Deficiency in plastidic glutamine synthetase alters proline metabolism and transcriptomic response in Lotus japonicus under drought stress


  • Pedro Díaz,

    1. Departamento de Bioquímica Vegetal y Biología Molecular, Facultad de Química, Universidad de Sevilla, Calle Profesor García González 1, 41012-Sevilla, Spain
    2. Laboratorio de Bioquímica, Departamento de Biología Vegetal, Facultad de Agronomía, Av. Garzón 780 CP12900, Montevideo, Uruguay
    Search for more papers by this author
    • These authors contributed equally to this work.

  • Marco Betti,

    1. Departamento de Bioquímica Vegetal y Biología Molecular, Facultad de Química, Universidad de Sevilla, Calle Profesor García González 1, 41012-Sevilla, Spain
    Search for more papers by this author
    • These authors contributed equally to this work.

  • Diego H. Sánchez,

    1. Max Planck Institute for Molecular Plant Physiology, Wissenschaftspark Golm, Am Mühlenberg 1, Potsdam-Golm, D-14476, Germany
    Search for more papers by this author
  • Michael K. Udvardi,

    1. The Samuel Roberts Noble Foundation, 2510 Sam Noble Pky., Ardmore, OK 73401, USA
    Search for more papers by this author
  • Jorge Monza,

    1. Laboratorio de Bioquímica, Departamento de Biología Vegetal, Facultad de Agronomía, Av. Garzón 780 CP12900, Montevideo, Uruguay
    Search for more papers by this author
  • Antonio J. Márquez

    1. Departamento de Bioquímica Vegetal y Biología Molecular, Facultad de Química, Universidad de Sevilla, Calle Profesor García González 1, 41012-Sevilla, Spain
    Search for more papers by this author

Author for correspondence:
Antonio J. Márquez
Tel: +34 954 557145
Fax: +34 954 626853


  • The role of plastidic glutamine synthetase (GS2) in proline biosynthesis and drought stress responses in Lotus japonicus was investigated using the GS2 mutant, Ljgln2-2.
  • Wild-type (WT) and mutant plants were submitted to different lengths of time of water and nutrient solution deprivation. Several biochemical markers were measured and the transcriptional response to drought was determined by both quantitative real-time polymerase chain reaction and transcriptomics.
  • The Ljgln2-2 mutant exhibited normal sensitivity to mild water deprivation, but physiological, biochemical and massive transcriptional differences were detected in the mutant, which compromised recovery (rehydration) following re-watering after severe drought stress. Proline accumulation during drought was substantially lower in mutant than in WT plants, and significant differences in the pattern of expression of the genes involved in proline metabolism were observed. Transcriptomic analysis revealed that about three times as many genes were regulated in response to drought in Ljgln2-2 plants compared with WT.
  • The transcriptomic and accompanying biochemical data indicate that the Ljgln2-2 mutant is subject to more intense cellular stress than WT during drought. The results presented here implicate plastidic GS2 in proline production during stress and provide interesting insights into the function of proline in response to drought.


Proline is one of the most common compounds produced in plant cells in response to different kinds of abiotic stress (Szabados & Savouré, 2010). This molecule is suitable for osmotic adjustment in cells because it can accumulate at high concentrations without interfering with cell metabolism (Bray, 1993). Proline has been suggested to play different roles in response to abiotic stress: a molecular chaperone that can stabilize the structure of proteins; a cellular pH buffer; a nitrogen and carbon source to be used in stress recovery; and a scavenger of hydroxyl radical or singlet oxygen (Verbruggen & Hermans, 2008; Szabados & Savouré, 2010). The accumulation of proline under stress conditions may also help to avoid photoinhibition of the photosystems (Hare & Cress, 1997).

Proline accumulation under stress results from a stimulation of proline biosynthesis, as well as an inhibition of its oxidation (Szabados & Savouré, 2010). It is mainly synthesized from glutamate by the sequential action of pyrroline-5-carboxylate synthetase (P5CS) and pyrroline-5-carboxylate reductase (P5CR), with the former being a rate-determining step of the pathway (Székely et al., 2008). This pathway is present in both the cytosol and the chloroplast (Szabados & Savouré, 2010). Alternatively, proline can be produced from ornithine by the action of ornithine-δ-aminotransferase (OAT), following the conversion of arginine into ornithine by the action of arginase (Delauney et al., 1993). The importance of this alternative pathway for proline synthesis is controversial. In some plants, the arginase/OAT route seems to have an important role in stress-induced proline production (Xue et al., 2009 and references therein). However, analysis of Arabidopsis thaliana OAT knockout mutants, which show no alteration in proline levels, suggests that the main role of this enzyme is catabolic, namely in the recycling of nitrogen from arginine (Funck et al., 2008). It is possible that OAT may have a biosynthetic role only under particular conditions or developmental stages, as in A. thaliana where it is induced in seedlings, but not in adult plants, in response to salt stress (Roosens et al., 1998), and in Brassica napus, where it is induced only after prolonged and severe osmotic stress (Xue et al., 2009). The degradation of proline occurs in the mitochondria and involves oxidation to pyrroline-5-carboxylate (P5C) by proline oxidase (also called proline dehydrogenase, PDH), and subsequent conversion into glutamate by pyrroline-5-carboxylate dehydrogenase (P5CDH) (Székely et al., 2008 and references therein). Reciprocal regulation of P5CS and P5CDH during stress acclimation and recovery avoids the formation of a futile cycle (Kiyosue et al., 1996).

The accumulation of high levels of proline is possible only if a sufficient amount of its main precursor, glutamate, is available. This means either an increase in the flux through the glutamine synthetase/glutamate synthase (GS/GOGAT) cycle, or the conversion of α-ketoglutarate into glutamate by the action of glutamate dehydrogenase (GDH). Although an increase in the aminating GDH activity has been detected in salt-stressed rice roots and wheat leaves (Lutts et al., 1999; Wang et al., 2007), Brugière et al. (1999) pointed out a central role for cytosolic GS in the biosynthesis of proline via a constant pool of glutamate under stress conditions. Accordingly, in some plants, increases in GS and GOGAT activity have been described in response to salt and water stress (Berteli et al., 1995; Bauer et al., 1997; Borsani et al., 1999; Díaz et al., 2005).

The two major isoforms of GS, cytosolic GS (GS1) and plastidic GS (GS2), seem to play different roles during stress. GS1 has been implicated in the production of proline in the phloem (Brugière et al., 1999) and/or in the remobilization of nitrogen during chronic water stress (Bauer et al., 1997), whereas GS2 is associated with stress tolerance (Kozaki & Takeba, 1996; Hoshida et al., 2000). Published data on the response of GS to osmotic stress are not entirely consistent, however. Most of the work carried out on leaves has not determined the specific contributions of GS1 and GS2 to total GS activity (with few exceptions, such as Lutts et al., 1999; Santos et al., 2004) and, depending on the plant studied, total GS activity may decrease, increase or be unaffected by osmotic stress (Bernard & Habash, 2009).

Previous studies on Lotus species have identified proline levels as an indicator of drought and salinity stress, and genes linked to proline metabolism are typically induced by these stresses (Díaz et al., 2005; Sanchez et al., 2010). The species Lotus japonicus is a model legume (Handberg & Stougaard, 1992) with a draft genome sequence and numerous resources for functional genomics (Márquez, 2005; Udvardi et al., 2005; Sánchez et al., 2008; Guether et al., 2009; Omrane et al., 2009). In the past, we have isolated and characterized at the molecular level two different L. japonicus mutants that lacked plastidic GS2 activity, but showed normal levels of cytosolic GS activity (Orea et al., 2002; Betti et al., 2006). One of these mutants, named Ljgln2-2, showed normal levels of Gln2 mRNA, but sequencing of the mutant Gln2 cDNA revealed the presence of a single point mutation leading to L278H amino acid replacement (Betti et al., 2006). Moreover, it was also shown that the mutant GS2 protein was unable to acquire a proper quaternary structure and was rapidly degraded. Despite the lack of plastidic GS2, the mutant showed normal growth relative to the wild-type (WT) when grown in a CO2-enriched atmosphere that suppresses photorespiration. Therefore, the Ljgln2-2 mutant represents an ideal system to reveal the specific contribution of each GS type (plastidic or cytosolic) to different cellular processes.

In this article, we make use of the Ljgln2-2 mutant to determine the role of GS2 in proline metabolism in L. japonicus during drought. We provide data to show that proline levels and the expression of genes for different pathways of proline metabolism are substantially affected as a consequence of the lack of plastidic GS2. These results, together with comparative analysis of WT and mutant transcriptomes, delineate an important role for plastidic GS2 in proline biosynthesis and in the response of the plant to water deprivation.

Materials and Methods

Plant growth and drought treatments

Lotus japonicus (Regel) Larsen cv. Gifu was initially obtained from Professor Jens Stougaard (Aarhus University, Aarhus, Denmark) and then self-propagated at the University of Seville. The Ljgln2-2 mutant was isolated from photorespiratory mutant screening, as described previously (Orea et al., 2002; Márquez et al., 2005; Betti et al., 2006). The mutant progeny of two consecutive backcrosses with WT were used for the present work. WT and mutant seeds were scarified and surface-sterilized, germinated in 1% agar Petri dishes, and transferred to pots using a 1 : 1 (v/v) mixture of vermiculite and sand as solid support. Five seedlings were planted in each pot and grown during 35 d in a growth chamber under 16 h : 8 h day : night, 20 : 18°C, with a photosynthetic photon flux density of 250 μmol m−2 s−1 and a constant humidity of 70%. CO2 was automatically injected to a final concentration of 0.7% (v/v) to allow for normal growth of the Ljgln2-2 mutant in a photorespiration-suppressed atmosphere. Non-nodulated plants were watered with Hornum nutrient solution, containing 5 mM NH4NO3 and 3 mM KNO3 (Handberg & Stougaard, 1992). Drought was applied by withholding irrigation for the reported period of time, and sample plants or leaves were harvested for further analysis. Each biological replicate consisted of a pool of tissue from five plants that were grown in the same pot.

Measurement of leaf water content

The water status of the plants was expressed as the relative water content (RWC), calculated from the fresh weight (FW), dry weight (DW) and turgid weight (TW) of detached trefoils as follows: RWC (%) = 100 × (FW − DW)/(TW − DW). In some cases, the percentage of hydric deficit (HD) was used, defined as 100 – RWC (%). The capacity of detached leaves to gain water was expressed as the water gain (WG), defined as: WG (%) = 100 × (TW − FW)/FW. TW was obtained after incubation of the detached trefoil for 8 h in water in a closed Petri dish.

For experiments with detached leaves (Fig. 1d), trefoils from normally watered plants were harvested and immediately weighed, and then gradually dehydrated by incubating at room temperature (with a relative humidity of 45%), and weighed again at designated time intervals before and after rehydration with water for 8 h.

Figure 1.

 Response to drought stress of wild-type (WT) and Ljgln2-2 mutant Lotus japonicus plants. (a) WT (white bars) and Ljgln2-2 (grey bars) mutant plants were submitted to drought stress for the number of days indicated and the leaf relative water content (RWC) was determined as described in the Materials and Methods section. (b) WT (white bars) and Ljgln2-2 (grey bars) mutant plants were submitted to drought stress for the number of days indicated and tested for water gain (WG) ability (see Materials and Methods section) immediately after stress. (c) WT (white bars) and Ljgln2-2 (grey bars) mutant plants were submitted to drought stress for the number of days indicated and tested for WG ability after 1 d of water resupply. (d) Detached leaves of WT and mutant plants were assayed for their capacity to gain or lose water before (black bar), during (light grey bar) and after (dark grey bar) dehydration at room temperature, as described in the Materials and Methods section. Data are the mean ± SD of three independent biological replicates. *Indicates significant difference between WT and Ljgln2-2 determined by Tukey’s test (< 0.05).

RNA extraction, quantitative real-time polymerase chain reaction (qRT-PCR) analysis and GeneChip hybridization

Leaf material was immediately frozen in liquid nitrogen after harvest, homogenized with a mortar and pestle, and kept at −80°C until use. Three independent biological replicates were used for the transcriptomic and qRT-PCR analyses (two for the transcriptomic analysis of drought-stressed plants). Total RNA was isolated using the hot borate method (Sánchez et al., 2008). The integrity and concentration of the RNA preparations were checked using an Experion bioanalyzer (Bio-Rad, with RNA StdSens chips and a Nano-Drop ND-1000 (Nano-Drop Technologies,, respectively.

For the transcriptomics experiments, RNA samples were labelled using the One-Cycle Target Labelling Kit (Affymetrix,, hybridized to the Affymetrix GeneChip® Lotus1a520343 and scanned according to the manufacturer’s instructions. qRT-PCR analysis was carried out essentially as described in Sánchez et al. (2008) using 2 × SensiMix Plus SYBR (Quantace, and an ‘Eppendorf Mastercycler EP’ thermal cycler (Eppendorf, A list of all the oligonucleotides used in this work is provided in Supporting Information Table S1. MIAME compliant data are deposited at Array Express ( as E-MEXP-2690.

Measurement of enzyme activities and other analytical determinations

Aliquots from the homogenized leaf powder stored at −80°C were resuspended in extraction buffer (5 ml g−1) and used for Fd-GOGAT and GS biosynthetic activity determinations, as described previously (Pajuelo et al., 1997; Betti et al., 2006).

Proline was extracted from 200 mg of leaf tissue with a mixture of methanol–chloroform–water (12 : 5 : 1) and quantified according to Borsani et al. (1999). Total free amino acids were extracted from 200 mg of leaf tissue essentially as described by Borsani et al. (1999) and quantified with ninhydrin reagent. Lipid peroxidation was expressed as thiobarbituric acid-reactive species (TBARS) as described by Borsani et al. (2001). Chlorophyll a (Chla) and Chlb contents were determined as described by Borsani et al. (1999). Protein concentration was determined using the Bio-Rad protein assay (Bio-Rad, using bovine serum albumin (BSA) as standard.


Response to drought stress of the Ljgln2-2 mutant

An initial water deprivation experiment was carried out in order to compare the response to drought stress of WT plants and the Ljgln2-2 mutant lacking plastidic GS2. After 4 d without water, both plant genotypes showed the same HD (c. 40% HD, Table 1). As expected, the GS activity of Ljgln2-2 was c. 15–20% that of WT, with residual GS activity representing the activity of cytosolic isoforms (Orea et al., 2002; Betti et al., 2006). Drought stress did not affect the total GS activity of either mutant or WT plants, suggesting that plastidic GS2 and cytosolic GS1 were not modulated by water stress, and that the cytosolic GS isoforms were not able to compensate for the absence of the plastidic isoform in L. japonicus. Therefore, Ljgln2-2 mutants proved to be a useful tool to examine the specific consequences of plastidic GS2 deficiency in relation to water stress, a topic which has not been studied previously.

Table 1.   Comparison of different parameters of control and drought-stressed wild-type (WT) and Ljgln2-2 mutant Lotus japonicus plants
  1. For each determination, leaves from plants under normal watering or 4 d of drought stress were used. Numbers in bold indicate statistically significant difference between WT and mutant (< 0.05).

  2. *, Significant difference between control and drought treatment (< 0.05). **, Significant difference at < 0.1.

  3. Values are the mean ± SD of at least three different biological replicates. Statistical analysis was carried out using Tukey’s test.

  4. Chl, chlorophyll; DW, dry weight; GS, glutamine synthetase; GOGAT, glutamate synthase; TBARS, thiobarbituric acid-reactive species.

Hydric deficit (%)WT14.3 ± 2.5*43.1 ± 2.6*
Ljgln2-215.0 ± 5.5*43.9 ± 5.3*
GS activity (mU mg−1)WT633 ± 60671 ± 78
Ljgln2-290 ± 16124 ± 21
GOGAT activity (mU mg−1)WT134 ± 15119 ± 10
Ljgln2-2128 ± 19125 ± 16
Total protein (mg g−1 DW)WT115.0 ± 25.0106.9 ± 15.2
Ljgln2-2112.0 ± 17.0104.7 ± 13.8
Amino acids (μmol g−1 DW)WT250.3 ± 61.9295.8 ± 68.0
Ljgln2-2329.5 ± 136.1335.3 ± 123.8
Chla (mg g−1 DW)WT6.54 ± 0.585.79 ± 1.10
Ljgln2-26.16 ± 1.995.60 ± 0.60
Chlb (mg g−1 DW)WT3.14 ± 0.182.63 ± 0.60
Ljgln2-23.05 ± 0.172.48 + 0.72
a : b ratioWT2.14 ± 0.242.09 ± 0.20
Ljgln2-22.20 ± 0.212.11 ± 0.30
TBARS (nmol g−1 DW)WT162.6 ± 28.4*263.5 ± 22.9*,**
Ljgln2-2166.6 ± 30.5*307.1 ± 23.3*,**

Contrary to the situation reported for other Lotus species (Borsani et al., 1999), GOGAT enzyme activity did not change during the drought stress treatment in either the Ljgln2-2 mutant or WT (Table 1). Total protein and free amino acid levels remained unchanged in drought-stressed plants, demonstrating that water deprivation for 4 d did not result in net protein degradation (Table 1). Likewise, no significant alterations in chlorophyll content were observed. However, an increase in TBARS was detected, with a slight but significant difference between the Ljgln2-2 mutant and WT (Table 1), indicating a higher level of oxidative membrane damage in drought-treated mutant plants.

In a subsequent experiment, WT and Ljgln2-2 mutant plants were subjected to 5 d of water deprivation followed by re-watering, and RWC of leaves (%) was determined. RWC of Ljgln2-2 mutant plants decreased at a similar rate to that of WT during the first 5 d of water deprivation (Fig. 1a). Although the plants remained fully rehydratable after 5 d of treatment, indicating that drought stress was still reversible, longer periods of water deprivation (6 and 7 d) resulted in a large difference in the rehydration ability of Ljgln2-2 mutant leaves. Fig. 1(b) shows that WG ability after 6 and 7 d of drought stress treatment was reduced significantly in mutant plants relative to WT, a result that was also reproduced after 1 d of water resupply to the plants (Fig. 1c). A deficiency in rehydration of Ljgln2-2 plants relative to WT was confirmed using detached leaves that had been dehydrated for 8 h at room temperature (Fig. 1d).

Lower proline content and a change in the pattern of expression of genes for proline metabolism in the Ljgln2-2 mutant

As the cytosolic GS isoforms are known to play an important role in proline biosynthesis under osmotic stress (Brugière et al., 1999), it was interesting to test whether the plastidic GS2 isoform was also involved in the biosynthesis of this amino acid under drought stress. A time-course dehydration experiment over a period of 4 d was performed for WT and Ljgln2-2 mutant plants, and the proline content was measured at different time points. A steady increase in proline levels was observed in both plant genotypes in response to drought (Fig. 2). However, the amount of proline accumulated by the Ljgln2-2 mutant was much lower than that by WT plants showing HD in the range 20–50% (Fig. 2). At c. 40% HD (4 d of drought), the proline content in the mutant was significantly lower (< 0.05) than that in WT (42.2 ± 12.6 μmol g−1 DW vs 66.0 ± 10.3 μmol g−1 DW, respectively), whereas the proline content under control conditions was equally low in both cases (4.8 ± 2.2 μmol g−1 DW for mutant vs 6.2 ± 0.7 μmol g−1 DW for WT). Proline represented 17 ± 3.7% of the total amino acid content in WT after 4 d of drought, whereas it represented only 9.3 ± 2.3% in the mutant. It is important to note that the lower content of proline was detected for Ljgln2-2 mutants under conditions in which dehydration was still reversible and full rehydration could be achieved (Fig. 1a). As the amount of total protein did not decrease under these stress conditions (Table 1), proline accumulation must have been a result of de novo biosynthesis and/or decreased degradation, rather than protein degradation.

Figure 2.

 Kinetics of proline accumulation in leaves of Lotus japonicus wild-type (WT) and Ljgln2-2 mutant. Leaf samples were taken at different times of water deprivation in order to measure the hydric deficit (HD) and proline content. WT, closed circles; Ljgln2-2, open circles.

We examined further whether the lack of plastidic GS2 and the lower amount of proline accumulation detected in the Ljgln2-2 mutant were associated with transcriptional changes of the genes for proline metabolism. The expression levels of genes involved in the main (P5CS/P5CR) and alternative (arginase/OAT) proline biosynthetic pathways, as well as the degradative route (P5CDH and proline oxidase), were determined by qRT-PCR analysis of watered plants and plants submitted to 4 d of water deprivation. Gene-specific oligonucleotides were synthesized based on sequences found in available databases. Eight different nonoverlapping sequence fragments corresponding to LjP5CS were found, as well as a single sequence for LjP5CR and LjOAT, two sequences corresponding to arginase (LjArg), one for LjP5CDH and another two for proline oxidase (Ljpro_ox). In the case of LjP5CS, three different categories of putative genes were defined according to their expression patterns: the first category was called P5CS-1 and the corresponding sequences were not induced by drought stress in mutant or WT plants; P5CS-2 sequences were moderately induced by drought in WT and mutant plants, and P5CS-3 sequences were highly induced in both plant genotypes (Fig. 3a). Drought induction of P5CS-2 and P5CS-3 genes was always much stronger in Ljgln2-2 than in WT plants at the same level of HD (Fig. 3a). By contrast, no drought-induced change in expression was observed for LjP5CR in either WT or mutant plants. For the alternative biosynthetic pathway, only one of the two arginase genes was induced by drought, and induction was greater in the mutant than in WT (Fig. 3b). The second arginase gene was weakly expressed in leaves and not induced by drought stress in either genotype (not shown). Expression of the unique LjOAT gene was found to be induced only in Ljgln2-2 plants after water deprivation (Fig. 3b). In the proline degradation pathway, the only modulation observed was the induction of LjP5CDH by drought in WT but not mutant plants (Fig. 3c). Neither of the proline oxidase genes were induced by drought in either genotype.

Figure 3.

 Effect of drought on the transcription of genes for proline metabolism. White bars represent the transcript level (expressed in relative units, r.u.) in Lotus japonicus control plants, normalized to unity, and grey bars indicate the transcript level in drought-stressed plants. Data are the mean ± SD of three independent biological replicates. The different probesets used were taken from the Affymetrix Lotus GeneChip or the sequences available in the database of the Kazusa DNA Research Institute ( (a) Main proline biosynthetic pathway. The different pyrroline-5-carboxylate synthetase (P5CS) probesets available were grouped into three categories according to their response to water deprivation. P5CS1 comprehended probesets chr2.CM0011.16, chr4.CM0179.12.1, Ljwgs_068114.2 and Ljwgs_136569.1; P5CS-2 comprehended probesets chr1.CM0147.99 and Ljwgs_006172.2; and P5CS-3 comprehended probesets Ljwgs_032463.1 and Ljwgs_053689.1. For simplicity, the expression levels of one probeset for each group are reported: probeset chr4.CM0179.12.1 for group P5CS1, probeset chr1.CM1047.99 for group P5CS2, and probeset Ljwgs_053689.1 for group P5CS3. The expression levels for pyrroline-5-carboxylate reductase (P5CR) corresponded to probeset Ljwgs_102146.1. (b) Alternative proline biosynthetic pathway. The data reported are for probeset Ljwgs_027017.1 (arginase) and probeset Ljwgs_090726.1 (ornithine-δ-aminotransferase, OAT). Probeset Ljwgs_046230.1 corresponding to another arginase sequence was not modulated by drought and is not included for simplicity. (c) Proline degradative pathway. The data reported are for probesets chr4.CM0170.37 (pyrroline-5-carboxylate dehydrogenase, P5CDH) and chr1.CM0215.44 (proline oxidase). Probeset Ljwgs_056129.1, corresponding to another proline oxidase sequence, was not modulated by drought and is not included for simplicity.

In summary, an altered pattern of expression for genes of proline metabolism was observed in the Ljgln2-2 mutant (Fig. 4), concomitant with its lower proline content, under drought relative to WT. In the Ljgln2-2 mutant, both the main (via P5CS) and alternative (via arginase and OAT) proline biosynthetic pathways showed a much higher level of induction relative to WT (Fig. 4a,b). At the same time, the mutant did not show the induction of the LjP5CDH gene in the degradative pathway observed for WT (Fig. 4c). These data support a role for plastidic GS2 in the generation of proline pools in response to drought.

Figure 4.

 Summary of transcriptional changes of the genes for proline metabolism in drought-stressed wild-type (WT) and Ljgln2-2 mutant Lotus japonicus. Solid arrows represent genes up-regulated by drought, the width of which reflects the extent of increase in the level of expression. Dashed arrows represent genes not modulated by drought in the conditions utilized in this work. (a) Main biosynthetic pathway. (b) Alternative biosynthetic pathway. (c) Proline degradation pathway.

Transcriptomic analysis of WT and Ljgln2-2 mutant

Given the lower than normal proline content of the Ljgln2-2 mutant under drought stress, as well as the transcriptional differences found in the genes involved in proline metabolism, transcriptomic analysis was performed to gain further insight into the global responses to drought stress in WT and the Ljgln2-2 mutant. The newly available Affymetrix Lotus GeneChip® (Sánchez et al., 2008; Guether et al., 2009; Høgslund et al., 2009) was used to profile leaf RNA from plants exposed to drought for 4 d, after which both genotypes presented identical HD (Table 1), and from watered control plants. First of all, transcript levels of the genes involved in proline metabolism were analysed to confirm the changes in expression found previously by qRT-PCR (Fig. 3). Good agreement between both technologies was obtained (Fig. 5).

Figure 5.

 Comparison of microarray and quantitative real-time polymerase chain reaction (qRT-PCR) data for the genes of proline metabolism. Each point represents a probeset corresponding to genes of proline metabolism. The values reported in the graph are the differences between the log2 expression levels of the probeset in normally watered Lotus japonicus plants and drought-stressed plants. Linear regression analysis of microarray data vs qRT-PCR data gave a regression coefficient of r2 = 0.70 and r2 = 0.57 for transcript levels from wild-type (WT) and Ljgln2-2 plants, respectively. Values are the mean of three independent biological replicates.

Changes in the whole transcriptome were analysed by a significance-based comparison of control and drought-treated WT and Ljgln2-2 mutant plants, applying a false discovery rate (FDR) of < 0.05. Gene transcript levels that were statistically altered under stress in each genotype were examined and functionally characterized using MapMan software (Usadel et al., 2005). About three times as many genes responded to water deprivation in Ljgln2-2 plants than in WT. In total, genes corresponding to 7915 probesets were modulated by drought in the mutant genotype, compared with 2608 in WT plants. Remarkably, c. 80% of genes that responded to drought in WT also responded in the mutant. However, 538 and 5845 genes were modulated specifically in WT and mutant plants, respectively (Fig. 6a). We describe below four different groups of genes according to their response to drought stress in the different genetic backgrounds.

Figure 6.

 Transcriptomics of drought stress in wild-type (WT) and Ljgln2-2 mutant Lotus japonicus. (a) Venn diagram representing the number of probesets elicited by drought (false discovery rate < 0.05) exclusively in WT or Ljgln2-2 plants, as well as those shared between both genotypes. (b) Comparison of the log2 of fold change value for the probesets that are significantly elicited by drought in both genotypes [equivalent to the 2070 probesets in the intersection of the first Venn diagram (a)].

Genes modulated by drought exclusively in Ljgln2-2 This group of genes is of particular interest as it may reflect a specific response of this genotype and/or greater stress experienced by mutant plants as a result of plastidic GS2 deficiency. Among the transcripts induced by drought specifically in the Ljgln2-2 mutant were many known stress-regulated genes (Table S1). These included several genes of the late embryogenesis abundant (LEA) family, MYB transcription factors, arginine decarboxylases and thaumatin-like proteins. Moreover, some of the modulated genes have been recognized previously as responsive to salinity in L. japonicus, such as a MYB transcription factor (TM1624.23), the protein phosphatase type 2C LjNPP2C1 (probeset gi4336433) and the LEA-like protein LjRD29B (probeset CM0148.30) (Sánchez et al., 2008). Thus, these genes were up-regulated by both drought and salt stress. In addition, several genes involved in the metabolism of abscisic acid were transcriptionally up-regulated by drought (LjABA3, probeset chr3.CM0634.85; LjAAO3, probeset chr2.CM0545.5; LjNCED1, probeset Ljwgs_049641.1), indicating the possible induction of abscisic acid synthesis.

Key genes of nitrogen metabolism were up-regulated, including those encoding the ammonium transporter LjAMT1;1 and the asparagine synthetase isoform LjAS2. Several genes of the GS/GOGAT cycle were significantly modulated by drought in Ljgln2-2. Both components of the plastidic GS2/GOGAT cycle were repressed by drought: the plastidic LjGS2 gene (probes TM1765.11 and gi18266052) was repressed about two-fold and a gene for ferredoxin-dependent GOGAT (Fd-GOGAT) was also down-regulated about two-fold. Interestingly, two genes encoding for different cytosolic GS1 isoforms (probes Ljwgs_098311.1 and gi1246767) were induced about two-fold by drought. Genes encoding several types of transporter were also induced by drought specifically in the mutant, including several ABC transporters, and phosphate and potassium transporters. A proline transporter gene (probesets Ljwgs_021753.1 and Ljwgs_043624.1) was also induced in mutant leaves, but not in the WT genotype, under drought. This proline transport has been reported to be induced in WT L. japonicus plants subjected to salt stress (Sánchez et al., 2008). By contrast, other nitrogen metabolism genes were down-regulated, such as LjAMT2 (probeset gi15799271), and several other putative ammonium transporter genes (TC12412, Ljwgs_114175.1, Ljwgs_037968.1) were repressed two- to five-fold.

Other genes for amino acid catabolism were up-regulated exclusively in the mutant, including genes encoding enzymes for lysine degradation, such as the key enzyme LKR/SDH (lysine ketoglutarate reductase/saccharopine dehydrogenase, a bifunctional enzyme). Lysine breakdown generates glutamate, the main precursor of several stress-related metabolites, such as proline and γ-aminobutyric acid.

Many genes related to photosynthesis were specifically down-regulated during drought in the Ljgln2-2 mutant, including genes for photosystem I (PSI) and PSII components, Chla/b-binding proteins, several enzymes of the Calvin cycle, the small Rubisco subunit and Rubisco activase. Interestingly, although transcription of most structural components of the photosystems was repressed, the chlorophyll content of Ljgln2-2 was equivalent to WT and no different from control plants (Table 1).

Among the genes induced exclusively in the Ljgln2-2 mutant during drought were several related to oxygen radicals (such as lipooxygenases, glutathione-S-transferases, ascorbate oxidases), which possibly indicated greater oxidative stress in the mutant than in WT under drought. This idea was supported by higher TBARS levels measured in the mutant genotype (Table 1). Curiously, genes for the cytosolic and plastidic isoforms of Fe-dependent superoxide dismutase (FeSOD; probesets chr5.CM0909.65 and gi46402889, respectively) were repressed by water deprivation only in the mutant.

In brief, we can say that several pieces of transcriptomic evidence indicate that the Ljgln2-2 mutant is subject to a more intense cellular stress than WT, such as the down-regulation of photosynthetic metabolism and the modulation of genes involved in secondary metabolism, together with the up-regulation of lipid degradation and down-regulation of lipid biosynthesis, which suggest increased membrane damage.

An overview of the different genes modulated by drought in the mutant in relation to their corresponding metabolic pathways is available as Fig. S1. A full list of the genes regulated in response to drought only in the mutant can be found in Table S1.

Genes modulated by drought in both WT and Ljgln2-2 Twenty-six per cent of the genes affected by drought stress in Ljgln2-2, corresponding to 2070 probesets, were also modulated in WT under the same conditions. These genes probably represent basic/core responses of L. japonicus plants to drought. Among these shared genes, many were related to oxidative stress metabolism, including several glutathione-S-transferase genes. Related to this is the observation that TBARS levels increased in both genotypes during drought (Table 1). Two glutathione peroxidase genes, namely LjGPX1 and LjGPX3 (probesets chr4.CM0558.28.1 and chr4.CM0042.46), were up-regulated in both genotypes. By contrast, a gene encoding the plastidic isoform of dehydroascorbate reductase (probeset gi66732626), another redox enzyme, was repressed in both genotypes. Other drought-induced genes shared by both WT and the mutant included the well-known stress-responsive genes P5CS (probes Ljwgs_053689.1 and Ljwgs_032463.1), and 1-aminocyclopropane-1-carboxylate (ACC) synthase and ACC oxidase genes, the latter possibly indicating an increase in ethylene synthesis in response to drought. Moreover, the up-regulation of chalcone synthase, the first enzyme in flavonoid biosynthesis, suggests increased flavonoid production in both genotypes in response to drought.

Among the shared down-regulated genes were several for aminoacyl-tRNA synthetases and protein synthesis initiation and elongation. Transcripts for several chloroplast structural proteins also declined. First of all, a group of genes related to the photosynthetic apparatus was significantly down-regulated in both genotypes. Transcript levels for several plastidic ribosomal proteins were also reduced, together with those for plastidic protein translocases and the large subunit of Rubisco. Taken together, these data indicate that chloroplast activities were substantially altered in both Ljgln2-2 and WT plants in response to water deprivation. A gene for an uncharacterized NH4+ transporter of the LjAMT1 family (probeset Ljwgs_028040.1 and Ljwgs_054494.1) was amongst those most repressed by drought.

The vast majority of genes responsive to drought in both genotypes were more responsive in the Ljgln2-2 mutant. A strong linear correlation (r= 0.95) was found between log2 of the fold change in gene expression between the two genotypes (Fig. 6b). The slope of the regression line was 1.6, indicating that transcript levels changed an average of 3.0-fold (21.6) more in the mutant than in WT. These data support the idea that the mutant perceived or actually experienced higher cellular stress under drought. Only 1% of genes, corresponding to 22 probesets, showed an opposite response to drought in the two genotypes (i.e. up-regulated in WT and down-regulated in Ljgln2-2, or vice versa).

The analysis of this set of genes highlighted a signalling component mainly related to ethylene and a redox component as the major responsive elements to drought in the two L. japonicus genotypes analysed. The increased fold change of these shared genes in Ljgln2-2 (Fig. 6b), however, indicated that this basic/core response of L. japonicus to drought acts at different intensities in the two genotypes.

Fig. S1 shows the functional importance of shared stress-responsive genes at the level of metabolic pathways, using MapMan software. A full list of this group of genes can be found in Table S1.

Genes modulated by drought exclusively in WT  Genes corresponding to 538 probesets were drought modulated in WT but not in the Ljgln2-2 mutant (Fig. 6a). Among the drought-induced genes was one encoding an ACC oxidase, the enzyme responsible for the last stage of ethylene production. Genes for the drought-responsive protein RD22 and some enzymes linked to redox metabolism (such as glutathione-S-transferases and a glutathione peroxidase isoform) were also induced in WT, but not in the mutant. Genes repressed by drought in WT included several encoding enzymes of the glycolysis and tricarboxylic acid cycles, such as aconitase and pyruvate dehydrogenase. This group of genes is probably related to the response to drought stress through the action of plastid GS. The presence of genes related to carbon metabolism in this data subset confirms previous results involving plastidic GS2 in the regulation of the C : N balance in L. japonicus (García-Calderón, 2009), and indicates that it may play a similar role under drought stress conditions. An overview of this group of probesets using MapMan software is presented in Fig. S1, and a full list of genes modulated specifically in drought-stressed WT plants can be found in Table S1.

Genes expressed differently in well-watered WT and Ljgln2-2 plants  Transcriptional differences between Ljgln2-2 and WT plants in the absence of drought were also analysed, to identify genes that were modulated by the lack of plastidic GS2 activity under nonphotorespiratory conditions. Genes corresponding to 594 probesets displayed statistically different levels of expression between the two genotypes (FDR < 0.05). Transcripts were higher for 168 and lower for 426 of these in the mutant vs WT (Table S1). Classification of these genes into functional categories revealed differential expression of genes for protein modification and degradation, which accounted for about 14% of genes (Fig 7a). Approximately 9% of genes were related to protein degradation (proteases and components of the proteasome and of the ubiquitination pathway) and 22 genes were related to biotic/abiotic stress signalling, including genes of phosphoinositide metabolism/signalling, receptor-like kinases and the LjMyb4 transcription factor.

Figure 7.

 Comparative transcriptomics of normally watered WT and Ljgln2-2 Lotus japonicus plants. (a) Classification of the differentially expressed probesets (false discovery rate < 0.05) in nonredundant functional categories according to MapMan software. (b) Comparison of the log2 of fold change value for the probesets that are differentially (FDR < 0.05) expressed between wild-type (WT) and mutant plants under control conditions (Ljgln2-2 – WT) and, at the same time, elicited by drought in the Ljgln2-2 mutant (Ljgln2-2 drought – Ljgln2-2). Linear regression analysis of the data gave a regression coefficient of r2 = 0.84.

Interestingly, 77% of the genes differentially expressed in well-watered mutant and WT plants were elicited by drought stress specifically in the Ljgln2-2 background (Table S1). The change in expression of these genes as a consequence of drought in the Ljgln2-2 mutant (transcript level in Ljgln2-2 drought vs water) was plotted against the differential expression of the genes between the two plant genotypes under control conditions (Ljgln2-2 water vs WT water) (Fig. 7b), yielding a strong inverse linear correlation (r2 = 0.91) with a slope of −1.83. The negative slope of the regression line indicated that this set of genes was regulated in an opposite fashion in mutant plants under control conditions relative to water deprivation, with only six probesets showing a deviation from this trend. Therefore, most genes differentially expressed as a consequence of plastid GS2 deficiency in normally watered conditions were modulated under drought. This set of data confirms the existence of a relationship between GS2 and the stress responsive machinery in L. japonicus, which is greatly magnified by drought conditions.


On the role of proline in the response to drought

Proline accumulation is a common response of many plants to abiotic stress. Considering that proline is essential for protein biosynthesis, loss-of-function mutants in genes for proline biosynthesis generally have a severe phenotype even in the absence of abiotic stress (Székely et al., 2008). The L. japonicus Ljgln2-2 mutant provided an opportunity to study stress-induced proline accumulation, as the proline levels of this plant are significantly different from those of WT only under stress (Fig. 2). Therefore, proline availability for protein biosynthesis is normal in the mutant under nonstress conditions.

The biological significance of proline accumulation during abiotic stress is still an open question, as this amino acid could play multiple roles. Indeed, proline has been proposed as a compatible osmolyte, a molecular chaperone, a way to store carbon and nitrogen to be used during recovery from stress and a buffer for cytosolic pH (Verbruggen & Hermans, 2008). Moreover, proline can provide protection against oxidative damage in two ways: directly by acting as a reactive oxygen species (ROS) scavenger, and indirectly by consuming NADPH during its biosynthesis and producing NADP+, the final acceptor of photosynthetic electrons, to avoid the transfer of excess reducing power to water that generates oxygen radicals (Verbruggen & Hermans, 2008). The results presented here provide some insight into the role of proline in drought stress in L. japonicus. Data from Figs 1(a) and 2 suggest that higher levels of proline are important for the rehydration ability of plants, but do not result in a lower rate of water loss. The lower ability of mutant plants to recover water after 5 d of drought (Fig. 1b and 1c) may be a result of increased membrane damage. Indeed, the higher TBARS content detected in the mutant (Table 1) is a symptom of lipid peroxidation, and may be indicative of greater membrane damage in the mutant relative to WT. This and other differences observed between the drought-stressed mutant and WT are coherent with increased oxidative damage in the mutant. Therefore, our results are consistent with previous studies that reported a role for proline in the scavenging of ROS (Szabados & Savouré, 2010). Moreover, proline has been shown to reduce lipid peroxidation in response to treatment with heavy metals (Mehta & Gaur, 1999).

Proline biosynthesis occurs in both the cytosol and the chloroplast (Szabados & Savouré, 2010). The smaller amount of proline accumulated by the Ljgln2-2 mutant, which has normal levels of cytosolic GS isoforms (Orea et al., 2002), suggests that the chloroplastic stress-induced proline may account for c. 40% of total proline accumulation, whereas cytosolic synthesis accounts for the rest. In A. thaliana, it has been shown recently that ‘housekeeping’ proline biosynthesis is probably cytosolic, whereas stress-induced synthesis is chloroplastic (Székely et al., 2008).

Changes in proline metabolism in the Ljgln2-2 mutant

The transcriptional response to dehydration of the genes for proline metabolism was markedly different between WT and the Ljgln2-2 mutant. Mutant plants showed transcriptional activation of both the main and alternative proline biosynthetic pathways (Fig. 4). Thus, our work provides novel data that support a role in proline biosynthesis for the arginase/OAT pathway. The recent work of Funck et al. (2008) argued for a role of OAT in arginine catabolism, but not in proline biosynthesis, as A. thaliana OAT knockout mutants did not show altered levels of proline. It is possible that the arginase/OAT pathway may play a biosynthetic role only under special stress conditions, such as those present in the drought-stressed Ljgln2-2 mutant. Consistent with this idea, transcriptional up-regulation of OAT has been reported in B. napus plants only after prolonged and severe osmotic stress (Xue et al., 2009). Regarding proline degradation, it was found that P5CDH was up-regulated by drought in WT plants, but not in the mutant (Fig. 3c). Reciprocal regulation of the arginase/OAT biosynthetic pathway and the degradative pathway probably avoids a futile cycle, bearing in mind that all of these enzymes are located in the mitochondria (Funck et al., 2008). Previous work has reported that P5CDH is induced by osmotic stress concomitantly with the induction of the proline biosynthetic pathway and an increase in cellular proline levels (Deuschle et al., 2001).

It is possible that the abnormal transcriptional pattern of proline metabolism genes in the drought-stressed mutant is caused by proline itself, which may act in a feedback loop to regulate the transcription of this set of genes. Indeed, intermediates in proline metabolism, such as P5C, are able to modulate the expression of osmotically regulated genes (Iyer & Caplan, 1998). Furthermore, feeding external proline to Arabidopsis plants modulated the transcription of genes of proline metabolism (Deuschle et al., 2001). Additional work is needed to determine the extent to which changes in proline metabolism in Ljgln2-2 mutant plants are a result of lower proline content and/or of their higher sensitivity to drought.

The transcriptional response of the genes for ammonium assimilation and proline metabolism was different in WT and mutant plants. In WT plants, GS and GOGAT transcripts and enzymatic activities were not increased under drought conditions. This suggests that the basal levels of GS and GOGAT are sufficient in order to support the increased biosynthesis of proline under drought conditions. However, an increase in the transcription of the genes of the main proline biosynthetic pathway was observed (Fig. 3a), suggesting that the enzymes for proline biosynthesis rather than those supplying proline precursors are the main control point of the route.

With regard to the Ljgln2-2 mutant, an increase in the transcription of genes for the cytosolic GS isoforms was detected in the transcriptome from drought-stressed plants. However, no increase in GS activity was reported for the mutant plants (Table 1), indicating that the increased transcription of cytosolic GS genes is not paralleled by increased cytosolic GS enzyme activity, at least at the initial stages of drought analysed. Nevertheless, an altered expression of genes for proline metabolism was observed in the mutant. Considering the lower levels of proline and the lack of plastidic GS2 characteristic of Ljgln2-2, this transcriptional response may be an attempt to compensate for the absence of the plastidic GS isoform. The fact that the mutant showed lower proline levels despite the upregulation of the proline biosynthetic pathways further confirms that plastidic GS2 is necessary for proline biosynthesis, and that an important fraction of proline biosynthesis occurs in the chloroplast under stress conditions.

Transcriptomic changes in the Ljgln2-2 mutant

The differential transcription of genes for proline metabolism in WT and the Ljgln2-2 mutant, identified initially by qRT-PCR, hinted at more widespread differences in the transcriptomes of the two genotypes under drought conditions. Transcriptomic analysis using the Affymetrix Lotus GeneChip revealed massive increases in the number and sensitivity of genes responding to drought in the mutant relative to WT. Genes corresponding to 5845 probesets were modulated by drought only in the Ljgln2-2 mutant (Fig. 6a), and these provide an insight into the relationship between the lower proline levels of this mutant and the regulation of gene expression. The massive transcriptional changes observed in the mutant are possibly related to the fact that plastidic GS2 is a crucial enzyme for stress tolerance in higher plants. In fact, plants that overexpress GS2 showed increased tolerance to high light intensity (Kozaki & Takeba, 1996) and to salt stress (Hoshida et al., 2000), in both cases through a protective action on the photosystems. Some of the genes induced by drought stress specifically in the Ljgln2-2 mutant are known stress-responsive genes, such as several involved in oxidative stress alleviation (Fig. S1, Table S1). The down-regulation of photosynthetic metabolism and the modulation of genes involved in secondary metabolism, together with the up-regulation of lipid degradation and the down-regulation of lipid biosynthesis, were other defining features of the mutant under drought stress (Fig. S1, Table S1). The down-regulation of the genes encoding for the structural components of the photosynthetic apparatus is a common response to high stress levels (Saibo et al., 2009). Taken together, these data indicate that the Ljgln2-2 mutant was subject to more intense cellular stress than WT. However, the modulation of antioxidant genes in both WT and Ljgln2-2 suggested that a basal level of oxidative stress was perceived by the two plant genotypes. Accordingly, the TBARS content was increased in both WT and Ljgln2-2 after water deprivation (Table 1). The transcriptomic data also suggested an involvement of plastidic GS2 in the regulation of the C : N balance in drought-stressed WT plants, indicating that alteration in plastidic GS2 might control a number of other metabolic processes in addition to proline metabolism.

A seemingly important group of genes differentially expressed between well-watered WT and Ljgln2-2 mutant plants was related to biotic/abiotic stress signalling even in the absence of external stress conditions (Fig. S1). As most of these genes were also highly modulated by drought stress in the mutant plants, it can be inferred that GS2 is a crucial protein in the stress-responsive machinery of the plant.

In summary, in this article, we have shown that the lack of plastidic GS produces four major consequences in the response to drought in L. japonicus plants: (1) mutants lacking plastidic GS2 show compromised recovery (rehydration) following re-watering; (2) the level of proline accumulation by mutant plants is reduced under drought stress; (3) the pattern of expression of genes for proline metabolism is altered, producing a stimulation of the main (P5CS) and alternative (arginase/OAT) proline biosynthetic pathways, in parallel with a lack of induction of genes for proline degradation (P5CDH); (4) massive changes in the transcriptome are produced. Therefore, a novel link was established between the plastidic GS isoform with the production of proline and the response to drought stress in plants.


This work was supported by EU project LOTASSA (FP6-517617), Spanish Ministry of Science (BIO2008-03213), and Junta de Andalucía/EU-FEDER-FSE (P07CVI-3026 and BIO-163) (Spain). P.D. was partially supported by PEDECIBA-UDELAR. We would also like to thank Modesto Carballo and the CITIUS Facility of the University of Seville for qRT-PCR measurements.