Sequences and transcriptional analysis of Coffea arabica var. Caturra and Coffea liberica plant responses to coffee berry borer Hypothenemus hampei (Coleoptera: Curculionidae: Scolytinae) attack

Abstract

The coffee berry borer (CBB) is the most prevalent pest of coffee plantations. Within the Coffea genus, C. arabica is susceptible to CBB and C. liberica shows a lower susceptibility. Two EST libraries were constructed from the total RNA of C. arabica and C. liberica fruits artificially infested with CBBs for 24 h. Using 6000 clones sequenced per library, a unigene database was generated, obtaining 3634 singletons and 1454 contigs. For each contig, the proportion of sequences present in both species was determined and a differential gene expression between the species was detected. C. arabica displayed a higher relative expression of proteins involved in general stress responses, whereas C. liberica showed the induction of a higher number of insect defense proteins. In order to validate the results, quantifications through real-time PCR were done. A hevein-like protein, an isoprene synthase, a salicylic acid carboxyl methyltransferase and a patatin-like protein gene were highly upregulated in C. liberica at 24 and/or 48 h after insect infestation compared to C. arabica. The identification of metabolic pathways induced by this pest insect provides tools to take advantage of the genetic resources available for the control of CBB.

Keywords:

• cDNA libraries
• differential expression
• ESTs
• real-time PCR
• CBB

Introduction

Coffee is one of the most important world commodities. Colombia has the highest yield of Coffea arabica in the world, with a planted area of near 850,000 ha and an average production of 680,000 ton. Hypothenemus hampei (Ferrari) (Coleoptera: Curculionidae: Scolitidae), better known as the coffee berry borer (CBB), is the most prevalent pest of coffee plantations. In Colombia, most of the economic losses due to an insect are caused by the attack of the CBB. Regarding coffee plants, very little is known about the biochemical and molecular interactions between this insect and its only natural host, Coffea spp., especially concerning its relation to the coffee berry.

Coffee berries start a vulnerability period from 120 to 150 days after flowering, and as they reach ripeness, they become more attractive to the insect because of the release of volatiles (Mathieu et al. 1996, 1998; Ortiz et al. 2004; Toci and Farah 2008). The females penetrate the berry, feed and lay their eggs. The larvae hatch and feed on the seed endosperm, decreasing its quality and often causing the detachment of small fruits, weight loss of the berries and a decrease in production (Duque et al. 1997).

The life cycle and habitat of the insect complicate its control since most of its life cycle occurs inside the fruit; because of this, it is necessary to adopt an integrated pest control strategy that includes cultural, biological and chemical control (Bustillo 2005). In addition, the climate conditions in Colombia make possible for trees to have berries all the year around, therefore supplying a constant source of food for the development of insect populations.

A variety with reduced susceptibility or moderate resistance to the CBB could be part of integrated pest management practices. Varieties with high incomplete resistance or complete resistance to CBB would mean a great competitive advantage. These varieties can alleviate the balance of expenses of the coffee growers and the negative effects that chemical control practices may have in the environment and the families who inhabit within the farm premises. Therefore, as a strategy for improving the integrated CBB control, works have been initiated to add a genetic control to this practice. To reach this goal, it is necessary to identify contrasting genotypes: a genotype that is susceptible to CBB, with agronomic characteristics of interest such as quality and production, and a genotype that is resistant to the insect.

To date, genetic factors that confer resistance have not been detected within the Coffea genus. However, preliminary studies have shown differences in the susceptibility of some species to CBB (Villagran 1991). Romero and Cortina (2004, 2007) reported a reduction of 30% in the insect's intrinsic growth rate when it fed on C. liberica fruits as compared to C. arabica var. Caturra fruits. These differences can be related to the presence of antimetabolic substances, the absence of appropriate nutrients or differences in the structure of the cell wall that directly affect the insect's digestion.

Based on these findings, differential expression studies began to characterize candidat e genes whose expression patterns correspond with the differences in the oviposition that the insects exhibited when they ate berries of C. liberica as opposed to C. arabica var. Caturra. Expression libraries of full-length cognate DNA (cDNA) copies from fruits infested with CBB for 24 h were constructed per each genotype (C. liberica and C. arabica var. Caturra) to determine transcriptional changes in the coffee berries that were induced by the attack of the insects and to compare the responses between these two genotypes under that condition. Based on the results using CBB berries 24, 48, and 72 h postinfestation, five candidate genes showing a significantly higher number of transcripts in the C. liberica library and previously reported in the literature as being involved in plant responses to insect attack were selected for expression validation by real-time PCR. Gene identification and the specific metabolic pathways of plant responses to insects in which they are involved complement and support coffee breeding efforts against the CBB.

Methodology

Biological material for the construction of cDNA libraries

Plants of the genotypes C. arabica var. Caturra and C. liberica (accession CCC1025) were selected from the Coffee Germplasm Bank at Naranjal, the Cenicafé Central Experimental Station at Chinchiná, Caldas, Colombia. Branches with healthy berries aged to approximately 150 days after flowering (DAF) were chosen from C. arabica and from 180 DAF for C. liberica because it takes longer time for this genotype to develop and reach the maturation stage (Medina et al. 1984; Eira et al. 2006).

An artificial infestation with CBBs in a proportion of three insects per berry was carried out. The infested branches were covered with entomological sleeves for 24 h. The berries penetrated by the CBB were collected and immediately stored in liquid nitrogen until their laboratory processing.

RNA isolation

Total RNA was extracted from each genotype using a kit for RNA isolation from plants (RNeasy^® Plant Mini Kit; Qiagen, Chastworth, CA, USA). Buffers RLT and RLC, provided by the manufacturer, were used for C. arabica var. Caturra and C. liberica, respectively. The RNA integrity was assessed by denaturing agarose gels and quantified by fluorometry.

Real-time PCR

Total RNA (250–750 ng) was reverse transcribed to its cDNA using MMLV reverse transcriptase (Promega, Madison, WI, USA) for one hour at 42°C. The primers used were CDS III/3′ and SMART IV oligo; the former recognizes the poly(A) tail, and the latter covers the region 5′ of the transcript. The CDS III/3′ and 5′ primers were used to synthesize the second chain. The amplification conditions were the following: 95°C for 1 min, followed by cycles of 95°C for 15 s and 68°C for 6 min. For C. arabica, 22 cycles were performed; and for C. liberica, 20 cycles.

Construction of cDNA libraries

The construction of cDNA libraries was carried out following the manufacturer's instructions of the Creator SMART cDNA Library Construction Kit (Clontech; Palo Alto, CA, USA). After the synthesis of the double-stranded cDNA, it was digested with the enzyme Sfi I. The digestion product was fractioned according to size using Chroma Spin 400 columns (Becton Dickinson, Franklin Lakes, NJ, USA). The fractions that exhibited the largest sizes were selected and inserted in the transformation vector pDNR-Lib. Finally, Escherichia coli DH 10B cells were transformed with this construct by electroporation. The titer recombination percentage and the number of independent clones of the genomic library were determined.

cDNA sequencing

Sequencing of the cDNA libraries was performed in two phases by two different organizations. In the first phase, 3072 sequences per genotype (Group 1) were obtained from REXAGEN, Seattle, WA, USA. In the second phase, 2500 sequence reactions per genotype (Group 2) were carried out by MACROGEN, Korea. All of the sequencing reactions were initiated from the extreme 5′-end, using the universal M13F primer.

Sequence analysis

Sequence quality was assessed using the program Codon Code Aligner (version 3.5, 2009) and with a Phred quality value of 30 (Ewing et al. 1998). Vector and contaminant sequences were eliminated, as well as sequences smaller than 200 bp. The number of reads corresponding to full-length sequences or to partial sequences was determined through the program Target Identifier (Min et al. 2005). The sequences were assembled with CAP3 (Huang and Madem 1999) using the information of both genotypes to identify shared sequences. Unigenes that corresponded to grouped sequences (contigs) and to nongrouped ones (singletons) were generated. The proportion of sequences present from both genotypes was determined for each contig. Protein translation was conducted using the program ESTScan (version 2.0) (Iseli et al. 1999). The putative function of the unigenes was determined through comparisons among the protein databases available in InterProScan (version 4.0) (Zdobnov and Apweiler 2001). BLASTn and BLASTx comparisons were also made (Altschul et al. 1990).

Statistical analysis of the cDNA libraries

Once the representative categories were determined, a comparison of the relative abundance of sequences in the cDNA libraries was carried out between orthologs of both genotypes, including sequences with and without known function, using the R statistic proposed by Stekel et al. (2000) and defined as R= − 2 log(L), where L is the ratio of the maximized likelihoods. For each gene a p value was calculated for the R statistic using the IDEG6 web tool (Romualdi et al. 2003).

Real-time PCR assays

In order to validate the expression of some of the genes identified in the cDNA libraries, new coffee artificial infestations with CBBs in a proportion of three insects per fruit were carried out in the same plant material used for the cDNA libraries. For this, three adult trees planted in the field with berries in the same developmental stages that were used for the libraries were selected. In each tree, two branches were selected and they were covered with entomological sleeves: in one, CBBs were added in a 3:1 insect–berry relation, and in the other branch no insects were added (control treatment). In the first tree, the berries from the infested and noninfested branches were collected after 24 h. In the other trees, the berries were collected 48 and 72 h postinfestation. After collection, berries were stored in liquid nitrogen. The setup of this experiment was done in October of 2009.

The frozen berries were ground in a mortar. The powder obtained from two berries (1 g) was used. Total RNA extraction was performed using the RNeasy® Plant Mini Kit (Qiagen, Valencia, CA, USA). Samples were treated with RNase-free DNase I (Promega, Madison, WI, USA). Finally, RNA integrity and purity were evaluated using agarose gel electrophoresis and a TBS-380 fluorometer (Turner Biosystems, Sunnyvale, CA, USA).

The first-strand cDNA synthesis was performed with oligo(dT) and a Sensiscript kit (Qiagen, Valencia, CA, USA) using 50 ng of the total RNA as a template in a 25 µL reaction mixture. The cDNA samples were then diluted (1:25), and 2 µL of the dilution was used in a SYBR Green PCR. Gene-specific primers were designed (Primer3plus; Untergasser et al. 2007) using a stringent set of criteria: predicted melting temperature of 61–62°C, primer length of 19–23 nucleotides, guanine–cytosine contents of 40%–60% and PCR amplicon lengths of 70–150 bp.

The genes chosen for real-time PCR and the primers used for the amplification are shown in Table 1.

Table 1. List of primers used in real-time PCR.

Putative name	Clone	Sequence (forward and reverse)
Hevein-like protein	CEN72529	5′TGCATGTGGGAGGTGCCTAA3′ 5′TCTAGCCCATTGCTGCAT3′
Isoprene synthase	CEN71857	5′TCCTCGACGGCGGAGATAGA3′ 5′TCGGGCAAACTCCTCAGACA3′
Trypsin inhibitor Kunitz	CEN73476	5′CACGAACCACGCCTTCTTCC3′ 5′TGCCCTCTGGCAGTTGTTCA3′
S-Adenosyl-L-methionine:salicylic acid carboxyl methyltransferase	CEN73008	5′TGGTAGCAAGGGGTATGGCG3′ 5′TGGACCTGGAAAACCACCCC3′
Patatin-like protein	CEN73488	5′CGACTTCGCCTCTGTCGGTT3′ 5′GTCCCCACTTGGCAACATCCT3′

PCR was carried out in an optical 96-well plate with chromo 4 (MJ Research, Waltham, MA, USA) sequence detection. The following is the standard thermal profile for all PCRs: 95°C for 10 min, 40 cycles at 95°C for 15 s, 60°C for 30 s, 72°C for 30 s and 72°C for 10 min.

The fluorescence signal was captured at the end of each cycle, and melting curve analysis was performed from 65 to 95°C, with data captured every 0.2°C during a 1-s hold. Data were analyzed using the Opticon monitor analysis software version 2.02 (MJ Research, Waltham, MA, USA). Quantification of each transcript in each cDNA source consisted of at least three independent technical replicates. The housekeeping gene for cyclophylin (Romano et al. 2004) was selected for reference expression.

Transcript abundance was normalized to cyclophilin by subtracting the Ct value of cyclophilin from the Ct value of the transcript, that is Ct = Ct_transcript−Ct_cyclophilin. The efficiency was estimated for each gene using absolute fluorescence data captured during the exponential phase of amplification of each reaction using the equation (1 + E) = 10^(−1/slope) (Ramarkers et al. 2003). Efficiency values were taken into account in all subsequent calculations. The transcript abundance in control and treated samples was obtained from the equation (1 + E)^(−Ct).

Results

Libraries analysis

Approximately 6000 clones per genotype were sequenced (6048 and 5952 from C. arabica and C. liberica, respectively) in two sequencing groups. Table 2 shows and summarizes the distribution of the clones in each sequencing round and each genotype. Using this data set, 10,603 sequences were assembled, representing 88±5% of the total. The discovery rate of genes is related to the degree of redundancy of the clones obtained per genomic library (Susko and Roger 2004), reaching, in this case, a value of 63% for both genotypes and an average length of 534 bp per sequence.

Table 2. General information regarding the C. arabica var. Caturra and C. liberica clones obtained from the sequencing of the full-length cDNA libraries.

	Genotype
	C. arabica var. Caturra		C. liberica
	Group 1^a	Group 2^b	Group 1	Group 2	Total
Sequenced clones	3072	2976	3072	2880	12,000
Good quality sequences	2843	2562	2572	2626	10,603
% Good quality sequences	92.3	86.1	83.7	91.2	N/A
Rate of gene discovery	63	56	58.7	63.5	N/A
Average size of inserts	572	526	482	556	534
Notes: ^aSequencing in REXAGEN, Seattle, WA, USA.
^bSequencing in Macrogen, Korea.

After quality examination, the percentage of full-length sequences for both genotypes was 52% and 31%, which are partial-length sequences exhibiting an open reading frame (Table 3).

Table 3. Full-length sequence description obtained from the cDNA libraries of C. arabica var. Caturra and C. liberica.

Target identifier report	C. arabica	C. liberica	Total
Total number of sequences	5405	5198	10,603
Full-length sequences	2506	2382	4888
Short full-length sequences	307	320	627
Ambiguous sequences	351	319	670
Partial sequences
5′-partial sequence	946	901	1847
3′-partial sequence	151	100	251
Full-length sequences and open reading frame	1815	1514	3329

The assembling with CAP3 was performed with all of the sequences from both genotypes to find redundancies among them and differences in the number of clones expressed per gene or protein (Table 4). From the 5088 unigenes obtained from this assemblage, 1754 corresponded to contigs and 3634 to singletons, with an average size of 654 bp per unigene. The redundancy obtained from the assembly was 4.8 ESTs/contig. Forty percent of the unigenes (contigs and singletons) are unique sequences of C. arabica var. Caturra, 41% corresponded to unique sequences of C. liberica and 19% were contigs with sequences shared between both genotypes.

Table 4. Good-quality sequence assemblage corresponding to both genotypes.

	Quantity	Average size (bp)	Maximum length	Redundance (EST/contig)
Contigs	1754	749	4246	4.8
Singletons	3634	558	880	NA
Total	5088	654	NA
Note: NA, not available.

Sequence translation into proteins using the ESTScan program was achieved for 75% of the unigenes, meaning that open reading frame candidates for these sequences were found. Once the amino acid sequence for each unigene was obtained, putative functions were assigned by making comparisons among the databases available in InterProScan. The information was complemented through BLASTx and BLASTn comparisons with the nonredundant GenBank databases. Figure 1 shows the unigene distribution obtained through this process. Of the total unigenes analyzed, 7% revealed a functional prediction (amino acids prediction), but they did not exhibit homology with the proteins reported in InterProScan and GenBank. Thirteen percent corresponded to hypothetical proteins that do not have previous experimental evidence for their assigned function (Eisenstein et al. 2000). Two percent of the sequences corresponded to ribosomal proteins, and 53% of the unigenes exhibited homology with a reported protein. Among the unigenes with known functions, several were reported with the same function. Therefore, groupings were carried out according to these annotations.

Figure 1.  Distribution of the unigene annotations from the C. arabica var. Caturra and C. liberica assemblage.

Once the unigenes were annotated, the protein assignments were classified on the basis of Gene Ontology (The Gene Ontology Consortium 2000) information. Table 5 contains the categories with the highest representation in the annotated unigenes.

Table 5. Representative categories related to the annotation of the assembled unigenes for C. arabica var. Caturra and C. liberica.

		Number of clones
Categories	Number of unigenes	C. arabica var. Caturra	C. liberica
Response to stress	519	1106	1070
Other metabolisms	445	354	357
Proteins synthesis	124	132	130
Transport proteins	89	70	57
Nucleases	85	93	56
Membrane proteins	83	81	77
Photosynthesis	78	83	81
Transcription factors	70	74	66
Linking proteins	70	182	155
Related to seed	68	120	135
Metabolism sugars	45	32	25
Genomic DNA	44	61	50
Lipids metabolism	39	126	116
Domains	38	30	21
Secondary metabolites	22	19	28
Wall proteins	15	16	20
Respiratory chain factors	2	3	1

Three expression groups were obtained after the statistical analysis of the cDNA libraries defined by p values less than or equal to 0.1: orthologs with similar expression patterns in both genotypes C. liberica and C. arabica var. Caturra (Table 6); proteins that are more abundantly expressed in C. arabica var. Caturra as compared to C. liberica (Table 7); and proteins with a higher expression in C. liberica (Table 8).

Table 6. Proteins from C. arabica var. Caturra and C. liberica with similar expression patterns (R p-value > 0.1).

	Number of sequences
Proteomic function	C. arabica	C. liberica	Description	References
Translation initiation factors	11	11	Growth and development	Thompson et al. (2004)
Profilin	6	6	Cellular elongation	Ramachandran et al. (2000)
Peroxisomal oxidase	4	4	Related to senescence	Schrader and Fahimi (2006)
Cell division	3	3	Growth and development	Dewitte and Murray (2003)
Pectin methylesterase inhibitor	2	2	Cell wall; physiology of the plant	Giovane et al. (2004)
Cellulose biosynthesis	2	2	Cell wall biogenesis	Richmond and Somerville (2000)
Ripening-related	14	15	Ripening-related	Moore et al. (2005)
Microtubules	12	11	Cellular elongation	Rajangam et al. (2008)
ATPase	12	11	Diverse cellular functions	Williams and Mills (2005)
Lignin biosynthesis	7	6	Cell wall	Ehlting et al. (2005)
Phenylpropanoid metabolism	6	5	Response to induced environmental stimuli and induced by wound	Bernards and Båstrup-Spohr (2008)
Shikimic acid pathway	6	5	Production of phenolic components	Bernards and Båstrup-Spohr (2008)
Lipid transfer protein	96	83	Related to cell wall, response to pathogens and chitin; seed production	Boutrot et al. (2005); Salcedo et al. (2007)
Miraculin-like protein	34	29	Related to the response to mechanical damage and change of flavor in the fruits	Tsukuda et al. (2006)
OLE-3	14	11	Oil storage in seeds	Simkin et al. (2007)
Alpha-galactosidase-like protein	12	9	Endosperm protein	Feurtado et al. (2001)
Glutaredoxin-related protein	12	8	Diverse metabolic functions; oxidative stress	Rouhier et al. (2008)
MADS-box protein	12	6	Flowering-related	Favaro et al. (2003)
Dirigent protein	7	3	Production of induced or constitutive lignin	Ralph et al. (2006)
Hypersensitive-induced response protein	6	2	Response to biotic factors	Jung et al. (2008)
Fructose-bisphosphate aldolase-like	6	3	Response to biotic and abiotic stresses	Taji et al. (2004)
Caffeoyl-CoA O-methyltransferase	6	3	Lignin production	Burlat et al. (2001)
Peroxidase	6	13	Multiple functions	Passardi et al. (2005)
S-Adenosyl methionine synthetase 3	2	6	Metabolic pathway	Lin et al. (2005)
Heat shock protein 90	1	4	Response to external stimuli; plant disease resistance	Kansaki et al. (2003); Sangster and Queitsch (2005)
Ethylene-responsive transcription factor	1	4	Fruit development and defence	Bustamante-Porras et al. (2005, 2007); Pereira et al. (2005); Dinesh et al. (2010)
Avr9/Cf-9 rapidly elicited protein	3	7	Induced by pathogens	Coram and Pang (2005)
Nonspecific lipid transfer protein	5	10	Storage	Marion et al. (2007)
N-Methyltransferase	4	7	Caffeine and jasmonic acid	McCarthy and McCarthy (2007)
NtEIG-E80 protein	8	11	Induced by pathogens	Takemoto et al. (2003)
Catalase	21	25	Hydrogen peroxide; response to pathogens	Thaler et al. (1999)
2S albumin	46	54	Storage structural; insect resistance	Chen et al. (2004); Salmona et al. (2008)
Senescence-associated nodulin 1A	1	3	Senescence and defense	Birch et al. (1999)
Luminal-binding protein/HSP70	1	3	Defense	Kansaki et al. (2003)
Polyphenol oxidase II	2	5	Defense	Thaler et al. (1999)

Table 7. Proteins with higher expression in C. arabica with respect to C. liberica (R p-value < 0.1).

	Number of sequences
Protein function	C. arabica	C. liberica	Description	Reference
Metallothionein-like protein	134	115	Response to different types of stresses	Moyle et al. (2005); De Nardi et al. (2006)
Early light-induced protein	56	38	Bind chlorophyll and photosynthesis	Hutin et al. (2003)
Glutathione S-transferase	21	11	Chemical defense	Edwards et al. (2000)
Auxin-repressed protein (ARP1)	11	5	Related to cellular division, elongation, differentiation; secondary root and vascular tissue formation	Kim et al. (2007)
Embryo-specific protein 3	10	4	Seed development	Girke et al. (2000)
F-box family protein	9	3	Auxin receptor	Dharmasiri et al. (2005)
Cysteine protease	8	1	Defense	Van der Hoorn and Jones (2004)
Cupin family protein	6	1	Nutrient reservoir, storage protein; insect defense	Dunwell et al. (2000); Liu et al. (2005)
Mannose/glucose-specific lectin	5	1	Carbohydrate metabolism; seed production	Wong et al. (2008); Kuku et al. (2009)

Table 8. Proteins with higher expression in C. liberica with respect to C. arabica (R p-value < 0.1).

	Number of sequences
Protein function	C. arabica	C. liberica	Observation	Reference
Hevein-like protein	0	8	Induced by insect	Reymond et al. (2000); Falco et al. (2001)
Isoprene synthase	0	4	Terpene synthases; induced by insects, repellent	Ferrieri et al. (2005); Laothawornkitkul et al. (2008a, 2008b)
Cytokinin-repressed gene (CR9)	1	8	Cytokinin	Alfenas-Zerbini et al. (2009)
Annexin-like protein	1	7	Tolerance to biotic and abiotic stresses	Jami et al. (2008)
Cytochrome B5	2	12	Flowering and fruit development	Fan et al. (2009)
Succinate dehydrogenase (ubiquinone)	0	4	Respiration	Millar et al. (2004)
Trypsin inhibitor Kunitz	0	7	Induced by insect defense	Mondego et al. (2005, 2011)
MYB transcription factor	4	13	Transcription regulation	Lee et al. (2001)
Arachidonic acid-induced DEA1	0	3	Fruits development	Weyman et al. (2006)
Tyrosine aminotransferase	0	3	Induction by herbivores and pathogens	Lopukhina et al. (2001)
E-class P450, group I CYP71E1	0	3	Defense	Schuler and Werck-Reichhart (2003); Ganjewala et al. (2010)
Mary storys protein (eukaryotic initiation factor 5A2)	3	9	Storage	Feng et al. (2007); Ma et al. (2010)
S-Adenosyl-l-methionine:salicylic acid carboxyl methyltransferase	0	2		Thaler et al. (1999); Moffatt and Weretilnyk (2001)
Patatin-like protein	6	14	Storage and defense protein	Alidhai and Rydel (2010)
Catechol O-methyltransferase	0	2	Response to external stimuli; antimicrobial alkaloid	Lam et al. (2007)
Trans-Cinnamate 4-monooxygenase CYP73A1	0	2	Cytochrome P450 superfamily; defense	Batard et al. (1997)

Real-time PCR

For the real-time PCR validation, the genes that corresponded to an insect defense function annotation were chosen. The real-time PCR amplification of the sequence CEN72529, annotated as a hevein-like protein, in each genotype and each infection treatment is shown in Figure 2. For this sequence, there is an early overexpression in C. liberica after 24 h of the insect attack. CBB induces gene expression almost four times higher than the control noninfested berries, and the overexpression continues and increases at 48 h. The gene repression is observed only at 72 h postinfestation. On the contrary, in C. arabica, the level of gene expression is low when compared to the controls and there is no difference in the expression among the infestation times tested.

Figure 2.  Relative expression of hevein-like protein gene CEN72529 by quantitative real-time PCR. Quantitative real-time PCR was performed with cDNA from green coffee berries at 24, 48, and 72 h after infestation of CBB in C. liberica and C. arabica.

In the case of the sequence CEN71857, annotated as a putative isoprene synthase gene (Figure 3), C. liberica showed a high overexpression at 24 h after infestation, with almost 20 times overproduction as compared to the control. This induction decreases at 48 h, and at 72 h, the levels are similar to the controls. In C. arabica, no overexpression of these genes is observed at any time.

Figure 3.  Relative expression of isoprene synthase gene CEN71857 by quantitative real-time PCR. Quantitative real-time PCR was performed with cDNA from green coffee berries at 24, 48 and 72 h after infestation of CBB in C. liberica and C. arabica.

The sequence CEN73476, annotated as a putative trypsin inhibitor Kunitz gene (Figure 4), has a relatively low expression compared to the other genes evaluated. In C. liberica, a low expression is observed at 24 h after infestation, which increases twice at 48 h and goes down again at 72 h. In C. arabica, the expression of this gene is similar at 24 and 48 h and increases at 72 h.

Figure 4.  Relative expression of trypsin inhibitor Kunitz gene CEN73476 by quantitative real-time PCR. Quantitative real-time PCR was performed with cDNA from green coffee berries at 24, 48, and 72 h after infestation of CBB in C. liberica and C. arabica.

The validation of the sequence CEN73008 (Figure 5), a putative S-adenosyl-l-methionine:salicylic acid carboxyl methyltransferase (SAMT), showed an overexpression of almost seven times in C. liberica at 24 h after infestation, with the levels decreasing at 48 h and induced again three times at 72 h. In C. arabica at 24 h the overproduction is lower than in C. liberica, but at 48 and 72 h the levels are higher than those in C. liberica at the same times. However, the overall highest expression is observed in C. liberica at 24 h.

The expression pattern of sequence CEN73488 (Figure 6), annotated as a patatin-like protein, is shown as a fivefold overexpression in C. liberica at 24 h as compared to the controls. After 48 h, the overproduction increased 30 times and the gene was then repressed at 72 h. It is the gene that overexpresses the most of all evaluated. On the contrary, C. arabica showed very low levels of expression at 24 and 48 h as compared to the controls, and at 72 h the expression is even lower.

Figure 5.  Relative expression of S-adenosyl-L-methionine:salicylic acid carboxyl methyltransferase gene CEN73008 by quantitative real-time PCR. Quantitative real-time PCR was performed with cDNA from green coffee berries at 24, 48, and 72 h after infestation of CBB in C. liberica and C. arabica.

Figure 6.  Relative expression of patatin-like protein gene CEN73488 by quantitative real-time PCR. Quantitative real-time PCR was performed with cDNA from green coffee berries at 24, 48 and 72 h after infestation of CBB in C. liberica and C. arabica.

Discussion

The Coffea genus is becoming the subject of increasing genomic research because of the importance of the crop around the world. Around 10,000 sequences coming from coffee seeds in an active transcriptional stage from two different genotypes – C. arabica and C. liberica – and induced by infestation with the CBB were obtained in this work. Previously, Salmona et al. (2008) studied the transcriptional expression of 266 genes involved in seed maturation, but this is the first study involving gene expression of a coffee seed under an insect attack.

In general, most of the previous work focused on the plant response to phloem-feeding insects such as aphids and whiteflies (Voelckel et al. 2004; Thompson and Goggin 2006) or the response of plant leaves to chewing lepidoptera insects (see review by Wu and Baldwin 2010), but only a few studies focused in the response of plant seeds to a coleoptera.

For this study, coffee berries with 150 DAF from C. arabica and with 180 DAF from C. liberica were chosen. Those fruits are in an early active developmental stage, as defined by De Castro and Marracini (2006). In this stage, genes that are involved not only in seed development but also in response to early insect infection can be identified. Previous research showed that during the first 24 h of infestation, the CBBs are found in position B, which means penetrating the pericarp (Bustillo 2006). De Castro and Maraccinni (2006) summarized that the pericarp is formed by the exocarp (peel) and the mesocarp, commonly referred to as the ‘true pulp’, rich in sugar and formed from the parenchyma cells with compact and dense cell walls in green fruits and the endocarp (also called parchment layer or ‘pergaminho’) that is a hard and lignified tissue. During this first 24 h of infection, the insects identify the berries and start a chewing of the tissue that forms the pericarp, penetrating the exocarp and the mesocarp without arriving to the endocarp. The oral secretions of the CBB that have been in contact with the tissue can act as elicitors that stimulate all of the major defense pathways, in a situation similar to the one reported in other insects (Felton and Korth 2000; Kahl et al. 2000). In addition to the chewing process, there is a mechanical stimulus caused by the insect in its effort to penetrate the pericarp that could be sensed by the cells of the pericarp in a fashion similar to the one reported in other crop plant tissues (Reymond et al. 2000, 2004). Here, a response in the coffee berry is induced because of the insect attack.

Table 5 shows that a high number of sequences are involved in plant responses to stress from the 1836 total number of unigenes belonging to a specific functional category; almost 30% of the sequences, the highest number found, correspond to stress response. This suggests that the insect is capable of trigging an intense response, engaging more that 2000 clones from both genotypes in C. arabica (1106) and C. liberica (1070) related to stress. It is important, however, to point out that the berries are coming from plants located in the field and, in addition, to be exposed to the insect attack they may been sensing other biotic and abiotic stress stimuli. The second category in importance corresponds to the sequences related to general metabolism; this category accounts for 25% of the sequences. The other groups are related to seed development and sugar biosynthesis, which are abundant; this feature is expected to take into account that the coffee berries observed were still under development. A similar representation from this category was obtained in both genotypes. There is an additional group related to photosynthesis. This group confirms the active photosynthesis process produced in the fruits already reported by Geromel et al. (2006) and Salmona et al. (2008).

Our interest was to identify the differences between the sequences coming from each g enotype that can account for the differences in susceptibility against the CBB. On the basis of this, the proteins were grouped into three groups: (1) proteins with similar levels of expression on both genotypes (p>1) (Table 6), (2) proteins overexpressed in C. arabica (p<1) (Table 7) and (3) proteins overexpressed in C. liberica (p<1) (Table 8).

The first group (Table 6) includes proteins related to cellular maintenance, those involved in cellular elongation, fruit growth and development, as well as in fruit ripening. The results observed in this sequence group allowed comparisons between the assessed coffee species because the proteins related to cellular maintenance were expected to show a similar expression in both genotypes, as was observed.

Among them were found seed storage proteins highly expressed, such as albumins and lipid transfer proteins (Marion et al. 2007; Salmona et al. 2008), and NtEIG-E80, a protein-elicitor-inducible gene described in Nicotiana tabacum (Takemoto et al. 2003) whose function is photoassimilate-responsive and an eukaryotic initiation factor eIF5A-2 that supports growth (Feng et al. 2007) but also plays a regulatory role in the response of plants to sublethal osmotic and nutrient stresses (Ma et al. 2010), as well as lipid transferases, oleosins OLE3 protein and alpha-galactosidases genes involved in upstream fatty acid metabolism. Some of those proteins were reported in C. arabica seed developmental stage by Salmona et al. (2008).

Similarly, flower and seed developmental proteins were found: MADS-box protein (Favaro et al. 2003) as well as a catalase, peroxidase and polyphenol oxidase, proteins related to ethylene production (Constabel et al. 1995; Thaler et al. 1999; Ward et al. 2001; Montavo et al. 2007) that participate in fruit maturation. A ‘Senescence-associated Nodulin 1A’ (Birch et al. 1999), two types of SAM-dependent N-methyltransferases superfamily enzymes, which are involved in diverse metabolic activities (Moffatt and Weretilnyk 2001), and an S-adenosyl methionine synthetase 3 SAMS3, associated with methionine regulation (Goto et al. 2002). Ethylene-responsive transcription factor implicated in seed development (Bustamante-Porras et al. 2005, 2007; Pereira et al. 2005), but also, able to induce the chitinase activity triggering a defense response against pathogens (Dinesh et al. 2010).

The overexpression of proteins related to photosynthesis and photooxidative stress was also observed with the identification of aldolases (Taji el al. 2004) and redox enzyme glutaredoxin, which additionally participate in flower development and salicylic acid signaling (Rouhier et al. 2008).

A group of proteins is involved in plant responses to biotic and abiotic stresses, such as miraculin-type proteins, generally reported in injury or pathogen responses (Tsukuda et al. 2006); a hypersensitive response protein induced by diseases and osmotic stress (Jung et al. 2008); several pathogenesis-related proteins (Jami et al. 2008); two types of heat shock proteins: HSP90 and HSP70, which frequently function together in pathogenic defense signal transduction pathway (Kansaki et al. 2003; Sangster and Queitsch 2005); and dirigent protein and caffeoyl-CoA O-methyltransferase, which are involved in the induction of lignin for plant protection (Burlat et al. 2001; Ralph et al. 2006) and other genes reported to be induced by pathogens such as an Avr9/Cf-9 rapidly elicited protein (ACRE gene) that is immediately activated upon perception of the pathogen (Rowland et al. 2005). The induction of those genes suggests that both genotypes are able to sense the stress produced by the insects.

However, differential expression between the genotypes was also observed in response to the biotic and abiotic stresses present at the moment of collecting the fruits.

Table 7 includes proteins that showed higher expression in C. arabica var. Caturra. Here are found proteins such as cysteine proteases involved in a large variety of function – plant growth and development – but also in senescence and programmed cell death, in accumulation of storage proteins in seeds, as well as in storage protein mobilization and in pathways signaling in response to biotic and abiotic stresses (Van der Hoorn and Jones 2004). Also, there were metallothioneins, cysteine-rich proteins that bind to heavy metals. The metallothioneins are upregulated during the development of fruits and they are a systemic response for abiotic stress (Moyle et al. 2005; De Nardi et al. 2006). Also, an embryo-specific protein 3 that participates in seed development (Girke et al. 2000), an auxin receptor F box protein that regulates auxin response, implicated in every aspect of plant growth and development (Dharmasiri et al. 2005), and a cupin family protein, which includes many of the storage proteins from higher plants and nutrient reservoir activity (Dunwell et al. 2000), were found.

Also noted is the overexpression of proteins related to photosynthesis and photooxidative stress with the induction of an early light-induced protein (Hutin et al. 2003).

In addition, a group of proteins involved in plant responses to biotic and abiotic stresses were observed to be upregulated; in this group, a glutathione S-transferase complex regulated for protective functions by environmental stimuli (Edwards et al. 2000), an auxin-repressed protein (ARP19) that allows salicylic acid inhibiting the pathogen growth in plants (Kim et al. 2007; Wang et al. 2007) and a mannose/glucose-specific lectin with antifungal activity in seeds (Kuku et al. 2009) were found.

In general, it seems that C. arabica responds to the CBB attack with a plethora of genes involved in general biotic and abiotic stress responses and they are not very specific to insect defense; the genes induced are more related to plant–pathogen interactions.

Unigenes with sequences corresponding only to C. arabica var. Caturra (data not shown) were identified; however, proteins of interest related to defense were not found in this group.

Regarding the group with the highest expression in C. liberica (Table 8), this protein set contained in general more sequences related to biotic and abiotic stress responses.

Among the proteins that the literature reports to respond to general metabolism were found: a CR9 protein involved in plant growth and development by regulating cell cycle (Alfenas-Zerbini et al. 2009), a succinate dehydrogenase related to mitochondrial respiration (Millar et al. 2004) and cytochrome B5 implicated in sugar transportation interviewing with flowering and fruit development (Fan et al. 2009).

Genes overexpressed by the presence of biotic and abiotic general stresses were also found in this group, such as an annexin related to drought, salt and oxidative stress responses (Huh et al. 2010) and also to the increasing message levels for several pathogenesis-related proteins (Jami et al 2008); a DEA1 protein induced by arachidonic acid that has a protective function against abiotic and biotic stresses including weather change, pathogens and pests (Weyman et al. 2006); a catechol O-methyltransferase that plays a key role in lignin biosynthesis and also in disease resistance (Lam et al. 2007); two types of cytochrome P450: the CYP73A1 and the CYP71E1, that responses to different environmental conditions, wounding and it is involved in plant–insect interactions (Batard et al. 1997; Ganjewala et al. 2010).

The genes induced by pathogen s such as a tyrosine aminotransferase involved in the plant defense response to pathogens attack (Lopukhina et al. 2001), and MYB transcriptor factor inducible within 1 day after fungal infection (Lee et al. 2001) were also identified.

In this group, there were proteins more specifically related to herbivory and insect attack response (i.e. genes related to the jasmonic acid pathway), such as a hevein-like protein (Reymond et al. 2000), isoprene synthetase (Loivamäki et al. 2008), a trypsin inhibitor Kunitz gene (Mondego et al. 2005, 2011), an S-adenosyl-l-methionine:salicylic acid carboxyl methyltransferase (Köllner et al. 2010) and patatin-like protein (Alidhai and Rydel 2010) that has been reported as specifically induced by insects. These genes induced or upregulated by insects that show overexpression in C. liberica, the genotype with lower susceptibility to CBB attack when compared to C. arabica var. Caturra, may explain the negative effect that this genotype causes on the CBB. Because of that, the expression of them was validated by real-time PCR.

The real-time PCR results showed three genes with very high expression in C. liberica at 24 or 48 h and low or none expression in C. arabica. The gene hevein-like protein was upregulated at 24 and 48 h after insect infestation (Figure 2), results similar to the ones obtained by comparing the number of transcripts in the libraries (Table 8). This gene is involved in the salicylic acid pathway (Medeiros et al. 2010); it has antifungal and antibacterial activities, but also it is induced by methyl jasmonate treatment in plants (Kiba et al. 2003) and it can be induced by insect herbivory (Reymond et al. 2000; Falco et al. 2001). Similar results were obtained with the gene that encodes the isoprene synthase enzyme, which was upregulated almost 20 times over at 24 h. The isoprene synthase is involved in the production of isoprene terpene by the plants. The isoprene is produced by some plants to protect its photosynthetic apparatus from the high temperatures and oxidative stress (Ferrieri et al. 2005), but also, recently, the isoprene has been reported as an insect repellent (Laothawornkitkul et al. 2008a, b; Loivamäki et al. 2008). The upregulation of isoprene and isoprene emission has been reported before in response to insect herbivory and jasmonic acid treatment (Ferrieri et al. 2005).

The other gene that was highly upregulated in C. liberica was the patatin-like protein, which showed between 5 and 30 times overproduction during the first 48 h of infestation; this result also corresponds to the ones obtained in the libraries. The patatin gene encodes a lipid acyl hydrolase protein that affects lipids metabolism (Renier et al. 2004) and inhibits the growth of coleoptera larvae (Strickland et al. 1995). When applied in appropriate levels in artificial diets, potato patatin is lethal to spotted cucumber beetle Diabrotica undecimpunctata larvae and will stunt the growth of survivors; so, maturation is prevented or severely delayed, resulting in no reproduction (Alidhai and Rydel 2010). So far this gene has not been described in coffee, which is an important founding, because it can cause similar effects in the CBB larvae. These tree genes are good candidates to be overexpressed in C. arabica using constitutive promoter and genetic transformation or transient expression experiments in order to confirm their function as related to CBB coffee defenses.

There were another group of genes that showed upregulation in C. liberica and higher overexpression in this genotype as compared to C. arabica, but in C. arabica, induction of those genes was also observed, which is the case of S-adenosyl-l-methionine:salicylic acid carboxyl methyltransferase and the trypsin inhibitor Kunitz gene.

SAMT is an enzyme that converts salicylic acid to methyl salicylate, an important volatile emitted from plant herbivore damage, detectable after 2 h of insect feeding (Köllner et al. 2010). In coffee berries, SAMT showed high induction in C. liberica at 24 h as compared to C. arabica, results similar to the ones obtained with the libraries; however, the gene was also upregulated in C. arabica at 48 and 72 h postinfestation. C. liberica upregulated this gene earlier than C. arabica.

The trypsin inhibitor Kunitz gene showed higher upregulation in C. liberica at 48 h as compared to C. arabica at a similar time. However, this gene was also upregulated in C. arabica. The trypsin inhibitor Kunitz gene is already reported in coffee by Mondego et al. (2005, 2011); this is the same miraculin-like gene, CoMir. Structural protein modeling indicated that CoMir had structural similarities with the Kunitz STI proteins but suggested specific folding structures (Mondego et al. 2011). The gene is also upregulated after coffee leaf miner (Leucoptera coffella) oviposition in resistant plants of a progeny derived from crosses between C. racemosa (resistant) and C. arabica (susceptible) and downregulated during coffee leaf miner oviposition and larvae eclosion (herbivoria) in susceptible plants. In this experiment, C. liberica behaved in a way similar to that for the resistant genotype.

Finally, the genes that have been identified in this study increase our understanding of functional genomics. The genes overexpressed in the C. liberica genotype in response to the CBB attack can be turned into new tools to widen the genetic basis of the cultivable varieties in the effort to improve their resistance to pests and diseases as a complementary alternative for the integrated pest management strategy of the CBB control.

Acknowledgements

This research was cofinanced by the Scientific and Technical Cooperation Agreement between the Ministry of Agriculture and Rural Development and the National Federation of Coffee Growers of Colombia.