Differential regulation of grain sucrose accumulation and metabolism in Coffea arabica (Arabica) and Coffea canephora (Robusta) revealed through gene expression and enzyme activity analysis


Author for correspondence:
Isabelle Privat
Tel:+33 2 47 62 83 83
Fax:+33 2 47 49 14 14
Email: isabelle.privat@rdto.nestle.com


  • • Coffea arabica (Arabica) and Coffea canephora (Robusta) are the two main cultivated species used for coffee bean production. Arabica genotypes generally produce a higher coffee quality than Robusta genotypes. Understanding the genetic basis for sucrose accumulation during coffee grain maturation is an important goal because sucrose is an important coffee flavor precursor.
  • • Nine new Coffea genes encoding sucrose metabolism enzymes have been identified: sucrose phosphate synthase (CcSPS1, CcSPS2), sucrose phosphate phosphatase (CcSP1), cytoplasmic (CaInv3) and cell wall (CcInv4) invertases and four invertase inhibitors (CcInvI1, 2, 3, 4).
  • • Activities and mRNA abundance of the sucrose metabolism enzymes were compared at different developmental stages in Arabica and Robusta grains, characterized by different sucrose contents in mature grain.
  • • It is concluded that Robusta accumulates less sucrose than Arabica for two reasons: Robusta has higher sucrose synthase and acid invertase activities early in grain development – the expression of CcSS1 and CcInv2 appears to be crucial at this stage and Robusta has a lower SPS activity and low CcSPS1 expression at the final stages of grain development and hence has less capacity for sucrose re-synthesis. Regulation of vacuolar invertase CcInv2 activity by invertase inhibitors CcInvI2 and/or CcInvI3 during Arabica grain development is considered.

Coffea Arabica


Coffea canephora


Coffea arabica invertase 3


Coffea canephora invertase 1


Coffea canephora invertase 2


Coffea canephora invertase inhibitor 1


Coffea canephora invertase inhibitor 2


Coffea canephora invertase inhibitor 3


Coffea canephora invertase inhibitor 4


Coffea canephora sucrose synthase 1


Coffea canephora sucrose synthase 2


Coffea canephora sucrose phosphatase 1


Coffea canephora sucrose phosphate synthase 1


Coffea canephora sucrose phosphate synthase 2


Coffee is a very important crop with more than seven million tons of green beans produced every year. Moreover, after oil, coffee is ranked second on international trade exchange markets. The two main species are cultivated throughout the tropics; these are Coffea arabica and Coffea canephora, representing 70 and 30% of world coffee production, respectively. In terms of cup quality, C. arabica (Arabica) is more appreciated by consumers because it less bitter and has a better flavour compared with C. canephora (Robusta). Cup quality is a vast subject, the determinants of which are not well understood.

Sucrose is one of the major sources of the free reducing sugars participating in the Maillard reaction that occurs during the roasting of the coffee grains. This reaction generates a significant number of products, including caramel, sweet and burnt-type aromas, and dark colors, which are typically associated with coffee flavor (Russwurm, 1969; Holscher & Steinhart, 1995). Although glucose and fructose are present in the mature grain, together they only represent approx. 0.03% of the dry weight of C. arabica grains. Sucrose is by far the most important free nonreducing sugar present in mature coffee grains (Rogers W. John et al., 1999). Furthermore, the lowest-quality Arabica grains have appreciably more sucrose (7.3–11.4%) than the highest-quality Robusta grains (4–5%; Russwurm, 1969; Chahan et al., 2002). While sucrose is significantly degraded during roasting, significant amounts (0.4–2.8% dry weight; DW) still remain in the roasted grain. Hence, the sucrose content of the roasted grain adds to the sweetness of the coffee and improves cup quality. Since sucrose plays a key role in determining coffee cup quality, it is important to develop a better understanding of the factors that control sucrose metabolism and accumulation in the different coffee varieties.

In higher plants, sucrose is the principal form in which carbon is translocated between photosynthetic (source) and nonphotosynthetic (sink) cells (Ziegler, 1975). Sucrose metabolism has been extensively studied in tomato (Solanum lycopersicum), a close relative of coffee. They are both members of the asterid I class and share a large number of common genes, especially during fruit development. The key reactions regulating sucrose homeostasis in tomato fruit are as follows: (i) the continuous rapid degradation of sucrose in the cytosol by sucrose synthase (SuSy) and cytoplasmic invertase (neutral); (ii) sucrose synthesis by either SuSy or sucrose-phosphate synthase (SPS) together with sucrose phosphatase (SP); (3) sucrose hydrolysis in the vacuole and/or in the apoplast by acid vacuolar and/or cell wall-bound invertase; and (iv) the rapid synthesis and breakdown of starch in the amyloplast (Nguyen-Quoc & Foyer, 2001; Rausch & Greiner, 2004; Roitsch & Gonzalez, 2004). As in other sink organs, the pattern of sucrose unloading is not constant during tomato fruit development. In the early stages of fruit development, sucrose is unloaded directly into cells from the phloem by a symplastic pathway involving direct plasmodesmatal connections between cells, and hence sucrose is not degraded to its composite hexoses during unloading at this stage. The SuSy and acid invertase activities of the sink cells are highest at this stage and are directly correlated with sucrose unloading capacity from the phloem, contributing directly to the phenomenon called sink strength (Klann et al., 1993, 1996; Godt & Roitsch, 1997; Nguyen-Quoc & Foyer, 2001). However, the symplastic connections are severed as the fruit develops and aploplastic pathways then become the predominant means of sucrose unloading and uptake. Sucrose is rapidly hydrolyzed by the cell wall-bound acid invertases during apoplastic unloading. Therefater, the glucose and fructose products are imported into the cells by hexose transporters. Sucrose is subsequently synthesized de novo in the cytoplasm by SuSy or SPS (Miron & Schaffer, 1991; Dali et al., 1992). SPS catalyzes an essentially irreversible reaction in vivo as a result of its close association with the enzyme sucrose phosphate phosphatase (SP; Echeverria et al., 1997). Fruit SuSy activity declines in parallel with the loss of the symplastic connections and it eventually falls below detectable levels at the onset of ripening (Robinson et al., 1988; Wang et al., 1993). Thereafter, SPS and SP become the predominant enzymes of sucrose synthesis, particularly at the late stages of tomato fruit development.

Research over the past two decades has confirmed the crucial roles of the sucrose-metabolizing enzymes in organ growth and development. For example, studies on mutants with reduced SuSy activity in maize (Chourey & Nelson, 1976; Chourey et al., 1998), pea (Craig, 1996), potato (Zrenner et al., 1995) and tomato (D’Aoust et al., 1999) have demonstrated unequivocally that SuSy activity exerts a major control over tissue import capacity and it directly influences the partitioning of carbohydrates, particularly with regard to starch accumulation. The vacuolar and cell wall-bound invertases are also key proteins in the regulation of sucrose metabolism during fruit development. Red-fruited tomato species, such as the commercial Solanum lycopersicum and the wild species S. pimpinellifolium, do not store sucrose but rather accumulate hexoses (glucose and fructose). Evidence from crosses of red-fruited species with sucrose-accumulating green-fruited species (Yelle et al., 1991) has shown the crucial role of acid invertase in preventing sucrose accumulation in red-fruited tomato species. Subsequently, genetic analysis located the locus conferring high amounts of soluble solids in S. pimpinellifolium fruit. This was coincident with the known position of vacuolar invertase TIV1 (Grandillo & Tanksley, 1996; Tanksley et al., 1996). Studies involving antisense TIV1 cDNA expression in transgenic tomatoes confirmed that vacuolar invertase was responsible for high amounts of fruit-soluble solids (Klann et al., 1993, 1996). Thus, vacuolar invertase is considered to play a major role in both the regulation of tissue hexose concentrations and in the mobilization of sucrose stored from the vacuoles of mature fruit (Klann et al., 1993; Yau & Simon, 2003). The constitutive expression of cell wall-bound invertases in transgenic tomato (Dickinson et al., 1991) and tobacco (von Schaewen et al., 1990) plants led to decreased sucrose transport capacities between sink and source tissues and caused stunted growth as well as more general alterations in plant morphology. Similarly, the absence of cell wall invertase has been shown to cause dramatic effects on plant and seed development in several species. For example, transgenic carrots with reduced amounts of cell wall invertase (CWI; Tang et al., 1999) showed poor tap root formation, particularly during the early elongation phase. Similarly, mutations in the maize miniature-1 (mn1) locus, which encodes CWI-2, cause aberrant pedicel formation and a drastic reduction in endosperm size (Miller & Chourey, 1992; Cheng et al., 1996). Interestingly, global acid invertase activities are dramatically reduced in the mn1 mutant, suggesting coordinate control of both the vacuolar and cell wall-bound isoforms.

Invertases are subject to a complex network of transcriptional and posttranscriptional controls that are regulated by developmental, environmental and carbohydrate signals (Sturm, 1999). In addition, plants produce small (< 20 kDa) inhibitory proteins, called invertase inhibitors, that contribute to the posttranscriptional control of invertase activity. Sequences encoding invertase inhibitors have been characterized in tobacco NtINVINH1 (Greiner et al., 1998) and maize ZM-INVINH1 (Bate et al., 2004). The specificities of these invertase inhibitors was demonstrated in vitro by using recombinant proteins (Bate et al., 2004) and in vivo via studies on transgenic plants (Greiner et al., 1999).

In order to improve coffee bean quality by increasing the final sucrose content of ‘green’ coffee, it is necessary to have a better understanding of the factors that control sucrose metabolism and accumulation during coffee grain maturation. Over the last few years, as genomic studies in Coffea have progressed, partial and full-length sequences encoding proteins involved in sucrose metabolism have begun to appear in public databases and in the literature. For example, sequences encoding SuSy (SUS1, SUS2), vacuolar (CaVAC1, partial sequence) and cell wall invertase (CaCWI1, partial sequence) have been progressively characterized and mapped (Leroy et al., 2005). More recently, data from pulse-chase and feeding experiments have revealed that the grain perisperm and endosperm tissues have the capacity to synthesize and accumulate sucrose (Geromel et al., 2006). Moreover, these authors suggested that sucrose synthesis and accumulation resulted from high SuSy activities late in coffee grain development.

The objective of the current work was to identify potential targets (genes and enzymatic activities) that could explain the differences in sucrose accumulation in the ‘green’ coffee grain of the two major commercially exploited species, C. canephora (Robusta) and C. arabica (Arabica). The most appropriate model for sucrose metabolism during fruit development has been established in the tomato pericarp (Nguyen-Quoc & Foyer, 2001). Based on this scheme, our aim was to compare sugar contents and sucrose-metabolizing enzyme activities during development, together with the identification and analysis of the expression patterns of the genes encoding the different isoforms of the sucrose-metabolizing enzymes. The comparative analysis of sucrose metabolism was undertaken during the development of Arabica and Robusta grains that have high and low sucrose accumulation capacities, respectively, during coffee grain maturation. In this way we were able to identify genes and associated enzyme activities that can account for the differences in sucrose accumulation that are generally observed between Arabica and Robusta grains.

Materials and Methods

Plant material

Leaves and fruit at different stages of development (small green fruit (SG), large green fruit (LG), yellow fruit (Y) and red fruit (R)) (Fig. 1) were harvested from coffee trees cultivated in Quito, Ecuador. Four genotypes were used for this study: Coffea arabica L. cv. Caturra (Arabica) CCCA12, CCCA02 and Coffea canephora var. robusta (Robusta) FRT05, FRT64. Owing to the fact that robusta cherries develop over a period of 9–11 months while arabica fruit develop over a period of 6–8 months (Wintgens, 2004), classification of the ripening stages was made on the relative parameters of size, weight and color change, rather than by weeks after flowering (Bargel & Neinhuis, 2005). Harvested samples were frozen immediately and then packaged in dry ice or frozen at –25°C for transportation, then stored at –80°C until use. These samples were used for analysis of sugars, gene expression and enzyme activities. Tissues (roots, young leaves, stem, flowers and fruit at different maturation stages) were also harvested from C. arabica L.cv. Caturra T 2308 grown under glasshouse conditions in Tours (France) and from Coffea canephora var. robusta BP 409 grown in the field at the Indonesian Coffee and Cacao Research Center (ICCRI), Indonesia. Fresh tissues were frozen immediately in liquid nitrogen and then stored at –80°C until use. These samples were used for RNA extraction and genomic DNA extraction for full-length cDNA isolation.

Figure 1.

Full maturation of coffee cherries. Representative coffee cherries of Coffea canephora var. robusta and Coffea arabica at four distinct ripening stages. SG, small green; LG, large green; Y, yellow and R, red. Bars, 1 cm. WAF (wk after flowering) is indicated for each stage and each species.

Extraction of total RNA and cDNA preparation

Total RNA was extracted from powdered samples as described previously (Rogers W. John et al., 1999). Samples were treated with DNase to remove DNA contamination, using the Qiagen RNase-Free DNase kit according to the manufacturer's instructions (Qiagen, Valencia, CA, USA). All RNA samples were analyzed by formaldehyde agarose gel electrophoresis and visual inspection of the ribosomal RNA bands upon ethidium bromide staining. Using oligo (dT20) as a primer, cDNA was prepared from samples (approx. 1 µg) of total RNA using the Superscript II Reverse Transcriptase kit (Invitrogen, Carlsbad, CA, USA).

Quantitative real-time PCR

Since inter- and intraspecific polymorphisms are observed in Coffea, the presence of polymorphisms was analyzed for each gene in each genotype before the design of primers/probe couples for the Q-PCR experiments. Primers were designed for each gene, based on available cDNA sequences. These were used to amplify the corresponding 600 pb sequence from genomic DNA extracted from each genotype. After sequence analysis, polymorphisms were identified and primers/probe couples (Supplementary material, Table S1) were designed using primer express software (Applied Biosystems, Perkin-Elmer, Foster City, CA, USA) based on a nonpolymorphic coding region (data not shown). TaqMan-PCR was performed according to the manufacturer's instructions (Applied Biosystems, Perkin-Elmer, Foster City, CA, USA). Standard curves were generated for all primers and probe couples using serial dilutions of plasmid DNA containing the appropriate target gene sequences. Ct values were determined and plotted against the natural logarithm of DNA concentration. Regression analysis provided a linear function from which the PCR efficiency could be calculated using the term E = e−1/m – 1, where E is the PCR efficiency, e is Euler's number and m is the slope of the regression function. Efficiencies were determined between 90 and 100%.

Reactions involving these cDNA samples contained 1 × TaqMan buffer (Applied Biosystems, Perkin-Elmer, Foster City, CA, USA) and 5 mm MgCl2, 200 µm each of dATP, dCTP, dGTP and dTTP, 100-fold dilution of cDNA corresponding to 0.001 µg of original RNA, and 0.625 units of AmpliTaq Gold polymerase. PCR was carried out using 800 nm of each gene-specific primer, forward and reverse, and 200 nm TaqMan probe. Reaction mixtures were incubated for 2 min at 50°C and 10 min at 95°C, followed by 40 amplification cycles of 15 s at 95°C/1 min at 60°C. Samples were quantified in the GeneAmp 7500 Sequence Detection System (Applied Biosystems). Transcript abundance was determined using rpl39 (large ribosomal subunit 39) as a basis of comparison. Values are the mean of three repetitions ± SE.

PCR amplification of a partial coffee SPS sequence

Degenerate oligonucleotides SPS-3 (5′ggNcgNgaYtctgaYacNggtgg3′) and SPS-4 (5′tggacgacaYtcNccaaaNgcYttNac3′) were deduced from conserved sequences of SPS and used as primers for PCR amplification. PCR reactions were performed in a 50 µl reaction volume with 100 ng genomic DNA from C. canephora BP 409, 0.5 µm of each primer, 200 µm of dNTPs, 1X buffer and 1 U of LA Taq polymerase (TAKARA BIO INC., Otsu, Shiga, Japan). After a pre-denaturing step at 94°C for 5 min, the amplification consisted of 30 cycles of 1 min at 94°C, 1 min at 12 different temperatures (from 45°C to 56°C) and 2 min at 72°C. The resulting PCR fragments were separated and purified by agarose gel electrophoresis. The PCR fragments from the major bands were purified, cloned and sequenced.

Universal genome walker

Genomic DNA was extracted from leaves harvested from C. canephora (Robusta) BP 409 trees (Crouzillat et al., 1996). Genomic DNA was digested with four different restriction enzymes (DraI, EcoRV, PvuI, StuI) and the resulting fragments were ligated blunt-end to the GenomeWalker Adaptor provided by the Universal GenomeWalker kit (BD Biosciences, San José, CA, USA). Both sets of reactions were carried out in accordance with the kit user manual. The four libraries were then employed as templates in a PCR reaction using gene-specific primers (GSP) (Table S2). The reaction mixtures contained 1 µl of GenomeWalker library template, 10 nmol of each dNTP, 50 pmol of each primer and 1 U of LA Taq polymerase in a final volume of 50 µl with the appropriate buffer. The following conditions were used for the first PCR: after denaturing at 95°C for 2 min, the first seven cycles were performed at 95°C for 30 s, followed by an annealing and elongation step at 72°C for 3 min. A further 35 cycles were carried out, with 95°C for 30 s, followed by the annealing/elongation step at 67°C for 3 min. Products from the first amplification using the primer pair AP1/GSP-GW served as template for the second PCR using AP2 and GSP-GWN as primers. The second PCR used 2 µl of the first amplification reaction (undiluted and different dilutions up to 1 : 50), and was performed as described for the first reaction, with the exception that the second reaction used only 25 cycles of amplification. The resulting PCR fragments were separated and purified by agarose gel electrophoresis. PCR fragments from the major bands were purified, cloned and sequenced.

3′ RACE FOR Ccinv1 cDNA isolation

Ribonucleic acid was extracted from various tissues (roots, young leaves, stem, flowers and fruit at different maturation stages) from C. arabica L. cv. Caturra T 2308 and from Coffea canephora var. robusta BP 409 as described previously (Simkin et al., 2006). Then cDNA was prepared from approx. 1 µg total RNA using dT(18)-Tail (5′cttccgatccctacgctttttttttttttttttt3′) primer according to the protocol in the Superscript II Reverse Transcriptase kit (Invitrogen). The cDNA samples generated were used in a PCR reaction with Inv1-3′a1 (5′gacgtgaatggttgctggtcagg3′) and Tail-3′RACE (5′cttccgatccctacgc3′) as primers for the first PCR and Inv1-3′a2 (5′tacagtgggtgctgagctttggt3′) and Tail-3′RACE as primers for the second PCR. The PCR reactions were performed in 50 µl reactions as follows: 5 µl of cDNA, 1 × buffer, 800 nm of each gene-specific primer, 200 µm each dNTP, 0.5 U of LA Taq polymerase. After denaturing at 94°C for 5 min, the amplification consisted of 35 cycles of 1 min at 94°C, 1 min at 55°C and 2 min at 72°C. An additional final step of elongation was performed at 72°C for 7 min. The resulting PCR fragments were separated and purified by agarose gel electrophoresis, and then cloned and sequenced.

Full-length cDNA amplification

In order to amplify full-length Inv1, Inv3 and SPS cDNA, the primer sets INV1-ATG (5′atggctagcttttacctctggctaatgtg3′), INV1-STOP (5′tcaattctttcgattgatactggcattct3′); INV3-ATG (5′atggagtgtgttagagaatatcaact3′), INV3-STOP (5′tcagcaggtccacgaggaggatctct3′) and cDNAC1-am3 (5′atggcgggaaatgactggataaacagttac3′), cDNAC1-am4 (5′ctagcttttgagaacccctagcttttccaac3′) were designed from INV1, INV3 and SPS sequences, respectively, that had been obtained from the genome primer walking or 3′RACE experiments. These three primer sets have been used to perform RT-PCR reactions using cDNA samples prepared with RNA extracted from various tissues of C. arabica L. cv. Caturra T 2308 and C. canephora var. robusta BP 409, as described previously (Simkin et al., 2006). PCR reactions were performed in 50 µl reaction volumes with the following composition: 5 µl cDNA; 1 × buffer, 800 nm of each gene specific primer, 200 µm each dNTP, 0.5 U of LA Taq polymerase. After denaturation at 94°C for 5 min, the amplification consisted of 35 cycles of 1 min at 94°C, 1 min at 55°C and 2 min at 72°C. An additional final step of elongation was done at 72°C for 7 min. Fragments were purified as previously done from agarose gels, and then cloned and sequenced.

DNA sequencing and sequences analysis

Recombinant plasmid DNA was prepared and then sequenced by GATC (Konstanz, Germany). Computer analysis was performed using DNA Star (Lasergene) software. Sequence homologies were verified against GenBank databases using BLAST programs (Altschul et al., 1990).

Determination of soluble sugars

Pericarp and hull tissues were separated at each maturation stage. Between 15 and 20 grains were homogenized in a cryogenic grinder with liquid nitrogen and the powder was then lyophilized for 48 h (Lyolab bII, Secfroid SA, Aclens-Lausanne, Switzerland). Sugar concentrations were determined according Rogers et al., 1999 and expressed in g 100 g−1 DW. Duplicate experiments were performed for each sample and mean values for each sugar are provided.

Enzyme activity measurements

Separate frozen coffee grain samples (20–25 grains per sample) were ground to a fine powder in liquid nitrogen and suspended in extraction buffer containing 100 mm Tricine (pH 7.5), 200 mm KCl, 5 mm DTT, 5 mm MgCl2, 1.3 mm EDTA, 4% (w/v) Polyclar AT, and 1% (v/v) protease inhibitor cocktail (Sigma Aldrich Corp., St Louis, MO, USA). The soluble protein fractions were separated by centrifugation and desalted on Sephadex G-25 columns (PD-10, GE Healthcare Bio-Sciences Corp., Piscataway, NJ, USA) in buffer containing 100 mm Tricine (pH 7.5), 200 mm KCl, 5 mm MgCl2, and 1.3 mm EDTA. SPS and SuSy activities were then measured according to Guy et al. (1992). SPS activities were measured under Vmax conditions in a buffer containing 50 mm MOPS-NaOH (pH 7.5), 15 mm MgCl2, 1.3 mm EDTA, 10 mm uridine 5′-diphosphoglucose, 40 mm glucose-6-phosphate, and 10 mm fructose-6-phosphate. SuSy activities were measured in the direction of sucrose synthesis in the same buffer, except that 10 mm fructose replaced the fructose-6-phosphate. Control reactions were performed in the absence of uridine 5′-diphosphoglucose. Soluble invertase activities were measured in buffer containing 100 mm sodium citrate, 100 mm NaH2PO4, and 100 mm sucrose at either pH 4.8 (acid invertase activity) or pH 7.0 (neutral invertase activity). Reaction mixtures were incubated for 2 h at 25°C. The reactions were then stopped by boiling for 5 min. Glucose and fructose were then determined according to Foyer & Ferrario (1994). Protein was determined according to Bradford (1976). For each experiment, between two and four independent samples were extracted and assayed per developmental stage and each assay was performed in duplicate.

Statistical analysis

Statistical analysis of enzyme activities was performed using a two-way ANOVA (Microsoft Excel). Statistical significance is denoted at the α = 0.05 level as indicated in the text, the absence of any significant difference being indicated by n.s.d.

Accession numbers

Nucleotide sequences published in this article have been deposited at NCBI under accession numbers DQ826510 (CcSS1), DQ834312 (CcSS2), DQ834313 (CcSP1), DQ834314 (CcInv1), DQ834315 (CcInv2), DQ834316 (CaInv3), DQ842235 (CcInv4), DQ834317 (CcInvI1), DQ834318 (CcInvI2), DQ834319 (CcInvI3), DQ834320 (CcInvI4), DQ842233 (CcSPS1 gene), DQ842234 (CcSPS2, partial genomic sequence), DQ834321 (CcSPS1 cDNA).


Sucrose is the main free sugar in the ‘green’ coffee grain

Glucose, fructose, and sucrose were determined during the development of two Robusta (Coffea canephora; FRT05 and FRT64) and two Arabica (Coffea Arabica; CCCA02 and CCCA12) genotypes. These genotypes were selected for the present study because they had been shown previously to have significantly different sucrose concentrations in the mature grain (C. Lambot, unpublished). Developmental patterns of grain glucose, fructose, and sucrose accumulation were similar during the development of both Arabica and Robusta grains (Table 1). At the earliest stage of maturity examined (stage SG), glucose was the main free sugar. Glucose concentrations were much higher in Arabica (CCCA12, 14.42%; CCCA02, 12.46%) than in Robusta (FRT05, 1.5%; FRT64, 2.82%) grains at this stage. Fructose concentrations were lower than glucose but they were also highest in Arabica grains. Glucose and fructose contents decreased during grain development in both species, falling to very low values at the mature red stage (R). The decrease in hexoses was accompanied by an increase in sucrose as the grain matured. Sucrose concentrations were significantly higher in mature Arabica grains (CCCA12, 9.82%; CCCA02, 10.96%) than in the Robusta grains (FRT05, 6.71%; FRT64, 6.60%). Interestingly, the extent of grain sucrose accumulation was not constant during the maturation process. Arabica grains at the LG stage had accumulated < 40% of the final sucrose concentration determined at the R stage. Sucrose accumulated rapidly in the Arabica grains between the LG and Y stages, such that 90% of the final sucrose content was present by the end of the Y stage. Sucrose is accumulated more gradually during Robusta grain maturation (Table 1). This suggests that cherry maturity is a much more important determinant or requirement for sucrose accumulation in Robusta than in Arabica.

Table 1.  Glucose, fructose and sucrose quantification during coffee grain development of two Coffea arabica genotypes, CCCA12 and CCCA02, and two Coffea canephora genotypes, FRT05 and FRT64
GenotypeStageGlucoseFructoseSucrose% sucrose
  1. Coffee cherries at four different maturation stages characterized by size and color have been used for this study: SG, small green; LG, large green; Y, yellow; R, red. Sugar concentration is expressed in g per 100 g DW. Accumulation of sucrose is also indicated as the percentage of the final amount in the red stage. Values are the mean of two replicates.

CCCA12SG14.421.522.65 27
LG5.620.493.11 32
Y0.100.128.04 94
CCCA02SG12.460.892.83 26
LG5.090.44  4 37
Y0.230.309.61 88
FRT05SG1.540.330.73 11
LG1.710.091.46 22
Y0.1003.13 47
FRT64SG2.820.401.79 27
LG2.480.271.94 29
Y0.0404.46 68

Sucrose-metabolizing enzyme activities during coffee grain maturation

The activities of SuSy, soluble acid and neutral invertase and SPS were determined throughout maturation in Arabica (CCCA12) and Robusta (FRT05 and FRT64) grains. These experiments were designed to determine whether the differences in the extent of sucrose accumulation during grain maturation observed in the Arabica and Robusta genotypes (Table 1) were related to changes in enzyme activities. SuSy activity was constant during CCCA12 grain maturation (Fig. 2a) and no significant differences were observed at any stages of development that were studied. By contrast, significant variations (P < 0.001) in SuSy activity were observed during the maturation of grains from both Robusta genotypes. SuSy activities were similar (n.s.d.) in the both Robusta and Arabica (CCCA12) genotypes at the early development (SG) stage. However, SuSy activity increased strongly between the SG and LG stages in both robusta genotypes and then declined drastically between LG and Y stages. It increased again slightly at the final maturation (Y to R stages) step. Interestingly, SuSy activity has been shown to be highly correlated with sink strength (i.e. the capacity of sink tissues to import sucrose) in different species (Robinson et al., 1988; Miron & Schaffer, 1991; Micallef et al., 1995). We thus conclude that sink strength is higher in Robusta than Arabica during the early stages of grain development.

Figure 2.

Enzymatic activities involved in sucrose metabolism during grain development in Coffea arabica and Coffea canephora. Sucrose synthase (SuSy) (a), sucrose-phosphate synthase (SPS) (b), soluble acid (c) and neutral invertase activity (d) have been determined in Coffea arabica CCCA12 and two C. canephora var. robusta genotypes, FRT05 and FRT64. Activity was analyzed in the grain at four maturation stages: SG, small green grain; LG, large green grain; YG, yellow grain; RG, red grain. In each case, values are the mean of at least four estimations ± SE.

Acid invertase activities were very low throughout the period of grain maturation in Arabica (Fig. 2c). By contrast, acid investase activity varied drastically between the two Robusta varieties examined (P < 0.001). The FRT05 grains had fivefold higher levels of acid invertase activity at the early stages of grain development than FRT64 (stage SG). However, acid invertase activity decreased significantly in both Robusta genotypes at the Y and R stages, reaching the very low values observed for Arabica at these stages (n.s.d.). A highly significant development stage × genotype interaction (P < 0.001) was observed. Neutral invertase activity was fairly constant throughout coffee grain development for all three genotypes and there was no significant genotypic effect. However, neutral invertase activities were highest at the SG stage and decreased significantly at the LG stage (P < 0.001), then staying constant through the final stages of maturation. Global neutral invertase activities were low in the Robusta genotypes compared with soluble acid activities, the highest values detected for neutral isoform corresponding to the lowest values detected for the acid invertase isoform, suggesting a predominant role for the vacuolar enzyme (Fig. 2d). By contrast, neutral and acid invertases have similar activities early in the development of Arabica grains.

Sucrose-phosphate synthase activities were similar in the Arabica and Robusta genotypes during grain maturation. While no genotypic effects were observed, SPS activity varied significantly (P < 0.001) with developmental stage (Fig. 2b). SPS activity increased between the SG and LG stages and decreased drastically between LG and Y stages (Fig. 2b), increasing once more at the end of grain maturation. Interestingly, SPS activity was higher in the Arabica genotype than in the two Robusta genotypes at the last stage of grain maturation.

Isolation and characterization of sucrose metabolism-related genes

Enzymes directly involved in the synthesis and degradation of sucrose have been studied in leaves and in developing fruits, tubers, and seeds from species such as tomato (S. lycopersicum), potato (S. tuberosum) and corn (Zea mays). DNA sequences encoding these key enzymes are available in GenBank. We therefore used homologous proteins, especially from organisms closely related to coffee (tomato and potato), to identify similar sequences in different coffee cDNA libraries (Lin et al., 2005). Protein sequences from tomato and potato were used in a tBLASTn search. In silico‘unigenes’, whose open reading frames showed the highest degree of identity with the ‘query’ sequence, were selected for further study. In some cases, the selected ‘unigenes’ contained at least one EST sequence that potentially represented a full-length cDNA clone, and this clone was then selected for re-sequencing to confirm both its identity and the ‘unigene’ sequence (Table 2). Percentage identities and closest annotated protein for each sequence in the databases are presented in Table 3, where each gene annotation is preceded by Cc and Ca to indicate whether the sequence was identified from C. canephora (Robusta) or C. arabica (Arabica).

Table 2.  Identification of cDNA-encoding proteins involved in sucrose metabolism in the Nestlé/Cornell coffee expressed sequence tag (EST) databases
Sucrose synthasecccl27e7NoCcSS1
Sucrose-phosphate synthaseAbsent
Sucrose phosphatasecccl19n15YesCcSP1
Vacuolar invertasecccl20f11YesCcInv2
Cytoplasmic invertasecccp28p22NoCcInv3
Cell wall invertasecccs46w27d20NoCcInv4
Invertase inhibitorcccp2d1YesCcInvI1
Table 3.  Characteristics of cDNA and the corresponding protein involved in sucrose metabolism in coffee
GenescDNAORFProteinIdentity Accession number
  1. cDNA and ORF are indicated in bases, protein size in kDa and the percentage of identity with the closest annotated homolog is mentioned.*, genes or cDNA marked with an asterisk are not full-length.

CcSS1*2543220399C. arabicaCAJ32597
   86S. tuberosumAY205302
CcSS23048242792.699C. arabicaCAJ32596
   89S. tuberosumAY205084
CcSPS131503150117.984N. tabacumAAF06792
CcSPS2*89N. tabacumABA64521
CcSP11721124846.781S. lycopersicumAAO33160
CcInv11731173164.498C. arabicaCAE01317
   66V. viniferaAAT09980
CcInv22212176164.693C. arabicaCAE01318
   78D. carotaCAA77267
CcInv3* 568 32193L. corniculatusCAG30577
CaInv31675167563.890M. esculentaDQ138370
CcInv4* 687 22273C. intybusCAA72009
CcInvI1 762 55820.738P. × acerifoliaCAD20556
   33Z. maysCAC69335
CcInvI2 703 53719.636N. tabacumCAA73333
CcInvI3 828 49518.419.3D. melanogasterAAF45837
CcInvI4 617 55520.243A. deliciosaAB091088
   23N. tabacumCAA73333
   25Z. maysCAC69335

Sucrose synthase, SPS and SP  Two distinct cDNA clones, cccl27e7 (CcSS1) and A5-1540 (CcSS2), encoding sucrose synthase, were identified in the EST coffee databases (Tables 2, 3). The partial clone cccl27e7 encodes only the C-terminal region of the protein that shares 99% identity with a SuSy previously identified in C. canephora (SUS2) (Geromel et al., 2006). The full-length CcSS2 cDNA (clone A5-1540) is 3048 pb long and encodes a protein of 92.6 kDa. This protein has 99% identity with a SuSy identified in C. arabica (SUS1) (Leroy et al., 2005; Privat et al., 2007). The isolated full-length cDNA encoding SP, CcSP1 (clone cccl19n15) is 1721 pb, and encodes a protein with a molecular mass of 46.7 kDa that shares 81% of identity with S. lycopersicum SP (AAO33160) (Lunn, 2003). No cDNA sequences were found that could potentially encode a SPS protein in the available coffee EST databases.

Invertases: cytoplasmic, vacuolar and cell wall isoforms Three distinct ESTs (cccl20f11, cccp28p22 and cccs46w27d20) encoding different invertase isoforms, were isolated (Table 2). The full-length sequence encoded by cccl20f11 clone was annotated as CcInv2 (C. canephora invertase 2) because it has 93% identity to the vacuolar invertase Inv2 (CAE01318, partial sequence) (Leroy et al., 2005) that was previously identified in C. arabica. The partial invertase sequence corresponding to the clone cccp28p22 was annotated as CcInv3 (C. canephora invertase 3) because it shares 93% identity with L. corniculatus neutral invertase (CAG30577; Table 3). The partial invertase sequence corresponding to the clone cccs46w27d20 was annotated as CcInv4 (C. canephora invertase 4) because it shares 73% identity with a C. intybus cell wall invertase (CAA72009). A partial cDNA sequence Inv1 (AJ575257) identified in C. arabica has been previously identified and it encodes a cell wall invertase. The partial sequences of CcInv4 and Inv1 did not overlap so this information could not be used to determine whether these cDNAs represent one gene or two different genes. However, further work showed these sequences represent different invertase genes (see later). Thus CcInv2, CcInv3 and CcInv4 cDNA encode vacuolar, neutral and cell wall invertase, respectively, in C. canephora.

Invertase inhibitors  Four full-length clones (cccp2d1, cccs30w14i24, cccs30w24n8 and A5-1462) which encode putative C. canephora invertase inhibitors were identified. These clones were annotated as CcInvI1, 2, 3 and 4, respectively (Tables 2, 3). The different CcInvI protein sequences have relatively weak homologies with the maize (ZM-INVINH1) or tobacco (NtINVINH1) sequences (Table 3). However, they contain the four ‘conserved’ Cys residues that are essential for function (Hothorn et al., 2003, 2004; Scognamiglio et al., 2003; Rausch & Greiner, 2004; Fig. S1).

Full-length cDNA amplification

SPS  Since SPS is a central enzyme in sucrose metabolism and no SPS genes were found in any of the coffee databases, we decided to use classical molecular biology techniques to isolate a cDNA encoding this enzyme. Using degenerated primers corresponding to two highly conserved SPS domains, two distinct PCR fragments (2000 and 1500 bp, respectively) were amplified from the BP 409 genome (C. canephora). After sequencing of both genomic sequences, an alignment of the encoded protein sequences with the S. lycopersicum protein sequence (AAC24872) indicated that we had isolated partial sequences from two different coffee SPS genes, which we have annotated as CcSPS1 and CcSPS2. The fragments corresponding to CcSPS1 and CcSPS2 partial genomic sequences were 1935 and 1550 bp long, respectively (Fig. S2). The protein sequences encoded by the partial CcSPS1 and CcSPS2 genomic sequences exhibited 70% identity. The CcSPS1 partial protein sequence shares 88% of identity with N. tabacum SPS form A while the CcSPS2 partial protein sequence shares 89% of identity with the N. tabacum SPS form B (Chen et al., 2005). Preliminary expression analysis showed that CcSPS1 was more highly expressed than CcSPS2 in various tissues, including coffee grain. Therefore, we primarily focused on the isolation of the complete sequence for the CcSPS1 gene. Using several rounds of primer directed genome walking using the Genome Walker™ technique, a full-length genomic sequence for the CcSPS1 gene was generated. The CcSPS1 gene is 7581 bp long, and is characterized by 13 exons and 12 introns. Using specific primers deduced from the CcSPS1 genomic sequence, a full-length cDNA encoding CcSPS1 was then amplified by RT-PCR. This cDNA is 3150 bp long and encodes as protein of 1049 aa with a predicted molecular weight of 117.9 kDa. The protein sequence encoded by the full-length CcSPS1 cDNA shares 84% of identity with the full-length sequence N. tabacum SPS form A (Table 3).

CaInv3  As noted earlier, the protein sequence encoded by the cccp28p22 (CcInv3) clone, which has 93% identity with the neutral cytoplasmic invertase from L. corniculatus (CAG30577) is not full-length. Using the Genome Walker technique, a genomic sequence from C. canephora corresponding to the missing 5′ region of the CcInv3 coding sequence was isolated and characterized. Subsequently, using specific primers deduced from the new genomic sequence data and the existing cDNA data, a full-length cDNA was amplified by RT-PCR. Several RNA samples from C. arabica T2308 and C. canephora BP 409 were used, but amplification corresponding to the full-length cDNA sequence was only obtained using RNA extracted from T2308. The protein sequence encoded by this new cDNA, CaInv3 (C. arabica invertase 3) (Table 3), shares 90% identity with a neutral invertase (DQ138370) from M. esculenta.

CcInv1  The full-length cDNA encoding the Inv1 protein in C. canephora was characterized using 3′RACE and Genome Walker techniques. The full-length cDNA for CcInv1 was found to be 1731 b long and the deduced protein was 576 aa long with a predicted molecular weight of 64.6 kDa. CcInv1 shares 98 and 66% of identity with Inv1 and CcInv4, respectively. There are at least two different genes encoding cell wall invertase in the C. canephora genome.

Comparative expression of sucrose metabolism-related genes

A comparative expression analysis of the sucrose metabolism-related genes identified earlier was performed using the same samples as those used for biochemical analysis. Transcript-specific assays based on fluorescent real-time RT-PCR (TaqMan, Applied Biosystems) were developed for each gene, and the relative transcript abundances in each RNA sample were quantified in relation to the expression of a constitutive transcribed gene (rpl39) in the same sample. Figures 3–6 depict the results of the quantitative RT-PCR analyses obtained. Furthermore, in order to determine whether some of these genes show grain-specific expression, we conducted a parallel analysis on transcript accumulation in mature FRT05 leaves (Fig. 4).

Figure 3.

Expression of sucrose-cleaving enzyme genes during grain development in Coffea canephora and Coffea arabica determined by quantitative RT-PCR. Transcript abundances of sucrose synthase CcSS1, CcSS2 (a) and invertases CcInv1–4 (b) were analyzed in grains of Coffea arabica CCCA 02 and CCCA12 and C. canephora var. robusta FRT64 and FRT05 at four maturation stages: SG, small green grain; LG, large green grain; YG, yellow grain; RG, red grain. The expression levels are determined relative to the expression of transcripts of the constitutively expressed rpl39 gene in the same samples. In each case, values are the mean of three estimations ± SE.

Figure 4.

Expression of genes involved in sucrose metabolism in Coffea canephora FRT05 leaves. Transcript abundances of sucrose synthase (CcSS1–2), invertase (CcInv1–4), sucrose-phosphate synthase (CcSPS1–2), sucrose phosphatase (CcSP1) and invertase inhibitor (CcInvI1–4) were analyzed in leaves relative to the expression of transcripts of the constitutively expressed rpl39 gene. In each case, values are the mean of three estimations ± SE.

Figure 5.

Expression of sucrose synthesis enzyme genes during grain maturation in Coffea canephora and Coffea arabica determined by quantitative RT-PCR. Transcript abundances of sucrose-phosphate synthase (CcSPS1–2) and sucrose phosphatase CcSP1 were analyzed in grains of Coffea arabica CCCA02 and CCCA12 and C. canephora var. robusta FRT64 and FRT05 at four maturation stages: SG, small green grain; LG, large green grain; YG, yellow grain; RG, red grain. The expression levels are determined relative to the expression of transcripts of the constitutively expressed rpl39 gene in the same samples. In each case, values are the mean of three estimations ± SE.

Figure 6.

Expression of invertase inhibitor genes during grain maturation in Coffea canephora and Coffea arabica determined by quantitative RT-PCR. Transcripts abundances of CcInvI1, CcInvI2, CcInvI3 and CcInvI4 were analyzed in grains of Coffea arabica CCCA 02 and CCCA 12 and C. canephora var. robusta FRT64 and FRT05 at four maturation stages: SG, small green grain; LG, large green grain; YG, yellow grain; RG, red grain. The expression levels are determined relative to the expression of transcripts of the constitutively expressed rpl39 gene in the same samples. In each case, values are the mean of three estimations ± SE.

Sucrose synthase and invertase mRNAs accumulate during the early stages of coffee grain maturation  Sucrose synthase (CcSS2) transcripts accumulate early during the development of both Robusta and Arabica grains. CcSS2 transcripts decrease at later maturation stages in both Robusta and Arabica grains, but the decrease was greatest in the two Robusta genotypes (Fig. 3a). The second SuSy gene (CcSS1) is poorly expressed compared with CcSS2 at all the stages analyzed in Robusta and Arabica grains, suggesting a predominant role for CcSS2 during coffee grain development. The CcSS1 and CcSS2 mRNAs were more abundant in fully expanded leaves than in grains (Fig. 4). However, CcSS2 mRNA is still predominant over CcSS1 in fully expanded leaves.

The two cell wall invertase isoforms CcInv1 and CcInv4 show different patterns of expression during coffee grain maturation (Fig. 3b). CcInv1 transcripts are barely detectable relative to CcInv4 transcripts during coffee grain maturation, while CcInv4 and CcInv1 are expressed to the same degree in leaves (Fig. 4). CcInv4 mRNA accumulates only at the early SG and LG stages in all genotypes, with slightly higher transcript abundance in Arabica than in Robusta grains. The vacuolar invertase transcript CcInv2 is expressed at a higher level than the two cell wall invertases. Furthermore, CcInv2 expression was restricted to the early SG and LG stages in all genotypes. However, CcInv2 mRNA was not accumulated at the same abundance in all genotypes, being sixfold higher in CCCA12 than in CCCA02 and threefold higher in FRT05 than in FRT64. The neutral cytoplasmic invertase CcInv3 was expressed at all stages examined, with relatively constant amounts throughout grain development. Nevertheless, CcInv3 expression was generally lower than that of CcInv2 but higher than CcInv4 (Fig. 3). Overall, these results indicate that the vacuolar invertase CcInv2 and the neutral invertase CcInv3 are the predominant transcripts at the earliest stage of grain development. Interestingly, a different pattern of expression was observed for the leaf invertases (Fig. 4). All four genes appear to be expressed at equivalent levels in leaves compared with the variable expression pattern observed during grain development. Moreover, all four invertase mRNAs accumulate to much higher amounts in leaves than in the grain. One notable exception is the vacuolar invertase CcInv2 mRNA, which is slightly higher at the SG stage of grain development than the amount present in leaves.

Sucrose phosphate synthase and SP transcripts accumulate during the later stages of coffee grain maturation. The expression patterns of CcSPS1 and CcSPS2 transcripts are markedly different from those of genes encoding the sucrose cleaving enzymes (SuSy and invertase) in the developing grain (Fig. 5). Only low abundances of CcSPS1 transcripts were detected at the earliest stages of grain development in all genotypes. Increases in CcSPS1 transcripts were observed between the LG and Y stages, being highest in the Arabica genotypes compared with the Robusta genotypes. The CcSPS1 mRNA abundances were high and constant during the final stages of Arabica grain maturation. By contrast, CcSPS2 mRNA accumulated to a lesser extent than CcSPS1 mRNA in Arabica genotypes at the final stages of grain maturation. However, CcSPS1 and CcSPS2 transcripts were present at about the same abundance in both Robusta genotypes. CcSP1 transcripts encoding SPS were present at very low levels (equivalent to the lower values for CcSPS1 and CcSPS2 transcripts; Fig. 5). However, there was a clear trend towards lower CcSP1 transcript abundance as grain development progressed in all genotypes tested (Fig. 5). Similar to the SuSy and the invertase genes, the CcSPS1, CcSPS2 and CcSP1 genes were more highly expressed in leaves (Fig. 4) than in grain (Fig. 5).

Potential impact of invertase inhibitors on vacuolar invertase activity

It is interesting to note that differences observed for acid invertase activity between CCCA12 and FRT05 (Fig. 2c) do not correlate with CcInv2 mRNA transcript abundance (Fig. 3). At the earliest stage of grain maturation, acid invertase activity was drastically (10 times) higher in FRT05 than in CCCA12, while CcInv2 mRNA accumulation was similar in both genotypes. By contrast, differences in acid invertase activity observed between FRT05 and FRT64 (Fig. 2) were directly correlated with the relative abundance of CcInv2 transcripts at the early developmental stages. Recent publications have shown that invertase activity is inhibited by interaction with small proteins called invertase inhibitors (Greiner et al., 1998, 2000). As noted earlier, four full-length cDNAs potentially encoding invertase inhibitors (Table 2) were found in the coffee databases. Data on the relative expression patterns of these genes are presented in Fig. 6. The invertase inhibitor transcripts are generally more abundant than the different invertase genes. The expression patterns of CcInvI2, 3 and 4 were similar, with expression largely restricted to the earliest stages (SG and LG stages) of grain maturation. Of the transcripts encoding invertase inhibitors, CcInvI2 were the most abundant in both Arabica genotypes, followed by CcInvI4, CcInvI3 and finally CcInvI1. Conversely, CcInvI4 transcripts are the most abundant in Robusta genotypes, followed by CcInvI1, CcInvI3 and finally CcInvI2. Higher abundances of CcInvI2 and CcInvI3 transcripts were detected in Arabica genotypes compared with the Robusta genotypes, CcInvI2 showing the highest expression level. It is interesting to note that there is a strong correlation between high transcript abundance for the invertase inhibitor genes CcInvI2 and CcInvI3 and the absence of acid invertase activity in the Arabica CCCA12 genotype at the earliest stages of grain development. The observation that both Arabica genotypes exhibit high abundances of CcInvI2 and CcInvI3 transcripts during the early phases of grain development, while both Robusta genotypes do not, suggests that both these genes/proteins may play a role in the regulation of acid invertase activity in Arabica grains at the earliest stage of development.


There is a strong correlation between the sucrose concentration of mature coffee grains and cup quality (Russwurm, 1969; Holscher & Steinhart, 1995). Strategies designed to increase the sucrose content of coffee grains either by marker-assisted selection, classical breeding or more direct genetic manipulation methods require a precise understanding of how sucrose metabolism is regulated during coffee grain development. The present study was designed to identify genes and associated enzyme activities that are important in determining sucrose metabolism and accumulation in Arabica and Robusta grains. Using the C. canephora (Robusta) and the C. arabica (Arabica) genomes, we first set out to isolate the full set of genes that are central to grain sucrose metabolism. The first group of sequences that we characterized encode sucrose-cleaving enzymes, including two cDNAs, CcSS1 and CcSS2. The second group is the invertase family. The three cDNAs isolated in this study, CcInv2, CcInv3 and CcInv4, encode the vacuolar, cytoplasmic and cell wall isoforms, respectively. Finally, we have isolated cDNAs encoding sucrose synthesis enzymes and characterized two different SPS sequences: CcSPS1 and CcSPS2 as well as a coffee SP cDNA, CcSP1. The identification of this new set of genes in the C. canephora or C. arabica genomes is important as it allows a more precise analysis of the gene expression patterns that underpin sucrose accumulation during grain maturation.

Sucrose homeostasis and net accumulation in sink organs involve the interaction of several ‘futile’ cycles incorporating import, degradation and re-synthesis (Nguyen-Quoc & Foyer, 2001) and encompassing reactions in the cytosol, the vacuole and the apoplast. In order to understand the relative importance of these processes in the developing coffee grain, we measured the gene expression patterns and corresponding activities of sucrose-metabolizing enzymes during the maturation of two high sucrose Arabica genotypes CCCA12 and CCCA02 and two low sucrose Robusta genotypes, FRT05 or FRT64. The Arabica genotypes accumulate approx. 30% more sucrose in the mature grain than the Robusta varieties (Table 1). The results obtained here allow us to draw the following conclusions:

  • • Sink strength is higher in Robusta than Arabica grains in early development. Developmental changes in SuSy activity were different in Robusta and Arabica grains. While SuSy activity was not significantly different during Arabica grain development, it showed a marked increase between the SG and LG stages in both Robusta varieties, decreasing at the yellow stage, but rising again late in development (Fig. 2a). Hence, high SuSy activity is associated with the growth phase of coffee fruit development and, more specifically, with endosperm development (De Castro & Marraccini, 2006; Geromel et al., 2006). Since SuSy activity is correlated with sink strength during the early stages of development (Sun et al., 1992; Wang et al., 1993; Zrenner et al., 1995; N'tchobo et al., 1999), our data suggest that Robusta has a greater sink strength than Arabica during grain expansion. Of the two SuSy genes characterized here, CcSS2 is highly expressed early in grain development and then decreases in all four genotypes, the decrease being most pronounced in the Robusta genotypes (Fig. 3a). By contrast, CcSS1 mRNA is present at very low levels at all stages. These results suggest that CcSS2 encodes the predominant isoform implicated in coffee endosperm development. The abundance of CcSS2 transcripts and pattern of change during development are similar in Arabica and Robusta. Variations in CcSS2 transcripts cannot therefore account alone for the significant differences in SuSy activities observed between the two species. However, SuSy activity is known to be regulated by multiple parameters, particularly substrate availability (Nguyen-Quoc & Foyer, 2001) and post-translational modification through covalent modification (Anguenot et al., 2006). The results presented here differ from those published by Geromel et al. (2006) in one aspect only; we did not observe a massive increase in SuSy activity late in Arabica grain development. This difference is explained by differences in the last maturation stage used in both studies. The last coffee cherry (RG stage) that was studied here may not be as mature as the one used by Geromel et al. (2006). Furthermore, the site location for the growth of the coffee plants was different (Brazil versus Ecuador) in the two studies. Environmental factors are important determinants of fruit maturation, and growth environment may have had an impact on sucrose metabolism.

  • • The expression of invertase inhibitors exerts control over acid invertase activity. Sink strength is also driven by the activity of cellular invertases, especially the vacuolar isoforms (Nguyen-Quoc & Foyer, 2001). Soluble acid invertase activity was much higher in Robusta (especially the FRT05) than in the Arabica genotype, particularly in the early stages of grain development (Fig. 2c). In the Arabica genotype CCCA12, acid invertase activity was low and relatively constant at all of the four grain developmental stages analyzed. These results show that sink strength (determined by acid invertase in association with SuSy) is higher in Robusta than in Arabica during the early stages of grain development, which rely on the symplastic sucrose import. The data presented here show that sucrose degradation is considerably higher in the Robusta than in the Arabica genotypes. The vacuolar invertase is encoded by a single gene, CcInv2, in coffee (D. Crouzillat, unpublished). Our data show that CcInv2 is a high-abundance transcript during early grain development in CCCA12 and in both Robusta varieties (Fig. 3b). Hence, differences in observed acid invertase activities between Arabica and Robusta genotypes cannot be attributed to differences in CcInv2 gene expression. Thus, we explored the possibility of post-translational controls of vacuolar invertase activity in coffee. Invertase activity has been shown to be inhibited by interaction with small proteins (< 20 kDa) (Greiner et al., 1999; Bate et al., 2004; Godt & Roitsch, 2006). Based on homologies, we isolated and characterized four cDNA-encoding potential invertase inhibitors (CcInv1–4 from the coffee databases; Table 2). CcInvI2 and CcInvI3 mRNA were more abundant in Arabica than in Robusta genotypes, with expression highly restricted to the earliest stages of grain maturation, CcInvI2 transcripts attaining the highest accumulation. The appearance of the CcInvI2 and CcInvI3 mRNAs correlates closely with the absence of detectable acid invertase activities at the earliest stages of grain maturation in CCCA12. By contrast, CcInvI2 and CcInvI3 transcripts were not detected in either Robusta genotypes at the same stages, when acid invertase activities were high and CcInv2 mRNA was most abundant. This result is interesting because it demonstrates a direct connection between the expression of specific invertase inhibitors and the inhibition invertase activity. To our knowledge, this is the first time that it has been possible to show such a relationship in the invertase inhibitor family. The cellular localization of the different invertase inhibitors could not be determined using available bioinformatic tools (TargetP, iPSORT). Further experiments are underway to establish the function and specificity of each potential invertase inhibitor, especially CcInvI2 and CcInvI3.

  • Neutral invertase activities, which are considered to reflect the cytoplasmic isoform, were high early in grain development and decreased in a similar manner in Arabica and Robusta grains during maturation (Fig. 2d). The changes in enzyme activities occur in parallel with changes in the abundance of CcInv3 mRNA (Fig. 3b). Neutral and vacuolar invertase activities were similar in both Arabica genotypes. However, the vacuolar isoform activity was predominant in both Robusta genotypes. The abundance of CcInv3 transcripts was much lower than that of CcInv2 transcripts (Fig. 3b), suggesting that CcInv3 gene-product was more abundant than the CcInv2 gene-product. However, we cannot rule out the possibility that there is a second gene encoding cytoplasmic invertase, as has been shown in other species.

  • The method used here (King et al., 1997) is effective for the extraction and assay of soluble acid and neutral invertases. The extraction of cell wall-bound invertases is much more difficult and requires very high salt concentrations (Husain et al., 2001). We therefore only assessed the expression of cell wall invertases CcInv1 and CcInv4 (Fig. 3b). CcInv1 transcripts were barely detectable at any development stage. CcInv4 transcripts were more abundant than CcInv1 transcripts early in development, and abundance was higher in Arabica than in Robusta. Studies on the expression of tomato LIN family sequences and orthologous genes in A. thaliana (Godt & Roitsch, 1997; Fridman & Zamir, 2003) have revealed that each gene is specifically expressed in vegetative and reproductive tissues with hardly any overlap. For example, LIN5 transcripts are exclusively expressed in the conductive tissues of the ovary, in the fruit placenta and in the pericarp during the early stages of cell division (Fridman et al., 2004). It may be that CcInv1 mRNA and, to a lesser extent, CcInv4 mRNA are of too low abundance to be detected in the grain studied here. It is probable that more sophisticated techniques such as in situ hybridization will be required to detect the tissue-specific expression patterns of the different cell wall invertases (Fridman et al., 2004).

  • Interestingly, the high sucrose-cleaving enzyme activities (SuSy, acid and neutral invertase) observed in early grain development did not result in a marked accumulation of hexoses in either Robusta genotype. Conversely, high glucose contents were detected in both Arabica genotypes at the SG stage, even though SuSy, acid and neutral invertase activities were lower than those measured in Robusta. These results suggest strongly the existence of ‘futile cycles’ of sucrose degradation and re-synthesis (Nguyen-Quoc & Foyer, 2001) in coffee grain development, as has been suggested previously (Nguyen-Quoc & Foyer, 2001; Geromel et al., 2006). Furthermore, it is possible that sucrose, glucose and fructose are translocated between different tissues in the coffee cherries (perisperm and pericarp). This would explain the observed differences between sugar contents and enzyme activities in the endosperm during early grain development.

  • • Higher CcSPS1 and CcSPS2 transcripts and SPS activities at the end of development account for the final sucrose content of the mature Arabica grains. High SPS activities are accompanied by sucrose accumulation during the maturation of Arabica and Robusta grains. Overall SPS activities were generally higher in Robusta than in Arabica at the earliest stages of grain development (Fig. 2b). However, both genotypes showed similar fluctuations in activity during grain development (Fig. 2b). SPS activity increased between the SG and LG stages and decreased drastically between the LG and Y stages. However, Arabica grains had significantly higher SPS activities than Robusta grains, particularly at the last developmental stages. This is associated with the higher final sucrose content of the mature Arabica grains. The changes in SPS activity at the end of grain maturation were much less marked in Robusta. CcSPS1 and CcSPS2 transcripts accumulated to the same abundance in both Robusta and Arabica grains at the early stages of grain development (Fig. 5). However, while the abundance of CcSPS1 and CcSPS2 mRNAs was constant in the Robusta genotypes, both CcSPS1 and CcSPS2 transcripts increased during the final stages of Arabica grain development, with CcSPS1 showing the greatest accumulation. CcSPS1 transcripts were predominant in Arabica and their abundance is consistent with the increase in SPS activity observed during grain development in the CCCA12 genotype.

A major aim of the coffee research program is to identify and map quantitative trait loci (QTL) involved in sucrose accumulation in the coffee grain, as well as those related to size and fruit development. It is probable that genomic locations of such QTLs will coincide with the candidate genes characterized in this study, or their related regulatory proteins. The results presented here demonstrate that C. canephora grains accumulate less sucrose than C. arabica for two reasons: higher SuSy and invertase activities early in grain development that prevent sucrose accumulation in Robusta; and lower SPS activities in Robusta than in Arabica later in grain development. While sucrose accumulation during coffee grain maturation is a complex process involving multiple genes and enzymatic activities, as shown in other species such as tomato (Fridman et al., 2004), maize (Thevenot et al., 2005), and sugarcane (Aitken et al., 2006), the data presented here have pinpointed targets for increasing the final sucrose content of ‘green’ grains and hence improving Robusta coffee quality.


We are grateful to Blandine Landré for the sugar content analysis. We wish to thank Maud Lepelley for the initial screening of sucrose metabolism-related genes in Cornell/Nestlé databases. We would also like to thank Cécile Hinniger for assistance in CcSPS1 cDNA isolation. Finally, we would like to thank Milton Alvarez and Samuel Van Rutte from Orecao farms (Ecuador) for supplying the coffee fruit samples used for this publication. We would like also to thank Charles Lambot for helpful discussions on sucrose content variability over C. Arabica and C. canephora cultivated varieties.