4.1. Physiological Characterisation
Tef has previously been classified as moderately drought tolerant in comparison to species within the
Eragrostis genus [
61]. Ginbot and Farrant [
15] have confirmed that this species has some measure of tolerance to water-deficit under drought stress, with brown seeded varieties (as used in this study) being tolerant of slightly higher amounts of water loss than white seeded varieties. In the current study, similar trends were observed in pre-flowering, brown seeded tef plants, where dehydration to below 40% RWC (
Figure 1A) resulted in increased electrolyte leakage rates (
Figure 2A) and considerable evidence of subcellular damage (
Figure 3). Increased membrane permeability with continuous dehydration stress has been linked to the enhanced synthesis of reactive oxygen species (ROS), a consequence of metabolic processes in chloroplasts, mitochondria and peroxisomes in particular, which can cause the breakdown of proteins, membrane lipids and photosynthetic pigments that function in maintaining cell membrane stability [
62,
63]. Photosynthesis is particularly susceptible to excess ROS formation under water deficit conditions and this has frequently been cited as a primary cause of damage and resultant plant death in most species [
64,
65]. The sensitivity of PS II activity to abiotic and biotic factors has resulted in the use of chlorophyll fluorescence, and particularly the measure of quantum efficiency of PS II (
Fv/
Fm) as an indicator of how plants respond to environmental change. Data from
Figure 2B show maintenance of
Fv/
Fm at values indicative of healthy, non-stressed leaves until 50–55% RWC, with a sharp drop (values declining below 0.4, indicative of possible damage related to photosynthetic shutdown) below 30% RWC. Ultrastructural analysis showed considerable decline in vacuolar area as water was lost from tissues, with some evidence of plasmalemma withdrawal and autophagosome formation upon dehydration to 50% RWC (
Figure 3B). Autophagy has been associated with cellular survival by removal of damaged organelles and cellular toxins and recycling of the breakdown products for the maintenance of cellular homeostasis. Furthermore, it has been proposed as being essential for drought stress tolerance [
66,
67]. We propose that tef is able to survive loss of up to 50% RWC, in part, due to such a strategy. However, drying to lower RWC, suggested increased evidence of subcellular damage, including breakage of cell walls, plasmalemma and loss of integrity of organelles (
Figure 3C,D) all further signs of stress-induced injury.
In summary, physiological studies performed here indicate that six week-old plants from a brown seeded tef variety are able to tolerate drying to ca 50% RWC (loss of 1.5 g H2O) before irreversible damage is initiated. We were thus interested in understanding the nature of protection afforded during initial drying to 50%, by investigating the tef proteome changing in response to dehydration stress.
4.2. Tef Proteomics
The starting point of the iTRAQ analysis was the examination and refinement of the list of peptides generated from database searching as opposed to the list of generated proteins. This is not an uncommon approach and has been used by many researchers in the field of mass spectrometry-based proteomics [
50,
68,
69,
70]. A potential concern with working with a list of peptides instead of proteins is the challenge of protein inference [
71], where the generated list contains both unique and non-unique (shared) peptides matched against the chosen database for protein identification. This concern is adequately addressed by using appropriate FDR thresholds and employing stringent estimation of error rates, so that only valid peptide identities meeting the FDR threshold requirements are detected and used for subsequent protein analysis [
69,
72]. Furthermore, the analysis of both unique and non-unique peptide mass spectra scans that meet FDR thresholds would be more representative of the proteins changing in a particular study. Interestingly, a total of 57 out of the 211 proteins (27%) found to be differentially regulated within the TE dataset (
Table 1 and
Table 2) were spliced variants arising from the alternative splicing of 25 potential splice events (genes). During this regulatory mechanism, primary transcripts or precursor-mRNAs with introns undergo alternative splicing to produce multiple transcripts from a single gene within the genome by using differential splice sites [
73]. In this regard, the functional complexity of the transcriptome and diversity of the proteome are increased between plant cells and tissues [
73,
74], particularly during plant development and in response to environmental stimuli, such as biotic and abiotic stress conditions [
75,
76]. In the TEU differentially regulated datasets (
Tables S2 and S3) and MU differentially regulated datasets (
Tables S4 and S5), however, no occurrences of spliced variants were present, presumably because only uniquely-matched peptides were used for protein identification, resulting in only one definitive protein entity per entry. Because iTRAQ experiments on the whole do not usually produce large amounts of peptide reads per protein [
69], the use and manipulation of only uniquely scanned peptides for protein identification has been shown to drastically limit the number of confidently proteins identified [
71,
72]. This is especially evident by the marginal difference observed in the amount of proteins identified between the TE and TEU differentially regulated datasets, 211 and 111 proteins, respectively. Because tef is considered to be a non-model crop species whose genome has only been recently sequenced [
18], the amount of annotated information therein cannot compare to that of model plant organisms. It is important to note that the tef genome and transcriptome have only been moderately-annotated, and this consequently, would lead to not all tef proteins being identified during database searching (as shown in
Table 1 and
Table 2;
Tables S2 and S3). Nevertheless, a significant amount of proteins within the TE and TEU datasets do contain protein annotations and therefore can be used to make protein inferences through bioinformatics analyses, while those unidentified proteins may lead to discovery of some unique new targets within the tef genome.
It could be suggested that a ‘cross-species identification’ approach would be better for non-model plant systems such as tef, where a generic (non-specific plant species) but well-annotated database is used for protein identification [
77,
78,
79]. This would potentially increase the amount of annotated and identified proteins, as in the case with the proteins identified by use of the
Liliopsida database (the MU dataset) available from UniprotKB, where 4328 tef proteins were identified during database searching, and 174 proteins were found to be differentially regulated (
File S3, Tables S4 and S5). Although this approach is widely used for non-model plant systems [
77,
78] such as tef and many others [
80,
81], using the same approach is not ideal as the number and confidence of identified proteins is reduced [
79]. The MU dataset was generated using only uniquely scanned peptides during database searching and contained more proteins with usable descriptions and annotations for bioinformatics inference (
Tables S4 and S5). The use of the TE database, however, provided identification of 5727 tef proteins in total (File S1), of which 211 were differentially regulated. The difference in the total amount of proteins detected can be explained by the fact that either some species-specific proteins will not be present during cross-species identification or those homologous proteins that are present will show small evolutionary differences in their sequences [
79]. Thus, the use of a very specific but moderately-annotated database (the TE database), would detect more proteins present, potentially highlight more proteo-bioinformatics changes that are unique to the organism under study, and also improve annotation and curation within the existing tef database.
4.3. Tef Protein Validation
Biological validation of the upregulated protein FBA by Western blotting, showed increased protein accumulation and band intensity with dehydration stress to 50% RWC (
Figure 4). Although FBA displayed negligible increases in fold change in the iTRAQ data (
Table 1 and
Table S2, fold change = 1.02), statistical testing based on
p-value showed it to be highly significant (FBA,
p-value = 0.005 in
Table 1 and
Table S2). Since an overall increase in protein accumulation is observed with dehydration stress, this result supports the iTRAQ findings and show that protein change is due to a biological consequence and not experimental variation.
Proteins tested for biological validation by enzymatic methods, MDHAR, POX and FBA, all showed increased enzyme activities at 60–65% RWC in response to dehydration stress (
Figure 5). FBA catalyses the reversible conversion of glyceraldehyde-3-phosphate and dihydroxyacetone phosphate to fructose-1, 6-bisphosphate during glycolysis/gluconeogenesis or in the reaction where erythrose-4-phosphate and dihydroxyacetone phosphate is converted to sedoheptulose-1,7-bisphosphate in the Calvin cycle [
82,
83]. Furthermore, FBA has been classified as one of the six non-regulated enzymes in the Calvin cycle that have been suggested to have a potential role in controlling photosynthetic carbon flux through the Calvin cycle [
83]. A significant increase in FBA activity was observed in tef at 60–65% RWC (
Figure 5B). It has been proposed that increased activity of FBA may function in the regeneration of ribulose 1,5 bisphosphate and increased CO
2 fixation, contributing to enhanced photosynthesis, increased growth rates and biomass yields [
83]. Furthermore, an increase in FBA activity has been observed in stress response for various other crop plants such as rice, in response to drought stress and increased salinity [
84,
85]; wheat seedlings, in response to anaerobic conditions [
86,
87]; wheat roots, in response to increased aluminium concentrations[
87,
88] and in Indian mustard, in response to increased cadmium concentrations [
89].
The stress responsive antioxidant enzymes known to offer protection against free radical accumulation, MDHAR and POX [
90,
91,
92,
93], displayed a large increase in enzymatic activity at 60–65% RWC (
Figure 5A,C). A second increase in enzyme activity was observed by both MDHAR and POX at 35–40% and 25–30% RWC, respectively, towards the latter stages of dehydration (
Figure 5A,C). These results support protein presence and accumulation according to iTRAQ findings for MDHAR (
Table 1 and
Table S2, fold change value = 1.03,
p-value = 0.027 and 0.04) and POX (
Table 1, fold change value = 1.08,
p-value = 0.002;
Table S4, fold change = 1.02,
p-value = 0.011) and also confirms that the large increases in quantitative expression observed in these protein isoforms in
Table 1 (fold change = 6.54,
p-value = 0.006) for MDHAR and POX (fold change = 6.82,
p-value = 0.008) are due to a biological change in response to dehydration stress and not experimental error. An increase in POX activity has been related to many oxidative and abiotic stresses [
94,
95], particularly in response to dehydration stress conditions in the crop plants wheat [
93,
96], oilseed rape [
97], sunflower [
98], horse gram beans [
99] as well as in response to salt stress in fox-tail millet and rice [
55,
95]. The increased production of free radicals as a consequence of stress conditions has been proposed to be the main reason for membrane lipid peroxidation, whereby the extent of peroxidation-induced damage is regulated by the antioxidative peroxidase enzyme system [
92,
95]. This could be due, in part, to the ability of POX acting on increased levels of H
2O
2 in cells as dehydration stress proceeds, even towards the final stages of dehydration stress (25–30% RWC) (
Figure 5C). The free radical, H
2O
2, has been postulated to have a dual role in plant cells, by either acting as a signalling molecule at low concentrations during non-stress conditions or as an activator of programmed cell death (PCD) at high concentrations during stressed conditions [
90,
100]. Dehydration to 50% RWC resulted in evidence of autophagy (
Figure 3B) and increased electrolyte leakage measurements (
Figure 2A) at RWCs below this, suggesting increased membrane damage. This is perhaps due to the extenuating effects of H
2O
2 build-up.
4.4. Tef Bioinformatic Analysis
Functional enrichment analysis of the GO-terms of tef proteins regulated in response to dehydration stress yielded a wealth of protein ontological information (
Table S6;
Figure 6A). Monodehydroascorbate reductase (NADH) activity was the most significantly changed GO-term in the category MF (
Figure 6A; 8.1% protein sequences). MDHAR was also significantly increased in quantitative expression during iTRAQ analysis in response to dehydration stress in the TE (
Table 1,
p-value = 0.006) and TEU datasets (
Table S2,
p-value = 0.027) and showed a considerable increase in enzymatic activity at low RWCs (35–40%) (
Figure 5A). MDHAR is one of the key enzymes involved in ascorbate reduction [
101] and functions in reducing the oxidised form of ascorbate (monodehydroascorbate) before being returned to the ascorbate pool [
91,
101]. MDHAR has been proposed to be an indicator of oxidative stress within plant tissues, playing an important role against accumulation of ROS due to increasing stress conditions [
90,
91]. This suggests that the generation of ascorbate as well as the regulation and maintenance of the ascorbate–glutathione cycle are important in the response to initial dehydration stress.
The Rab family of cellular processes active in the regulation of vesicular membrane traffic [
102] and regulatory membrane protein transport processes were equally over-represented in response to dehydration stress (
Figure 6A). The flow of membrane constituents between endomembrane structures and the plasmalemma is critical for the maintenance of cellular homeostasis in response to signal transduction [
103]. This is also important in autophagy, which has been linked to the restoration and maintenance of cellular homeostasis through the recycling and removal of damaged cellular constituents through protein degradation [
66,
67], where drought has been reported to induce PCD [
67]. Furthermore, GO-terms allocated to biological processes responsible for regulating membrane trafficking and the flow of proteins and other macromolecules to numerous endpoints inside and outside the cell through a signalling cascade[
104,
105] were over-represented in response to dehydration stress (
Figure 6A). These Rab family of small GTP-binding proteins function as molecular alterations that cycle between ‘active’ and ‘inactive’ states within the cell through the binding and hydrolysis of GTP [
105], thereby controlling the endocytic network in plants [
106]. Interestingly, the stress-inducible small GTP-binding protein Rab7 gene (
PgRab7) isolated from
Pennisetum glaucum, a relatively drought-stress tolerant food grain crop grown in India, has been reported to increase tolerance to abiotic stresses such as drought and increased salinity in transgenic tobacco[
106]. Similarly, the Rab7 gene (
TaRab7) isolated from wheat leaves infected with the wheat stripe rust pathogen (
Puccinia striiformis f. sp.
tritici), was proposed to play an important role in early stages of wheat-stripe rust fungus interaction and stress tolerance [
107]. In tef, the regulation of autophagy with dehydration stress may enhance drought stress tolerance until plant viability is compromised and PCD pathways are triggered.
During dehydration stress, tef responses to biotic challenges such as fungal or bacterial infections are also important, as the GO-terms response to symbiont and symbiotic fungus and regulation of symbiosis encompassing mutualism through parasitism were highly over-represented (
Figure 6A; 8.1% and 11.3% protein sequences, respectively). Although tef has been proposed to be relatively resistant to damage by insects or competition from weeds [
108], at least 22 species of fungi and three pathogenic nematodes have been previously associated with tef [
38,
108]. The GO-term pentose metabolic process was also significantly over-represented in response to dehydration stress (
Figure 6A; 9.7% protein sequences). The pentose phosphate pathway has been reported to have a dual role in oxidative stress response in plants [
109]. Firstly, by providing an available source of soluble-sugars that can either be involved in ROS-producing metabolic pathways [
109,
110] or, alternatively, by being involved in the active production of NADPH, a major co-factor required in the antioxidant ascorbate-glutahione cycle [
90,
109]. In addition, these soluble sugars have been proposed to act as nutrient and metabolite signalling molecules that activate specific signalling pathways leading to imperative gene modification and proteomic changes in response to a number of stresses [
109].
A substantial amount of GO-terms were enriched in tef downregulated proteins (
Table S6B;
Figure 6B). The functional enrichment of GO-terms found to be over-represented in downregulated proteins, were commonly linked to quinone cycling in the plastoquinone pool during oxidative phosphorylation (
Figure 6B). The complexes NADH dehydrogenase and NAD(P)H dehydrogenase, both function in reducing plastoquinones during the flow of electrons when ATP is generated [
111,
112]. While NADH dehydrogenase functions in cellular respiration in the mitochondria [
112], NAD(P)H dehydrogenase is localised in the thylakoid membranes of chloroplasts, participating in cyclic electron transport reactions around photosystem I and chlororespiration (interactions linking respiratory electron transport chain and photosynthetic electron transport chain in thylakoid membranes of chloroplasts) [
113,
114]. NAD(P)H, in particular, has been proposed to lessen oxidative stress in plants [
114]. Increased supplying of ATP for photosynthesis has been reported during environmental stress conditions, particularly during drought stress [
115]. However, since photosynthetic metabolism under water-deficit stress is reported to be responsible for the production of large amounts of free radicals [
90], these processes, in effect, are decreased in tef in an attempt to perhaps minimise ROS production. In further support that reduced ROS production is important in the tef dehydration stress response, GO-terms involved in photosynthetic processes such as light harvesting and chlorophyll binding as well as GO-terms linked to ROS-producing processes through the generation of additional ATP, such as the transfusion of solutes in the form of cations and protons across membranes, were over-represented in downregulated proteins (
Figure 6B).
The categories, transport and response of metal ions in the form of manganese, were well over-represented in downregulated proteins (
Figure 6B). The positively charged micronutrient, manganese, is required during the splitting of water in photosystem II, when photosynthesis occurs [
116,
117] and has been reported to play important roles as a co-factor and activator of enzymes in various sub-cellular compartments [
103,
116]. To avoid toxicity within cell tissues, cytosolic manganese concentrations need to be kept low [
116] and are usually transported out of the cytosol by metal transporters where they are either localised to the plant cell membrane or to the vacuolar membrane where metals are sequestered into large moderately inert compartments [
117]. If manganese concentrations are not carefully monitored in plant cells, toxicity is usually indicated by chlorosis, brown specks, necrosis and crinkled leaves, which arise due to the inhibition of chlorophyll synthesis [
117]. The disruption of manganese ion transport and homeostasis and consequent decreased protein abundance in tef, comes as no surprise in response to dehydration stress as photosynthetic potential has been shown to decrease at water contents below 55% RWC (
Figure 2B). The decrease in photosynthesis and inhibition of chlorophyll synthesis, would ultimately lead to increased manganese concentrations and toxicity within tef plant cells due to metal transport and cellular manganese homeostasis disruption.
Potential modification of the cell wall, particularly in the form of the terms cellulase activity, cellulose catabolism, beta-glucan catabolism and cell wall modification involved in multidimensional cell growth, were over-represented in tef downregulated proteins (
Figure 6B). The effect of cell wall re-structuring and modification during stress conditions is a common phenomenon in plant cells [
118] as a consequence of turgor loss during dehydration stress [
119,
120]. Many plants curtail the growth of their stems and leaves when subjected to low water potential [
121] and continue to elongate the root tissues for deeper soil penetration and water mining as a result of adapting to drought conditions [
120,
121]. Previous observations in tef with regards to increased primary root lengths and decreased shoot growth in response to drought conditions have been reported [
13] and have been proposed to be an adaptive morphological response of tef in water-limiting environments [
13]. Lastly, the GO-term, sucrose-phosphate synthase activity, was over-represented in downregulated proteins (
Figure 6B). Sucrose phosphate synthase (EC 2.4.1.14) plays an important role in the synthesis of sucrose using substrates derived from glycolysis such as fructose-6-phosphate and UDP-glucose. In correlation to being functionally enriched in downregulated tef proteins (
Figure 6B), the enzyme has been previously shown to decrease in activity in the leaves of other C4 species as well, such as maize [
122] and sugarcane [
123], in response to dehydration stress. The decline in sucrose accumulation has been proposed to be due to the decline in readily available photosynthetic triose phosphate, which ultimately leads to a decline in the enzyme activity of sucrose phosphate synthase [
124].