Advances in genomics for the improvement of quality in Coffee. Accepted Article

Transcription

1 Advances in genomics for the improvement of quality in Coffee Hue T.M. Tran ab, L. Slade Lee c, Agnelo Furtado a, Heather Smyth a, Robert Henry a* a Queensland Alliance for Agriculture and Food Innovation (QAAFI), The University of Queensland, Australia b Western Highlands Agriculture & Forestry Science Institute (WASI), Vietnam c Southern Cross University, Australia * Correspondence to: Robert Henry, Queensland Alliance for Agriculture and Food Innovation (QAAFI), The University of Queensland, St Lucia QLD 4072, Australia, Tel , robert.henry@uq.edu.au This article has been accepted for publication and undergone full peer review but has not been through the copyediting, typesetting, pagination and proofreading process, which may lead to differences between this version and the Version of Record. Please cite this article as doi: / jsfa.7692

2 Abstract Coffee is an important crop that provides a livelihood to millions of people living in developing countries. Production of genotypes with improved coffee quality attributes is a primary target of coffee genetic improvement programs. Advances in genomics are providing new tools for analysis of coffee quality at the molecular level. The recent report of a genomic sequence for robusta coffee, Coffea canephora, is a major development. However, a reference genome sequence for the genetically more complex arabica coffee (C. arabica) will also be required to fully define the molecular determinants controlling quality in coffee produced from this high quality coffee species. Genes responsible for control of the levels of the major biochemical components in the coffee bean that are known to be important in determining coffee quality can now be identified by association analysis. However, the narrow genetic base of arabica coffee suggests that genomics analysis of the wild relatives of coffee (Coffea spp.) may be required to find the phenotypic diversity required for effective association genetic analysis. The genomic resources available for the study of coffee quality are described and the potential for the application of next generation sequencing and association genetic analysis to advance coffee quality research are explored. Key words: Coffee quality, genetics, genomics, biochemical compounds, next generation sequencing, association studies.

3 GENERAL INTRODUCTION Coffee is an important crop and the second most traded commodity in the world (after petroleum) providing a living to more than 125 million people 1. Coffee belongs to the Rubiaceae family and the Coffeeae tribe and consists of more than 124 species spread across two genera Coffea L. and Psilanthus Hook.f, each of which in turn consist of two sub-genera, Coffea and Baracoffea (J.-F. Leroy) J.-F. Leroy, and Psilanthus and Afrocoffea (Moens) summarized by Anthony et al. 2 respectively. The grouping of Coffea and Psilanthus genera has been examined in several studies Commercial coffee production is dominated by only two species belonging to the Coffea genus: C. arabica and C. canephora (the latter generally referred to as robusta coffee). All coffee species are diploid (2n=2x=22) and generally self-incompatible, except for C. arabica which is a self-fertile tetraploid (2n=4x=44) derived from a spontaneous hybridization between C. canephora (as paternal progenitor) and C. eugenioides (as maternal progenitor) 8 9. Although C. arabica is considered to have better cupping quality than C. canephora, improving the quality of both commercial species remains a target for most coffee improvement programs. With advances in genomic and sequencing technology, it is feasible to understand the coffee genome and the molecular inheritance underlying coffee quality, thereby helping improve the efficiency of breeding programs. This review will discuss current knowledge regarding the genetics of those biochemical compounds which are considered quality determinants, for use as a foundation for improving coffee quality by breeding. Available genomic resources for the study of the genetics of biochemical compounds that are likely to be playing a role in coffee flavour will also be discussed. In addition, the potential value of the study of genetics of coffee

4 quality using next generation sequencing and association genetic analysis will be considered. BIOCHEMICAL CONTROL OF COFFEE QUALITY AND ITS VARIATION Coffee quality is assessed by evaluating both the physical attributes of the coffee beans and the organoleptic properties of the coffee. Physical quality reflects moisture content, defects (e.g. sticks, stones, damaged beans and black beans ), bean size and bean colour; while organoleptic quality reflects aroma, taste, flavour, body (a feeling of the heaviness or richness on the tongue), acidity and the preference of tasters 10. A range of biochemical compounds in coffee beans are important contributors to the quality of the coffee in the cup. Biochemical compounds influencing the organoleptic quality of coffee beans include non-volatile and volatile components. Important non-volatile components of the bean include carbohydrates and fibre (sucrose, reducing sugars, cell wall polysaccharides, lignin), nitrogenous compounds (protein, free amino acids, caffeine, trigonelline), lipids (coffee oil, diterpene esters), minerals (potassium and phosphorus), acids and esters (total chlorogenic acids, aliphatic acids and quinic acid) 11. All of these biochemical compounds are considered to play a role in roasting chemistry 12. Proteins and amino acids, for example, react with reducing sugars to produce aroma precursors. Chlorogenic acids (CGAs) and caffeine, are responsible for bitterness 13. Among these components, sucrose, caffeine, trigonelline, lipids and CGAs are major biochemical compounds that contribute to the flavour of the beverage after the roasting of the beans 11. While sucrose, trigonelline and lipid have a positive correlation with coffee cup quality, caffeine and some sub classes of CGAs have a negative correlation with coffee cup quality (often referred to simply as cup ). There are about three hundred volatiles detected so far in green coffee bean 14 of which twenty one volatiles,

5 including 3-isobutyl-2-methoxypyrazine and 2-methoxy-3,5-dimethylpyrazine, have been identified as key compounds for coffee cup 14. In the roasted coffee bean, there are more than a thousand volatiles which can be grouped into twenty key aroma compound-groups 15. The aroma of the brew is different from that of roasted coffee. Several notes in roasted coffee become more intensive in the brew such as caramel and phenolic odour. This change in aroma profile of coffee is not caused by the formation of new odorants, but by a shift in the concentrations of existing ones as summarised by Grosch 16. Coffee flavour and its formation is extremely complex; however, numerous studies on coffee flavour and its constituents have been reported. How the volatile and nonvolatile compounds contribute to coffee quality and relationships between sensory properties and the composition of coffee has been thoroughly reviewed by Flament 12 and more recently by Sunarharum et al. 17. The reactions involved in the formation of coffee aroma and its mechanisms have been reviewed by Buffo & Cardelli-Freire 18. However, the change in bean composition, especially in aroma profile, from green to roasted bean is complicated and not all formation pathways are fully understood under coffee roasting conditions 19. The use of green bean or roasted bean in the study of coffee quality genetics therefore should be considered carefully since aroma in roasted bean and brew is of customer's interest, while the green bean attributes are more directly affected by genetics. This is a challenge in the study of genetics of bean composition, that is, ascribing links to coffee quality. However, with the progress in functional genomic approaches, the identification of molecular determinants of coffee quality characteristics is feasible and it is possible to select coffee varieties with superior beverage quality 20.

6 The bean chemical composition and organoleptic characteristics are significantly different between species and to a lesser extend within species 10. A number of studies on the variation in the biochemical composition of the bean are summarised in Table Fatty acid and CGAs were used as indicators to differentiate arabica varieties Similarly, Tessema et al. 27 also found that the quality traits (organoleptic quality) and biochemical constituents (sucrose, fat, crude protein and minerals) were diverse in the arabica germplasm collections from Ethiopia. This is an encouraging result for the study of variation of quality traits in C. arabica for association mapping. In general, compounds that have a positive correlation with quality such as sucrose, trigonelline and lipids are at a higher concentration in C. arabica than in C. canephora and some other Coffea species. Significant differences in flavour between different coffee types have also been reported. Robusta coffee beans have a bitter, full bodied taste, but low acidity while those of arabica coffee are more aromatic with more perceptible acidity but less body. Within the robusta coffees, Moschetto et al 28 found important differences in cup quality between the two main genetic groups studied (Congolese and Guinean), for preference, aroma, acidity, body and bitterness. Within the arabica coffees, different varieties can be associated with specific flavour profiles. For example, genotype SL28 produces a milder brew than Kent, genotypes Blue Mountain and Bourbon produce finer coffees than Catimor, while genotype Liberica yields coffee that is bitter and without finesse 29. The composition of the beans that determines the properties and quality of coffee in the cup depends not only on genetic factors (species, varieties) but also on non-genetic factors such as the cultivation conditions (soil, air temperature, altitude, sun exposure

7 and rainfall), plant condition (plant age, bean maturity), cultivation practices (shade management, fertilisation and irrigation) and harvest and processing methods. In general, volcanic soils, low temperature, high altitude and growing under shade have positive correlation to coffee compositions which relates to coffee quality Postharvest techniques, roasting and the preparation of the beverage also have a strong influence on coffee quality Understanding the influence of each factor and their interaction would help with defining realistic breeding strategies. Since there have been many reports presented in detail on these issues , especially the most recent reviews of Joet et al. 39 on the relationships between coffee quality, bean chemical composition and environmental effects, or Santos et al. 40 on climate change impacts on bean quality, or Sunarharum et al. 17 on the impacts of processing, roasting and preparation methods to aroma profile, they are not reviewed in this paper. Another factor that may influence coffee quality is epigenetic effects. Epigenetics is modified gene expression not caused by alteration in the gene sequence or DNA code. This change is due to factors that control how and when each gene is expressed. Epigenetics is extremely complex and is not yet fully understood. Although there have been several studies on factors affecting coffee quality, there has not been research on epigenetics and coffee quality. Only a very few studies have reported on epigenetics in relation to size and shape of the tree 41, and epigenetic mechanisms (especially in phenolic compounds) in somatic embryo regulation in arabica coffee species Being an allotetraploid species formed by a recent natural hybridization between two diploid ancestors, C. canephora and C. eugenioides, and which may have gone through a single allopolyploidization event 44, arabica may have genetic redundancy allowing for sequence divergence resulting in development of functional variations between duplicated genes (homoelogues). Proteins with different properties may be encoded

8 by homoelogues leading to new genetic diversity in breeding populations. However, homoelogues are also subject to extensive epigenetic control and are therefore not always equally expressed 45. Landey et al. 43 confirmed the very limited genetic and epigenetic alternations in somatic embryogenesis-derived plants during coffee somatic embryogenesis while embryogenic suspensions have a high risk of genome and epigenome instability. However, Vidal et al. 46 found homoeologous genes displaying differential expression for arabica using Single Nucleotide Polymorphisms (SNPs) from Expressed Sequence Tags (ESTs) and gene expression from transcript redundancy between C. arabica and C. canephora. The divergence of function in homoeologous genes may result in phenotypic novelty with useful agronomic traits. Homoeologous gene expression in C. arabica grown in two contrasted conditions was also used to compare with those of its progenitors and to examine its ability to adapt to environmental variations 47. Recently, Lashermes et al. 44 confirmed that the genomic rearrangements involving homoeologous exchanges as well as gene silencing, elimination and conversion appear to have occurred in C. arabica and to be a major source of genetic diversity. This on one hand will make a potent new source of genetic diversity for different traits, but on the other hand make it difficult to conduct association genetics in coffee. GENETICS OF BIOCHEMICAL COMPOUNDS KNOWN TO BE RELATED TO COFFEE QUALITY Improving both yield and quality is the target for coffee breeding programs. In a study by Montagnon et al 48, genetic correlations between coffee yield and quality determinants and cup tasting components measured in two groups of C. canephora were not significant, indicating that the coffee quality is not negatively associated with yield in C. canephora. Understanding the genetic inheritance of quality traits can

9 benefit breeding programs, but there appears little research that has focussed on understanding the genetic control of key biochemical compounds that relate to coffee beverage quality. Known inheritance mode and heritability values of the key biochemical compounds determining coffee quality in intraspecific and interspecific hybrids are detailed in Table Male/female additive inheritance in sucrose and trigonelline was contrasted in two studies using intraspecific hybrids 51, 56. The heritability measured in interspecific hybrids was different from that of intraspecific hybrids due to the differences among species in relation to their fixation during evolution of specific alleles of some genes relating to the variability of quality components. As a result, the genetic variation in quality at the interspecific level is not similar to that in intraspecific hybrids 10. The high narrow-sense heritability of caffeine and fat content in intraspecific hybrids implies that these traits can be improved at the intraspecific level. The negative correlation between fat content and sucrose content indicates that care should be taken in selecting for a combination of fat and sucrose content 48. Both major and minor genes with additive gene effects appear to be involved in the variation of caffeine content in the seeds 57 implying the complexity of this trait. The inheritance of coffee cup quality was also reported by Leroy et al 51 where for bitterness of robusta coffee, female additive effects were predominant, while for acidity, only a male additive effect was observed 51. This information will assist the breeding (e.g. parent selection) for robusta coffee quality improvement. The inheritance of quality characters in introgressed lines, which resulted during arabica improvement to introduce the disease resistant genes from Timor hybrid to arabica, was mentioned by Bertrand et al and also reviewed by Leroy et al. 10.

10 In summary, highly heritable traits like fat or caffeine content could be improved effectively by crossing between parents of different favourable values. For traits with lower heritability like trigonelline, CGAs or sucrose, Marker Assisted Selection (MAS) would be necessary 4. With of the advances in sequencing technology and marker development coupled with trait phenotyping, it should be possible to easily identify trait-linked markers for use in MAS. For coffee genetic improvement, most of the past and recent research has focused on the improvement of canephora quality due its known lower quality compared to arabica. The focus of arabica improvement is more on transfer of disease resistance from Timor hybrid 10 or climate change adaptation 59 ; quality improvement is the second priority in arabica as its quality is considered superior. However, this does not mean that arabica quality improvement is unnecessary. Quality determinants are not the same between arabica and canephora. For canephora, apart from caffeine reduction, the bean composition of galactomannans 60 or other compounds (other polysaccharides) which influence the extraction during the industrial process of lyophilisation (i.e. instant coffee processing) is probably the most important target. Fresh brewing is popular for arabica, rather than instant coffee processing although it may be applied to arabica in the future. For arabica, the increasing consumer demand for speciality coffee has intensified research aimed at pinpointing small differences in quality factors as well as to quantify these organoleptic features 61. For more detail on the trend and strategies of canephora and arabica improvement breeding, there are excellent reviews by Leroy et al. 10 and Vossen et al. 20.

11 COFFEE GENOMIC RESOURCES Coffee genetic diversity in C. arabica and C. canephora A number of works on the assessment of arabica diversity have been done with differing results. In general, among three main types of material (cultivar/varieties; accessions/introgression/hybrids; and spontaneous/sub-spontaneous) almost all studies show a very low genetic variation amongst arabica cultivars using different marker systems (RAPD, AFLP, SSR, ISSR, SRAP, TRAP) and coffee collected from different regions (Brazil, Yemen, Indian, Australia and Vietnam, Tanzania, Hawaii) The second group (accessions/introgression/hybrids) shows low to significant variation, more diverse in the introgressed lines The last group, in which one may expect to see the most diversity, has diversity ranging from low to moderate 77 to high 78. The correlation of genetics with geography/origin clearly shows in some studies but not in the others In the most recent work presented in the World Coffee Research annual report 59, genetic diversity assessment of 800 arabica coffee accessions from the arabica collection at CATIE, Costa Rica shows the least genetic diversity of this species compared to other major crops. This study also found that cultivated coffee varieties contain approximately 45% of the genetic diversity found in the 800 aforementioned accessions (while only 2% of tomato diversity found in cultivated tomato) indicating the limitation of variability for breeding programs. This also is problematic for association mapping work in arabica for breeding improvement, and thus the selection of appropriate germplasm for association mapping is very critical. Therefore it is essential to understand the population structure and its genetic diversity. In contrast to C. arabica, C. canephora possess a high genetic diversity due to its origin, reproduction method and dissemination 80, and it is structured in two main groups

12 (Guinean and Congolese) with six subgroups The linkage disequilibrium assessed in C. canephora breeding populations was variable, indicating the possibility of applying association studies within C. canephora species and using it to improve C. arabica due to its parental relationship to this species 84. Understanding the genetic diversity of the two dominant species is critical for any genetic improvement scheme for any traits in coffee. Coffee whole genome and synteny The genome size of coffee has been estimated to be 1.3 Gb for C. arabica 85 and 710Mb for C. canephora 86. Using flow cytometry, the DNA content of arabica was estimated as 2.47 pg (equivalent to 1.2 Gb) and canephora as 1.43 pg (equivalent to 700 Mb), while other diploid coffee species varied from 0.96 pg (469 Mb) (C. mauritiana) to 1.84 pg (900 Mb) (C. humilis) Despite its economic importance, the first high-quality draft genome of robusta coffee has just been completed and reported in To generate a whole genome sequence of C. canephora, Roche 454 single and mate-pair reads and Sanger BAC-end reads were used, representing 29.5X coverage of the genome size (710 Mb). In addition, Illumina sequence data, corresponding to 60X of coverage of the Coffee genome (710 Mb), was also used to correct sequencing errors and fill gaps. The resulting genome assembly consists of 25,216 contigs and 13,345 scaffolds with a total length of Mb (Table 3). Using a high-density genetic map, a total of 349 scaffolds covering approximately 364 Mb were anchored to the 11 C. canephora chromosomes, among which 139 were both anchored and oriented. Coffee displays the most conservative chromosomal gene order among asterid angiosperms. Transposable elements (TEs) in the C. canephora genome were identified and classified accounting for almost half (49.2%) of the reference genome length. Organelle to nuclear genome transfers was also analysed with a typical amount of chloroplast-derived fragments

13 (2,014 insertions, accounting for 0.16% of the draft genome) and an unusually large 750kb mitochondrion-derived fragment more than 100 Kb longer than those known from other plant genomes. In total, 25,574 protein-coding gene models were obtained in which several gene functions involved in secondary compound biosynthesis and coffee flavour and aroma were characterized. Species-specific gene family expansions were examined such as N-methyltransferases (NMTs), defence-related genes, and alkaloid and flavonoid enzymes involved in secondary compound synthesis. This is the most valuable genomics resource for further genomics study in coffee, especially as all of the genomics data, transciptomic data, SNP polymorphisms and genotyping data, gene families and metabolic pathways are publicity available via the Coffee Genome Hub 89. The whole chloroplast genome sequence of Coffea arabica L. was first reported by Samson et al. 90 (Table 3) and this also provides an important reference for studying other species and helps build up the phylogeny among members of the Coffeeae tribe which is still questionable. Several research groups are working on an C. arabica genome sequence using different biological materials and sequencing platforms This genomic resource is extremely important for the study of genetics and genomics of the high quality coffee species, C. arabica. Several studies have compared genomes between coffee and other crops. The most recent study reported that coffee chromosomal regions showed unique one-toone correspondences with grape chromosomes, proving the lack of any events changing ploidy intervening in their separate histories. Coffee and grape genomes show one-to-three correspondence with the tomato genome 86. Among coffee species, comparison of genomes showed a high gene synteny between C. arabica (EaEaCaCa) and C. canephora (CC) and C. arabica (EaEaCaCa), C. eugenioides

14 (EE) and C. liberica (LL) , but numerous chromosomal rearrangements were detected 98. The two sub-genomes of C. arabica (Ca and Ea) indicated sufficient sequence differences, and thus obtaining a whole-genome sequence would be an important resource for the allotetraploid genome of C. arabica. Results also revealed that C. eugenioides was more closely related to C. arabica than to C. liberica, which was in agreement with the ancestral history of the allotetraploid C. arabica 100. Other genomic resources Molecular markers As for other crops, determination of molecular predictors for coffee quality traits would help reduce the length of breeding selection cycles and thus phenotypic evaluation cost. However, the use of DNA technology in coffee quality improvement is still in its infancy and therefore limited. Pot et al. 101 used polymorphisms generated from SNPs, INDELs (insertions and deletions) and SSRs (Simple Sequence Repeats) to identify the nucleotide diversity of four sucrose metabolism enzymes in C. canephora genotypes using direct sequencing. The variation of these genes was also analysed between different Coffea species to allow the identification of more polymorphic sites using parallel in silico analysis of EST resources 101. AFLP (Amplified Fragment Length Polymorphism) and SSR markers were used to construct a genetic map of an F2 population between C. arabica and C. canephora (artificial tetraploid). The number of markers associated with quality traits identified was nineteen for sugar content, eight for caffeine, eight for CGAs and one for caffeine and CGAs 102. These markers need to be validated in other genotypes for consistency before they can be used in marker assisted selection in coffee breeding. Recently, a total of 33,239 SNPs specific to C. arabica and 87,271 SNPs specific to C. canephora were developed using targeted genome capture strategies and next-generation sequencing, and were evaluated on 72

15 samples from C. canephora and 72 from C. arabica. These genomic resources will support for genome assemblies, accelerate breeding of interested traits as well as manage genetic diversity in coffee species 103. Bacterial artificial chromosome (BAC) libraries. Three BAC libraries were developed for each species of C. canephora and C. arabica in relation to sucrose metabolism 104, genome composition and evolution 105, nematode and leaf rust resistant 106, bean size and cup quality 107 and mannose-6-phosphate reductase gene (CaM6PR) 108 (Table 4). These available BAC libraries of Coffea are an important resource to support genomic studies such as comparative genomics and the identification of genes and regulatory elements controlling important traits in coffee. Expressed sequence tags (ESTs) The development of large EST sequencing projects has supported the identification of potential genes contributing to quality traits and their interplay with other genes through essential biochemical pathways. ESTs facilitate the development of whole transcriptome analysis and the identification of the biosynthetic pathways associated with the expression of quality and the genes within these pathways 10. The EST resources available in coffee have been reviewed by Kochko et al. 85 and are summarised in Table 5. A total of 246,500 ESTs have been reported with 69,801 for C. canephora, 166,133 for C. arabica and 10,566 for C. racemosa using different organs of the plant (leaves, flower, fruits and roots) at different stages of development and maturation However, not all these resources are published or available which limits their use. A number of studies have used the available resource of ESTs for further research such as microsatellite marker discovery 116, analysis of transcriptome divergence and analysis of genes involved in the biosynthesis pathways of lipids and main storage proteins 117, gene structure prediction

16 and functional annotation with the detection of a total of 345 pathways in C. arabica and 300 pathways in C. canephora 118. This genomic resource will be very important for further research on coffee quality improvement. QTL identification and genetic mapping The construction of genetic maps is often the first step for molecular dissection of complex traits (yield components, quality and disease). Genetic maps enable the mapping of quantitative trait loci (QTLs) controlling these traits and thus provide a framework for isolation of candidate genes followed by manipulation of genes for yield increase and quality improvement 119. In many cases, markers linked to QTLs are directly used in MAS without the need of understanding gene functionality underlying the QTLs. Linkage mapping in coffee in general requires more effort and cost than in annual crops due to its long life cycle, low polymorphism (in the case of C. arabica with the exception of Ethiopian cultivars and germplasm), and the absence of a large collection of DNA markers as well as genomic resources 119. The first genetic maps were constructed in the diploid C. canephora by Paillard et al. 120 followed by several others constructed for different coffee populations 51, 56, 96, 102, (Table 6). To date, a number of QTLs analyses relating to quality compounds and cup have been performed on C. canephora and other species, but none has been reported for C. arabica as indicated in Table 7 50, 51, 56, 131, 132 Co-location of some QTLs and genes involved in different metabolic pathways related to coffee quality were noted 51. Using such candidate genes in association mapping may determine the allelic variation of coffee quality 133. The co-localisation between QTLs relating to organoleptic traits and genes associating with caffeine or CGA

17 biosynthesis may explain the fact that both CGA and caffeine are involved in conferring bitterness in the coffee cup 51. Gene identification The identification of genes relating to quality is one of the main objectives of several coffee research groups around the world. Thanks to the concerted efforts on coffee genomics, a number of coffee candidate genes have been identified and some of them have been cloned and characterised. These results are useful to the coffee genetics community, especially those on genes encoding the enzymes of key metabolic processes. These genes are candidate genes which may control the variability of coffee quality 10. Genes regulating the main chemical components that are thought to be involved in the flavour and sensory quality of coffee are listed in Table 8 13, 51, 86, 104, 117, 131, Recently, 36,935 unigenes from leaves and fruits of C. eugenioides, one of C. arabica s ancestors, were identified by Yuyama et al A sub-set of these genes related to sugar metabolism and fruit ripening were examined for their expression. This will be a valuable resource for molecular and genetics studies of commercial coffee species. Currently there are works on transcriptomics to understand the transcriptional regulations during seed development or more specifically for caffeine accumulation 143 and chlorogenic acids 144 or galactomannan 60. These could facilitate the development of robust genomics/metabolomics fingerprints of coffee bean quality and ultimately be useful in breeding programs for coffee quality. Although several genes encoding the biosynthesis of biochemical compounds in coffee have been identified in C. arabica and C. canephora, there are no genes identified for trigonelline synthesis and no studies on allelic variation (e.g., SNPs) associated with low and high levels of the key biochemical compounds which can be utilised in MAS. However, sequences from the genes that are known can serve as useful references in

18 re-sequencing to detect polymorphisms for genetic mapping of the control of biochemical compounds in C. arabica. ASSOCIATION GENOMICS APPROACH USING NEXT GENERATION SEQUENCING (NGS) AND PERSPECTIVES FOR THE STUDY OF GENETICS OF COFFEE QUALITY Research on genetics and genomics for coffee, especially in relation to quality is relatively limited compared to other crops and thus belies its potential and economic contribution. The reasons could possibly be due to the lack of funding most coffee growing regions being developing countries the complexity of the quality traits and the limitations of the technology used. More studies have been conducted for robusta coffee than arabica coffee due to its lower ploidy level and greater genetic diversity. Previous studies showed considerable genetic variation in compounds defining coffee quality among coffee species, especially in diploids. In addition, a number of genes relating to the synthesis of sucrose, caffeine, several sub-groups of CGAs and lipids have also been identified. These studies provided gene sequences via ESTs that will facilitate further studies on bean composition relating to coffee quality. However, there has been no research focusing on allelic variation of these genes in natural populations except bi-parental mapping of QTLs for trigonelline and CGA contents. A number of previous studies involving genome-wide association studies (GWAS) using next generation sequencing (NGS) have used these genetic approaches to detect candidate genes/qtls of complex traits such as physical wood properties, resistant genes, drought tolerance, cold hardiness and timing of bud set summarised by Sexton et al. 145 and successfully applied in various crops including annuals (rice, maize, soybean, wheat) or perennial crops and forestry trees (pine, cottonwood) summarised by Hall et al Based on knowledge of biochemical compounds (and their reference gene sequences) that are thought to be involved in the flavour and sensory quality, it should be possible to apply

19 NGS and GWAS combined with bi-parental QTL mapping to detect valuable haplotypes from natural populations for use in breeding coffee varieties with better quality. Application of NGS in plant genomics studies has been extensively reviewed by a number of authors NGS lowers the cost of sequencing and generates a large amount of data in a single sequencing run, making it feasible to study genetics at the whole genome level. The steps involved in the NGS strategies (shearing of DNA, ligation, library preparation) as well as different sequencing platforms (Illumina, Life Technologies, PacBio) have been described and their advantages and disadvantages have been reviewed by several authors 151, 152. NGS can be used via an approach of whole genome sequencing or targeted sequencing. Two methods of whole genome sequencing can be used including (1) de novo sequencing - applied for species that do not have pre-existing sequence data, followed by the use of bioinformatics tools to assemble the sequences and obtain the genomic map for that species; and (2) re-sequencing performed on individuals of a species that has already known genome sequence 152. Targeted sequencing includes two approaches, amplicon sequencing and target enrichment. Amplicon sequencing will sequence amplified regions (PCR amplicons) of small and selected regions of the genome with lengths of hundreds of base pairs and with a high depth of coverage to identify common and rare sequence variations. Sequencing PCR amplicons of sets of candidate genes from DNA bulks will help to identify the available variation in these genes exploited in a population 148. Target enrichment is quite similar to amplicon sequencing in terms of using only selected regions or genes enriched in the library from genomic DNA. However, target enrichment allows for larger DNA insert sizes and enables a greater amount of total DNA to be sequenced per sample. For this method, regions of interest in the genome

20 can be targeted, making it an ideal approach for examining specific gene pathways, or as a follow-up analysis to GWAS 153. When making decisions on the approaches of candidate gene or of whole-genome, one of the most critical factors is the extent of linkage disequilibrium (LD) in the organism of interest. In genome-wide studies, the extent of LD determines the mapping resolution that is achieved and the numbers of markers covered 154. Genetic, biochemical, physiological and phenotypic information must be used to support the selection of candidate genes 155. Where a developmental or biochemical pathway is well-understood, candidate gene selection is straightforward; but the researcher is typically limited to the field of known genes, which presents the risk of overlooking causal gene mutations not previously identified 146 NGS technology has been used to assemble the whole genome sequence of C. canephora 86 and to examine genetic change in allopolyploidisation of arabica 44. To obtain the genomic sequence of C. arabica, as only the draft genome of C. canephora is available, a de novo approach can be applied. The other approach for obtaining the genomic sequence of C. arabica is whole genome re-sequencing as the available draft genomic sequence of C. canephora can be used as reference since it is one of progenitors of C. arabica. The whole genome sequence will be a key resource of data to identify gene sequences and SNP markers for most genes including novel and published genes. EST and BAC sequences can be used as reference sequences to assemble sequence reads from NGS data to identify genes/alleles and their positions on the chromosome. In addition, whole-genome re-sequencing and SNP genotyping data can be used to generate a dense SNP genetic map of the population for QTL mapping.

21 Amplicon sequencing in a larger set of samples could be an appropriate approach to genetic association study of quality traits, since the biosynthesis pathways, QTLs and candidate genes of sucrose, caffeine and several subgroups of CGAs are known. This will help to confirm or validate the candidate genes from the literature and identify the alleles that are useful for breeding. For those compounds for which candidate genes have never been identified or very few identified, for example trigonelline and some compounds belonging to the lipid group, the approach could be to use whole genome sequencing via NGS to detect SNPs, then using an association genomics approach to detect the link between SNPs and the traits of interest. Association studies differ from quantitative trait locus mapping by utilising samples from genetically diverse populations that contain short stretches of linkage disequilibrium due to generations of recombination 145. Such studies thus enable relationships between phenotypic traits and underlying DNA sequence variation to be understood. Association studies may be suitable for perennial species such as coffee because they can be carried out on pre-existing populations, in collections or in selection trials. This approach does not require specific populations created by controlled crossing (like conventional genetic mapping approaches) which is very difficult for perennial crops 155. Association or linkage disequilibrium mapping has become a very popular method for dissecting the genetic basis of complex traits in plants 146. For the study on volatile and non-volatile compounds in coffee and that are likely to be playing a role in coffee flavour, once the data on biochemical compounds and SNPs variation are available, association genomics will be applied to identify key genes associating coffee quality. The integration of NGS technologies and association study to analyse complex traits will be a powerful strategy complementing the traditional method of parent-hybrid map construction 152.

22 CONCLUSIONS While whole genome technology will provide a powerful resource, a good start has been made with AFLP and SSR markers associated with quality traits identified for sugar content, caffeine and CGAs in C. arabica. These are already helpful in studies aimed at genetic improvement of coffee quality, but importantly, they will provide useful reference points as the genomic approach is developed. Existing coffee BAC libraries are beneficial in comparative genomics, the identification of genes and regulatory elements controlling important traits in coffee, and will play an important role in validating new C. arabica genomic data as it emerges. With the current knowledge of biochemical compounds that govern coffee bean composition that relates to beverage quality and available genetic resources, it is possible to apply an association genomics approach using both whole genome sequencing and targeted sequencing to detect the link between SNPs and the traits of quality determinants from natural populations for use in quality improvement via breeding programs. However, caution should be taken when using this approach for C. arabica since this species has a narrow genetic base, lacks a reference genome, is a polyploid (4n) and lacks populations from controlled crosses for analysis of specific traits and the validation of markers or genes once detected. The aforementioned difficulties can be overcome by the use of wild arabica accessions from Ethiopia, development of a reference genome sequence for C. arabica, and the complementary use of the two approaches of targeted re-sequencing and whole genome re-sequencing in variant discovery respectively. There is no doubt that the availability of whole genome sequences will provide the greatest stimulus yet to the understanding of the genetic basis of coffee quality traits, and indeed of many other aspects of the plant s biology.

23 ACKNOWLEDGEMENTS This research was supported under Australian Research Council's Linkage Projects (project number LP ). We acknowledge Australia Awards Scholarship (AAS) and QAAFI/UQ for their ongoing support of this research. REFERENCES 1. ICO, ICO Annual Review 2011/12, Ed (2013). 2. Anthony F, Bertrand B, Etienne H and Lashermes P, Coffea and Psilanthus, in Wild Crop Relatives: Genomic and Breeding Resources. Springer, pp (2011). 3. Maurin O, Davis AP, Chester M, Mvungi EF, Jaufeerally-Fakim Y and Fay MF, Towards a phylogeny for coffea (rubiaceae): Identifying well-supported lineages based on nuclear and plastid DNA sequences. Annals of Botany 100: (2007). 4. Davis AP, Chester M, Maurin O and Fay MF, Searching for the relatives of Coffea (Rubiaceae, Ixoroideae): The circumscription and phylogeny of Coffeeae based on plastid sequence data and morphology. American Journal of Botany 94: (2007). 5. Tosh J, Davis AP, Dessein S, Block PD, Huysmans S, Fay MF, Smets E and Robbrecht E, Phylogeny of Tricalysia (Rubiaceae) and its relationships with allied genera based on plastid DNA data: Resurrection of the Genus Empogona. Annals of the Missouri Botanical Garden 96: (2009). 6. Anthony F, Diniz LEC, Combes M-C and Lashermes P, Adaptive radiation in Coffea subgenus Coffea L. (Rubiaceae) in Africa and Madagascar. Plant Systematics and Evolution 285:51 64 (2010). 7. Davis AP, Tosh J, Ruch N and Fay MF, Growing coffee: Psilanthus (Rubiaceae) subsumed on the basis of molecular and morphological data; implications for the size, morphology, distribution and evolutionary history of Coffea. Botanical Journal of the Linnean Society 167: (2011). 8. Charrier A and Eskes AB, Botany and Genetics of Coffee, in Coffee: Growing, Processing, Sustainable Production - A Guidebook for Growers, Processors, Traders, and Researchers, Ed by Wintgens JN. WILEY-VCH Verlag GmbH & Co. KCaA (2004). 9. Lashermes P, Combes M-C, Robert J, Trouslot P, D'Hont A, Anthony F and Charrier A, Molecular characterisation and origin of the Coffea arabica L. genome. Molecular Genetics and Genomics 261: (1999).

24 10. Leroy T, Ribeyre F, Bertrand B, Charmetant P, Dufour M, Montagnon C, Marraccini P and Pot D, Genetics of coffee quality. Brazilian Journal of Plant Physiology 18: (2006). 11. Farah A, Coffee as a speciality and functional beverage, in Functional and speciality beverage technology, ed. by Paquin P. Woodhead Publishing in Food Science, Technology and Nutrition, pp (2009). 12. Flament I, Coffee flavour chemistry. Chichester, UK (2002). 13. Joët T, Laffargue A, Salmona J, Doulbeau S, Descroix F, Bertrand B, Kochko Ad and Dussert S, Metabolic pathways in tropical dicotyledonous albuminous seeds: Coffea arabica as a case study. New Phytologist 182: (2009). 14. Holscher H and Steinhart H, Aroma compounds in green coffee, in Food flavours Generation, Analysis and Process Influence, ed. by Charalambous G. Elsevier Science, Amsterdam, pp (1995). 15. Oestreich-Janzen S, Chemistry of coffee, in Comprehensive Natural Products II: Chemistry & Biology, Ed by L. M and Liu H-W. CAFEA GmbH, Hamburg, Germany (2010). 16. Grosch W, Chemistry III: Volatile compounds, in Coffee Recent development, ed. by Clarke RJ and Vitzthum OG. Blackwell Science (2001). 17. Sunarharum WB, Williams DJ and Smyth HE, Review: Complexity of coffee flavor: A compositional and sensory perspective. Food Research International 62: (2014). 18. Buffo RA and Cardelli-Freire C, Coffee flavour: an overview. Flavour And Fragrance Journal 19: (2004). 19. Mestdagh F, Wocheslander S, Rodriguez A, Egli A, Davidek T and Blank I, Conversion of green coffee precursors into flavour during roasting of arabica and robusta coffees, in ASIC: The 25th International conference on coffee and science, Ed, Colombia (2014). 20. Vossen Hvd, Bertrand B and Charrier A, Next generation variety development for sustainable production of arabica coffee (Coffea arabica L.): a review. Euphytica (2015). 21. Campa C, Ballester JF, Doulbeau S, Dussert S, Hamon S and Noirot M, Trigonelline and sucrose diversity in wild Coffea species. Food Chemistry 88:39-43 (2004). 22. Campa C, Doulbeau S, Dussert S, Hamon S and Noirot M, Diversity in bean caffeine content among wild Coffea species: evidence of a discontinuous distribution. Food Chemistry 91: (2005a). 23. Ky C-L, Louarn J, Dussert S, Guyot B, Hamon H and Noirot M, Caffeine, trigonelline, chlorogenic acids and sucrose diversity in wild Coffea arabica L. and C. canephora P. accessions. Food Chemistry 75: (2001b). 24. Mazzafera P, Trigonelline in coffee. Phytochemistry 30: (1991).

25 25. Bertrand B, Villarreal D, Laffargue A, Posada H, Lashermes P and Dussert S, Comparison of the effectiveness of fatty acids, chlorogenic acids, and elements for the chemometric discrimination of coffee (Coffea arabica L.) varieties and growing origins. Agricultural and Food chemistry 56: (2008). 26. Villarreal D, Laffargue A, Posada H, Bertrand B, Lashermes P and Dussert S, Genotypic and environmental effects on coffee (Coffea arabica L.) bean fatty acid profile: impact on variety and origin chemometric determination. Journal of Agricultural and Food Chemistry 57: (2009). 27. Tessema A, Alamerew S, Kufa T and Garedew W, Genetic diversity analysis for quality attributes of some promissing Coffea arabica germplasm collections in Southwestern Ethiopia. Journal of Biological Sciences 11: (2011). 28. Moschetto D, Montagnon C, Guyot B, Perriot JJ, Leroy T and Eskes A, Studies on the effect of genotype on cup quality of Coffea canephora. Tropical Science 36:18 31 (1996). 29. Wintgens JN, Factors influencing the Quality of Green Coffee, in Coffee: Growing, Processing, Sustainable Production - A Guidebook for Growers, Processors, Traders, and Researchers, Ed by Wintgens JN. WILEY-VCH Verlag GmbH & Co. KCaA (2012). 30. Bertrand B, Vaast P, Alpizar E, Etienne H, Davrieux F and Charmetant P, Comparison of bean biochemical composition and beverage quality of Arabica hybrids involving Sudanese-Ethiopian origins with traditional varieties at various elevations in Central America. Tree Physiology 26: (2006). 31. Bertrand B, Boulanger R, Dussert S, Ribeyre F, Berthiot L, Descroix F and Joet T, Climatic factors directly impact the volatile organic compound fingerprint in green Arabica coffee bean as well as coffee beverage quality. Food Chemistry 135: (2012). 32. Joët T, Laffargue A, Descroix F, Doulbeau S, Bertrand B, kochko Ad and Dussert S, Influence of environmental factors, wet processing and their interactions on the biochemical composition of green Arabica coffee beans. Food Chemistry 118: (2010). 33. Barbosa JN, Borém FM, Cirillo MÂ, Malta MR, Alvarenga AA and Alves HMR, Coffee quality and its interactions with environmental factors in Minas Gerais, Brazil. Journal of Agricultural Science 4 (2012). 34. Muschler RG, Shade improves coffee quality in a sub-optimal coffee-zone of Costa Rica. Agroforestry systems 51: (2001). 35. Duarte GS, Pereira AA and Farah A, Chlorogenic acids and other relevant compounds in Brazilian coffees processed by semi-dry and wet post-harvesting methods. Food Chemistry 118: (2010). 36. Avelino J, Barboza B, Araya JC, Fonseca C, Davrieux F, Guyot B and Cilas C, Effects of slope exposure, altitude and yield on coffee quality in two altitude terroirs of Costa Rica, Orosi and Santa Mar ıa de Dota. Journal of the Science of Food and Agriculture 85: (2005).

26 37. Decazy F, Avelino J, Guyot B, Perriot JJ, Pineda C and Cilas C, Quality of different Honduran coffees in relation to several environments. Journal of Food Science 68: (2003). 38. Borém FM, Ribeiro DR, Taveira JHS, Prado MVB, Ferraz V, Tosta MF, Luz MPS and Cortez RM, Genotype and environment Interaction in chemical composition and sensory quality of natural coffees, in The 25th International conference on coffee and science ASIC 2014, Ed, Colombia (2014). 39. Joët T, Bertrand B and Dussert S, Environmental effects on coffee seed biochemical composition and quality attributes: a genomic perspective, in The 25th International conference on coffee and science ASIC 2014, Ed, Colombia (2014b). 40. Santos CAFd, Leitão AE, Pais IP, Lidon FC and Ramalho JC, Perspectives on the potential impacts of climate changes on coffee plant and bean quality. Emir J Food Agric 27: (2015). 41. Lecolier A, Verdeil J-L, Escoute J, Chrestin H and Noirot M, Laurina mutation affected Coffea arabica tree size and shape mainly through internode dwarfism. Trees 23: (2009). 42. Nic-Can GI and Pena CDl, Chapter 6: Epigenetic Advances on Somatic Embryogenesis of Agronomical and Important Crops, in Epigenetics in plants of agronomic importance: Fundamentals and applications transcriptional regulation and chromatin remodelling in plants, ed. by Alvarez-Venegas R, Peña CDl, Armando J and Casas-Mollano. Springer (2014). 43. Landey RB, Cenci A, Georget F, Bertrand B, Camayo G, Dechamp E, Herrera JC, Santoni S, Lashermes P, Simpson J and Etienne H, High genetic and epigenetic stability in Coffea arabica plants derived from embryogenic suspensions and secondary embryogenesis as revealed by AFLP, MSAP and the phenotypic variation rate. PLOS ONE 8 (2013). 44. Lashermes P, Combes M-C, Hueber Y, Severac D and Dereeper A, Genome rearrangements derived from homoeologous recombination following allopolyploidy speciation in coffee. The Plant Journal 78: (2014). 45. Bottley A, Chapter 3: Epigenetic variation amongst polyploidy crop species, in Epigenetics in plants of agronomic importance: Fundamentals and applications transcriptional regulation and chromatin remodelling in plants, ed. by Alvarez- Venegas R, Peña CDl, Armando J and Casas-Mollano. Springer (2014). 46. Vidal RO, Mondego JM, Pot D, Ambrosio AB, Andrade AC, Pereira LF, Colombo CA, Vieira LG, Carazzolle MF and Pereira GA, A high-throughput data mining of single nucleotide polymorphisms in Coffea species expressed sequence tags suggests differential homeologous gene expression in the allotetraploid Coffea arabica. Plant Physiol 154: (2010). 47. Combes M-C, Dereeper A, Severac D, Bertrand B and Lashermes P, Contribution of subgenomes to the transcriptome and their intertwined regulation in the allopolyploid Coffea arabica grown at contrasted temperatures. New phytologist 200: (2013).