A continuum of research projects to improve extraction of oil and proteins in oilseed plants

two main

which is highly basic and rich in sulphur containing amino acids, and the 12S globulintype which is neutral and of high molecular weight. The 2S albumin fraction contributes largely to reach the recommendations in digestible lysine and methionine for cattle feed which actually represents 70% of the animal feed outlet (Tostain, 2009). Efficient oil extraction from rapeseed appears rather difficult, by comparison to other seeds currently crushed at the industrial scale (soybean, sunflower). As rapeseed seeds are very rich in oil (45% like sunflower seeds), crushing requires a preparation step consisting in flaking, cooking, prepressing, pelletizing and an extraction step using hexane. The flaking-cooking step allows better efficiency of the prepressing and the solvent treatment lowers the residual oil in the expeller cake from around 15-20% to 2-4%. The deoiled cake is then desolventized by heat and live steam treatment. The meal therefore obtained is a very rich protein source but the various treatments for desolventisation have a negative impact on the protein solubility and digestibility, lowering its use and value, especially for animal feeding. Moreover, these proteins and especially 2S albumins present interesting functional properties (Malabat et al., 2001) such as foaming or emulsifying properties, which are partly or completely lost through the standard oil extraction process. Problems associated with hexane, an inflammable solvent recognized as a volatile organic compound responsible for air pollution, have made extraction plants expensive to build, run and maintain, due to environmental safety issues and regulations (Gros et al., 2003, Campbell et al., 2011. Finally, the emission of hexane and the energy consumption per ton of crushed rape seeds are around 1 liter and 280 kWh, respectively 3 . This energy is mainly thermal (85%), obtained from fossil sources (gas) and a large part (65%) is used to cook and dry the seeds, then to desolventize the meal. An improvement of the ability of rape seeds to be crushed could allow to decrease the necessary energy and consequently, the environmental and economical costs of the crushing operation. The increase of rapeseed economical value therefore relies in our ability to improve oil extraction while preserving the availability and stability of protein by-products, and saving energy during the process.

Organization of oil and protein reserves within seeds
Seed lipids and proteins are stored in specialized sub cellular organelles called oil and protein bodies (OBs, PBs) (Purkrtova et al., 2008a;Herman and Larkins, 1999). OBs are composed of a core of neutral lipids (mainly triacyl glycerols) surrounded by a phospholipid monolayer in which a limited number of proteins is found. The protein complement of seed OBs has been described in various botanical families, among them, Brassicaceae (Jolivet et al., 2004;Jolivet et al., 2006, Katavic et al., 2006Jolivet et al., 2009) and Euphorbiaceae (Eastmond, 2004;Popluechai et al., 2011). The number of proteins varies from 3 to 33, the reasons for this high variability remaining unknown. Oleosins are the most abundant proteins in seed OBs. These very hydrophobic proteins belong to a multigenic family (Kim et al., 2002). They are involved in seed OBs stability, size and oil yield (Siloto et al., 2006) and freezing tolerance (Shimada et al., 2008). Caleosin, a OB protein capable to bind Ca 2+ stabilizes OBs in vivo and in vitro (Froissard et al., 2009;Purkrtova et al., 2008b). Less is known on other minor proteins found at the surface of seed OBs. Biogenesis, senescence, lipid composition, structural organisation, and stabilization of OBs remain largely unknown.
In contrast to their oil counterparts, PBs are almost exclusively composed of proteins that serve as sources of nitrogen, sulfur, and carbon compounds during seed germination (Shotwell and Larkins, 1988). These proteins undergo controlled condensation starting in the endoplasmic reticulum. Storage proteins are found either as proforms in intermediate compartments, called precursor accumulating vesicles and dense vesicles or as maturated proteins in the final protein storage vacuole (Robinson et al., 2005).

Seed reserve extraction
Rapeseed oil extraction using pressing has been studied and optimized by testing various conditions of mechanical and thermal treatments of the seeds, according to  (Laisney, 1984). However, it has been observed that whenever a new constraint (quality of the seeds, temperature, etc) emerged, it led to a decrease in performance which was difficult to rally. The crushing of the new double-low rapeseed varieties in the eighties was a well known example. The use of twin-screw extruder for extracting sunflower oil was improved by addition of phosphoric acid and alcohol, which enhanced the lability of the oily spherosomes (Dufaure et al., 1999), thus releasing the oil more easily.
Aqueous Extraction Processing, and Enzyme Assisted Extraction Processing are very attractive. They lead to three distinct fractions. The residual, insoluble material, is rich in cellulose, proteins and entrained soluble materials. The liquid fraction (skim) contains soluble proteins, minerals, carbohydrates and dispersed OBs of small size (Campbell et al., 2011). The oil in water emulsion (cream) is stabilized by proteins and phospholipids. Stabilization by mucilage has also been reported for linseeds (Gros et al., 2003). Recovery of oil from the dispersed OBs and emulsions remains a challenge. High oil extraction yield (up to 99%) from soybean is reported in the literature (see Campbell et al., 2011 for review). However, oil is found either in skim (up to 23%) or in cream (up to 76%), which may need further destabilization for complete oil extraction. Enzyme Assisted Extraction of rapeseed oil and proteins with a set of commercial enzymes improved extraction, but the overall yield remained low (22.2-26% of oil, instead of 16.5% in the absence of enzymes) (Latif et al., 2008), even if the oil quality (in terms of oxidative stability parameters) was better than when solvent extraction was used.
A cognitive approach rather than an empiric one to predict the behaviour of the material during the process would lead to a more efficient fitting of the process to the seed.
According to ''reverse engineering'', a cognitive approach would also allow to suit the composition and the structure of the seed to the need (ability to be destructured, quality of the by-products). It is therefore extremely important to identify the molecular and cellular factors to understand the mechanisms involved in biogenesis of storage oil and protein bodies in seeds, in order to identify key factors for the stability of these storing organelles. This will allow to select rapeseed genotypes with the appropriate traits for easier oil extraction and to develop milder processes which should use as little as possible energy, and ideally no solvent, for safety and toxicity reasons. The ideal products of such extraction should be refined oil and meal devoid of solvent with proteins retaining their initial functional properties.
A continuum of research projects to improve oil and proteins extraction from oilseed plants Due to the economic and environmental issues associated with seed reserve extraction, it is necessary to have academic laboratories and industries work together. Since 2006, the French National Research Agency 4 has supported different projects aiming to improve extraction of oil and proteins from oilseed plants. Genobodies project (2006Genobodies project ( -2009 involved five academic laboratories and one industrial partner. It aimed to analyze oil and protein bodies in Arabidopsis thaliana and Brassica napus seedlings to serve as a basis of knowledge to further improve seed extraction procedure. Genergy project (2008-2012, six academics and two industries) focusses on oil yield increase, nitrogen input reduction and improvement of oil extraction while preserving availability and stability of by-products and saving energy during the extraction process. The genetic variability of a large panel of genotypes, studied in this project, is used to explore several traits (seed yield, oil yield, pressing. . .). The effect of N supply on the traits is also studied. SOPOL project (2008-2012, four academics) aims at producing generic knowledge, using various biological and physical approaches to solve the three dimensional structure of seed OBs ''structural proteins'', namely oleosins and caleosin, and give a molecular basis to OBs stability.

Selected results from the research projects
Extensive oil and protein extraction will not be achieved without substantially increasing the knowledge on the composition, structure and stability of these complex emulsions (containing lipids, proteins, and polysaccharides). Intraspecies variability, and their associated-biological processes associated (reserve accumulation and mobilization of stored material) will deserve special attention too. Results presented in the upcoming sections have been obtained within the frame of the ANR Genobodies, Genergy and SOPOL programs.

On oil bodies
Description of the protein composition of OBs from doublezero winter-type B. napus have been achieved by a combination of proteomic and genomic tools (Jolivet et al., 2006;Jolivet et al., 2009). By comparison with A. thaliana OBs, rapeseed OBs contains numerous integral protein isoforms displaying a high level of sequence conservation with their arabidopsis counterparts (Jolivet et al., 2009). This can be explained not only by the polyploidy nature of the B. napus genome but also by the presence of numerous duplications of chromosomal portions into the rapeseed genome. Genes coding for some OB proteins of interest are expressed during seed development in a pattern similar to that of oil accumulation, and a sequential deposition of integral OB proteins has been established (Jolivet et al., in press). Mutants for the major oleosins have been constructed in A. thaliana and B. napus in spite of the fact that the production of null mutants is challenging for polyploidy species such as B. napus. Solubilization of oleosins by specific polymers prior to structural determination using powerful Synchrotron Light has provided original data on their fold in solution (Gohon et al., 2011). Calcium ions were capable to affect the solubility of caleosin, and to strongly modify the shape and aggregation state of purified OBs (Purkrtova et al., 2008b). The presence of an hydroxysteroid dehydrogenase (HSD1) activity in A. thaliana and B. napus OBs has been detected but the biological function and the substrates of this enzyme remain unknown (d'Andr ea et al., 2007a).

On protein bodies
The mechanisms responsible for the high degree of reserve protein condensation in PBs remain a matter of debate (Herman and Larkins, 1999;Vitale and Denecke, 1999). We aimed to improve condensation of storage proteins by overexpression of candidate genes involved in transport and/or condensation of storage proteins. Overexpression in A. thaliana developing seeds of the receptor VSR1;1, a major vacuolar receptor for storage proteins (Shimada et al., 2003) had no massive impact on oil and on protein quantity and composition. Overexpression of the receptor like protein, RMR (Jiang et al., 2000), could not be tested. Moreover, the absence of homologue of RMR in the rapeseed EST library is not encouraging to pursue with RMR genes. In order to try to modulate expression of 12S globulins and 2S albumins, major constituents of PBs in Brassicaceae with contrasted nutritional values, 12S expression has been silenced. Both 12S and 2S protein expressions are impacted, while oil content was not affected.

Biotechnological outputs
Oleosins represent 2-3 % of the seed mass (d' Andr ea et al., 2007b). It is possible to selectively extract oleosins from seeds (d'Andr ea et al., 2007c) and to produce fractions enriched in oleosins from cakes using a mixture of organic solvents (d'Andr ea et al., 2007b). These fractions are better emulsifiers than phospholipids (PLs), as deduced from interfacial studies and reconstituted OBs studies (C. Lebon, unpublished). Induction of OBs coalescence using ions would be of great interest for the oil extraction. (Purkrtova et al., 2008b). The optimization of pressing in terms of industrial production remains difficult because the crushing facilities are equipped with screw presses with continuous flow capacities of several tons per hour. The need for miniature assays is evident to facilitate the screening of seeds and improve crushing capability. This is especially true for the seeds of the miniature model plant A. thaliana. Using a micropress, the static pressing of different seeds has been studied: B. napus, Linum usitasissimum and A. thaliana. During pressing, the behaviour of A. thaliana seeds differs according to the ecotype considered (Savoire et al., 2010;Savoire, 2008). The behaviour of the seeds during static and continuous pressing is different since in the latter the material must be rigid enough to form a plug which is necessary to increase the pressure in the barrel and the separation of the oil from the cake. Moreover, when applied to linseed, similar evolution of pressing yield according to harvesting date has been highlighted between static and continuous (Komet) presses. Work is in progress to model the continuous pressing with data from static pressing and other rheological characteristics.

Conclusions and perspectives
The issues of better understanding the reserves biogenesis in oleoproteaginous seeds and the molecular basis of their extractability are yet to be answered. Cognitive work, combining knowledge of genes and the structure of the seed at different levels of organization, study of model pressing, and associated mathematical representations, seems unavoidable to propose more gentle conditions of reserves extraction to obtain desired raw materials. A better comprehension of the biogenesis and the reserves accumulation and extractability in model plants as well as the rules to scale up that knowledge for extrapolation to crops are definite topics of research for the future. The recent publication of patents (i.e. DOW WO 2008/ 024840 A2) on the extraction of protein reserves from rapeseed for food purposes, or the emergence of new oilseeds for the bio fuel market (i.e. Jatropha, Camelina) are perfect illustrations of these new topics.