Temporal profiling of the heat-stable proteome during late maturation of Medicago truncatula seeds identifies a restricted subset of late embryogenesis abundant proteins associated with longevity


J. Buitink. Fax: +33 241 22 55 49; e-mail: julia.buitink@angers.inra.fr


Developing seeds accumulate late embryogenesis abundant (LEA) proteins, a family of intrinsically disordered and hydrophilic proteins that confer cellular protection upon stress. Many different LEA proteins exist in seeds, but their relative contribution to seed desiccation tolerance or longevity (duration of survival) is not yet investigated. To address this, a reference map of LEA proteins was established by proteomics on a hydrophilic protein fraction from mature Medicago truncatula seeds and identified 35 polypeptides encoded by 16 LEA genes. Spatial and temporal expression profiles of the LEA polypeptides were obtained during the long maturation phase during which desiccation tolerance and longevity are sequentially acquired until pod abscission and final maturation drying occurs. Five LEA polypeptides, representing 6% of the total LEA intensity, accumulated upon acquisition of desiccation tolerance. The gradual 30-fold increase in longevity correlated with the accumulation of four LEA polypeptides, representing 35% of LEA in mature seeds, and with two chaperone-related polypeptides. The majority of LEA polypeptides increased around pod abscission during final maturation drying. The differential accumulation profiles of the LEA polypeptides suggest different roles in seed physiology, with a small subset of LEA and other proteins with chaperone-like functions correlating with desiccation tolerance and longevity.


days after pollination




desiccation tolerance




dry weight


hydrophilic interaction chromatography


heat shock protein


liquid chromatography electrospray ionization tandem mass spectrometry


late embryogenesis abundant


relative humidity


trifluoroacetic acid


Anhydrobiosis (‘life without water’) is the remarkable ability of certain organisms to survive almost total dehydration. The tolerance to desiccation is a multifactorial trait that involves the protection of macromolecules and membranes from the deleterious effects of water removal. In nature, anhydrobiosis often bridges periods of adverse conditions. This phenomenon is encountered in so-called ‘orthodox’ seeds whose development includes the successive acquisition of the capacity to germinate and to withstand desiccation to low moisture content. Seeds acquire mechanisms to stay alive in this dry state for an extended period of time, ranging from decades to up to 2000 years (Sanhewe & Ellis 1996; Probert et al. 2007; Sallon et al. 2008). Seed longevity is a paramount factor on which seed banks rely to preserve biodiversity (Li & Pritchard 2009) and an important trait in maintaining the seed quality over time, allowing for rapid, homogeneous seed germination and seedling emergence. Longevity is conferred by the ability to stabilize the biological entity for long periods of time by suspending its metabolic activity and forming an amorphous highly viscous matrix (i.e. a glassy state) that severely slows down deteriorative reactions (reviewed by Buitink & Leprince 2004).

One of the hallmarks of seed development is the accumulation of a set of transcripts that were coined as ‘late embryogenesis abundant’ (LEA) (Galau, Bijaisoradat & Hughes 1987). In developing cotton seeds, these authors observed that transcript levels of several LEA genes increased during the post-abscission programme in association with the termination of seed filling (Galau et al. 1987; Hughes & Galau 1991). The specific temporal accumulation of a certain number of LEA mRNAs during seed maturation and desiccation has led to the suggestion that these abundant polypeptides may function to protect the embryo from desiccation damage (reviewed by Tunnacliffe & Wise 2007; Battaglia et al. 2008). It has been suggested that LEA proteins act as protein stabilizers, hydration buffers, membrane protectants, antioxidants and ion chelators (Tunnacliffe & Wise 2007; Battaglia et al. 2008). In vitro evidence is accumulating to support some of these roles, particularly that LEA proteins prevent water stress-induced aggregation of sensitive proteins (Chakrabortee et al. 2007; Boucher et al. 2010).

Most of the in vivo studies demonstrate a protective function of LEA proteins in relation to mild osmotic stress. Overexpression of LEA genes in vegetative tissues of a range of species leads to enhanced tolerance to osmotic stress, often being related to salt or drought stress (Figueras et al. 2004; Bahieldin et al. 2005; Roychoudhury, Roy & Sengupta 2007; Xiao et al. 2007; Olvera-Carrillo et al. 2010). Loss of function of a gene of the LEA_1 group in Arabidopsis leads to decreased tolerance in seedlings and plants upon water deficiency and reduced ability to recover after stress (Olvera-Carrillo et al. 2010). However, these environmental osmotic challenges to which plants are exposed are much milder than those experienced by developing seeds during maturation drying (estimated between −100 and −300 MPa). Indeed, the role of LEA proteins in seed development, including desiccation tolerance (DT) and longevity, remains largely unknown. Manfre & colleagues (2009) have shown that seeds of Arabidopsis em6 mutants show defects in maturation drying. A recent study in Arabidopsis on the role of seed-specific dehydrins demonstrated that down-regulation of LEA14, XERO1 and RAB18 reduced seed survival in the dry state (Hundertmark et al. 2011).

Plants contain a large number of different LEA genes (up to 51 in Arabidopsis) that are divided in seven groups according to their PFAM domains (Bies-Ethève et al. 2008; Hundertmark & Hincha 2008). An overview of LEA proteins present in Arabidopsis seeds after 2 h of cold imbibition using a gel-based one-dimensional (1D) proteome approach demonstrated that the imbibed seeds contain at least 10 different LEA proteins (Oliveira et al. 2007). Whether each of them plays a role in the survival in the dry state or whether they have other functions is unknown. A previous proteome study identified a specific subset of LEA proteins that were associated with the re-induction of DT in germinated radicles of Medicago truncatula seeds (Boudet et al. 2006). Here we investigated whether these LEA proteins also accumulate in relation to the acquisition of DT during maturation, and if there is a unique repertoire of LEA proteins that is linked to the acquisition of longevity. For this purpose, we took advantage of the long maturation of M. truncatula seeds, a model species for legumes. These seeds acquire their DT midway during seed filling (Rosnoblet et al. 2007), and a further 3 weeks of maturation is necessary to acquire their full longevity capacity. A complete reference map of heat-stable proteins was established in mature seeds to be able to monitor the entire repertoire of LEA polypeptides during seed development in a detailed spatial and temporal manner. A total of 35 LEA polypeptides that are encoded by 16 LEA genes accumulated with distinct patterns. The abundance of four LEA polypeptides increased in parallel to the acquisition of longevity. A total of 29 LEA polypeptides increased sharply upon pod abscission and final maturation drying, suggesting additional roles besides their implication in seed survival in the dry state. This work also identified additional heat-stable proteins with chaperone-like function that are associated with seed longevity.


Plant material and treatments

Plants of M. truncatula Gaertn. (A17) were grown in a mix of vermiculite and soil in a growth chamber at 24 °C/21 °C, 16 h photoperiod at 200 µm m−2 s−2. Flowers were tagged and developing seeds were removed from the pods at different time intervals. Seeds were rapidly dried over an airflow of 42% relative humidity (RH), generated by a saturated salt solution of K2CO3, for 3d. Fresh and dry weight (DW), water content and germination were determined according to Boudet et al. (2006). For ageing experiments, dried harvested seeds were kept over a saturated solution of NaCl (75% RH) at 35 °C in hermetically sealed boxes for different time spans, after which the germination percentage of 80 seeds was determined after imbibition in a solution of 20 µm fluridone at 20 °C in the dark to overcome the effect of residual dormancy that was acquired during final maturation (Bolingue et al. 2010), which would otherwise complicate the interpretation of the aging data. For protein extraction, dried seeds were stored at −80 °C before use.

Protein extraction

Sample preparation for 2D gel and Western blot analysis

Soluble proteins were extracted in triplicates from 50 seeds according to Boudet et al. (2006). After two consecutive centrifugations at 13 000 g at 4 °C, the resulting supernatant was diluted to 3 mg mL−1 and heated for 10 min at 95 °C in a heating block with shaking (200 r.p.m.), cooled for 15 min on ice and centrifuged at 13 000 g for 15 min at 4 °C. The resulting supernatant corresponded to the heat-stable fraction. Protein concentrations were assayed according to Bradford (1976).

Sample preparation for hydrophilic interaction chromatography (HILIC) fractionation

A batch of 30 mature seeds was ground in liquid nitrogen using mortar and pestle. To obtain a fraction containing both soluble and insoluble proteins, proteins were precipitated in 800 µL of acetone containing 10% trichloroacetic acid (TFA) and 0.07% dithiothreitol (DTT) during 1 h at −20 °C. Samples were centrifuged at 20 000 g for 30 min at 4 °C. Pellets were washed with 1 mL of acetone, 0.07% DTT and incubated for 30 min on dry ice. After centrifugation (20 000 g for 30 min) at 4 °C, pellets were dried for 5 min in a rotary evaporator. Proteins were resuspended at 4 °C in a rehydration buffer [6 m urea, 2 m thiourea, 4% CHAPS (w/v), 20 mm DTT, HALT protease and phosphatase inhibitor cocktail (Thermo Scientific, Courtaboeuf, France)]. One mg of this protein fraction was digested by sequencing grade modified trypsin (Promega, Madison, WI, USA) in 25 mm ammonium carbonate plus 10% acetonitrile at 37 °C. Digests were acidified to 1% TFA, desalted on a 200 mg UPTI-CLEAN SPE COLUMN C18 (Interchim, Montluçon, France) and lyophilized to dryness.

Two-dimensional (2D) electrophoresis

Hundred µg heat-stable proteins were separated according to their isoelectric point on 24 cm, non-linear 3–10 pH gradient strips (Bio-Rad, Hercules, CA, USA), followed by separation on 12% polyacrylamide gels according to Boudet et al. (2006). The experiments were set up in a randomized block design where six gels corresponding to three independent protein extractions from various maturation stages were run in parallel. Six to eight replicate gels per stage were accumulated independently.

Gel staining, image and statistical analysis of heat-stable proteome

After staining and scanning, digitalized gels were analysed using the PD-Quest 7.1 software (Bio-Rad) according to Boudet et al. (2006). After optimization of the parameters for background subtraction and spot detection, the spots that were not present in at least 50% of the gels and those exhibiting a quality below the set value of 20% (max. value being 100%) were discarded (Boudet et al. 2006). A group of 15 spots whose abundance did not vary were taken for normalization according to Gallardo et al. (2003). A paired t-test was performed using the Statgraphics software (StatPoint Inc, Herndon, VA, USA) to analyse differences in intensity between embryonic axes and cotyledons. Correlation between P50 and spot relative intensity was calculated according to Pearson and the significance of the correlation coefficient was determined from a two-tailed t-test (P < 0.01). Experimental molecular masses and pI were determined from digitalized gels using 2D marker proteins (Bio-Rad) and the calibration method of the PDQuest software (Bio-Rad).

Hydrophilic interaction chromatography

Preparative chromatographic separations were performed on a Waters 2695 HPLC system using a 4.6 × 250 mm TSK-gelAmide-80 5 µm particle column (Tosoh Bioscience, Interchim). One mg of desalted tryptic digest peptides was loaded in 80% solvent B (0.1% TFA in acetonitrile). Solvent A consisted of 0.1% TFA in water. Peptides were eluted with an inverse gradient of 80% B for 5 min followed by 80 to 55% B in 45 min and finally a step gradient to 100% A in 5 min at 0.5 mL min−1. Ten fractions of 2.5 mL were collected throughout the gradient and lyophilized for liquid chromatography electrospray ionization tandem mass spectrometry (LC-ESI-MS/MS) analyses.

Mass spectrometry and protein identification

2D gel analysis

Spots of interest were excised from the 2D gels and subjected to in-gel tryptic digestion. Briefly, gel slices were washed with 100 µL of 25 mM NH4HCO3 by incubation for 1 h at room temperature, followed by dehydration with 100 µL 50% (v/v) acetonitrile in 25 mm NH4HCO3 for 45 min at room temperature. Proteins were reduced and alkylated in 10 mm DTT at 57 °C for 1 h, after which 55 mm iodoacetamide was added, followed by another 45 min of incubation at room temperature. Gel slices were washed with NH4HCO3 and dehydrated with 100% acetonitrile. Gel slices were rehydrated with 10 µL of trypsin solution (15 ng µL−1 in 25 mm NH4HCO3) (Sequencing grade, Promega). After an overnight incubation at 37 °C, tryptic fragments were extracted with 1% formic acid in 70% acetonitrile and analysed by LC-ESI-MS/MS using a nanoscale HPLC (Famos-Switchos-Ultimate system, LC Packings, Dionex, San Francisco, CA, USA) coupled to a hybrid quadrupole orthogonal acceleration time-of-flight mass spectrometer (Q-TOF Global, Micromass-Waters, Manchester, UK) as described in Boudet et al. (2006). Mass data were analysed with the Protein Lynx Global Server software (Micromass-Waters). Protein identification was performed by comparing the data with the UniProt sequence databank (date of release: 13 August 2010) or with TIGR Medicago EST databank (date of release: 10 April 10 2010).

HILIC fractionation

Each fraction was resuspended in 2% acetonitile/0.04% TFA. Nano-HPLC-ESI-MS/MS analysis was performed on an UltiMate 3000 RSLCnano LC system (Dionex). Peptide separation was carried out on a C18 column (Acclaim PepMap C18, 2 cm × 100 µm × 5 µm, Dionex) at a flow rate of 200 nL min−1 using a gradient from 2 to 50% of 0.1% (v/v) formic acid in acetonitrile. The HPLC was coupled to an ESI- LTQ-Orbitrap Velos mass spectrometer (Thermo Fisher Scientific, San Jose, CA, USA). Mass data were recorded on a mass range 400–2000 m/z and MS/MS spectra acquired by using a TOP5 sequencing mode with CID fragmentation. The ESI-LTQ-Orbitrap Velos raw data were converted into mgf files using Proteome Discoverer software (Thermo Fisher Scientific Inc., version 1.2). Fragment ion data were interpreted using the Mascot 2.2 program (Matrix Science, London, UK).

Western blot analysis

One µg of heat-stable protein extract or 5 µg of soluble protein extract was separated by sodium dodecyl sulphate–polyacrylamide gel electrophoresis (SDS–PAGE) on 12% (w/v) polyacrylamide gel, transferred to nitrocellulose membranes (Schleicher and Schuell Inc., Dassel, Germany), and immunodetection was performed by chemiluminescence using a rabbit polyclonal antibody (1:10 000) raised against a recombinant PM25 protein as described by Boudet et al. (2006). Quantification of signal intensity was performed densitometrically using QuantityOne software (Bio-Rad, Hercules, CA, USA).


Characterization of the heat-stable proteome

The complete LEA gene annotation is only available for Arabidopsis (Bies-Ethève et al. 2008; Hundertmark & Hincha 2008). Therefore, we first obtained an inventory of LEA genes in M. truncatula and established the family classification for this species. The in silico analysis using both the IMGAG Mt3.0 version as well as the MtGI8 database revealed a total of 34 genes encoding LEA proteins that are divided in seven LEA families, based on the Pfam classification according to Bateman et al. (2002) (Table 1). Table 1 also presents the additional classification types as well as putative homologues in Arabidopsis to facilitate comparison (Bies-Ethève et al. 2008; Hundertmark & Hincha 2008).

Table 1.  Identification of LEA proteins in M. truncatula and corresponding annotation in the Medicago database
Pfam namePfam#Dureprotein nameTC (MTGI10)Mt 3.0AGIe-value2D spot numberHILIC # peptides
  1. Proteins were grouped according to the Pfam and Dure's classification (Dure et al. 1989). Identified polypeptides by LC-ESI-MS/MS after separation by 2D SDS-PAGE and HILIC are indicated. The homologs and corresponding e-values in Arabidopsis are also indicated.

  2. LEA, late embryogenesis abundant; LC-ESI-MS/MS, liquid chromatography electrospray ionization tandem mass spectrometry; NI, not identified; SDS–PAGE, sodium dodecyl sulphate–polyacrylamide gel electrophoresis.

dehydrinPF00257D11DHN3TC144427   11,12,138
PM12TC145410Medtr3g162790.1  NI1
BudCar5TC153209Medtr7g103100.1  NI3
RAB21TC167838Medtr4g156230.1  NINI
EM proteinAJ498523  7112
CapLEA.IITC164048AT1G526901.00E-3138,39,908 common CapLEA.I
CapLEA.IIITC147670AT1G526903.00E-19NI5 (4 common CapLEA.I)
LEA_1PF03760D113PM1TC143872Medtr7g111880.1AT5G067602.10E-34324 (1 common D-113.I)
D-113.ITC153057Medtr7g111860.1AT5G067604.00E-22434 common D-113.II
D113.IITC147896Medtr7g111870.1AT5G067601.00E-19434 common D-113.I
D-34.IVTC165526Medtr2g091970AT3G224902.00E-23NI2 common D-34.III
LEA_2PF03168D95LEA14-like cBQ124186At1g014705.00E-27NI9
LEA-like aTC148928Medtr5g095950.1AT2G440604.80E-74NINI
LEA-like bTC143875AT2G440604.60E-124NINI

Characteristically, only a few LEA polypeptides can be detected after 2D separation on SDS-PAGE gels using total protein extracts from seeds (Gallardo et al. 2003). Therefore, a protein fraction that was enriched in LEA polypeptides was obtained using their hydrophilic and unstructured character, which allows them to remain in solution following heating of the crude protein extract at 95 °C (Boudet et al. 2006). The polypeptides present in this so-called ‘heat-stable’ protein fraction in mature seeds were identified to establish a reference map for subsequent characterization of LEA proteome during seed maturation. For this purpose, the heat-stable protein fraction of mature seeds was separated by 2D SDS-PAGE (Fig. 1). Out of the 135 spots detected, 90 spots with a relative high intensity were sequenced using LC-ESI-MS/MS (Supporting Information Table S1) and identified based on the Medicago database (Fig. 1, Table 1 and Supporting Information Table S1). A total of 38 spots corresponded to LEA polypeptides that were encoded by 16 different LEA genes representing almost all families (Table 1). Seven LEA proteins were represented by several spots, belonging almost all to the LEA_4 family, except for a DHN (Table 1). Three spots were identified as CapLEA polypeptides, but no distinction could be made between which polypeptides were encoded by which gene (Table 1; Supporting Information Table S1). Furthermore, two spots were identified as LEA D-113 (LEA_1), but the peptide analysis did not allow for discrimination between the two genes LEA D-113.I and LEA D-113.II that were found in the MGI8 Medicago sequence database (Supporting Information Table S1). Both genes are adjacent on chromosome 7.

Figure 1.

Reference map of the heat-stable proteome of mature M. truncatula seeds. Hundred mg of the heat-stable proteins was separated by SDS-PAGE using 24 cm non-linear immobilized pH gradient strips (3–10). pI and molecular mass (in kDa) are indicated. Numbers indicate the spots that were identified (see Table 1 and Supporting Information Table S1). SDS–PAGE, sodium dodecyl sulphate–polyacrylamide gel electrophoresis.

Forty-nine spots in the heat-stable protein fraction correspond to polypeptides other than LEA proteins, such as seed storage proteins (16 spots), glycine-rich proteins (2 spots) and a number of stress proteins, such as heat shock proteins (HSP, 5 spots) and 1-cys-peroxyredoxin (Fig. 1, Supporting Information Table S1). Based on spot intensity, 54% of the total heat-stable protein fraction is represented by LEA polypeptides.

Identification of LEA proteins in total soluble protein extracts via LC-ESI-MS/MS

For the remaining 17 LEA genes identified in M. truncatula (Table 1), spots corresponding to the encoded proteins were not identified in the seed extracts (Table 1). To assess whether these proteins were missing in our 2D gels because they were below a detectable level in mature seeds or because of an incomplete enrichment by heating the sample, an additional LC-ESI-MS/MS analysis was performed on the total soluble protein fraction of dry mature seeds (Supporting Information Table S1). In order to decrease the complexity of the spectral analysis, and taking advantage of the hydrophilic nature of LEA proteins, the tryptic digest of the total soluble protein extract was first submitted to HILIC to separate the peptides according to their polarities (Alpert 1990). As expected, we found that a majority of LEA peptides could be separated this way from non-LEA peptides (Supporting Information Fig. S1). All the polypeptides encoded by the 16 LEA genes that were identified on the 2D gels on the heat-stable protein fraction were detected. In addition, four other LEA polypeptides were detected, of which two were identified as dehydrin-like proteins, namely PM12 and Budcar5 (Table 1). Budcar5 was previously detected in the heat-stable proteome and shown to accumulate only in young M. truncatula roots that were submitted to an osmotic stress (Boudet et al. 2006). Most likely, it was too faintly present on the 2D gels of mature seeds to be detected. A third polypeptide identified in the total protein extract was a LEA14-like protein, one of the four members of the LEA_2 (PF03168) family. Analysis of amino acid sequence, based on the normalized net charge and mean hydrophobicity, predicts that this family contains proteins that are relatively hydrophobic and structured (Supporting Information Fig. S2), and this was experimentally shown by Singh et al. (2005). This was also experimentally confirmed in our study as the fractionation of LEA14-like peptides by HILIC identified the nine LEA14-like peptides in the fractions containing the more hydrophobic peptides (Supporting Information Table S1). This explains why they were not detected in our heat-stable proteome. Our data give support to the contention that proteins from this family should no longer be considered as LEA proteins (Tunnacliffe & Wise 2007). The fourth polypeptide identified in the crude extract was CapLEA.III (Table 1).

For the three pairs of LEA isoforms (D113.I and D113.II, CapLEA.I and CAPLEA.II, D113.II and D-34.IV), LC-ESI-MS/MS peptide analysis did not permit to distinguish between encoding genes due to their high similarity (Supporting Information Table S1). Considering the good overlap between the different LEA identification methods, we considered that the heat-stable proteome gives an accurate representation of the LEA proteome in dry seeds and can be used for subsequent quantification of the LEA proteome.

Tissue and condition-specific gene expression of LEA proteins

To investigate whether the LEA proteins that are detected in mature seeds exhibit seed-specific expression, their gene expression in different tissues and under various conditions was investigated using the gene expression atlas of M. truncatula (Fig. 2). The genes encoding LEA proteins that were detected in the heat-stable proteome and LC-ESI-MS/MS are listed at the top of Fig. 2. For 11 genes, transcripts appeared to be seed specific. A second group of genes that are expressed in seeds were also expressed in roots during salt stress. The other LEA proteins that were not detected in seeds showed a much more general expression profile, including leaf tissue, cell suspension, root and shoot tissues. Transcripts of D113.II were only detected in seeds, whereas those of D113.I were detected in almost all tissues investigated. Possibly, spot 43 is encoded by D113.II and not D113.I. Transcripts of two CAPLEA genes were also detected in split roots under sufficient nitrogen.

Figure 2.

In silico expression analysis of 28 LEA genes in different tissues and different environmental conditions in M. truncatula using the Medicago gene expression atlas (http://mtgea.noble.org/v2) (Benedito et al. 2008). Expression data (log2) are depicted using the Heatmap Builder software (King et al. 2005); green = low intensity, red = high intensity. LEA, late embryogenesis abundant.

Organ specificity of LEA proteins in M. truncatula

To obtain an overview of the organ specificity of the different LEA polypeptides and other heat-stable polypeptides, their abundance was analyzed separately in cotyledons and embryonic axes that were dissected from mature seeds. The percentage of heat-stable proteins in the soluble protein fraction is similar in both organs (Supporting Information Table S2). The heat-stable protein profiles of cotyledons and axes are comparable (Supporting Information Fig. S3). Quantification of the spot abundance was performed after normalization on total density in valid spots (Supporting Information Table S3). Table 2 shows the most abundant heat-stable polypeptides (with intensity > 1% of the total intensity) that are present in either cotyledons or axes. Out of the 21 most abundant polypeptides, 16 heat-stable polypeptides represented LEA proteins (Table 2). The other heat-stable polypeptides were seed storage proteins (vicilin/convicilin), Hsp20 and a glycine-rich RNA-binding protein RPN-1, and ferritin/1-cys-peroxiredoxin.

Table 2.  List of the most abundant heat-stable polypeptides that are detected in embryonic axes and cotyledons of mature M. truncatula seeds
Spot #IdentificationPfam LEAEmbryonic axesCotyledonsA/CP-value
MeanSD% intMeanSD% int
  1. Abundance is expressed as normalized intensities [mean and standard deviation (SD)], percent of the total intensity of the heat-stable proteome (% int), and as the ratio of intensity between embryonic axes and cotyledons (A/C). The P-values (paired t-test) indicate the level of significance in intensity between axes and cotyledons.

44EM6LEA_5242 783439324.439 80223570.9724.771.08E-10
43D-113.I/IILEA_193 14663449.3731 22442393.092.980.00002
63D-34.IISMP30 84920813.111 5917321.152.660.00001
27Ferritin/1-cys-PrxNA18 63310581.877 01215310.692.660.00012
71EM proteinLEA_575 18045937.5637 95234053.751.980.00009
90Cap-LEA.I/IILEA_412 3845281.2512 5417311.240.990.86591
86RNP-1NA7 5283860.7610 63717091.050.710.95749
 8PM10LEA_415 33023211.5422 37920722.210.690.05328
32PM1LEA_17 02416520.7110 119203210.690.27122
39Cap-LEA.I/IILEA_498 55936929.92145 907811714.420.680.00036
21LEAM, D-34.ILEA_4/SMP15 69415881.5823 65632862.340.660.06074
62MP2LEA_411 2777581.1319 80543301.960.570.91887
22PM25SMP16 88722781.730 23923482.990.560.00176
12ConvicilinNA53 08930195.3499 86746099.870.530.00001
11DHN3Dehydrin15 38114081.5529 84337462.950.520.00342
70Hsp20NA1793019231.839 21627433.880.460.00011
60SBP65LEA_44 3517090.4411 71211841.160.370.00035
15PM18LEA_48 18611880.8226 87963472.660.30.00473
23VicilinNA5 0074270.527 27079892.70.180.00041
 4SBP65LEA_41 5427700.1620 07913451.980.080.00066
 3SBP65LEA_45522330.0611 66136971.150.050.00096

A total of 13 LEA polypeptides were over twofold more abundant in the cotyledons than in axes, out of which 11 were identified as SBP65, and the other two as an isoform of PM18 (spot 15) and MP2 (spot 55) (Supporting Information Fig. S3, Supporting Information Table S3). For SBP65, the total intensity of all the 11 isoforms was fivefold higher in cotyledons than in axes, with spot 3 over 21-fold higher in cotyledons. Additional spots that are overrepresented in cotyledons were identified as two storage proteins (spot 12 and 23) and a small heat shock protein (sHsp20, spot 70) (Table 2). The LEA polypeptides that had a higher intensity (P < 0.01) in the axes were identified as EM6 (spot 44), LEA D-34.II (spot 63) (Table 2) and D113.I/II (spot 43) (Table 2, Supporting Information Fig. S3, Supporting Information Table S3). Only EM6 was considerably more abundant in the axes, with a 25-fold difference in intensity between the two organs, whereas the difference was between two- and threefold for the other two polypeptides (Table 2). Other non-LEA protein polypeptides that are overrepresented in the axes were identified as a mixture of ferritin/1-cys-peroxyredoxin (spot 27) and two legumins (45 and 83) (Table 2, Supporting Information Fig. S3, Supporting Information Table S3).

Considering that only three LEA polypeptides are more abundant in the axes compared with cotyledons, and most of them only with a two- to threefold difference, the temporal expression of the LEA proteome during seed maturation was characterized using whole seeds.

Acquisition of DT and longevity during maturation of M. truncatula seeds

In order to link the appearance of particular LEA polypeptides to the different physiological processes that are acquired during maturation, we characterized their acquisition in detail during seed development of M. truncatula (Fig. 3). From 24 to 32 days after pollination (DAP), seeds were still green. Chlorophyll was completely lost at 44 DAP, which corresponded to the point of pod abscission (Fig. 3a). DW increased significantly between 20 and 28 d, then levelled off from 32 DAP onwards, indicating the end of seed filling (Fig. 3b). Seed water content decreased between 20 and 24 DAP due to the replacement of water by reserve deposition and reached a plateau between 32 and 40 DAP. The actual drying occurred from 40 DAP onwards, when pods were still attached to the mother plant and seeds contained 0.91 g H20 g−1 DW. At pod abscission (44 DAP), seeds contained 0.59 g H20 g−1 DW. Thereafter, they dried further in the detached pods and reached their equilibrium water content (i.e. 0.11 g H20 g−1 DW) around 48 DAP.

Figure 3.

Characterization of seed maturation of M. truncatula. (a) Pods and isolated seeds at the different days after pollination (DAP). (b) Changes in dry weight (DW) and fresh weight (FW) (mg per seed) and water content (WC) (g H2O per g DW). Data are the means ± SE of five replications using four seeds. Bars represent the standard deviation. (c) Acquisition of desiccation tolerance (open circles) and longevity (black triangles) during maturation. Desiccation tolerance was determined by scoring the percentage of germination of immature seeds after rapid drying. Longevity was expressed as the P50, corresponding to the time of storage (days) at which the dried seed lot reached 50% germination. (d) Changes in germination percentages of rapidly dried seeds harvested at indicated stages of maturation and stored at 35 °C and 75% relative humidity (RH). Third-order sigmoid curves were fitted to the data to obtain the P50 values.

At 20 DAP, seeds isolated from the pods germinated at 100% (data not shown). However, they were all desiccation sensitive as they failed to germinate after rapid drying over an airflow of 43% RH (Fig. 3c, open symbols). DT was acquired between 24 and 28 d, with germination percentages after drying of 36 and 91%, respectively. No obvious difference was found between the acquisition of DT in axes versus cotyledons, measured by growth or greening after drying, respectively (data not shown).

To assess longevity at the different maturation stages, harvested seeds were rapidly dried over an airflow of 43% RH for 3 d to halt development and stored over a saturated NaCl salt solution of 75% RH at 35 °C. After different intervals of storage, the percentages of germination were assessed following imbibition. Survival curves during storage indicated that longevity was progressively acquired during maturation (Fig. 3d). From these curves, the time to lose 50% viability during storage (i.e. P50) was determined to quantify the longevity. P50 values increased 30-fold from 1.1 d at 28 DAP to 31 d at the point of abscission (Fig. 3c). Longevity increased gradually over the 22 d period during which water contents decreased from 1.3 to 0.6 g H2O g−1 DW (compare Fig. 3b,c). Final drying of the seeds in the abscised pods was accompanied by a reduction in P50 to 25 d.

Analysis of LEA protein abundance during maturation of M. truncatula seeds

The amount of heat-stable proteins did not vary significantly during seed maturation, whereas the soluble proteins on a DW basis increased only slightly (Supporting Information Table S2). Few changes were observed in the soluble protein profile after separation on 1D SDS-PAGE (Supporting Information Fig. S4). However, profiles of the heat-stable protein fraction after separation by SDS-PAGE changed considerably, with numerous bands appearing during maturation (Supporting Information Fig. S4). The same increase in number and intensity of heat -stable polypeptides can be seen in 2D gels (Fig. 4). Because of the increase in spot number and their relative abundance during development, spot intensity was normalized based on a group of 15 spots whose abundance did not vary (Gallardo et al. 2003). The different normalized intensities of the detected spots are summarized in Supporting Information Table S4. To validate this method of normalization, the abundance of PM25, one of the LEA proteins detected, was measured by Western blot analysis (Boudet et al. 2006). The relative intensity of PM25 from the normalized 2D gel analysis was similar to the intensity measured using the immunodetected signal between 24 and 44 DAP (Fig. 5). At 48 DAP, the intensities of PM25 varied between replicates but were not significantly different between both methods.

Figure 4.

Evolution of the heat-stable proteome during seed maturation of M. truncatula. Representative 2D gels are shown from seeds at 24 DAP (a), 28 DAP (b), 32 DAP (c), 36 DAP (d), 40 DAP (e), 44 DAP (f) and 48 DAP (g) as described in Fig. 1. 44 DAP corresponds to the point of pod abscission, and 48 DAP corresponds to dry mature seeds. The position of PM25, used for the quantification by Western blot, is shown in panel (e). pI and molecular mass (in kDa) are indicated. DAP, days after pollination.

Figure 5.

(a) Quantification of PM25 LEA protein using image analysis of the intensity of spot 22 from 2D gels after normalization (2D, black bars), and by Western blot (WB) using the intensity of signals obtained from the immunoblot with an anti-PM25 antibody. The Western blot was performed using the total soluble protein (S) or heat-stable protein (HS) fraction. Quantification of the Western blot analysis on the HS fraction represents the mean of three replicates ± SE (b–c). Immunodetection of PM25 after one-dimension SDS-PAGE separation of 1 µg of heat-stable proteins (b) and 5 µg of total soluble proteins (c). Lanes correspond to the maturation stages indicated in panel (a). LEA, late embryogenesis abundant; SDS–PAGE, sodium dodecyl sulphate–polyacrylamide gel electrophoresis.

The number of detected spots increased significantly between the stages: from 50 spots at 24 DAP to 134 spots in mature, 48-day-old seeds, particularly above a pI of 5.5 (Fig. 4, Supporting Information Table S4). Spots with the highest intensity are both members of group LEA_5 (i.e. EM and EM6), one member of LEA_1 (D113.I/II), the DHN3 and one of the several isoforms of two genes belonging to the LEA_4 family (spot 39 of CapLEA.I/II and spot 61 of SBP65) (Supporting Information Table S4).

Comparison of the different expression profiles for 28 LEA polypeptides showed two patterns of expression (Fig. 6). The first pattern (Fig. 6a–e) is characterized by a gradual increase until 44 DAP followed by a slight decrease thereafter. It includes several of the most abundant LEA polypeptides (EM, CapLEA.I/II and D113.I/II), one of the D-34 family members (Fig. 6d) and the two most abundant spots of SBP65 (Fig. 6e). It also includes two non-LEA proteins, namely RNP-1 and Hsp20 (Fig. 6p,q). The second pattern of expression (Fig. 6f–o) can been seen for 22 LEA polypeptides that represent 9 out of 15 LEA genes. Their intensity increased sharply around pod abscission (44 DAP), concomitantly with the drop in water content from 0.91 to 0.11 g H2O g−1 DW (Fig. 3b). In mature seeds, these polypeptides represented 50% of the total LEA proteome.

Figure 6.

Changes in spot intensity of the LEA polypeptides during maturation. (a–e) LEA polypeptides with gradual increase in parallel with the acquisition of longevity (solid line). (f–o) LEA polypeptides that increase in intensity around the point of abscission (44 DAP). (p–q) Polypeptides other than LEA proteins that increase with longevity. When applicable, the different numbers correspond to different isoforms. Normalized intensity values are divided by 10 000. LEA, late embryogenesis abundant.

Identification of heat-stable polypeptides that correlate with the acquisition of DT or longevity

Nine heat-stable polypeptides appeared or increased in parallel to the acquisition of DT between 24 and 28 DAP (Fig. 7, Supporting Information Table S4). Two LEA_4 polypeptides (PM18 and an isoform of CapLEA) appeared whereas three LEA polypeptides, EM6 (LEA_5) and PM25 (SMP) and CapLEA (LEA_4), increased 5.6-, 3.3- and 4.2-fold, respectively (Fig. 6). The total intensity of all LEA polypeptides at the desiccation-tolerant stage represented only 6% of the total LEA proteins present in mature seeds. Among non-LEA proteins, RPN-1 (spot 86) and another glycine-rich protein (spot 31) also appeared at 28 DAP, whereas two legumins (spot 75 and 54) increased, respectively, at 1.7- and 4.6-fold during that period (Supporting Information Table S4).

Figure 7.

Changes in LEA polypeptide abundance during the acquisition of desiccation tolerance in developing seeds of M. truncatula. EM6 (a,b), PM25 (c,d), PM18 (e,f) and two polypeptides of CapLEA.I/II (g,h). a,c,e,g: representative 2D gels; pI and molecular mass (in kDa) are indicated. b,d,f,h, relative spot intensities at 24 and 28 DAP. Data represent the mean of six to eight replicates ± SE. DAP, days after pollination; LEA, late embryogenesis abundant.

The heat-stable polypeptides that increased in abundance in parallel to the acquisition of longevity were identified using a Pearson correlation analysis between normalized intensities and P50 values including the time point at 48 DAP for which longevity decreased (Table 3, Supporting Information Table S4). Seven spots were retained (two-tailed probability test, P < 0.01) and corresponded to four LEA proteins: EM (LEA_5; spot 71), D113.I/II (LEA_1; spot 43), D-34.III (SMP; spot 65) and CapLEA I/II (LEA_4; spot 39). An estimation of their abundance based on the absolute intensities shows that they are highly abundant in mature seeds, representing 14.4% of the total intensity of the heat-stable protein fraction and 35% of the intensity of the LEA proteome of mature seeds. The other two heat-stable polypeptides that correlated with longevity corresponded to RPN-1 (spot 86) and Hsp20 (spot 70) (Table 3). Interestingly, the common denominator of these two proteins is their chaperone activity (Prieto-Dapena et al. 2006; Kim et al. 2007; Tejedor-Cano et al. 2010).

Table 3.  Polypeptides present in the heat-stable proteome whose abundance (normalized intensity) is correlated with longevity (P50)
Spot #IdentificationLEA familyr2 P50P-value% total intensity
  1. An estimate of the polypeptide abundance based on the percentage of the total intensity of the heat-stable proteome is also indicated.

  2. LEA, late embryogenesis abundant; NA, not applicable; PEA, phosphatidylethanolamine.

86RNA-binding protein RPN-1NA0,9959,8e-61,50
70Heat shock protein Hsp20NA0,9590,000652,78
71EM proteinPF004770,9460.00544,35
66PEA-binding proteinNA0,8500.008890,41

Post-transcriptional regulation of LEA proteins

The observation that the abundance of the majority of polypeptides accumulated at final maturation drying was rather surprising. Indeed, a previous transcriptome study on the acquisition of desiccation tolerance in M. truncatula identified numerous LEA transcripts that accumulated around 20–24 DAP (Buitink et al. 2006), suggesting the possibility of post-transcriptional regulation. This was further investigated by assessing whether transcript and protein abundance are correlated using the M. truncatula gene expression atlas (Benedito et al. 2008). The gene atlas contains detailed transcriptome data of embryos between 10 and 24 DAP and an additional time point at 36 DAP which is comparable with our time point at 36 DAP (Gallardo et al. 2003; Benedito et al. 2008). Fig. 8 shows the transcript levels of 20 LEA genes during seed development. Transcripts accumulated mostly during seed filling up to 24 DAP, whereas the LEA proteome increased mainly after 36 DAP (Fig. 6). For 16 out of 20 genes, hardly any change in expression was detected over the 12 subsequent days, from 24 to 36 DAP. Transcript levels of LEA_5 EM increased later than those of its family member EM6, which was also observed for Arabidopsis (Bies-Ethève et al. 2008). Only three genes, of which two closely related LEA_4 genes (CapLEA.II and III) and a LEA_1 (D113) member, showed a gradual increase in transcript level between 16–24 and 24–36 DAP (Fig. 8). Interestingly, these three genes code for polypeptides that correlate with longevity (Table 3). These data suggest that during development, most LEA transcripts are accumulated long before they are mobilized for protein synthesis during maturation drying. A more detailed study including later time points of maturation is necessary to confirm this conclusion.

Figure 8.

Expression profile of LEA genes during seed development. Data were obtained from the Medicago gene expression atlas (http://mtgea.noble.org/v2) and transformed into Log2 values. LEA, late embryogenesis abundant.


The maturation phase is described as the time encompassing the accumulation of storage reserves and drying (Santos-Mendoza et al. 2008). However, when seed filling is terminated, seeds of most species do not proceed directly to final drying, but exhibit a so-called ‘late maturation phase’, which so far has received little attention (Sanhewe & Ellis 1996; Probert et al. 2007). Here we show that for M. truncatula seeds, the period after seed filling is an important phase in which longevity increases over 30-fold, and the majority of the 35 detected LEA polypeptides accumulate. Furthermore, the long time span of this late maturation phase provides an elegant model to dissect mechanisms related to longevity. Here, we found that the pattern of LEA protein accumulation varies considerably between gene families, but also between members of the same family and even among the different isoforms of the same LEA protein (Fig. 6), suggesting multiple mechanisms of regulation. This is also evident from the difference in transcript and protein appearance. Overall, our physiological data combined with the LEA proteome analysis reveal that the final phase of maturation following seed filling is an intricate and important phase for seeds during which they acquire further physiological quality.

From the 38 LEA polypeptides detected in mature seeds, a small subset of LEA polypeptides, representing only 6% of the intensity, accumulates during the acquisition of DT. Three of these LEA polypeptides (PM25, EM6 and PM18) also reappear upon the re-induction of DT in germinated radicles (Boudet et al. 2006), suggesting that their regulation is specifically linked to a desiccaton tolerance signalling pathway. The physico-chemical properties of EM6 and PM25 have been well characterized in vitro (Soulages et al. 2002; Gilles et al. 2007; Boucher et al. 2010). For both proteins, the removal of water induced a transition from a fairly disordered conformation to the formation of a considerable amount of ordered structures, at a hydration level around 0.2 to 0.3 g g−1, corresponding to the onset of the removal of the hydration shell. An in vitro study on the function of PM25 demonstrated that the protein has a strong capacity to bind water, and is capable of dissociating cold and desiccation-aggregated proteins (Boucher et al. 2010). The role of both proteins in DT remains, however, elusive; Arabidopsis mutants deficient in Em6 (Manfre et al. 2009) and M. truncatula mutants deficient in PM25 (Leprince, unpublished data) produce dry viable seeds.

The four LEA polypeptides that accumulate in parallel to the acquisition of longevity represent one third of the intensity of the LEA proteome at maturity, which would be expected if they were to play a direct role as a protective molecule. A broad stabilization function in the dry state, described as molecular shield (Tunnacliffe & Wise 2007), is well plausible. Another putative role would be that these polypeptides contribute to the stability of the glassy state, which plays a critical role in seed longevity (Buitink & Leprince 2004, 2008). There is now accumulating evidence that in vitro, LEA proteins of the LEA_4 group to which CapLEA belongs, and members of the LEA_1 family, increase the density of sugar glasses (Wolkers et al. 2001; Shih et al. 2010; Shimizu et al. 2010), a property known to increase its stability (Buitink & Leprince 2004). Another possible role for these LEA proteins is to protect membranes during water removal (Tolleter et al. 2007; Tunnacliffe & Wise 2007). It should be noted that transcript levels of CapLEA and D113.II also accumulate in roots that are submitted to salt stress. Possibly, these LEA proteins have several functions depending on the water content, as has been suggested for non-reducing sugars (Hoekstra, Golovina & Buitink 2001).

Temporal profiling of the heat-stable polypeptides suggests that different subsets of LEA proteins confer desiccation tolerance or improve stability in the dry state. However, it should be noted that PM25 and EM6 that are associated with DT are highly similar to D-34.III and EM, respectively (89 and 86%), both being correlated with longevity. Their high similarity might suggest that the molecular mechanisms by which the homologs act on the acquisition of DT or longevity might be comparable. This is also suggested by the accumulation of two CapLEA isoforms at the acquisition of DT, of which one is also correlated with longevity. Double knockout plants will be needed to ascertain their function to avoid redundancy. Another LEA protein that was detected at 28 DAP, but further increased 50- to 60-fold between 44 and 48 DAP, during final maturation drying, is LEAM. This mitochondrial LEA protein was shown to interact with membranes in the dry state and to protect liposomes subjected to drying (Tolleter et al. 2007), and it was suggested that it is involved in the protection of the inner membrane of mitochondria during desiccation.

The heat-stable proteome does not only contain LEA polypeptides, but also numerous polypeptides involved in stress responses, such as chaperonins, heat shock proteins (5 spots) and 1-cys-peroxyredoxin (Fig. 1A, Supporting Information Table S1). Two additional polypeptides that correlated in a highly significant manner with the increase in longevity were identified as a glycine-rich RNA-binding protein RPN-1 and sHsp20. RNP-1 shows very strong similarity to GRP7 of Arabidopsis (At2g21660). It has been shown that GRP7 helps Escherichia coli grow and survive better during cold shock, and was suggested to exhibit RNA chaperone activity (Kim et al. 2007). sHsp20 is a small heat shock protein that has been correlated to DT (Wehmeyer & Vierling 2000). Modulation of the expression of a heat shock factor HaHSFA9 demonstrated that small Hsps play a role in enhanced seed survival during controlled deterioration at high water content (24 and 15%) and high temperature (50 °C) conditions (Prieto-Dapena et al. 2006; Tejedor-Cano et al. 2010). Several LEA proteins are also acting against protein aggregation both in vitro (Boucher et al. 2010) and in vivo (Chakrabortee et al. 2007). Altogether, it can be inferred that protection against desiccation-induced aggregation might be an important mechanism conferring longevity to seeds, and that different sets of LEA and HSP might serve this purpose.

The majority of LEA polypeptides (from 11 out of 16 LEA genes) accumulated during the final maturation drying in relation to pod abscission, long after the acquisition of DT and longevity. This finding sheds new light on the function of LEA proteins besides their role in survival in the dry state. Indeed, at 40 DAP, these LEA polypeptides have not yet accumulated, but the longevity of these seeds is comparable with that of mature seeds, in which they are present. Considering that they accumulate during the drying phase, these proteins might regulate the dehydration rate of seeds after pod abscission through the high water-binding capacity of these hydrophilic proteins (Tunnacliffe & Wise 2007; Boucher et al. 2010). Alternatively, their late appearance during maturation suggests that they might play a protective role during seed imbibition instead of development. For example, chilling tolerance during seedling emergence of soybean is associated with the presence of a 35 kDa DHN (Ismail, Hall & Close 1999). In Arabidopsis, AtLEA4-5, the homolog of MtD113.I/II that is described here, enhanced germination efficiency under water deficit (Olvera-Carrillo et al. 2010).

In conclusion, this work shows that during maturation, there are two additional developmental phases after seed filling, the first leading to the acquisition of longevity and the second corresponding to final maturation drying during seed dispersal. These are substantiated by the dynamic changes of the LEA proteome. This study provides a detailed repository of seed-specific LEA proteins that accumulate in relation to the survival in the dry state and represents an important step towards the analysis of their functional properties in vivo and in vitro.


We thank B. Ly Vu for his technical assistance in taking care of the plants, Dr F. Montrichard for remarks regarding the proteomic approach, and Audrey Geairon and Yvan Choiset for their technical assistance in mass spectrometry analyses and Waters 2695 HPLC system, respectively. This work was funded in part by grants from the Région Pays-de-la-Loire [program PHOSPHOSAVE (2008–2011) and QUALISEM (2010–2014)].