Transcriptome reprogramming during severe dehydration contributes to physiological and metabolic changes in the resurrection plant Haberlea rhodopensis
- 581 Downloads
Water shortage is a major factor that harms agriculture and ecosystems worldwide. Plants display various levels of tolerance to water deficit, but only resurrection plants can survive full desiccation of their vegetative tissues. Haberlea rhodopensis, an endemic plant of the Balkans, is one of the few resurrection plants found in Europe. We performed transcriptomic analyses of this species under slight, severe and full dehydration and recovery to investigate the dynamics of gene expression and associate them with existing physiological and metabolomics data.
De novo assembly yielded a total of 142,479 unigenes with an average sequence length of 1034 nt. Among them, 18,110 unigenes were differentially expressed. Hierarchical clustering of all differentially expressed genes resulted in seven clusters of dynamic expression patterns. The most significant expression changes, involving more than 15,000 genes, started at severe dehydration (~ 20% relative water content) and were partially maintained at full desiccation (< 10% relative water content). More than a hundred pathways were enriched and functionally organized in a GO/pathway network at the severe dehydration stage. Transcriptomic changes in key pathways were analyzed and discussed in relation to metabolic processes, signal transduction, quality control of protein and DNA repair in this plant during dehydration and rehydration.
Reprograming of the transcriptome occurs during severe dehydration, resulting in a profound alteration of metabolism toward alternative energy supply, hormone signal transduction, and prevention of DNA/protein damage under very low cellular water content, underlying the observed physiological and metabolic responses and the resurrection behavior of H. rhodopensis.
KeywordsDesiccation tolerance Haberlea rhodopensis Hormone signaling pathway Primary and secondary metabolism Resurrection plant Transcriptome
Differentially expressed genes
Kyoto Encyclopedia of Genes and Genomes
Nucleotide excision repair
Principal component analysis
Environmental stress, especially drought, place limitations on plant distribution and crop production worldwide. Intensive breeding efforts, based on crosses with wild drought-tolerant relatives as potential sources of useful traits, are underway. At the same time, interest in a small group called resurrection plants is increasing, as these plants can withstand long periods of almost full desiccation and recover their vegetative system activities quickly and fully upon rehydration. This makes them useful models to study desiccation tolerance and potential sources of genes, related to these important phenomena [1, 2, 3, 4]. Further efforts are needed to elucidate the mechanisms of desiccation tolerance in resurrection plants as a sustained dynamic process.
The Balkan endemic Haberlea rhodopensis, one of the few species of resurrection plants, native to Europe, has been the subject of intensive genetic studies in recent years [5, 6, 7]. Transcriptomic analysis of H. rhodopensis gene expression under conditions of normal watering, dehydration (50% relative water content; RWC), full desiccation (5% RWC) and rehydration has been performed previously . We have shown that the dynamics of desiccation tolerance can be characterized by numerous changes in various processes at additional important time points during drying: the stages of slight (~ 65–75%) and severe (20–25% RWC) dehydration . For example, at the onset of drying (at 75% RWC), an approximately 25-fold increase in jasmonic acid (JA)  occurs, coinciding with the start of the decrease in photosynthetic performance . On the other hand, oxygen evolution in leaves of H. rhodopensis did not decrease until 20% RWC . Our studies revealed that the levels of numerous compounds changed dramatically during desiccation, particularly before or around the severe dehydration stage (25% RWC), including total phenols, sugars, total glutathione, malondialdehyde, sucrose/fructose ratio, abscisic acid (ABA), salicylic acid (SA) and starch levels [9, 11, 12]. Dramatic changes in photosynthetic performance and energy supply, such as ATP and glycolytic intermediates, were also observed at this stage of dehydration [9, 10, 11, 12]. This time point appears to be of crucial importance, as non-resurrection plants related to H. rhodopensis tolerated the loss of almost two-thirds of their water content, whereas further drying to 20% RWC was irreversible .
In an attempt to dissect the dynamic transcriptional regulation events preceding the dramatic physiological and metabolic changes that occur during desiccation in H. rhodopensis, we performed de novo transcriptome sequencing at new time points during drying from the fully hydrated to desiccated state: slight and severe dehydration stages.
Transcriptome sequencing and assembly
Transcriptomic analysis of gene expression in plants that were regularly watered (fresh control, F), slightly dehydrated to 75% RWC (D75), severely dehydrated to 20% RWC (D20), fully dehydrated to < 10% RWC (DT) and fully rehydrated after DT (R) was performed using Illumina next-generation sequencing technology. De novo assembly yielded a total of 142,479 unigenes, with an average sequence length of 1034 nt and an N50 value of 1664 nt (Additional file 1: Table S1). Approximately 75% of all unigenes were novel to H. rhodopensis. The sequence length distribution of the unigenes is indicated in Additional file 2: Fig. S1a. The numbers of unigenes annotated using several databases, NCBI non-redundant protein sequences, NCBI nucleotide collection, Swiss-Prot, Kyoto Encyclopedia of Genes and Genomes (KEGG), the Clusters of Orthologous Group (COG) and Gene Ontology (GO), ranged from 91,753 to 51,046. Among the matching sequences, 61% showed the closest match to sequences of Sesamum indicum, a species belonging to the family Pedaliaceae, order Scrophulariales (Additional file 2: Fig. S1b) . There were 46,922 transcripts with no sequence similarity to known genes (orphan genes).
Analysis of differentially expressed genes (DEGs)
Differential gene expression analysis was performed to monitor transcriptomic variations in the plant under five dehydration treatments. From the 10 libraries, 41,279,554 to 55,613,670 clean sequence reads were obtained after trimming (Additional file 3: Table S2), which were sufficient for quantitative analysis of gene expression. Based on the fragments per kilobase of transcript per million mapped reads (FPKM) of the unigenes in each sample, 18,110 unigenes were identified as DEGs, of which approximately 40% (6877) were novel.
A Venn diagram of the DEGs illustrates their occurrence in each dehydration treatment (Fig. 1b). The largest number of unique genes was found in the D20 (7065), followed by DT (1210), R (129) and D75 (153) treatments, whereas only 119 genes were present in all dehydration treatments. More than 80% of novel DEGs were either unique to D20 or shared by D20 and DT (Additional file 2: Fig. S1c). Based on their FPKM values, a random selection of 20 genes encoding enzymes, transcription factors and regulators involved in metabolism and stress responses were subjected to quantitative real-time PCR (qPCR) analysis to validate the DEGs identified by RNA-seq. The expression levels of these genes were significantly correlated with those determined by qPCR (r = 0.99, p < 0.01) (Additional file 5: Fig. S2a, Additional file 6: Table S4).
Principal component analysis of transcript profiling of dehydrated and rehydrated plants
Clustering of DEGs by expression profiles and their correlations with physiological changes during dehydration
The GO term enrichment of each cluster indicates that both metabolic processes (catabolism and biosynthesis) and developmental growth were repressed upon slight dehydration (Cluster 3), whereas responses to stimuli and redox processes/homeostasis were actively maintained until the water content decreased to 20% (Clusters 2, 4 and 5) (Fig. 3c). On the other hand, the transcripts activated at D75 (Cluster 7, translation and gene expression related) differed markedly from those activated at D20 and DT (Clusters 2 and 1, catabolic process, response to stimulus, redox process/homeostasis, localization, and regulation of seed germination).
To understand the significance of these gene expression clusters and to test the robustness of our cluster analysis, the first principle component from each stage was subjected to correlation analysis with four previously published physiological parameters . We found that the decreases in net CO2 assimilation, stomatal conductance, transpiration and Fv/Fm during dehydration showed negative correlations with Clusters 1 and 2 and positive correlations with Clusters 5 and 6 (Fig. 3d).
GO and KEGG enrichment of DEGs in dehydrated plants
Finally, we used these changes to construct a functionally organized GO/pathway term network based on the connectivity among GO terms (Fig. 4b, Additional file 8: Table S6). Numerous changes in this network were analyzed further, including primary and secondary metabolism, photosynthesis, hormone signal transduction, protein quality control and DNA repair.
DEGs related to energy and primary and secondary metabolism
The expression patterns of photosynthesis-related genes generally were in agreement with the inhibition of photosynthesis observed during D20 and DT, including genes encoding the subunits of photosystem I (PSI) and photosystem II (PSII), light-harvesting complexes II and I, cytochrome b6/f complex and electron transport. Approximately a quarter of these transcripts, including NADPH-quinone oxidoreductase subunits and ATP synthase, were up-regulated only during the early stage of dehydration (D75). However, a few transcripts encoding ferredoxin 3 (FDX3) and subunits of the PSI and PSII reaction centers (psaA and psbC) maintained elevated expression levels throughout the dehydration process (Additional file 9: Fig. S3).
With respect to secondary metabolism, flavonoid 3′-monooxygenase (F3’H) and SRG1, which are involved in the accumulation of flavonoids, and flavanone 7-O-glucoside 2”-O-beta-L-rhamnosyltransferase (I7GT1), from the phenylpropanoid metabolism pathway, were up-regulated during dehydration. Two enzymes from the glutathione cycle, glutathione reductase (GSHRC) and phospholipid hydroperoxide glutathione peroxidase (GPX4), were also increased during desiccation (Fig. 5b).
Positive or negative correlations were found between the expression levels of most genes and the previously measured accumulation of their corresponding metabolites, depending on the reaction type (catabolic or anabolic) (Fig. 5c). For example, the levels of L-valine and leucine were negatively correlated with the expression of BCAL3, which plays a role in the catabolism of these amino acids, whereas SUS2 expression was correlated positively with sucrose content but negatively with UDP-glucose (Fig. 5c).
DEGs involved in plant hormone signaling pathways
DEGs related to protein quality control and DNA repair
Base excision repair (BER), nucleotide excision repair (NER) and mismatch repair (MMR) are the three main pathways involved in DNA repair. The majority of transcripts encoding proteins that participate in NER, including DNA damage binding proteins 1 and 2 (DDB1/DDB2), decrease in DNA methylation (DDM1), replication protein A 1 and 2 (RPA1/RPA2), and DNA repair helicase (XPB1) showed prominent patterns of up-regulation in D20 and DT (Fig. 7b). In contrast, most transcripts involved in BER and MMR were down-regulated during dehydration (Fig. 7b).
Most plants are not of the resurrecting type and therefore can tolerate water loss down to approximately 30% RWC only for relatively short periods [11, 15]. Below this water content, they are unable to recover even if water becomes available again; in other words, this level is the “point of no return” for most plants. Our previous physiological and metabolic studies described numerous changes that occurred in H. rhodopensis during severe dehydration [10, 12]. One of the aims of our present study was to understand whether and when (in terms of before or around this stage of dehydration) alterations occur at the transcriptome level in association with respective physiological changes.
It was previously shown that signaling and transcriptional changes induced by dehydration in H. rhodopensis regulate growth and photosynthesis (protein kinases and early light-inducible proteins), sugar metabolism (sucrose and raffinose synthase) and activation of diverse protectants (catalases, LEA and HSPs) . Our de novo analysis not only supported these results but also identified novel genes that extend our understanding of the roles that cyclic electron flow (CEF), carbon turnover, phenylpropanoids, stress hormone signal transduction, protein quality control and DNA repair play in desiccation tolerance. The additional dehydration time points used here allowed us to outline the dynamic transcriptomic regulation during the dehydration and rehydration processes. The largest numbers of both up- and down-regulated DEGs were observed at this stage (Fig. 1). Most dehydration-repressed transcripts reached their lowest level at D20, regardless of whether these genes were transiently up-regulated during certain stages of dehydration or upon recovery (Fig. 3). A significant portion of the D20 up-regulated DEGs were annotated in the same biological processes and KEGG pathways as those up-regulated in DT, but did not always match those up-regulated in D75, as a large proportion of the DEGs showed opposing up- and down-regulation when the expression patterns of D75 and D20 were compared (Fig. 3). These data clearly indicate that desiccation tolerance mechanisms undergo reprogramming at the transcriptome level at the D20 stage. This finding extends the present understanding of desiccation tolerance mechanisms in H. rhodopensis.
Energy and primary and secondary metabolic processes
Over-reduction of the electron transport chain and carbon starvation are among the main negative impacts of drought stress that result in reduced carbon fixation. Approximately 40% of the novel unigenes reported here are related to metabolic pathways and photosynthesis (Additional file 2: Fig. S1d). This large proportion allowed us to investigate the correlations of our transcriptomic data with physiological and metabolomic data (Fig. 3d; 5c). In this analysis, we identified changes in the photosynthetic apparatus and the mobilization of free and lipid-derived fatty acids and branched amino acids to ensure the availability of energy and alternative carbon sources to allow accumulation of osmoprotectants and antioxidants during desiccation. Many of these changes are similar to those that have been observed in plant seeds or during carbon starvation in non-resurrection plants [10, 16, 17, 18, 19, 20, 21, 22].
Resurrection plants can reduce linear electron transport flux through reorganization of the photosynthetic apparatus, thus preventing oxidative stress [10, 23]. A dehydration-triggered decrease in PSII performance was accompanied by reduced expression of associated genes, which belonged mostly to clusters 4, 5 and 6 (Fig. 3d), including chlorophyll a/b-binding protein and the light-harvesting complex Lhcb1, as reported previously . In addition, the decrease in transcription of the PSI-K subunit of PSI (Additional file 9: Fig. S3) is directly responsible for the altered distribution of excitation energy and uncoupling between the photosystems , possibly explaining the observed decrease in electron flow between PSI and PSII [10, 25]. On the other hand, our present transcriptomic evidences indicating maintenance of CEF during dehydration corroborated the results of our previous study . The up-regulation of ferredoxin and NADPH-quinone oxidoreductase subunits not only suggests the presence of the chloroplastic NAD(P)H dehydrogenase complex in H. rhodopensis but also highlights its role in desiccation tolerance as a component of CEF [26, 27].
The process of lipid turnover is very important in both tolerant and sensitive plants under drought stress. The up-regulated expression of PLD during dehydration (Fig. 5a) is likely to correspond with increased glycerophosphorylethanolamine (GPE) and glycerophosphorylcholine (GPC) and decreased phosphatidylethanolamine levels, reported in our previous study . PLD activity induction by dehydration has been reported in the resurrection plant Craterostigma plantagineum , as well as in the drought-tolerant Arabidopsis ecotype Columbia . A drastic decrease in lipid levels, especially membrane lipids such as phosphatidylethanolamine, has been observed in Ramonda serbica, a close relative of H. rhodopensis  and in the African resurrection plant Sporobolus stapfianus . Thus, the reported correlation between PLD and related metabolites  (Fig. 5c), combined with the increased levels of malondialdehyde [5, 11, 30], strongly confirm the occurrence of lipid degradation in H. rhodopensis under stress. In comparison with the massive degradation of membranes in desiccation-sensitive species , these processes are quite limited in resurrection plants. Here, the turnover of damaged lipids may have a protective role, ensuring the accumulation of osmoprotectants such as GPC , signaling molecules and carbon precursors needed for biosynthetic processes during drought stress [10, 19, 33]. The up-regulation of the main enzymes responsible for fatty-acid degradation, MBCD and ACOX2 (Fig. 5a), supports the decreased free fatty acid content found during desiccation (Fig. 5c) .
Degradation of branched-chain amino acids provides an alternative carbon source in plants under unfavorable conditions [17, 21]. The up-regulation of BCAL3 and HMGL, which are involved in the degradation of valine, leucine and isoleucine (Fig. 5a), may explain the decreasing levels of these amino acids during dehydration (Fig. 5c) [10, 12].
Increased expression of CYSZ, TPIC, PGKH, HXK1, SUS2 and SPSA1 (Fig. 5a), which are related to the citric acid cycle, glycolysis/gluconeogenesis and sucrose biosynthesis, is in agreement with previous reports of increased metabolite levels during these biological processes and massive accumulation of sucrose during dehydration (Fig. 5c) [5, 10, 12].
We also found significantly elevated expression of genes involved in the synthesis of phenylpropanoids and flavonoids (Fig. 5b). I7GT1 encodes a UDP-glycosyltransferase, a type of enzyme that is related to the accumulation of p-coumaryl alcohol 4-O-glucoside, coniferin, syringin, and coniferyl and sinapyl alcohol 4-O-glucoside via glucosylation of soluble intermediates in Arabidopsis . The up-regulation of F3’H and SRG1 during dehydration (Fig. 5b) is correlated directly with the accumulation of flavonoids, including quercetin, aromadendrin, taxifolin, ephedrine, leucocyanidin, cyanidin, cis-3,4-leucopelargonidin and pelargonidin [35, 36]. These phenylpropanoids and flavonoids are common plant secondary metabolites that function both as structural and signaling molecules .
Glutathione is involved in plant desiccation tolerance as an antioxidant [11, 37, 38]. DEGs related to glutathione metabolism have been reported previously in H. rhodopensis  and another resurrection plant species from the same family Gesneriaceae, Boea hygrometrica . Over-expression of two transcripts in this study (Fig. 5b), GPX4 and GSHRC, accorded with these data and strongly confirmed the importance of glutathione as a major factor in desiccation tolerance.
Plant hormone signal transduction
The plant hormone signaling network plays an integral role in the perception of and response to unfavorable environments. Previously, transient induction of JA and ABA has been reported, reaching their highest levels at approximately 55 and 25% RWC, respectively, in the dehydrated leaves of H. rhodopensis, whereas SA accumulates to its highest level at 25% RWC and is maintained during full dehydration . The increased expression of biosynthesis- and signaling-related genes for these stress hormones contributes to plant drought tolerance by inducing biosynthesis of defense proteins and protective secondary metabolites. In this study, we found that the JA receptor COI1 was induced during dehydration (Fig. 6), but that the transcriptional activator of the JA signaling pathway, JAR1, was significantly repressed at the D20 and DT stages, indicating conversion of the JA signaling pathway under continuous stress, in accordance with the reported pattern of JA dynamics . In contrast, at the transcriptional level, ABA and SA signaling transduction is activated at D20 and DT through the induction of key transcription factors such as ABFs, GBFs and AIB factors, and NPR1, thus confirming the roles of ABA and SA as stress signaling molecules and that of SA in the antioxidant response of the plant to dehydration .
Dynamic changes of ethylene in desiccation tolerance have not been reported; however, 1-aminocyclopropane-1-carboxylic acid oxidase transcripts have been shown to accumulate during dehydration . In this study, the increased transcription of ethylene receptors (ETR1, ETR2 and EIN4) and the key transcription factor EIN3/EIL1 during the early stage of dehydration suggests a possible role of ethylene in the early response to dehydration, which is followed by repression of signal transduction in H. rhodopensis, as indicated by up-regulation of the key negative regulator CTR1 at the D20 and DT stages. Ethylene negatively regulates cell proliferation at the root apical meristem . Thus, the transient increase of ethylene biosynthesis- and signaling-related genes in H. rhodopensis may contribute to the rapid inhibition of cell expansion during dehydration.
Crosstalk among SA, JA and ethylene-dependent signaling pathways regulates plant responses to abiotic and biotic stresses, which are frequently associated with hypersensitive response (HR)-like cell death [41, 42, 43]. The final response depends on the extent of crosstalk among these molecules. Inhibition of the ethylene and JA signaling pathways under severe and full dehydration and the crosstalk of these pathways with SA signal transduction likely contribute to prevention of HR-associated cell death, which is required for desiccation tolerance in resurrection plants.
Auxins, cytokinins, GAs and BRs have been recognized as crucial signaling molecules that control plant growth and development. The positive regulators of the cytokinin signaling pathway AHP1 and AHP4 were only slightly induced in the early stage of dehydration, whereas negative regulators of auxin pathways were induced throughout the dehydration process (Fig. 6), indicating inhibition of both signal transduction processes. This finding is in agreement with the observation of suspended growth of plants during dehydration. Notably, receptor genes for auxin, GA and BRs, including TIR1, GID1B, BRL1 and BRL3, were up-regulated throughout the dehydration and rehydration processes, in parallel with IAA-acyl acid amido conjugators (group II GH3s)  and the BZR1 homolog BEH2, indicating a possible mechanism for the rapid recovery of the resurrection plant upon rehydration.
Protein-quality control and DNA repair
The ability to maintain protein and DNA integrity is indispensable to cellular survival under constant exposure to adverse stresses that may result in mutagenesis or cell death . Despite this importance, data on protein quality control and DNA repair are very limited or lacking, although recent molecular studies show their potential involvement in desiccation tolerance in the budding yeast Saccharomyces cerevisiae, Arabidopsis seeds and the resurrection plant B. hygrometrica [46, 47, 48]. The mechanisms controlling these processes, such as the unfolded protein response, aid in protein folding or in degradation of misfolded secretory proteins by producing protein-folding and other factors [49, 50]. Activation of the unfolded protein response in H. rhodopensis was reflected by up-regulation of two key regulators during D20 and DT (Fig. 7a): ERD2, a transmembrane ER lumen protein-retaining receptor that controls traffic in the Golgi body and retrograde transport to reclaim ER proteins , and ERO1, a protein associated with protein disulfide isomerase oxidoreductases, which play a key role in ER redox regulation via protein stabilization by forming disulfide bonds between structures .
The ubiquitin–proteasome system is the major pathway for proteasomal degradation of damaged and misfolded proteins. Activation of this system in D20 and DT, indicated by up-regulation of the majority of genes encoding E3 ubiquitin ligase (Fig. 7a), hypothetically results in proteomic redistribution; however, the reasons for this redistribution are unclear and remain to be explored in the future. A regulatory role of protein ubiquitination has been revealed, based on the accumulation of important metabolites during acquisition of rapid desiccation tolerance in B. hygrometrica . E3 ubiquitin ligase may cooperatively interact with HSP/C70 for quality control [54, 55]. This interaction may occur intensively in H. rhodopensis during the D20 and DT stages.
Damaged DNA can be repaired via the processes of BER, MMR, and NER. In this study, we observed predominate activation of NER pathway genes, such as DDB1/DDB2, DDM1, RPA1/RPA2, XPB1 and PCNAs , indicating the importance of proofreading DNA repair to maintain genome integrity during severe dehydration. Studies of the DNA-damage response in seeds have suggested that particular proteins are required for certain repair mechanisms during the early stage of desiccation [47, 57]. Considering the similarity of desiccation tolerance in vegetative tissues of resurrection plants and seeds of non-recalcitrant species , a dynamic mechanism for maintaining genome integrity is therefore reasonably predicted in H. rhodopensis under desiccation. NER has been found to play an important role in reversing UV-induced DNA damage in plants . Building upon known protective responses such as LEA and HSPs , this is the first report to associate NER with desiccation tolerance in resurrection plants. Further studies are needed to clarify the detailed mechanism of the DNA-repair process in H. rhodopensis under desiccation conditions.
Together, our results show that during severe dehydration, a reprograming of the transcriptome occurs that defines the desiccation tolerance and resurrection behavior of H. rhodopensis.
Materials and methods
Plant materials and stress treatments
In vitro-propagated plants of H. rhodopensis  were transferred to a standard soil–sand–gravel mixture in pots (6 cm diameter) and grown for approximately 1 year in a controlled environment at 22–24 °C, under a 16-h photoperiod, 60% relative humidity, and a photon flux density of 36 μmol m− 2 s− 1.
Dehydration was induced by withholding water from potted plants. Plants required at least 8 days to become fully desiccated. After 10 days in the fully dry state, the plants were re-watered. Rehydration was initiated by watering the plants, and full recovery took 1 week. A non-destructive method of monitoring water content based on in vivo fast fluorescence was applied to collect plant leaf samples, according to the established relationship between leaf water content and photosynthetic performance during stress and recovery . The Handy PEA (Hansatech Instrument Ltd., King’s Lynn, Norfolk, UK) device was used for measurement of the prompt chlorophyll a fluorescence of plants in the light-adapted state. Based on the SOM (self-organizing map) developed previously , we were able to collect samples at five time points corresponding to the following RWC stages: F (90% RWC), D75 (75% RWC), D20 (20% RWC), DT (6% RWC) and R (90% RWC). Mature, fully expanded leaves from the middle of rosettes of similar size and appearance were used in all measurements. At least four independent plants were used for each sampling point.
RNA extraction, cDNA library construction, and Illumina high-throughput sequencing
Plant leaves (0.4 g) were ground under liquid nitrogen to a fine powder using a cooled mortar and pestle. Samples were placed in 2-mL Eppendorf tubes, and total RNA was isolated following the protocol for the GeneMATRIX Universal RNA kit (EURx Ltd., Gdansk, Poland).
Extracted RNA was used to construct cDNA libraries for sequencing analysis. Ten cDNA libraries were prepared using the Truseq™ RNA sample preparation kit from Illumina (San Diego, CA, USA). Libraries were size-selected for cDNA target fragments of 200–300 bp on 2% low-range ultra-agarose gels followed by PCR amplification using Phusion DNA polymerase (NEB, Ipswich, MA, USA) for 15 cycles. After quantification using the TBS380 fluorometer, 150-bp paired-end RNA-seq libraries were constructed using the Illumina HiSeq 4000 platform (Illumina).
Transcriptome de novo assembly, sequence annotation, classification and alignment
The raw data were first processed using Seqprep (https://github.com/jstjohn/SeqPrep) and Sickle (https://github.com/najoshi/sickle) and the obtained high quality clean reads were subjected to de novo assembly using Trinity (https://github.com/trinityrnaseq/trinityrnaseq) . The unique assembled transcripts were then used for functional annotation using BLASTX and BLAST2GO (http://www.blast2go.com/b2ghome) . Metabolic pathway analysis was performed using KEGG (http://www.genome.jp/kegg) . Sequence alignment was performed using BLAT  with contig sequences from both this de novo assembly and published data  (doi: https://doi.org/10.1007/s00018-012-1155-6, 18_2012_1155_MOESM1_ESM.xlsx). Unigenes were considered to differ when the sequence homology was < 90% in identity and < 80% in coverage.
Differential expression analysis and functional enrichment
To identify DEGs between two different samples, the expression level of each transcript was calculated according to the FPKM method. RSEM (http://deweylab.biostat.wisc.edu/rsem/)  was used to quantify gene and isoform abundances. The R statistical software package Empirical analysis of Digital Gene Expression in R (http://www.bioconductor.org/packages/2.4/bioc/html/edgeR.html)  was utilized for differential gene expression analysis. Principle component analysis of DEGs was performed using SIMCA 13.0 (Umetrics, Satorius Stedim Biotech, Umeå, Sweden). In addition, functional enrichment analysis including GO and KEGG terms in comparison with the whole-transcriptome background was performed using a Bonferroni-corrected p-value < 0.05 threshold. GO functional enrichment and KEGG pathway enrichment were carried out using Goatools (https://github.com/tanghaibao/Goatools) and KOBAS (http://kobas.cbi.pku.edu.cn/download.php) . For visualization of the functionally grouped network of KEGG terms, the ClueGO plugin  was used with Cytoscape . To map all identified genes with altered expression in D20 and to visualize the pathway relationships among these genes, we used a general mapping method in which mapped genes represent a low percentage of all associated genes per pathway, and Bonferroni correction was omitted. Subsequently, the KEGG file for each pathway was downloaded, and all genes in the pathway were extracted from the ClueGO table. Then, network analysis was performed using Cytoscape software. The Pearson correlations between genes and metabolites were calculated using Cytoscape. The expression of each gene extracted from the ClueGO table for each pathway is expressed as the average log2 ratio for each treatment compared with the control.
All purified RNA samples described above were adjusted to the same concentration (100 ng/μL) using RNase-free ddH2O and then reverse-transcribed into cDNA using M-MLV reverse transcriptase (Promega, Madison, WI, USA), following the manufacturer’s instructions. Quantification of the transcripts was performed using SYBR® Green Realtime PCR Master Mix (TOYOBO, Osaka, Japan). Each reaction (10 μL) contained 0.4 μL of the cDNA template, 0.8 μL of the primer mixture (10 μM each, mixed to equivalent volume), 5.0 μL 2× PCR Master Mix and 3.8 μL ddH2O. Reactions were performed using a two-step method: 95 °C for 10 s, then 40 cycles of 95 °C for 10 s, 55 °C for 30 s, 72 °C for 30 s, and 95 °C for 15 s, followed by a final melt cycle from 55 °C to 95 °C. Gene-specific primers are shown in Additional file 6: Table S4. qPCR was performed for three technical replicates of each RNA sample (as well as two independent biological replicates). The normalized relative expression levels of the target genes were calculated by the 2-ΔΔCt (ΔCt = CtTarget – Ct18S, where Ct is the cycle threshold) using 18S rRNA as the internal standard.
The clustering of log-transformed expression data from each treatment was performed using Gene Cluster 3.0, with normalization of genes, the centered correlation similarity metric and complete linkage method. Clustered data were viewed, and the corresponding heatmaps were generated using JAVA TREEVIEW 1.6.4. Genes of interest were selected according to their relevance in particular biological processes or pathways, and then their expression data were pooled, log2 transformed and normalized to the untreated hydrated controls.
We thank Dr. Tao Xu (Institute of Botany, Chinese Academy of Sciences) for excellent assistance with data analysis.
This work was supported by China-Bulgaria’s Inter-Governmental S&T Cooperation project [grant number 14–10]; and Bulgarian Science Fund [grant number DNTS/China/01/7]. The funding bodies had no role in study design, data collection, analyses and interpretation, decision to publish or preparation of manuscript.
Availability of data and materials
The datasets reported in this paper are available at NCBI GEO (http://www.ncbi.nlm.nih.gov/geo/) with accession number GSE109958. All data generated or analyzed during this study are included in this article (and additional files) or available from the corresponding author on reasonable request.
DD and XD conceived and designed the study. DM was responsible for obtaining the plant materials under different treatments and extracted RNA. JL carried out the transcriptome sequencing and qPCR. JL, CL, PM and RS performed the data analysis. JL, DM, CL, PM, RS, DD and XD analyzed the results, interpreted the data and wrote the manuscript. All authors read and approved the final manuscript.
Ethics approval and consent to participate
Consent for publication
The authors declare that they have no competing interests.
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
- 8.Georgieva K, Maslenkova L, Peeva V, Markovska Y, Stefanov D, Tuba Z. Comparative study on the changes in photosynthetic activity of the homoiochlorophyllous desiccation-tolerant Haberlea rhodopensis and desiccation-sensitive spinach leaves during desiccation and rehydration. Photosynth Res. 2005;85(2):191–203.PubMedCrossRefGoogle Scholar
- 10.Mladenov P, Finazzi G, Bligny R, Moyankova D, Zasheva D, Boisson AM, Brugiere S, Krasteva V, Alipieva K, Simova S, et al. In vivo spectroscopy and NMR metabolite fingerprinting approaches to connect the dynamics of photosynthetic and metabolic phenotypes in resurrection plant Haberlea rhodopensis during desiccation and recovery. Front Plant Sci. 2015;6:564.PubMedPubMedCentralCrossRefGoogle Scholar
- 11.Djilianov D, Ivanov S, Moyankova D, Miteva L, Kirova E, Alexieva V, Joudi M, Peshev D, Van den Ende W. Sugar ratios, glutathione redox status and phenols in the resurrection species Haberlea rhodopensis and the closely related non-resurrection species Chirita eberhardtii. Plant Biol. 2011;13(5):767–76.PubMedCrossRefGoogle Scholar
- 15.Oliver MJ, Guo L, Alexander DC, Ryals JA, Wone BW, Cushman JC. A sister group contrast using untargeted global metabolomic analysis delineates the biochemical regulation underlying desiccation tolerance in Sporobolus stapfianus. Plant Cell. 2011;23(4):1231–48.PubMedPubMedCentralCrossRefGoogle Scholar
- 16.Avelange-macherel MH, Ly-Vu B, Delaunay J, Richomme P, Leprince O. NMR metabolite profiling analysis reveals changes in phospholipid metabolism associated with the re-establishment of desiccation tolerance upon osmotic stress in germinated radicles of cucumber. Plant Cell Environ. 2006;29(4):471–82.PubMedCrossRefGoogle Scholar
- 18.Dieuaide M, Couée I, Pradet A, Raymond P. Effects of glucose starvation on the oxidation of fatty acids by maize root tip mitochondria and peroxisomes: evidence for mitochondrial fatty acid β-oxidation and acyl-CoA dehydrogenase activity in a higher plant. Biochem J. 1993;296(1):199–207.PubMedPubMedCentralCrossRefGoogle Scholar
- 25.Goltsev V, Zaharieva I, Chernev P, Kouzmanova M, Kalaji HM, Yordanov I, Krasteva V, Alexandrov V, Stefanov D, Allakhverdiev SI. Drought-induced modifications of photosynthetic electron transport in intact leaves: analysis and use of neural networks as a tool for a rapid non-invasive estimation. Biochim Biophys Acta-Mol Cell Res. 2012;1817(8):1490–8.CrossRefGoogle Scholar
- 30.Solti Á, Mihailova G, Sárvári É, Georgieva K. Antioxidative defence mechanisms contributes to desiccation tolerance in Haberlea rhodopensis population naturally exposed to high irradiation. Acta Biologica Szegediensis. 2014;58(1):11–4.Google Scholar
- 33.Rakic T, Lazarevic M, Jovanovic ZS, Radovic S, Siljak-Yakovlev S, Stevanovic B, Stevanovic V. Resurrection plants of the genus Ramonda: prospective survival strategies-unlock further capacity of adaptation, or embark on the path of evolution? Front Plant Sci. 2014;4:550.PubMedPubMedCentralCrossRefGoogle Scholar
- 35.Saito Y. Bershad, Brian N, levy, Henry M. Manageability, availability and performance in Porcupine: a highly scalable, cluster-based mail service Acm Sigops Operating Systems Review. 1999;34(2):9–11.Google Scholar
- 37.Georgieva K, Rapparini F, Bertazza G, Mihailova G, Sárvári É, Solti Á, Keresztes Á. Alterations in the sugar metabolism and in the vacuolar system of mesophyll cells contribute to the desiccation tolerance of Haberlea rhodopensis ecotypes. Protoplasma. 2017;254(1):193–201.PubMedCrossRefGoogle Scholar
Open AccessThis article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.