Figures
Abstract
Plant genomes are massively invaded by transposable elements (TEs), many of which are located near host genes and can thus impact gene expression. In flowering plants, TE expression can be activated (de-repressed) under certain stressful conditions, both biotic and abiotic, as well as by genome stress caused by hybridization. In this study, we examined the effects of these stress agents on TE expression in two diploid species of coffee, Coffea canephora and C. eugenioides, and their allotetraploid hybrid C. arabica. We also explored the relationship of TE repression mechanisms to host gene regulation via the effects of exonized TE sequences. Similar to what has been seen for other plants, overall TE expression levels are low in Coffea plant cultivars, consistent with the existence of effective TE repression mechanisms. TE expression patterns are highly dynamic across the species and conditions assayed here are unrelated to their classification at the level of TE class or family. In contrast to previous results, cell culture conditions per se do not lead to the de-repression of TE expression in C. arabica. Results obtained here indicate that differing plant drought stress levels relate strongly to TE repression mechanisms. TEs tend to be expressed at significantly higher levels in non-irrigated samples for the drought tolerant cultivars but in drought sensitive cultivars the opposite pattern was shown with irrigated samples showing significantly higher TE expression. Thus, TE genome repression mechanisms may be finely tuned to the ideal growth and/or regulatory conditions of the specific plant cultivars in which they are active. Analysis of TE expression levels in cell culture conditions underscored the importance of nonsense-mediated mRNA decay (NMD) pathways in the repression of Coffea TEs. These same NMD mechanisms can also regulate plant host gene expression via the repression of genes that bear exonized TE sequences.
Citation: Lopes FR, Jjingo D, da Silva CRM, Andrade AC, Marraccini P, Teixeira JB, et al. (2013) Transcriptional Activity, Chromosomal Distribution and Expression Effects of Transposable Elements in Coffea Genomes. PLoS ONE 8(11): e78931. https://doi.org/10.1371/journal.pone.0078931
Editor: Leonardo Mariño-Ramírez, National Institutes of Health, United States of America
Received: March 25, 2013; Accepted: September 17, 2013; Published: November 11, 2013
Copyright: © 2013 Lopes et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Funding: This study was funded by grants provided by the Brazilian agencies Fundação de Amparo à Pesquisa do Estado de São Paulo (2008/05894-1 to CMAC and fellowship 2007/55985-0 to FRL) and Conselho Nacional de Desenvolvimento Científico e Tecnológico (to CMAC). The authors thank the Centre de Coopération Internationale en Recherche Agronomique pour le Développement for the grant (ATP project “Plasticité phénotypique des pérennes sous contrainte hydrique au champ” to PM and ACA) that permitted the study of C. arabica cultivars. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: The authors have declared that no competing interests exist.
Introduction
Transposable elements (TEs) are genetic entities with an intrinsic mobilization capacity. As a result of this characteristic, they are responsible for donating regulatory sequences [1] and transcription regulatory signals [2], as well as for creating considerable genomic instability, mediating chromosome rearrangements [3], altering both gene expression and function [1], and creating novel genes and exons [4]. Such mobilization can also result in host genome contraction and expansion [5], [6]. According to a unified classification system proposed for eukaryotic transposable elements [7], TEs can be grouped into two classes according to their transposition mode: Class I elements (retrotransposons), which use the enzyme Reverse Transcriptase (RTase) to transpose via an RNA intermediate to a new genome insertion site, and Class II elements, which are transposed directly via DNA molecule using a transposase (Tpase) enzyme. Class I elements are divided into five orders (LTR, DIRS, PLE, LINEs, SINEs), each of which is subdivided into superfamilies (LTR: Copia, Gypsy, Bel-Pao, Retrovirus, ERV; DIRS: DIRS, Ngaro, VIPER; PLE: Penelope; LINEs: R2, RTE, Jockey, L1, I; SINEs: tRNA, 7SL, 5S). Class II (DNA transposons) elements are split into two subclasses: subclass I contains superfamilies either with terminal inverted repeats (Tc1-Mariner, hAT, Mutator, Merlin, Transib, P, PiggyBac, PIF-Harbinger and Cacta) or without terminal inverted repeats (Crypton), whereas subclass II comprises the Helitron and Maverick superfamilies.
Plant genomes are massively invaded by repetitive DNA, primarily LTR retrotransposons [8]. Many of these retrotransposons are located near host genes and thus could impact the expression of these neighboring genes whose expression is mediated by a variety of mechanisms, such as RNAi [9], DNA methylation [10], and readout transcription [11]. While the transcriptional activity of TEs seems to be tightly controlled by host genomes [12], e.g. by transcriptional gene silencing mechanisms such as those that prevent the access of the host transcriptional machinery [13], reports also show that TEs can be activated under certain stress conditions, such as pathogen infection, physical injuries, abiotic stress [14] or cell culture [15], [16], [17], [18].
Of the approximately one hundred species in the Coffea genus, only C. arabica and C. canephora are used in commercial production, representing ∼70% and 30% of global coffee production, respectively [19]. C. arabica is a unique polyploidy species of the genus (4n = 4X = 44 chromosomes) and was derived from a recent (1 million years ago) natural hybridization between C. canephora and C. eugenioides [20]. C. canephora is a diploid (2n = 2x = 22 chromosomes) and is an auto-incompatible species that grows in humid and lowland habitats. It is usually more resistant to pests and diseases as well as to abiotic stresses like water deprivation and is also characterized by a higher productivity and bean caffeine content than C. arabica. However, the quality of the beverage is regarded as inferior to that of C. arabica [21].
The identification of transposable elements in Coffea was initiated only recently [22], [23], [24], [25], [26]. To our knowledge, detailed analyses of the abundance, activity and regulation of transcriptionally active TEs in Coffea genomes, as well as analyses of the relationship of these to their chromosomal distribution, have yet to be performed. We previously searched the Brazilian Coffee Genome Project database (LGE database, http://www.lge.ibi.unicamp.br/cafe) aiming to identify TE fragments within coding regions in expressed sequences (ESTs) of three Coffea species (C. arabica, C. canephora and C. racemosa). The ESTs and unigenes harboring TEs were analyzed regarding the size of the TE fragment, the functional categories to which they were assigned to and to their contribution to the proteome [22]. In the present study, we rescreened the LGE database using a more sensitive method, which allowed for a substantial increase in the number of unigenes harboring TEs, and used the gene sequences to identify paralogous unigenes that do not contain TEs. Expression levels of isoforms with and without TEs in C. arabica and C. canephora transcriptome were analyzed in cell culture and plants grown under different irrigation conditions. This approach was taken in order to understand the regulatory effects that exonized TEs may exert on Coffea host genes. The expression levels of TE transcripts themselves were also assayed across the same conditions in order to better understand how they are regulated and how they respond to various stresses including different drought and irrigation conditions as well as cell culture and polyploidization. The chromosomal distribution of Coffea TEs was interrogated genome-wide for the first time here using FISH.
Results
Frequency and Classification of Expressed TEs in the Transcriptomes of Coffea spp
A set of 181,405 ESTs derived from 39 cDNA libraries (31 from C. arabica and 8 from C. canephora) were used to identify, classify and quantify the number of expressed TEs. Sequence similarity searches allowed the identification of 320 EST sequences homologous to TEs in the two Coffea species (Table 1; Tables S1 & S2 in File S1). For C. arabica, the proportion of sequences that were homologous to DNA transposons (51%) and to LTR+NLTR retrotransposons (49%) were similar (P>0.05). Ty3/Gypsy was the most frequently identified superfamily among the LTR Retrotransposons (22%). However, in C. canephora, the proportion of transposons (13.2%) and retrotransposons (86.8%) was considerably different, as were the frequencies of Ty3/Gypsy (80.5%) and Ty1/Copia (3.6%) (both P<0.05).
The TEs were classified into families based on the best alignment against a completely characterized reference TE library (Table S3 in File S2). The 100 ESTs from C. arabica were classified into 24 families, 8 belonging to DNA transposons and 16 to Retrotransposons (LTR:13, NLTR: 1 and not classified: 2) and the 220 ESTs from C. canephora were classified into 18 families (DNA transposons: 7, LTR Retrotransposons: 9, NLTR: 1 and not classified LTR: 1). The main difference between diversity of TE families between the two species is due to higher number of Ty1-Copia families in C. arabica, which harbors seven families of this superfamily of LTR retrotransposon, but excepting Tst1, all occurring in just one EST (Figure 1, Tables S1 & S2 in File S1). Both species differ also regarding the TEs expression. For example, the most expressed families in C. arabica among the retrotransposons, considering their expression among the total TEs and among the TE classes, respectively, were Retrosat2 (11 ESTs: 11% of the TEs and 22% of Class I), Melmoth (8 ESTs: 8% of the TEs and 16% of Class I) and Tst1 (6 ESTs: 6% of the TEs and 12% of Class I). Regarding the DNA transposons, the most expressed families in C. arabica were MuDR (12 ESTs: 12% of the TEs and 24% of Class II), Jittery (11 ESTs: 11% of the TEs and 22% of Class II) and Soymar (11 ESTs: 11% of the TEs and 22% of Class II). Likewise, in C. canephora the most expressed retrotransposons were dea1 (94 ESTs: 43% and 49%), Retrosat2 (56 ESTs: 25% and 29%) and Del1 (18 ESTs: 8% and 9%) and, the most expressed DNA transposons were MuDR (7 ESTs: 3% and 24%), AtMu and Activator (6 ESTs each: 2.7% and 21%), and Jittery (5 ESTs: 2% and 17%).
Characterization of cDNA Clones from C. arabica Similar to TEs
After the 100 clones identified in C. arabica were classified into families, the redundant clones (identical ESTs and identified in the same library) were excluded and 64 clones remained, from which a sample of 27 were fully sequenced (Table S4 in File S3) and the remaining partially sequenced. Again, this set of sequences was compared against two TE banks: the reference elements cited above and the collection of consensus TEs stored in Repbase [27]. The results showed that these cDNAs were highly similar to elements from Repbase described in species closely related to the genus Coffea, such as Vitis vinifera and Solanum tuberosum. The occurrence of frame shifts and stop codons was distinct for some comparisons due to the differential choice of frame in the sequence translation. Putative complete transposase or polyprotein searches were performed by evaluating CDSs that spanned at least 60% of the reference or consensus TE and had no frame shifts and stop codons.
Expression Levels of TE Transcripts
Macroarray expression profiling was performed for 64 TE transcripts (31 DNA transposons and 33 retrotransposons) in six samples from the allotetraploid C. arabica (callus treated with cycloheximide versus untreated callus (CHX+, CHX–), irrigated versus non-irrigated leaves (_I, _NI) from drought tolerant versus drought sensitive cultivars (I59, Rubi)), and in two samples from the diploid C. canephora (14_). Many TEs exhibit relatively low expression levels (Figure 2A and Figure S1 in File S4), and overall TEs are expressed at significantly lower levels than genes measured for the same cultivars and conditions (Figure 2B).
(A) Heatmap showing the relative expression levels of TE transcripts for 31 DNA transposons and 33 retrotransposons. CHX+: C. arabica callus treated with cycloheximide, CHX–: C. arabica callus untreated, I59_I: C. arabica irrigated leaves from drought tolerant cultivar Iapar59, I59_NI: C. arabica non-irrigated leaves from Iapar59, 14_I: C. canephora irrigated leaves from drought tolerant cultivar, 14_NI: C. canephora non-irrigated leaves from drought tolerant cultivar, Rubi_I: C. arabica irrigated leaves from drought sensitive cultivar Rubi, Rubi_NI: C. arabica non-irrigated leaves from Rubi. (B) Overall expression level differences between TEs, genes with TE insertions (TE+Genes, n = 77) and genes without TE insertions (TE–Genes, n = 63) across the conditions measured here. Average expression levels ± standard errors were compared using the Students’ t-test and the Mann-Whitney U test (MWU) as indicated. (C) Average expression levels for retrotransposons versus DNA transposons. (D) Average Manhattan distances between expression profiles within versus between TE families. (E) Individual TE sequences that have significantly up-regulated upon cycloheximide treatment (CHX+).
TE expression patterns across the species, cultivars and conditions assayed here are highly dynamic and apparently unrelated to their classification at the level of TE class (DNA transposon versus retrotransposons) or at level of specific TE families (Figure 2C). When hierarchical clustering is used to relate the expression patterns of individual TE transcripts, there is no apparent coherence within TE classes or families; individual members of TE classes and families are dispersed throughout the resulting tree (Figure S2 in File S4). The coherence of expression patterns within and between TE families was also measured using average Manhattan distances between TE expression profiles. There is no statistical difference in the average distances between TE expression profiles within or between families (Figure 2D). This same lack of overall coherence in TE expression patterns can be seen at the level of individual TE families, where there is generally no difference in the distances between expression profiles within or between families (Figure S3 in File S4). Exceptions to this overall pattern can be seen for the Jittery and Tip100 families of DNA transposons, which have relatively coherent within family expression patterns (Figure 2A and Figure S3 in File S4). Interestingly, TE expression for CHX– (callus untreated) is not higher than that of the other tissues/conditions in C. arabica (Figure S4 in File S4), suggesting that cell culture conditions do not de-repress TE expression in this species.
The dynamic expression patterns seen for individual TE transcripts, along with the overall lack of TE expression pattern coherence within classes and families of elements, suggest that the expression of individual TE insertions is independently regulated based in part on the surrounding genomic environment. This may include both sequence-based and epigenetic factors for the surrounding genomic environment. However, it should be noted that changes in TE expression in response to cycloheximide treatment appear to be more consistent across individual TE insertions. On average, TEs are expressed at higher levels in callus treated with cycloheximide (CHX+ versus CHX– in Figure 3); although, the difference is only marginally significant owing to the high level of variation in TE expression under cycloheximide treatment. Nevertheless, there are a number of individual TE transcripts that show highly significant differences between CHX+ and CHX– conditions (Figure 2E and Figure S5 in File S4). Cycloheximide (CHX) is a compound that blocks translation and thereby inactivates the nonsense-mediated mRNA decay pathway (NMD). The NMD pathway provides for the detection and degradation of aberrant mRNAs that contain premature termination codons resulting from point mutations, rearrangements or errors during transcription or RNA processing [28]. More to the point, NMD may also represent a genome defense mechanism against the proliferation of TEs since mRNA sequences derived from TEs are frequent targets of NMD [29]. Thus, the up-regulation of TE expression upon CHX treatment would seem to reflect the mitigation of the plant genome’s NMD based defense against TE expression. The consistency of this pattern seen across TE transcripts assayed here underscores the likely importance of NMD in genome defense against plant TE expression.
Overall TE expression levels are compared for cycloheximide treated (CHX+) versus untreated (CHX–) C. arabica callus and irrigated (I) versus non-irrigated (NI) leaves for drought tolerant C. arabica and C. canephora as well as drought sensitive C. arabica. Average expression levels ± standard errors were compared using the Students’ t -test and the Mann-Whitney U test (MWU) as indicated.
We performed similar comparative analyses in order to evaluate whether there are overall differences in TE expression between paired samples of irrigated (I) and non-irrigated (NI) cultivars that are either drought sensitive or drought tolerant. TE transcripts tend to be expressed at significantly higher levels in non-irrigated cultivars for the drought tolerant samples; this effect is marginally significant for C. arabica leaves and more highly significant for C. canephora leaves (Figure 3). Drought sensitive cultivars from C. arabica show the opposite pattern with irrigated samples showing significantly higher TE expression (Figure 3). Considered together, these results suggest the interesting possibility that TE expression levels may be tuned to the drought response proclivities of their host genomes.
TE–Derived CDSs
The unigenes containing TE cassettes (i.e. exonized TEs) previously identified in the C. arabica and C. canephora ESTs using RepeatMasker [22] are presented in Tables S5 & S6 in File S5, respectively. Each unigene cluster contains sequences that represent a unique gene resulting from assembling of various ESTs. Rescreening the ESTs of C.arabica against the library of 2,503 consensus TEs from RepBase yielded 421 TE–containing ESTs. Due to the high number of matches, only those with E-values<e−10 were analyzed, resulting in 303 matches. A comparison of these TE–containing ESTs against the database of unigenes (EST clusters) in the LGE database revealed 27 new TE–derived protein CDSs in C. arabica (Table S5 in File S5). All of the CDSs matched with proteins with a clearly defined function, namely protein factors related to transcriptional and spliceosomal machinery, chaperones and alcohol dehydrogenase.
Effect of Exonized TEs on the Expression and Regulation of TE–containing Transcripts by NMD
Screening the unigenes containing TE cassettes from C. arabica (86) and C. canephora (59) against their respective sets of unigenes allowed the identification of 111 and 47 paralogous unigenes, respectively, based on their high sequence similarity (Tables S5 & S6 in File S5), as illustrated in Figure S6 (File S4). Two sets of alignments showed evidence of putative alternative splicing in C. arabica. The first had three transcripts (uni_CA_046, uni_CA_125 and uni_CA_127) that were similar to a rust resistance Rp1-D-like protein (GB: AAK18308) and the second had two transcripts (uni_CA_055 and uni_CA_138) that were similar to a universal stress protein (USP) family protein (GB: NP_850562). Additionally, this analysis is not exhaustive because 1) the EST libraries were built based only on the representation of the 5′ end of mRNAs, preventing the identification of TEs in other portions of the transcript; 2) several libraries were not targeted for full sequencing (saturation), particularly for C. canephora, which meant that fewer tissues were represented and fewer ESTs were obtained; and 3) the limited Coffea genomic resources do not enable a complete analysis of gene structure and TE insertion sites or the identification of transcripts related to each expressed locus.
The expression levels of 77 gene transcripts containing TE cassettes (TE+Genes) and 63 paralogous gene transcripts that lacked TEs (TE–Genes) were analyzed using macroarrays across the same eight cultivars as previously described for C. arabica and C. canephora (Figure S1 in File S4). TE+Genes have significantly lower levels of overall expression than TE–Genes (Figure 4A & Figure S7A in File S4), and in fact TE+Gene expression levels are not distinguishable from those of TE transcripts themselves (Figure 2B). This result indicates that the presence of TEs in genic transcripts leads to their down-regulation and suggests the possibility that mechanisms similar to those used to repress TE transcript, such as NMD, may be involved in this process. To test this possibility, we compared the effects of CHX treatment on TE+Genes versus TE–Genes. TE+Genes show significantly greater overall levels of expression in CHX+ conditions compared to CHX–, whereas TE–Genes show no such difference in expression across the different CHX treatments (Figure 4B & Figure S7B in File S4). CHX+ conditions, which are seen facilitate the expression of TE+Genes, inactivate the NMD pathway for aberrant transcript repression. Thus, these results are consistent with the activation of the NMD pathway by the presence of TE cassettes within expressed gene transcripts.
(A) Comparison of overall expression levels of genes with TE cassettes (TE+Genes, n = 77) versus genes with no TE cassettes (TE–Genes, n = 63). Average expression levels ± standard errors were compared using the Students’ t-test and the Mann-Whitney U test (MWU) as indicated. (B) Differences in overall expression levels between CHX+ and CHX– conditions for TE+Genes versus TE–Genes. Average expression levels ± standard errors were compared between CHX+ and CHX– conditions for TE+Genes and TE–Genes individually using the Students’ t-test and the Mann-Whitney U test (MWU) as indicated.
Chromosomal Distribution of TEs in C. arabica and its Parental Species
Three TEs (two transposons: MuDR and Tip100 and one LTR-retroelement: Del1) with average low expression levels were selected, and their chromosomal distribution was evaluated using FISH in C. arabica var. typica, C. eugenioides and C. canephora (Figure 5). The transposon signals were preferentially clustered in chromosomal terminal positions in the two ancestors (C. eugenioides and C. canephora) of C. arabica. Interstitial and/or proximal signals were observed in larger numbers in C. arabica var. typica, especially using the Tip100 probe, indicating an increase in transposition activity in this species compared to the parental species.
FISH using sequences of transposons MuDRA (GI311206994), Tip100 (GI 315896428) and of retrotransposon Del1 (GI 315862857) in the chromosomes of C. arabica var. typica, C. eugenioides and C. canephora. The MuDRA probe hybridized in 14 locations in C. arabica var. typica (A), with terminal, interstitial and proximal signals. Arrows indicate interstitial/proximal sites. This same probe hybridized preferentially clustered signals in C. eugenioides, with scattered signals in two pairs (D) and only clustered terminal signals in C. canephora (G). The Tip100 probe showed 36 hybridization sites in C. arabica var. typica (B), with chromosomes containing three sites (arrows) and two hybridization sites in terminal and proximal/interstitial regions (arrowheads). The same probe showed only eight chromosomes with terminal sites in C. eugenioides (E) and 14 chromosomes with signals in C. canephora (H). Note that four chromosomes exhibit two signals, being terminal and interstitial (arrows) and double terminal (arrowheads). The Del1 probe hybridized in 20 chromosomes in C. arabica var. typica (C). In only eight of them clustered signals were observed (arrows). From 12 chromosomes with signals observed in C. eugenioides (F), only two presented scattered ones (arrows). For C. canephora (I), this probe showed two pairs with scattered signals and evident terminal signals in six chromosomes (arrows). Bar represents 10 µm.
The probe containing the LTR-retrotransposon Del1 showed a different hybridization profile, exhibiting signals that were more scattered. The number of positions hybridized with the Del1 probe in C. arabica was nearly the same as that of the sum of signals of C. eugenioides and C. canephora. Furthermore, C. canephora chromosomes showed preferentially clustered signals, and most C. eugenioides chromosomes showed dispersed signals. These results are in agreement with previous FISH results [30] using another probe containing a fragment homologous to a gag-like element from a Del1 LTR-retrotransposon isolated from the germplasm of C. arabica var. typica. Similarities in Ty1-copia-like retroelement among the different Coffea genomes were also reported [25].
Discussion
This study provides a preliminary understanding of the TE regulatory dynamics in the allotetraploid and complex genome of C. arabica. The findings presented here show that this species is an interesting study organism because most DNA transposons and retrotransposons seem to be submitted to fine transcriptional control. Differences and similarities with other plant genomes were observed. For instance, it has been reported that cell culture conditions increase the expression levels of some TEs in plants, as for example Tnt1 from tobacco [31]; Tos10, Tos17 and Tos19 from rice [32]; several DNA transposons and retrotransposons from sugarcane [17]; and LTR retrotransposons from rice [18]. In contrast to these previous observations, our results show that cell culture conditions per se do not elevate TE expression levels in C. arabica. On average, TEs in untreated cell culture conditions (CHX–) show no significant difference in expression levels compared to TEs analyzed from C. arabica plant tissue (Figure S4 in File S4). These results suggest that TE regulatory and/or suppression mechanisms remain largely intact in C. arabica cell culture conditions. This conclusion is supported by the observation that treatment of C. arabica cell culture with CHX, a repressor of NMD, leads to a significant increase in TE expression (Figure 3). In other words, the NMD regulatory systems that suppress TE expression remain active in C. arabica cell culture.
Results reported here also underscore the importance of NMD-related TE repression systems for the regulation of host genes. Indeed, NMD not only represses the expression of the TEs themselves (Figure 3) but also appears to repress the expression of host genes that contain exonized TE sequences (Figure 4). This finding represents a novel molecular mechanism by which TE sequences can influence the regulation of Coffea host gens.
Polyploid genomes are subjected to extensive changes, such as insertions/deletions, inversions and translocations, as well as alterations in gene expression patterns [33], [34]. Although the mechanisms of these changes are poorly understood, increasing TE transpositional activity is a possibility since quiescent TEs in a diploid genome can become activated in the new polyploid genetic environment. Additionally, the genetic redundancy in a polyploid genome can mitigate the deleterious effects of transposition [35], thereby allowing TEs to proliferate in allotetraploids and insert within gene-rich chromosomal regions. Another possibility is the relaxation of host silencing mechanisms (e.g., methylation) in allotetraploids, which should also allow for increased transposition rates [36]. These factors may explain the increase in transposon copy number and their more prevalent interstitial chromosomal location we observed in the allotetraploid C. arabica compared to its parents, C. canephora and C. eugenioides.
Coffea arabica and C. canephora showed a low TE–like mRNA abundance; only 0.17% of ESTs were expressed TEs (320 out of 181,405 ESTs). This low abundance has also been observed in other plant genomes. For an example, 60% of the Z. mays genome is composed of retroelements, but only 0.014% of these retroelements (56 out of 407,000 ESTs) were identified as expressed [37]. However, a recent systematic search in the maize transcriptome showed that 1.5% of its ESTs (25,282 ESTs out of more than 2 million) were similar to 56 well characterized TE families [18]. In Saccharum officinarum, out of 260,781 ESTs, 276 (0.1%) were considered to be expressed TEs [38]. Finally, in Eucalyptus grandis, out of 123,889 ESTs, 124 (0.1%) were identified as transcriptionally active TEs [39]. Our data reinforce the fact that TEs are poorly represented in the Coffea transcriptome, although plant genomes are enriched by those repetitive sequences. This paradox reflects the strong host repression of TE transcriptional and transpositional activity in plants, as is illustrated by the paucity of TEs in Coffea transcriptomes and the heterochromatic distribution of most Coffea TE sequences.
Nevertheless, transcriptional activation of several plant retrotransposons under stress has been shown and it seems that these mobile sequences have adapted to their host genomes through the evolution of highly regulated promoters that mimic those of the stress-induced plant genes (see [14], for a review). Moreover, it has been also shown, as for example with Tnt1 in tobacco, that subfamilies of the same retrotransposon show different stress-associated patterns of expression [40]. Here, in a broader analysis of TE expression, we demonstrate that differing plant drought stress levels relate strongly to the changes in TE expression levels observed upon changes in irrigation conditions. Drought stress conditions were evaluated here in terms of predawn leaf water potentials Ψpd. In coffee (as in many other plants), Ψpd values close to 0 (>−0.4 MPa) are observed for unstressed plants, while more negative values (<−0.4 MPa) characterize drought stress. In other words, more highly negative Ψpd values reflect higher levels of drought stress. Thus, the observed Ψpd value of −0.80 MPa for C.arabica cv. Iapar59 is considered to show moderate drought stress, whereas the −3.02 MPa Ψpd value of C. canephora clone 14 is considered to show severe drought stress [41], [42]. Both of these drought tolerant Coffea cultivars show higher expression in the non-irrigated conditions (Figure 3). Interestingly, the opposite pattern was observed for drought sensitive cultivars from C. arabica, which showing significantly higher levels of TE expression in irrigated conditions (Figure 3). Given the presumed effects of genomic environment on the expression of individual TE transcripts noted above, these divergent phenomena may relate to the overall state of the particular plant genome with respect to its ideal drought-related growth and regulatory conditions. In drought sensitive plants, TE expression levels go down in drought conditions consistent with an overall depression of genomic regulatory activity. On other hand, TE expression is up-regulated upon drought conditions in drought tolerant plants presumably consistent with the ideal growth/regulatory conditions of these cultivars.
Materials and Methods
The ESTs used in our study are derived from libraries of the Brazilian Coffee Genome Project, hereafter called PGCB (http://www.lge.ibi.unicamp.br/cafe), which contain partial sequences of cDNA of a wide range of tissues (e.g., seeds, embryogenic calli, roots, leaves, flowers), developmental stages and plant material submitted to biotic (e.g., stems infected with Xylella spp and nematodes) and abiotic (e.g., water deficit) stress conditions [43], [22]. They comprises 131,150 ESTs from thirty-one cDNA libraries of C. arabica and 50,255 ESTs from eight cDNA libraries of C. canephora (12,332 obtained from 2 Brazilian Coffee Genome Project libraries plus 37,923 of six libraries from Lin et al. 2005 [44], deposited at the SOL Genomics Network http://www.sgn.cornell.edu/content/coffee.pl). These ESTs were analyzed with two main objectives: (1) to characterize the classes, types and numbers of expressed TEs, and to investigate their expression; and 2) to investigate the impact of fragments of TEs (TE–cassettes) inserted in coding regions of both species comparing the expression of sequences harboring TE–cassettes and homologous sequences not harboring TE–cassettes (likely paralogous sequences).
Construction of a Permanent cDNA Library of Clones of Interest
To characterize the expression profiles of active TEs and to evaluate putative differences in expression of transcripts containing TE–cassettes compared to their isoforms without TEs, the 242 cDNA clones of interest (64 of expressed TEs, 86 of unigenes with TE–cassettes and their 77 of isoforms without TEs) were trimmed from the PGCB libraries. Multiplication of 3 µL of bacteria culture containing each cDNA insert, cryopreserved in glycerol (50% v/v) and kept at −80°C was allowed in deepwell plates with 1.2 ml LB liquid culture medium for. After growth, 75 µL was removed to construct a permanent library with only the cultures of interest, followed by the purification and cloning of DNA plasmids.
Identification of Expressed TEs
Expressed TEs were identified in 181,405 ESTs (libraries of C. arabica plus of C. canephora). The transcripts were considered likely to represent a transcriptionally active element when the TE sequence occupied more than 70% of an EST or unigene. EST clusters that were similar to expressed TEs were not considered because they may represent the mRNA assemblies of distinct insertions. Searches for the transcriptionally active TEs were performed using keywords such as “transposon”, “transposase”, “polyprotein”, “retrotransposon”, “retroposon”, “MITEs”, “LINEs”, “SINEs” and family names (e.g., “hAT”, “MuDR”, “En/Spm”) of the ESTs. A BLASTx [45] comparison of the ESTs with TE annotations against the protein sequences in the NCBI NR (non-redundant) database was then performed. Many BLAST hits were obtained. To eliminate spurious and unreliable results, a stringent cut-off (E = 1e−30) was applied. The resulting transcripts were classified into families according to the best alignment by BLASTx or tBLASTx [45] against a completely characterized library of 96 reference TEs (Table S3 in File S2), as well as against 840 consensus TEs of phylogenetic species close to Coffea (e.g., Vitis vinifera, Populus trichocarpa, Solanum lycopersicum, and Solanum tuberosum) obtained from the dicotyledonous plant library Repbase (www.girinst.com). When ESTs or unigene sequences were annotated by the alignment with more than one reference TE from different plant TE databases, the matches between Coffea transcripts and reference TEs with the higher RM score was chosen. The frequencies of retrotransposons and transposons, as well as expressed TEs of the Ty-Copia and Ty3-Gypsy superfamilies were compared using a χ2 test.
From the set of 100 clones homologous to TEs of C. arabica (Table S1 in File S1), 64 were cloned into pSPORT1 vector, and sequenced using the BigDye® Terminator v3.1 Cycle Sequencing kit (Applied Biosystems, Foster City, CA, USA) using the universal primers M13F (5′-GTAAAACGACGGCCAG-3′) and M13R (5′-CAGGAAACAGCTATGAC-3′) as well as through internal primers specific to each clone and run on a 3730xl DNA Sequencer (Applied Biosystems). The sequences were clustered using CodonCode Aligner v.3.5.6 (www.codoncode.com), and bases had a Phred quality ≥20. The identification notation of these active TEs was “Ca_” (for C. arabica), “TE–” (for transposable elements) plus “three numerical digits”, for example Ca_TE_031. Full sequences were obtained for 27 of these clones and partial sequences (sizes over 50% of the total length) for the remaining 37.
Identification of Novel Cases of TEs Incorporated into Mature mRNAs from C. arabica and of Paralogous Sequences without TEs
It has been shown that the choice of sequence similarity search methods to detect TE–derived sequences strongly influences the estimate of TE-cassettes that can be identified in protein coding regions [46]. Hidden Markov model based searches followed by BLAST methods (tBLASTx → tBLASTn → BLASTx → BLASTp → BLASTn) and RepeatMasker are more sensitive in identifying exonized TEs. We used that protocol to identify novel TE–derived sequences in protein coding sequences of Coffea in addition to those previously identified by RepeatMasker alone [22]. A total of 131,150 ESTs from C. arabica were compared by tBLASTx [45] against 2,503 plant consensus TEs from Repbase [27]. To avoid spurious results, only the best E≤e−10 matches were accepted, without imposing additional scores or length thresholds. ESTs containing TEs were then compared to EST clusters of each species for the identification of their respective unigenes. They summed 145 cDNA clones containing TE–cassettes (59 from C. arabica obtained in our previous study [22] plus 27 novel ones obtained in this study, and 59 from C. canephora), which were compared by BLASTn [45] against all cDNA clones of each species. This procedure allowed the identification of highly similar and thus likely paralogous unigenes without TEs. Examples of alignments between unigenes containing TEs and the highly related unigenes using sequence similarity searches by BLASTn are given in Figure S6 (File S4).
Expression Analyses
The expression analyses were carried out for the 64 individual transcriptionally active TEs characterized in this study (Table 2) and for transcripts of CDSs harboring TE–cassettes (77) identified in this and in a previous study [22] and corresponding CDSs without TE insertions (63), identified in this study (Table 3).
Plant material.
For probe synthesis, total mRNA was extracted from the following samples of C. arabica: a) Drought stress: leaves of cultivars tolerant (Iapar59) and sensitive (Rubi) to drought grown in field conditions (Cerrado Agricultural Research Center, Planaltina-DF, Brazil) with (predawn leaf water potentials Ψpd−0.38±0.10 and −0.22±0.07 MPa for Iapar59 and Rubi cultivars, respectively) and without (Ψpd = −0.80±0.12 and −1.88±0.36 MPa for Iapar59 and Rubi cultivars, respectively) irrigation [47], b) Cell culture: embryogenic callus from C. arabica cv. Catuaí Vermelho maintained in a multiplication medium for ∼4 months; c) Inhibition treatment: the same embryogenic callus was treated for 4 h with the protein biosynthesis inhibitor cycloheximide (CHX: 10 and 30 mg/mL in alcohol) added to cell culture for a final concentration of 100 and 300 µg/mL. C. canephora var. conilon drought stress: clone 14, tolerant to drought was selected by the INCAPER [41] and grown in a greenhouse with (unstressed condition, Ψpd leaves = −0.02±0.03 MPa) or without (stress, Ψpd leaves = −3.02±0.12 MPa) water [42]. For fluorescent in situ hybridization (FISH), slides were prepared with root tips of C. arabica var. typica, C. canephora and C. eugenioides pretreated with a saturated solution of paradichlorobenzene for 24 h at 14°C, without acid hydrolysis.
RNA isolation, DNAse treatment and reverse transcription.
RNA of all samples was extracted from cells using Concert™ reagent (Invitrogen, Carlsbad, CA, USA) according to the manufacturer’s protocol. Total RNA samples (10 µg) were incubated for 30 min at 37°C with 3 U of RQ1 RNAse-Free DNAse (Promega, Madison, WI, USA) in a final volume of ∼10 µl. Each total RNA sample was mixed with 1.5 µl Oligo(dT)12–18 (Invitrogen), heated at 75°C for 10 min, and then cooled for 5 min on ice. The reaction mixture for the reverse transcription contained 5 µl of 5×first-strand buffer, 2.5 µl of 0.1 mol/l DTT, 40 U of RNAseOUT, 50 µCi (α-33P)-dCTP and 2.5 µl of a 10 mM mixture of unlabeled dNTPs (dATP, dTTP and dGTP) and was heated at 42°C for 5 min. The reverse transcription was performed at 42°C for 20 min with 300 U of SuperScript III First-Strand Synthesis System for RT-PCR (Invitrogen). Then, 1.25 µl of unlabeled dCTP (10 mM) was added and maintained for 1 h, terminated by heating at 94°C for 5 min and cooled for 5 min on ice. The total volume (30.25 µl) was used in the hybridization experiments.
Amplification of target DNA.
Each target cDNA (100 ng) was amplified by PCR in a volume of 25 µl with 1.25 U of Platinum® Taq DNA Polymerase (Invitrogen) in 10× polymerase buffer, 2 mM MgCl2, 200 µM each dNTP and 10 µM of each universal primer, M13F and M13R. The solutions were heated to 94°C for 2 min, followed by 35 cycles of denaturation (94°C for 30 sec), annealing (50°C for 30 sec), extension (72°C for 4 min), and final extension at 72°C for 7 min. The target DNAs were used to make the membrane arrays.
Macroarray experiments and analysis.
The PCR products of the target DNAs were denatured in DMSO (50% v/v) for 30 min at 37°C, arrayed in a 384-well microtiter plate and then spotted twice in the same position onto Performa II nylon filters (Genetix Limited, Hampshire, UK) using the high-throughput robot system Q-BOT (Genetix Limited). To increase the signal homogeneity among spots and filters, the set of 64 cDNAs was spotted in duplicate (2×2 array) onto two identical arrays using the same nylon filters (222×222 mm). Additionally, 16 spots containing cDNA of the reference gene ubiquitin were applied to delimit the two arrays. After sample deposition, the filters were dampened with a denaturant solution (NaCl 0.13 M and 0.5 M NaOH) for 10 min and a neutralization solution (NaCl 1.5 M and Tris 1 M) for 5 min, then fixed by UV light exposition (1,200 µj/cm2) for 12 sec and stored at −80°C. The filters were pre-hybridized for 2 h at 65°C in Modified Church and Gilbert Buffer (0.5 M Na Phosphate Buffer pH 7.2, 7% SDS, 10 mM EDTA) and hybridized overnight with cDNA sample probes. Membranes were washed for 15 min three times with 0.1% SDS/1×SSC and three times with 0.1% SDS/0.1×SSC at 65°C. After washing, the filters were exposed on imaging plates BAS-MS 2340 (Fujifilm, Tokyo, Japan) for 72 h in a BAS 2340 cassette (Fujifilm) and scanned using a fluorescent image analyzer FLA3000 (Fujifilm). The radioactive intensity of each spot was quantified by Array Gauge software (Fujifilm), corrected by the level of the local background, normalized to the average intensities of the reference gene ubiquitin (except for callus treated with CHX, in which the reference gene expression was completely suppressed). For differences in probe labeling, normalization was by use of the average signals of all genes studied. The homogeneity of the spot replicates were evaluated and represented by average values using limma of the Bioconductor package [48] from R (http://www.r-project.org).
The resulting normalized expression levels for individual probes, expressed as signal intensity values, were visualized, clustered and statistically analyzed across all conditions assayed here. Signal intensity values of individual transcripts were visualized and hierarchically clustered using the TIGR Multiexperiment viewer (MeV) program (http://www.tm4.org/mev.html). Average expression levels between conditions were compared using both parametric (Student’s ttest) and non-parametric (Mann-Whitney U test) statistical tests. Differences in condition-specific expression profiles for individual transcripts were computed using Manhattan distances between signal intensity vectors across conditions. The resulting distances were averaged within and between TE classes and families to measure TE expression coherence.
Fluorescent in Situ Hybridization
FISH was performed as described elsewhere [49] with modifications. Three expressed TE cDNA clones (GI 311206994, GI 315896428 and GI 315862857 similar to MuDR, Tip100 and del1, respectively) were used to synthesize the probes with biotin-14-dATP by nick translation. The reaction mixture (total volume 33 µl) contained 15 µl of 100% formamide, 6 µl of polyethylene glycol, 3 µl of 20×SSC, 1 µl of calf thymus DNA (100 ng), 4 µl of water and 4 µl of each probe (200 ng). The samples were denatured at 70°C for 10 min, and hybridization was performed at 37°C overnight in a humidified chamber. The washes were carried out in 6×SSC and 4×SSC/0.2% Tween 20 at room temperature. The probes were detected with avidin-FITC, followed by post-detection washes in 4×SSC/0.2% Tween 20 at room temperature. Slides were mounted with 25 µl of antifade, composed of glycerol (90%), 1,4-diaza-bicyclo(2,2,2)-octane(2.3%), 20 mM Tris-HCl pH 8.0 (2%), water and 1 µl of 2 µg/ml 4,6-diamidino-2-phenylindole (DAPI). The images were acquired using a Leica DM4500 B Microscope (Leica Microsystems, Wetzlar, Germany) equipped with a DFC 300FX Digital Color Camera (Leica Microsystems), and the image was overlapped with red color for DAPI using the Leica IM50 4.0 image management software (Leica Microsystems).
Supporting Information
File S1.
Tables S1 & Table S2. List of CDSs similar to expressed TE families identified in the transcriptome from C. arabica (Table S1) and C. canephora (Table S2).
https://doi.org/10.1371/journal.pone.0078931.s001
(PDF)
File S2.
Table S3. Completely characterized transposable elements library used for the classification into families of the expressed TEs identified in the Coffea transcriptome.
https://doi.org/10.1371/journal.pone.0078931.s002
(PDF)
File S3.
Table S4: List of fully cloned 27 cDNA from Coffea arabica similar to plant TEs.
https://doi.org/10.1371/journal.pone.0078931.s003
(PDF)
File S4.
Figures S1–S7. Supplementary Figures for the TE and gene expression analyses.
https://doi.org/10.1371/journal.pone.0078931.s004
(PDF)
File S5.
Tables S5 & Table S6. List of unigenes containing or not TE–cassette insertions in C. arabica (Table S5) and C. canephora (Table S6).
https://doi.org/10.1371/journal.pone.0078931.s005
(PDF)
Acknowledgments
We thank Mário A. P. Saraiva and Felipe Vinecky for making the array membranes and for manipulating the BCGP’s cDNA libraries, respectively, to Gustavo C. Rodrigues (EMBRAPA CNPTIA) for his assistance during the field trial experiments and to Elias A. G. Carnelossi for technical support.
Author Contributions
Conceived and designed the experiments: FRL CMAC. Performed the experiments: FRL PM CRMS MFC. Analyzed the data: FRL CMAC DJ KJ LW CRMS ALLV. Contributed reagents/materials/analysis tools: CMAC ACA ALLV GAGP JBT LFPP. Wrote the paper: CMAC FRL DJ KJ PM ALLV.
References
- 1. Jordan IK, Rogozin IB, Glazko GV, Koonin EV (2003) Origin of a substantial fraction of human regulatory sequences from transposable elements. Trends Genet 19: 68–72.
- 2. Thornburg BG, Gotea V, Makalowski W (2006) Transposable elements as a significant source of transcription regulating signals. Gene 365: 104–110.
- 3. Zhang J, Peterson T (1999) Genome rearrangements by nonlinear transposons in maize. Genetics 153: 1403–1410.
- 4. Kidwell MG, Lisch D (1997) Transposable elements as sources of variation in animals and plants. Proc Natl Acad Sci U S A 94: 7704–7711.
- 5. Fedoroff N (2000) Transposons and genome evolution in plants. Proc Natl Acad Sci U S A 97: 7002–7007.
- 6. Bennetzen JL (2002) Mechanisms and rates of genome expansion and contraction in flowering plants. Genetica 115: 29–36.
- 7. Wicker T, Sabot F, Hua-Van A, Bennetzen JL, Capy P, et al. (2007) A unified classification system for eukaryotic transposable elements. Nat Rev Genet 8: 973–982.
- 8. Kumar A, Bennetzen JL (1999) Plant retrotransposons. Annu Rev Genet 33: 479–532.
- 9. Zaratiegui M, Irvine DV, Martienssen RA (2007) Noncoding RNAs and gene silencing. Cell 128: 763–776.
- 10. Kinoshita Y, Saze H, Kinoshita T, Miura A, Soppe WJ, et al. (2007) Control of FWA gene silencing in Arabidopsis thaliana by SINE-related direct repeats. Plant J 49: 38–45.
- 11. Kashkush K, Feldman M, Levy AA (2003) Transcriptional activation of retrotransposons alters the expression of adjacent genes in wheat. Nat Genet 33: 102–106.
- 12. Biemont C (2009) Are transposable elements simply silenced or are they under house arrest? Trends Genet 25: 333–334.
- 13. Okamoto H, Hirochika H (2001) Silencing of transposable elements in plants. Trends Plant Sci 6: 527–534.
- 14. Grandbastien MA (1998) Activation of plant retrotransposons under stress conditions. Trends Plant Sci 3: 181–187.
- 15. Pouteau S, Huttner E, Grandbastien MA, Caboche M (1991) Specific expression of the tobacco Tnt1 retrotransposon in protoplasts. EMBO J 10: 1911–1918.
- 16. Hirochika H (1993) Activation of tobacco retrotransposons during tissue-culture. EMBO J 12: 2521–2528.
- 17. de Araujo PG, Rossi M, de Jesus EM, Saccaro NL Jr, Kajihara D, et al. (2005) Transcriptionally active transposable elements in recent hybrid sugarcane. Plant J 44: 707–717.
- 18. Vicient CM (2010) Transcriptional activity of transposable elements in maize. BMC Genomics 11: 601.
- 19.
Lashermes P, Andrade AC, Etienne H (2008) Genomics of coffee, one of the world’s largest traded commodities. In: Moore H, Ming R, editors. Genomics of Tropical Crop Plants: Springer. 203–226.
- 20. Lashermes P, Combes MC, Robert J, Trouslot P, D'Hont A, et al. (1999) Molecular characterisation and origin of the Coffea arabica L. genome. Mol Gen Genet 61: 259–266.
- 21. Leroy T, Ribeyre F, Bertrand B, Charmetant P, Dufour M, et al. (2006) Genetics of coffee quality. Brazilian Plant Physiology 18: 229–242.
- 22. Lopes FR, Carazzolle MF, Pereira GA, Colombo CA, Carareto CM (2008) Transposable elements in Coffea (Gentianales: Rubiacea) transcripts and their role in the origin of protein diversity in flowering plants. Mol Genet Genomics 279: 385–401.
- 23. Guyot R, de la Mare M, Viader V, Hamon P, Coriton O, et al. (2009) Microcollinearity in an ethylene receptor coding gene region of the Coffea canephora genome is extensively conserved with Vitis vinifera and other distant dicotyledonous sequenced genomes. BMC Plant Biol 25: 9–22.
- 24. Dubreuil-Tranchant C, Guyot R, Guellim A, Duret C, de la Mare M, et al. (2011) Site-specific insertion polymorphism of the MITE Alex-1 in the Genus Coffea suggests interspecific gene flow. Int J Evol Biol 2011: 358412.
- 25. Hamon P, Duroy PO, Dubreuil-Tranchant C, Mafra D'Almeida Costa P, Duret C, et al. (2011) Two novel Ty1-copia retrotransposons isolated from coffee trees can effectively reveal evolutionary relationships in the Coffea genus (Rubiaceae). Mol Genet Genomics 285: 447–460.
- 26. Cenci A, Combes MC, Lashermes P (2012) Genome evolution in diploid and tetraploid Coffea species as revealed by comparative analysis of orthologous genome segments. Plant Mol Biol 78: 135–145.
- 27. Jurka J, Kapitonov VV, Pavlicek A, Klonowski P, Kohany O, et al. (2005) Repbase update, a database of eukaryotic repetitive elements. Cytogenet Genome Res 110: 462–467.
- 28. Brogna S, Wen J (2009) Nonsense-mediated mRNA decay (NMD) mechanisms. Nat Struct Mol Biol 16: 107–113.
- 29. Mendell JT, Sharifi NA, Meyers JL, Martinez-Murillo F, Dietz HC (2004) Nonsense surveillance regulates expression of diverse classes of mammalian transcripts and mutes genomic noise. Nat Genet 36: 1073–1078.
- 30. Yuyama PM, Pereira LF, Santos TB, Sera T, Vilas-Boas LA, et al. (2012) FISH using a gag-like fragment probe reveals a common Ty3-gypsy-like retrotransposon in genome of Coffea species. Genome 55: 825–833.
- 31. Courtial B, Feuerbach F, Eberhard S, Rohmer L, Chiapello H, et al. (2001) Tnt1 transposition events are induced by in vitro transformation of Arabidopsis thaliana, and transposed copies integrate into genes. Mol Genet Genomics 265: 32–42.
- 32. Hirochika H, Sugimoto K, Otsuki Y, Tsugawa H, Kanda M (1996) Retrotransposons of rice involved in mutations induced by tissue culture. Proc Natl Acad Sci U S A 93: 7783–7788.
- 33. Feldman M, Levy AA (2005) Allopolyploidy - a shaping force in the evolution of wheat genomes. Cytogenet Genome Res 109: 250–258.
- 34. Feldman M, Levy AA (2012) Genome evolution due to allopolyploidization in wheat. Genetics 192: 763–774.
- 35. Voytas DF, Naylor GJ (1998) Rapid flux in plant genomes. Nature Genetics 20: 6–7.
- 36. Madlung A, Tyagi AP, Watson B, Jiang H, Kagochi T, et al. (2005) Genomic changes in synthetic Arabidopsis polyploids. Plant J 41: 221–230.
- 37. Meyers BC, Tingley SV, Morgante M (2001) Abundance, distribution, and transcriptional activity of repetitive elements in the maize genome. Genome Res 11: 1660–1676.
- 38. Rossi M, de Araujo PG, Van Sluys MA (2001) Survey of transposable elements in sugarcane expressed sequence tags (ESTs). Genet Mol Biol 24: 147–154.
- 39. Bacci Jr M, Soares RBS, Tajara E, Ambar G, Fischer CN, et al. (2005) Identification and frequency of transposable elements in Eucalyptus. Genet Mol Biol 28: 634–639.
- 40. Beguiristain T, Grandbastien MA, Puigdomènech P, Casacuberta JM (2001) Three Tnt1 subfamilies show different stress-associated patterns of expression in tobacco. Consequences for retrotransposon control and evolution in plants. Plant Physiol 127: 212–221.
- 41. Ferrão RG, Fonseca AFA, Ferrão MAV, Silveira JSM, Bragança SM, et al. (2000) EMCAPA 8141 - Robustão Capixaba, variedade clonal de café conilon tolerante à seca, desenvolvida para o estado do Espírito Santo. Revista Ceres 273: 555–560.
- 42. Marraccini P, Vinecky F, Alves GS, Ramos HJ, Elbelt S, et al. (2012) Differentially expressed genes and proteins upon drought acclimation in tolerant and sensitive genotypes of Coffea canephora. J Exp Bot 63: 4191–4212.
- 43. Mondego JM, Vidal RO, Carazzolle MF, Tokuda EK, Parizzi LP, Costa GG, Pereira LF, Andrade AC, Colombo CA, Vieira LG (2011) Pereira GA; Brazilian Coffee Genome Project Consortium (2011) An EST-based analysis identifies new genes and reveals distinctive gene expression features of Coffea arabica and Coffea canephora. BMC Plant Biol 11: 30.
- 44. Lin C, Mueller LA, Mc Carthy J, Crouzillat D, Pétiard V, et al. (2005) Coffee and tomato share common gene repertoires as revealed by deep sequencing of seed and cherry transcripts. Theor Appl Genet 112: 114–130.
- 45. Altschul SF, Madden TL, Schäffer AA, Zhang J, Zhang Z, et al. (1997) Gapped BLAST and PSI-BLAST: a new generation of protein database search programs. Nucleic Acids Res 25: 3389–3402.
- 46. Piriyapongsa J, Rutledge MT, Patel S, Borodovsky M, Jordan IK (2007) Evaluating the protein coding potential of exonized transposable element sequences. Biol Direct 2: 31.
- 47. Freire LP, Marraccini P, Rodrigues GC, Andrade AC (2013) Análise da expressão do gene da manose-6-fosfato redutase em folhas de plantas de café (Coffea arabica) submetidas ao estresse hídrico no campo. Coffee Sci 8: 17–23.
- 48. Gentleman RC, Carey VJ, Bates DM, Bolstad B, Dettling M, et al. (2004) Bioconductor: open software development for computational biology and bioinformatics. Genome Biol 5: R80.
- 49. Heslop-Harrison JS, Schwarzacher T (1991) In situ hybridization with automated chromosome denaturation. Technique 3: 109–116.