)

al., 2005). However, more elaborate work using highly accurate HPLC-MS techniques suggests that some flavonoids and flavonoid catabolites may be more bioavailable than once thought (Crozier et al., 2009).

Flavan-3-ols are the most readily absorbed of all the flavonoids (Donovan et al., 1999; het Hof et al., 1999; Gonthier et al., 2003; Henning et al., 2005; Tsang et al., 2005; Schroeter et al., 2006). For example, in two associated studies, volunteers who consumed a green tea extract or drank a green tea beverage rich in epicatechin and epigallocatechin metabolized and excreted the equivalent of up to 47 % and 26% of each of these

respective flavan-3-ols (Auger et al., 2008; Stelmach et al., 2009). Similar findings were reported from human studies involving consumption of a cocoa drink, also rich in epicatechin and epigallocatechin (Schroeter et al., 2006).

Most research to date has found that whole polymeric PAs are not absorbed at all (Donovan et al., 2002; Manach et al., 2005; Espin et al., 2007; Crozier et al., 2009). Instead, dimers, trimers and some PA oligomers may be degraded into hydroxyphenolic acids by bacteria in the colon and then absorbed, with relative percentage of degradation decreasing with increasing degree of polymerization (Deprez et al., 2000; 2001; Rios et al., 2003; Ward et al., 2004). However, degradation of larger PAs appears to be limited, and they are generally not depolymerized into their monomeric constituents (Donovan et al., 2002; Rios et al., 2003; Gonthier et al., 2003; Tsang et al., 2005). In addition, work in both rats and humans has shown that very small amounts of dimers, trimers and even pentamers can be absorbed intact (Holt et al., 2002; Sano et al., 2003; Tsang et al., 2005; Shoji et al., 2006; Serra et al., 2010). Following two recent reviews of the current

literature on PA bioavailability, Crozier et al. (2009) and Serrano et al. (2009) both concluded that much more work needs to be done on PA metabolism and PA catabolite absorption before we can know their true bioavailability and bioactivity.

Flavonols and anthocyanins appear to be absorbed intact in much lower

concentrations than flavan-3-ols. While some flavonol-glycosides accumulate in human blood as glucuronated, sulfonated and methylated quercetin-conjugates quickly after ingestion, total excretion amounts to less than three percent of total flavonol intake (Day et al., 2001; Jaganath et al., 2006; Mullen et al., 2006; Krogholm et al., 2010). However, a number of clinical studies have since found that the equivalent of greater than 20

(19)

percent of the total flavonols ingested are absorbed and excreted as hydroxyphenolic acid catabolites, suggesting that many of the flavonols are degraded by colonic bacteria in the large intestine (Jaganath et al., 2006; Vorsa et al., 2007; Mullen et al. 2008; Lehtonen et al., 2010). Anthocyanins are absorbed in even lower quantities, typically being excreted from urine in amounts less than one percent of the total ingested anthocyanins

(Matsumoto et al., 2001; Wu et al., 2002; Manach et al., 2005; Espin et al., 2007; Crozier et al., 2009; Garcia-Alonso et al., 2009; Milbury et al., 2010). However, as is the case with flavonols, a high concentration of hydroxyphenolic acids are found in urine and plasma of rats following ingestion of anthocyanin-rich juice, suggesting that some anthocyanins are degraded by colonic microflora as well (Borges et al., 2007). The role of anthocyanin-derived phenolic acids in health effects is supported by studies which find that while less than 0.1 % of ingested anthocyanins are detectable in urine after 24 hours, a substantial increase in plasma antioxidant status occurs after anthocyanin ingestion (Garcia-Alonso et al., 2009).

In summary, despite the vast number of promising results from in vitro cell culture experiments, most flavonoids are not directly absorbed intact into human plasma in appreciable amounts. Therefore, the significance of these types of studies is unclear. However, both short-term human intervention and long-term epidemiological studies strongly suggest that flavonoids have either a direct or indirect protective function against cardiovascular disease, and probably cancer and neurodegenerative diseases as well. To fully establish roles for specific flavonoid classes in human disease prevention, more short-term clinical trials with purified extracts should be performed, because at this point it is difficult to discern a clear advantage for any one flavonoid class (Crozier et al., 2009). Furthermore, given the issue of low bioavailability, it seems appropriate for future mechanistic studies to focus instead on the metabolic and cellular effects of flavonoid-derived metabolites and catabolites (Donovan et al., 1999; McGhie and Walton, 2007; Crozier et al., 2009).

(20)

1.4 Importance of gene-metabolite correlations in understanding the molecular genetics of flavonoid biosynthesis

In order to answer fundamental questions in biology, it is often efficacious to study the expression of genes, as gene expression regulation is a main determinant of cell form, function and plasticity in every known organism. In addition, observing the

expression of specific genes or small gene networks can provide important clues to the function of individual gene products (i.e. proteins). In agricultural science, genes and proteins of interest to researchers are often those related to crop yield and growth rate, stress tolerance, adaptability, and synthesis of nutritionally-important metabolites. Developmental, physiological and metabolic processes (e.g. flavonoid synthesis) of crop plants can be altered following the identification and characterization of genes and proteins that are most integral to these processes (Tanksley and McCourt, 1997). The development of germplasm banks along with the advancement of marker-assisted breeding and genetic modification technologies can allow for faster, specific and more effective alterations of crop plants (Francia et al., 2005; Varshny et al., 2005).

A common method to understand transcriptional regulation of flavonoid

biosynthesis in plants is to first profile the expression of known or putative biosynthetic and regulatory genes (Winkel-Shirley, 2001; Tohge et al., 2007). Genes of interest can be identified by measuring their messenger RNA or transcript levels in a whole organism, tissue or cell type under various conditions, such as different developmental stages, altered growth conditions, or between wild type and extreme/mutant phenotypes (Hazen et al., 2003). This allows for identification of the best candidate genes that might impact important metabolic or physiological processes (Alba et al., 2004). In some cases, phenotypes or physiological responses may not be macroscopically observable. And in other situations, transcript abundance may not be exactly predictive of functional outcome. For example, some abundant transcripts may not form functional proteins (Owens et al., 2008), and the proteins they encode can also have unpredictable turnover rates or stability (Sullivan and Green, 1993). Therefore, it is useful to combine gene expression profiling with a chemical or cytological phenotype (Tohge et al., 2007). Chemical phenotypes can be approximated by measuring levels of specific or diverse classes of metabolites, and then inferring gene-metabolite correlations from these data (Fiehn et al., 2000; Kaplan et al., 2007). The function of the protein encoded by a

(21)

candidate gene can then be demonstrated in subsequent experiments if necessary. The major genes and enzymes that have demonstrated roles in flavonoid biosynthesis are reviewed in the next section.

1.4.1 Molecular genetics of flavonoid biosynthesis: major genes and enzymes responsible for biosynthesis

The genetics, biochemistry and chemistry of flavonoid synthesis have been thoroughly investigated, especially in Antirrhinum majus (snapdragon), Arabidopsis thaliana (thale cress), Petunia hyrbida (petunia), and Zea mays (corn) (see e.g. Holton and Cornish, 1995; Shirley et al., 1995; Winkel-Shirley, 2001; Marles et al., 2003; Dixon et al., 2005; Lepiniec et al., 2006; Zhao and Dixon, 2010). Seed coat pigmentation (transparent testa or tt) mutants of Arabidopsis have been particularly useful for

identifying the biosynthetic and regulatory genes of PA synthesis (Abrahams et al., 2002; Winkel-Shirley, 2002; Xie et al., 2003; Lepiniec et al., 2006). Once seeds have

approached maturity, their coats (testa) turn brown due to oxidative cross-linking of PAs with other cell components. When a biosynthetic or regulatory PA gene has been

mutated, the parent plant produces seeds that either fail to turn brown or are severely delayed in browning, which causes the seeds to appear transparent. Therefore, absence of browning serves as a phenotypic marker to identify PA mutants (Debeaujon et al., 2000).

All three classes of fruit flavonoids are synthesized from chalcone precursors, which are formed by condensation of p-coumaroyl-CoA with three malonyl-CoA molecules (Fig. 1-2). This reaction is catalyzed by the type III polyketide synthase enzyme chalcone synthase (CHS; TT4), followed by isomerization by chalcone isomerase (CHI; TT5; not shown) to form flavanones. Most species have one to three copies of CHS and a single copy of CHI in their genomes (Goto-Yamamoto et al., 2002; Cominelli et al., 2008; Schijlen et al., 2007; van den Hof et al., 2008). However, in some species they each can exist as larger multigene families, such as the strawberry (Almeida et al., 2007) and soybean (Yi et al., 2010) CHS gene families and the Lotus japonicus CHI family (Shimada et al., 2003). Expression can vary widely in the larger gene families and functional characterization has generally been limited (Koes et al., 1989;

(22)

Flavanones are hydroxylated at the 4′ position on their B-ring due to the fact that the B-ring originates from p-coumarate (4-hydroxycinnamate). Flavanones are also hydroxylated at the 3 position of the C-ring by flavanone-3β-hydroxylase (FHT; F3H; TT6) to form dihydrokaempferol. This product is often further hydroxylated at the 3′, or 3′ and 5′ positions of the B-ring by cytochrome P450-dependent monooxygenases to form dihydroquercetin (flavonoid 3′-hydroxylase; F3′H; TT7) and dihydromyricetin (flavonoid 3′5′-hydroxylase; F3′5′H), respectively (Fig. 1-2). F3′5′H can also catalyze the 5′ hydroxylation of 3′4′ hydroxylated flavonoid intermediates (Halbwirth, 2010). These hydroxylation steps can occur instead on flavanones and flavonols. Most, if not all, flavonoid-producing angiosperms have a functional F3′H, but many species lack F3′5′H, including Arabidopsis, apple and rose, and therefore they cannot make 3′5′ hydroxylated flavonoids (Han et al., 2010). In contrast with CHS and CHI, flavonoid hydroxylases have been functionally characterized in multiple species, and most species studied to date have one functional copy of each or a just a single F3′H (Holton et al., 1993; Kaltenbach et al., 1999; Bogs et al., 2006). The hydroxylation pattern of flavonoids can impact function. For example, ortho-dihydroxylated (3′4′) flavonols are the most efficient UV-B absorbers (Ryan et al., 1998), while ortho-dihydroxylated PAs have a high in vitro antioxidant capacity (Halbwirth, 2010). Orange pelargonidin-glycosides are 4′ hydroxylated anthocyanins, red cyanidin-glycosides are 3′4′ hydroxylated, and dark purple delphinidin is trihydroxylated (Holton and Cornish, 1995). In Petunia, a unique cytochrome b5 type-B was shown to be necessary for full F3′5′H enzymatic activity (de

Vetten et al., 1999). This protein has not been identified in any other species, although a candidate homolog in grapevine was predicted to be involved in flavonoid hydroxylation based on its’ gene expression pattern (Bogs et al., 2006).

(23)

p-coumaro yl-CoA + 3 malon yl-CoA CHS FHT F3'H F3'5'H DFR leucocyanidin s ANS UFGT Anthocyanins anthocyanidins ANR LAR

epicat echin catechin

Proanthocyanidins

cyanid in-3-glyc osi de (R1=OH)

delph in id in-3-glyc osi de (R1,R2=OH) peoni di n-3-glycos ide (R1=OCH3) petuni din-3-g lycos ide (R1=OH,R2=OCH3)

malv idi n-3-glyc oside (R1,R2= OCH3)

Ext ensi on units

Terminal unit

Flavonols

kaempfero l-3-glycos ide (R1,R2=H) quercetin-3-g lycos ide (R1= OH) myriceti n-3-glyc oside (R1,R2= OH)

Figure 1-2. Simplified flavonoid biosynthesis pathway leading to flavonols, anthocyanins and PAs.

The diversity of flavonoid structures found in highbush blueberry are shown, along with the key biosynthetic enzymes. CHS: chalcone synthase, FHT: flavanone-3β-hydroxylase, F3'H: flavonoid 3'-hydroxylase, F3'5'H: flavonoid 3'5'-hydroxylase, DFR: dihydroflavonol reductase, ANS: anthocyanidin synthase, LAR: leucoanthocyanidin reductase, ANR: anthocyanidin reductase, UFGT: UDP-glucose:flavonoid-3-O-glycosyltransferase.

(24)

Dihydroflavonols can be directly used as substrate by flavonol synthase (FLS; not shown), which produces flavonols. Flavonols are typically stored as 3-O-glycosides and the addition of the sugar is catalyzed by UDP-glucose: flavonoid-3-O-glycosyltransferase (UFGT/F3GT). Some UFGTs that glycosylate flavonols are reported to be bifunctional in that they can glycosylate other flavonoids as well (Tohge et al., 2005; Greisser et al., 2008b). Regulation of FLS gene expression is the major determinant of whether dihydroflavonols will be directed into flavonols (Czemmel et al., 2009; Ferreyra et al., 2010). The B-ring of flavonols can be further modified by methyltransferases, some of which may bifunctionally act on anthocyanins as well (Lücker et al., 2010). Flavonoid methyltransferases have only rarely been identified and functionally characterized, such as in petunia flowers (Jonsson et al., 1984) and grape berries (Lücker et al., 2010).

The alternative route for dihydroflavonols is through dihydroflavonol reductase (DFR; TT3), which catalyzes a reduction of a ketone at the 4 position of the C-ring to a hydroxyl, producing the leucoanthocyanidins (Fig. 1-2). Both PAs and anthocyanins are produced from leucoanthocyanidins, and therefore DFR is a key control point in the flavonoid pathway. Most species have one or two DFRs (Shirley et al., 1995; Peters and Constabel, 2002; Xie et al., 2004; Dixon et al., 2005; Tsai et al., 2006), but rarely more (Yoshida et al., 2010). Based on results to date, there is no evidence that plants utilize one copy of DFR to produce anthocyanins and a second one for PAs (Dixon et al., 2005).

Leucoanthocyanidins can either be catalyzed into anthocyanidins by

anthocyanidin synthase/leucoanthocyanidin dioxygenase (ANS; TT18; LDOX), or 2,3-trans-flavan-3-ols (catechins) by leucoanthocyanidin reductase (LAR). As with DFR, most species investigated appear to have one or two copies of ANS (Abrahams et al., 2003; Bogs et al., 2005; Tsai et al., 2006; Pang et al., 2007). While all species that make anthocyanins and PAs necessarily have a functional ANS, some species (e.g.

Arabidopsis) do not have an LAR gene and therefore do not produce catechins. In some species such as Medicago truncatula, LAR gene expression does not correlate well with PA synthesis, and the LAR enzyme does not seem to affect the catechin content of PAs (Pang et al., 2007). Therefore, the specific role of LAR has been questioned in some species (Akagi et al., 2009a; 2009b). By contrast, functional support for one isoform of LAR catalyzing the formation of catechins has been found in Desmodium uncinatum

(25)

(Tanner et al., 2003), grape (Bogs et al., 2005; Gagne et al., 2009), and Lotus (Paolocci et al., 2007). Other, more divergent genes with similarity to LAR are present in many

species, however they have yet to be successfully characterized biochemically (Bogs et al., 2005).

Anthocyanidins can be glycosylated and then transported to the vacuole as colored anthocyanins. Glycosylation is most common at the 3-O, 5-O and 7-O positions and these are catalyzed by anthocyanin-O-glycosyltransferases (UFGT/A3GT), which are usually similar in sequence to flavonol glycosyltransferases (Fig. 1-2; Ford et al., 1998; Tohge et al., 2005). Plants tend to have numerous putative UFGT genes and it is often difficult to predict their specific in vivo function based solely on phylogenetic analysis. Therefore, it is hard to estimate a typical anthocyanin UFGT gene copy number (Ono et al., 2010). Methoxylations are catalyzed by anthocyanin methyltransferases (Hugueney et al., 2009; Lücker et al., 2010), and acylations are performed by anthocyanin

acyltransferases (Nakayama et al., 2003). The exact series of events that occur in transporting anthocyanins to the vacuole have not been fully elucidated. However, the evidence to date suggests that anthocyanins are synthesized by a multienzyme complex on the cytosolic side of the endoplasmic reticulum surface (Saslowsky and Winkel-Shirley, 2001). The anthocyanins may then be bound by a glutathione-S-transferase (GST; TT19; Kitamura et al., 2004; Conn et al., 2008) and transported to the vacuolar membrane (tonoplast) in vesicles (Zhang et al., 2006; Poustka et al., 2007; Grotewold and Davies, 2008). Once at the tonoplast, two transport mechanisms have been demonstrated to date. The first was identified in Z. mays, in which an ATP-binding cassette- (ABC) type multidrug resistance-associated protein (MRP) actively transports anthocyanins across the tonoplast (Goodman et al., 2004). In the second mechanism, which has been demonstrated in grapes (Gomez et al., 2009), transport is carried out by H+-dependent multidrug and toxic extrusion (MATE) transporters (Gomez et al., 2009). In the

experiments by Gomez et al., prior acylation of anthocyanins was necessary for transport. Work in petunia has identified a P-type H+-ATPase that acidifies the vacuole for normal anthocyanin accumulation (Verweij et al., 2008), which could in theory provide the proton gradient necessary for MATE-mediated anthocyanin transport (Gomez et al., 2009).

(26)

During PA synthesis anthocyanidins can also be converted into 2,3-cis-flavan-3-ols (epicatechins) by anthocyanidin reductase (ANR; BANYULS). The gene encoding ANR was only identified and characterized in 2003 by Xie and colleagues. Since the PAs of most species are composed primarily of 2,3-cis-flavan-3-ols, ANR gene expression has successfully been used as a marker for PA synthesis (Xie et al., 2003; Greisser et al., 2008a). ANR proteins have been biochemically characterized in a variety of species, in which they are typically encoded by one or two gene copies (Xie et al., 2003; Xie and Dixon, 2004; Bogs et al., 2005; Almeida et al., 2007; Akagi et al., 2009a). Recent evidence suggests that, at least in Medicago and Arabidopsis, epicatechin is synthesized in the cytosol (Pang et al., 2007), glucosylated at the 3′-O position (Pang et al., 2008), and then transported across the vacuole by a MATE transporter (TT12), which is similar in identity to the anthocyanin MATE proteins (Marinova et al., 2007; Zhao and Dixon, 2009). TT19 may bind the flavan-3-ols prior to transport by TT12 (Kitamura et al., 2010). A tonoplastic H+-ATPase (AHA10) has been identified in Arabidopsis that is required for flavan-3-ol and PA accumulation in seed coat vacuoles (Baxter et al., 2005). It is quite possible that AHA10 provides the proton gradient necessary for PA-specific MATE transporter function. Once in the vacuole, flavan-3-ols can exist as monomers, or they can be polymerized into PAs, which can range in size from two to over fifty subunits

(Monagas et al., 2003). It is not known how PAs are polymerized, how PA length is determined and whether enzymes catalyze these processes (Zhao et al., 2010). Upon seed maturation, PAs in the seed coat become cross-linked and form hard, brown protective structures with other cell components. In Arabidopsis, this oxidative process is apparently catalyzed by at least TT10, which is a laccase-like enzyme (Pourcel et al., 2005; 2007). The proteins described in this section are summarized in Table 1-1.

(27)

17

Table 1-1. Summary of proteins known to be directly involved in flavonoid biosynthesis.

Protein Classification Required for Function/Product Recommended citation

CHS; TT4 Type III polyketide synthase all flavonoids chalcones Burbulis and Winkel-Shirley, 1999

CHI; TT5 Isomerase all flavonoids flavanones Burbulis and Winkel-Shirley, 1999

FHT; F3H; TT6

2-oxoglutarate-dependent dioxygenase

all flavonoids dihydroflavonols Pelletier and Shirley, 1996 F3'H; TT7 P450-dependent

monooxygenase

3’ hydroxylated flavonoids

dihydroflavonols Bogs et al., 2006 F3'5'H P450-dependent

monooxygenase

3’5’ hydroxylated flavonoids

dihydroflavonols Bogs et al., 2006

DIF-F Cytochrome b5 B 3’5’ hydroxylated

flavonoids (3’?)

dihydroflavonols de Vetten et al., 1999 FLS 2-oxoglutarate-dependent

dioxygenase

flavonols flavonols Downey et al., 2003b

F3GT Glycosyltransferase flavonols flavonol glycosides Jones et al., 2003

FOMT Methyltransferase flavonols methylated flavonols Lücker et al., 2010

DFR; TT3 NADP+ oxidoreductase anthocyanins/PAs leucoanthocyanidins Peters and Constabel, 2002 ANS; LDOX;

TT18

2-oxoglutarate-dependent dioxygenase

anthocyanins/PAs anthocyanidins Saito et al., 2002

ANR; BAN NADP+ oxidoreductase PAs epicatechins (EC) Xie et al., 2003

LAR NADP+ oxidoreductase PAs catechins Tanner et al., 2003

F3'GT Glucosyltransferase PAs 3'-glucosylated EC Pang et al., 2008

TT19 Glutathione-S-transferase-like

anthocyanins/PAs carries flavonoid to tonoplast? Kitamura et al., 2010

TT12 MATE transporter PAs transport flavan-3-ol into

vacuole

Miranova et al., 2007

AHA10 H+-ATPase PAs H+ gradient for MATE Baxter et al., 2005

(28)

18

Table 1-1. Continued from previous page.

Protein Classification Required for Function/Product Recommended citation

? Enzyme? PAs polymerization Zhao et al., 2010

TT10 Laccase PAs cross-linked PAs Pourcel et al., 2005

A3GT Glycosyltransferase anthocyanins anthocyanins Ford et al., 1998

AOMT Methyltransferase anthocyanins methylated anthocyanins Hugueney et al., 2009; Lücker et al., 2010 AATs Acyl-CoA transferases anthocyanins acylated anthocyanins Nakayama et al., 2003

MRP3 MRP ABC transporter anthocyanins transport into vacuole Goodman et al., 2004 AnthoMATE MATE transporter anthocyanins transport into vacuole Gomez et al., 2009 PH5 P-type H+-ATPase anthocyanins H+ gradient for MATE Verweij et al., 2008 MYBA type R2R3 MYB transcription

factor

anthocyanins ↑ UFGT Walker et al., 2007

MYB5a,b R2R3 MYB transcription factor

PAs,

anthocyanins/PAs

↑ general flavonoid genes Deluc et al., 2006; 2008 MYBF type R2R3 MYB transcription

factor

Flavonols ↑ FLS Czemmel et al., 2009

MYBPA1 type R2R3 MYB transcription factor

PAs

(anthocyanins?)

↑ general flavonoid genes +

ANR/LAR

Bogs et al., 2007 TT2 type R2R3 MYB transcription

factor

(29)

1.4.2 Molecular genetics of flavonoid biosynthesis: transcriptional regulation

The following two sections will provide an overview of the current understanding of the genetic regulation of flavonoid synthesis in the context of fruit development and ripening. To understand the importance of regulating flavonoid synthesis coordinately with stages of fruit maturation, the biochemical changes that occur during fruit

development will be introduced first.

1.4.2.1 Biochemistry and genetic regulation of fleshy fruit ripening

The majority of the changes responsible for producing edible fruit occur in parallel with the final stages of seed maturation during the ripening phase. While seeds are beginning to develop, the surrounding carpel and ovarian tissues first expand by cell division and expansion (Ozga and Reinecke, 2003). These tissues likely serve as

protection from pathogens, herbivores and abiotic stresses early in development. They are usually tough as a result of strong cell walls, and unpalatable due to a buildup of acids, bitter PAs and other phenolic compounds (Brady, 1987). Successful fertilization (termed anthesis) triggers expression of MADS-box transcription factors that induce changes associated with expansion of the ovary (Liljegren et al., 2000; Giovannoni et al., 2004; Branbilla et al., 2007; Gonzalez et al., 2007).

Once seeds have approached maturity, fruit physiology shifts into the ripening phase. Ripening is marked by a reduction in astringency, bitterness, acidity, green coloration and cell wall strength, and an increase in sugar concentration, pigments and volatile compounds. Many of these changes are biochemical in nature. For example, acidity is reduced by enzymes that utilize organic acids (e.g. citrate, malate, tartrate) as substrates (Etienne et al., 2002), while sugar is synthesized de novo via activation of the sucrose synthesis pathway as well as from starch breakdown. Cell growth during ripening is primarily due to accumulation of massive amounts of sucrose, glucose and/or fructose, which can make up 10-20 % of the cell volume (Coombe, 1976). Cell walls are actively weakened by enzymes that break down (polygalacturonases, β-galactosidases, cellulases, pectin esterases, α-L-arabinofuranosidases) and reorganize (β-1,4-glucanases, pectate lyases, xyloglucan hydrolases, β-xylanases, expansins) polysaccharides that comprise the

(30)

cell wall (Goulao and Oliveira, 2008). The bright pigmentation of many fleshy fruits is due to biosynthesis of pigment compounds, usually anthocyanins and/or carotenoids, with a subsequent degradation of green chlorophyll (Allen et al., 2008). Unique aromas and flavours are due to biosynthesis of a diverse collection of terpenoid, aromatic and aliphatic compounds (Dunlevy et al., 2009; Klee, 2010).

In climacteric fruits, the onset of these changes is associated with a burst in respiration and ethylene synthesis. It is thought that ethylene stimulates the initiation of ripening in this class of fruits (Barry and Giovannoni, 2006), provided that they have reached a certain developmental stage. Recent work in the model climacteric fruit tomato suggests that the onset of ethylene synthesis and responsiveness is controlled in part by MADS-box transcription factor-regulated gene expression (Vrebalov et al., 2002; 2009; Manning et al., 2006; Lin et al., 2008; Cantu et al., 2009).

In contrast with climacteric fruit ripening, non-climacteric fruit (e.g. grape, blueberry, citrus fruits) are named as such because they generally do not undergo a substantial upsurge in ethylene-associated respiration at ripening initiation and respond only incrementally to exogenous ethylene if harvested pre-maturely. Due to this

difference, it is often suggested that the initiation of non-climacteric fruit ripening may result from different mechanisms, which are currently unknown. Preliminary evidence suggests that MADS-box transcription factors also function in initiating non-climacteric fruit ripening (Boss et al., 2001; 2002; Vrebalov et al., 2009; Jaakola et al., 2010). Given the likelihood of common developmental ripening initiation signals, differences in ripening initiation physiology between the two classes of fleshy fruit are more likely to result from alterations downstream of these transcription factors.

Work primarily in grapes has implicated abscisic acid (ABA) as a possible downstream growth regulator of non-climacteric ripening. ABA is most well-known for its roles in inducing embryo growth and subsequent seed dormancy, regulating stomatal opening, and mediating responses to abiotic stress. The amount of physiologically available ABA is determined by the balance between its biosynthesis and metabolism (Fig. 1-3; Nambara and Marion-Poll, 2005). In grapes, free ABA and ABA biosynthetic genes and enzymes increase substantially at ripening initiation (Deluc et al., 2007; Giribaldi et al., 2007; 2010; Lund et al., 2008; Koyama et al., 2009; Gambetta et al.,

(31)

2010). The most noticeably induced gene at ripening initiation in grapes is

9-cis-epoxycarotenoid dioxygenase (NCED), which encodes an enzyme that catalyzes the rate-limiting step in ABA biosynthesis (Iuchi et al., 2001; Zhang et al., 2009a): cleavage of the C40 carotenoids 9′-cis-neoxanthin and violaxanthin to form C15 xanthoxin (Fig. 1-3).

Some experiments have shown positive effects of ABA application on certain ripening parameters, especially sugar synthesis, anthocyanin-based pigmentation and induction of related genes and enzymes (Coombe and Hale, 1973; Pirie and Mullins, 1976; Chervin et al., 2004; Jeong et al., 2004; Peppi et al., 2008; Koyama et al., 2009; Owen et al., 2009; Wheeler et al., 2009; Gambetta et al., 2010; Giribaldi et al., 2010). It is not known if a similar mechanism also operates in other non-climacteric fruit, such as highbush blueberry.

Flavonoids have a number of important functions in fruit. PAs are part of the defensive system that protects immature seeds and fruit tissue, flavonols protect against UV damage, and anthocyanins are the main visual cue of maturation in many fruits. Given their importance to plant fitness, the biosynthesis of each class of flavonoids must be tightly regulated in coordination with different stages of fruit maturation.

(32)

Zeaxanthin (C₄₀) Xanthoxin (C₁₅) NCED Abscisic acid ABA-GE 7'-hydroxy ABA Dihydrophaseic acid 8'-hydroxy ABA Phaseic acid

Figure 1-3. Simplified pathway of abscisic acid (ABA) metabolism in plants.

ABA is synthesized from xanthoxin, the product of NCED, which cleaves carotenoids. ABA-glucose ester (ABA-GE) is thought to be an inactivated storage form of ABA. The main catabolic pathway in most plants is through 8′-hydroxy ABA to inactive phaseic acid and dihydrophaseic acid. Other catabolic pathways to 7′-hydroxy-ABA and neo-phaseic acid (not shown) are also possible. NCED: 9-cis-epoxycarotenoid dioxygenase.

(33)

1.4.2.2 Temporal and spatial regulation of flavonoid biosynthesis in fleshy fruits

Ever since research began accumulating on potential health benefits of fruit flavonoids, much work has been devoted to understanding the genetic regulation of flavonoid synthesis in fleshy fruits. The most attention has been given to grape berries (Vitis vinifera L.) due to their economic and organoleptic importance in wine-making. In 1996, Boss and colleagues used Northern blot analysis to estimate the expression of several flavonoid genes (CHS, CHI, FHT, DFR, and ANS) in a developmental time series of grape tissue from flowering to ripening. They found that the five genes were highly expressed up to four weeks following flowering, after which they decreased to low levels. At the initiation of ripening (“véraison”), the expression of these genes increased again to a maximum, paralleling the appearance of anthocyanin-based pigmentation. This was accompanied by an induction in berry skin of UFGT transcript, which based on its expression profile, was predicted to encode a glycosyltransferase that could glycosylate anthocyanidins. The predicted function of this GT was later biochemically confirmed by the same group (Ford et al., 1998). The early phase of expression was unexpected by the authors, in that no anthocyanins were detected in these tissues. They wondered if the general flavonoid genes might instead be expressed in order to produce flavonols and proanthocyanidins.

In recent years, research groups have employed quantitative real-time PCR to more finely study the expression of flavonoid genes in developing grapes. For example, Downey et al. (2003b) found that flavonol concentration was highest in grape flowers and then steadily decreased through ripening. However, on a per fruit basis, the amount of flavonols continued to increase until two weeks after véraison. Expression of an FLS gene was highest in flowers and also had an expression peak from two-to-six weeks after véraison. Therefore, there appeared to be two phases of flavonol synthesis in grape berries, which is probably based on the physiological needs of the fruit, assuming that the flavonols are functioning as UV-screens.

Furthermore, Bogs et al. (2005) revealed a uniphasic expression pattern of ANR and LAR1, which were highly expressed only in flowers and smaller green fruits up to véraison as was predicted by Boss et al. (1996). From flowering to nine weeks after véraison, the total PA concentration in whole fruits decreased by roughly one half.

(34)

However, in skin the flavan-3-ol and PA concentrations increased until véraison, during a period when ANR and LAR1 expression was negligible. Skin flavan-3-ols were primarily catechin, whereas the PAs were almost exclusively epicatechin and epigallocatechin. The reason for an increase in skin PAs at véraison despite limited ANR expression remains an unresolved contradiction. In grape seeds, ANS, ANR and LAR2 expression remained high until shortly after véraison, which corresponded with a spike in flavan-3-ol concentration, and a continual increase in seed PAs on a per fruit basis (Downey et al., 2003a).

A second paper from this group investigated the developmental regulation of F3′H and F3′5′H gene expression. They found that F3′H expression was high in seeds up until véraison, while some expression was also detected in skin during ripening (Bogs et al., 2006). By contrast, F3′5′H was most highly expressed during ripening, with low transcriptional abundance in young fruit. They reasoned that the differential expression of each of the flavonoid hydroxylases was related to flavonoid structure. For example, seed flavan-3-ols and PAs were composed only of epicatechin and catechin, which are

hydroxylated at the 3′ and 4′ positions. In addition, grape flavonols were identified as being primarily quercetin-glycosides, which are also dihydroxylated. Therefore, the early accumulation of F3′H transcripts was related to flavonoid structure. However, PAs in fruit skin were made up of epigallocatechin units in addition to (epi)catechin. These units are 3′4′5′ hydroxylated, suggesting that F3′5′H is actively involved in skin PA synthesis. By contrast, the ripening-specific phase of F3′5′H expression corresponded to

anthocyanin structure. Other groups have confirmed the flavonoid hydroxylase expression patterns found by Bogs et al. (2006) in a variety of other grape cultivars (Castellarin et al., 2006; Jeong et al., 2006).

Therefore, for the most part, PA and anthocyanin synthesis appear to be separated developmentally in grape berries via differential gene expression. General flavonoid genes as well as ANR and LAR are highly expressed from flowering until the initiation of ripening, and 3′4′ hydroxylated flavan-3-ols and PAs are synthesized during this period in both seeds and skin. Flavonols are also synthesized in this period and accumulate mostly in skin. At the initiation of ripening, PA synthesis largely ceases while a second upsurge in general flavonoid gene expression occurs. This second phase of expression includes an anthocyanin-specific GT gene and F3′5′H and is correlated with the appearance of

(35)

delphinidin-type (3′4′5′ hydroxylated) anthocyanins. By contrast, flavonoid synthesis in apple and strawberry seems to be less strictly controlled at the transcriptional level with respect to specific phases during the growth season (Takos et al., 2005; Almeida et al., 2007; Carbone et al., 2009).

The closest relative of blueberry that has been studied in terms of flavonoid molecular genetics is the European blueberry or bilberry (Vaccinium myrtillus). Using Northern blot analysis, Jaakola and colleagues (2002) found that the expression of CHS, FHT, DFR and ANS was most intense in colored ripening fruits and also in flowers. Since these tissues all contain anthocyanins, the expression was therefore associated with anthocyanin synthesis. However, very little expression was detected in small green stages and only minute concentrations of PAs were found throughout development. As a result, the differential genetic regulation of PA, flavonol and anthocyanin synthesis in blueberry species remains unclear.

1.4.2.3 The role of R2R3 MYB transcription factors in regulating flavonoid synthesis

The distinct separation of PA and anthocyanin synthesis particularly in grapes led some authors to suggest that specific transcription factors might be controlling the

differential regulation of flavonoid gene expression. The first gene encoding a putative flavonoid-specific transcription factor identified in grape was found by searching for the closest homolog of R2R3-MYB transcription factors from maize and petunia that were shown to regulate anthocyanin synthesis (Grotewold et al., 1991). The authors found that the expression pattern of the gene (VlMYBA1) paralleled that of UFGT during ripening in red grapes, but not white grapes (Kobayashi et al., 2002). Expressing VlMYBA1 in

somatic embryos of Kyoho grape resulted in the appearance of blue, anthocyanin-rich spots via induction of UFGT transcript specifically, which was absent in control embryos. A second group later showed that a similar gene, VvMYBA2, also controls anthocyanin synthesis by producing a functional protein that modulates UFGT gene expression (Walker et al., 2007). These two transcription factors do not appear able to regulate expression of general flavonoid genes, rather only UFGT and anthocyanin-modification genes (Table 1-1; Cutanda-Perez et al., 2009). Anthocyanin-specific homologs of

(36)

VvMYBA1/2 have since been characterized in a number of other fruit species, including tomato (Mathews et al., 2003), apple (Takos et al., 2006; Ban et al., 2007b; Espley et al., 2007), mangosteen (Palapol et al., 2009), bayberry (Niu et al., 2010) and pear (Feng et al., 2010).

R2R3 MYB transcription factors that can activate general flavonoid pathway genes have also been identified in grapes. For example, overexpression of the VvMYB5a and VvMYB5b genes in tobacco led to accumulation of both PAs and anthocyanins (Deluc et al., 2006; 2008). Both transcription factors were found to strongly transactivate general flavonoid gene promoters in vitro, but only weakly transactivate the ANR and UFGT promoters. The VvMYB5a gene is expressed most highly in the skin, flesh and seeds of young pre-véraison fruit, and therefore its expression pattern is associated with PA and flavonol accumulation. By contrast, VvMYB5b was most highly expressed in the weeks before and after véraison (Deluc et al., 2008). The authors concluded that in vivo VvMYB5a likely activates expression of general flavonoid genes for PA synthesis, whereas VvMYB5b activates general flavonoid genes for both PA and anthocyanin synthesis (Deluc et al., 2008).

R2R3 MYB regulators of PA synthesis have also been functionally characterized in grapes. The first one is VvMYBPA1, which led to an increase in PAs when

overexpressed in the Arabidopsis transparent testa 2 (tt2) mutant (Bogs et al., 2007). The authors showed that VvMYBPA1 could in vitro very strongly activate promoters of all of the general flavonoid pathway genes as well as ANR and LAR, but not UFGT. The

VvMYBPA1 gene was expressed during early development and most highly expressed in seeds from two weeks before to two weeks after véraison. A second R2R3 MYB

transcription factor called VvMYBPA2 also regulates PA synthesis in grapes.

Overexpression of VvMYBPA2 and VvMYBPA1 in grapevine hairy roots each led to a five-fold increase in PAs (Terrier et al., 2009). Microarray analysis revealed that general flavonoid genes as well as ANR and LAR1 genes were upregulated to similar levels following overexpression of either transcription factor. Expression of the UFGT and FLS genes were unaffected by either transcription factor, corresponding to no differences in anthocyanin and flavonol concentrations compared to control roots. Unexpectedly, despite the fact that VvMYBPA1 could not activate the UFGT promoter in vitro, the

(37)

VvMYBPA1 gene was significantly expressed in grape berry skin up to four weeks after véraison. Deluc et al. (2008) suggested that the protein may therefore activate general flavonoid gene promoters during anthocyanin synthesis in a fashion similar to

VvMYB5b. By contrast, expression profiling revealed that VvMYBPA2 transcript was highly abundant only in the skin of pre-véraison fruit. The authors suggested that early in fruit development, VvMYBPA1 likely regulates seed PA synthesis whereas VvMYBPA2 primarily regulates skin PA synthesis.

A slightly different approach has been taken to understand the genetic regulation of PA synthesis in persimmon (Diospyrus kaki) fruit. Shortly after fruit set PA

concentration and flavonoid gene expression in edible low-PA persimmon varieties gradually decreases to negligible levels, whereas in inedible high-PA varieties synthesis remains high for many weeks (Akagi et al., 2009a). In addition to differences in total PA concentration, those PAs produced in the low-PA varieties are shorter and significantly reduced in epigallocatechin and epigallocatechin-gallate units (Fig. 1-1) compared to the high-PA varieties. Targeted qPCR revealed that the structural differences were largely due to minimal F3′5′H, DFR, ANS and ANR expression in the low-PA fruits. In a second paper, the authors identified and characterized the gene at the Ast locus, which is

responsible for determining whether the fruit will become low or high in PAs. This gene (DkMYB4) is most similar to the VvMYBPA1 gene of grape. Using qPCR they found that DkMYB4 gene expression in the low PA varieties followed a nearly identical pattern as the PA biosynthetic genes, whereas expression of other candidate MYB genes did not correlate well with PA content (Akagi et al., 2009b). Overexpression of this gene in kiwi calluses resulted in enhanced PAs, while antisense DkMYB4 expression in persimmon calluses significantly reduced biosynthetic gene expression and reduced epicatechin and epigallocatechin content in PAs. In a subsequent paper, this group found that another candidate MYB, DkMYB2, increased PA content by up to 10 times and mean degree of PA polymerization from four to 22 units when overexpressed in kiwi calluses (Akagi et al., 2010). It also enhanced expression of CHS, DFR, ANS and ANR genes, and in transient activation studies it more strongly activated the ANR (16 fold vs. 7 fold) and LAR (13 fold vs. 2 fold) promoters than DkMYB4. These two transcription factors appear to coordinately regulate PA synthesis and structure in persimmon fruit flesh.

Flavonoid gene expression and metabolite profiling during fruit development in highbush blueberry (Vaccinium corymbosum L.)

Supervisory Committee

Abstract

Table of Contents

List of Tables

List of Figures

List of Supplementary Tables (Appendix B)

List of Supplementary Figures (Appendix B)

Acknowledgments

1. Chapter One: General Introduction