Value of hydroalcoholic treatment of rapeseed for oil extraction and protein enrichment

– This study investigated alternative solvents: ethanol and isopropanol, to replace hexane and enhance the quality and value of oil and meal. Rapeseed oil extraction was carried out using ethanol (92 wt.% or 96 wt.%), isopropanol (84 wt.% or 88 wt.%) or hexane (as reference). Results show that hydroalcoholic extraction increased meal protein content by 13% compared to hexane extraction, but without signi ﬁ cant in ﬂ uence of alcohol and water content. However, increasing water content improved glucosinolate extractability. Isopropanol 84 wt.% eliminated most glucosinolates from the seeds, decreasing glucosinolate concentration by 49 – 73% compared to meals extracted by the other alcohols.


Introduction
Vegetable protein production and use is a major agrifoodsector challenge for the future. On one hand, the everincreasing global demand for animal proteins is driving codemand for vegetable proteins as animal feed. On the other hand, the vegetable protein market for human consumption is booming, bringing with it a number of imperatives in terms of functional and nutritional quality.
Rapeseed meal is a source of protein, characterized by a well-balanced amino acid composition that offers potentially excellent nutritional value (Campbell et al., 2016). Rapeseed meal comes from the oilseed crushing industry, where it is the main co-product, accounting for around 55% of rapeseed mass. After oil extraction, the rapeseed meal contains around 35-40% proteins, plus carbohydrates (30-35%), crude fiber (10-15%), minerals (5-10%), and up to 10% secondary plant metabolites (von der Haal et al., 2014;Carré et al., 2016). However, the current rapeseed oil production-line configurations limit the production of a digestible protein-rich meal (Grala et al., 1994;Adem et al., 2014;Mosenthin et al., 2016). The presence of anti-nutritional components (fibers, glucosinolates, phytates and phenolics) and the loss of protein solubility during the crushing process hinders further value recovery (Bell, 1993;Quinsac et al., 1994;Adem et al., 2014). Rapeseed meal is used as animal feed, mostly for cattle and pigs, but it is less suitable for poultry or fish. Rapeseed meal is currently the world's second-leading source of vegetable protein, way behind soybean.
Increasing concerns for health, safety and environmental care have prompted to study alternative solvents in order to substitute hexane in oil extraction (Li et al., 2014;Sicaire et al., 2015;Baümler et al., 2016;Breil et al., 2017). Sicaire et al. (2015) and Breil et al. (2017) used physical properties of solvents (viscosity, boiling point, vaporization enthalpy and toxicity), theoretical solvent-lipids solubility and extraction experimentations in order to compare different solvents. They demonstrated that extraction efficiency varies depending on the lipid class or component (TAG, phospholids, tocopherols...), and the solvent choice depends on lipids composition of the meal.
Also, regarding the principles of Green Chemistry and Green Extraction (Chemat et al., 2012), isopropanol and ethanol have many benefits as compared to other solvents. They are less toxic than other "green" solvents (Chemat et al., 2012;Prat et al., 2016), already authorized for food extraction by the European Directive 2009/32/EC, readily available and bio-sourced (Breil et al., 2017;Perrier et al., 2017). In term of energy, alcohols are not favorable in the conventional extraction process because of the energy required to evaporate solvent. However, extraction process can be adapted to the use of these solvents. For example, the distillation step of miscella can be replaced by cold demixing method to recover oil and solvent phase. This contributes to save energy and to lead to a more economically viable process .
Isopropanol and ethanol are made more polar solvents than hexane, which has negative consequences on oil selectivity and miscibility (Sicaire et al., 2015;Breil et al., 2017). However, they are known to simultaneously solubilize oil and some nonlipid components of seeds, which helps to concentrate the proteins and detoxify the meal. According to Berot and Briffaud (1983), the extraction of de-hulled rapeseed flour by ethanol or isopropanol at 60 v/v% alcohol increased protein concentration from 53 to 61-63 g per 100 g of de-oiled dry matter and removed of 96-97% of polyphenols and more than 99% of glucosinolates, but not phytin phosphorus. Sinichi and Diosady (2014) used absolute isopropanol to extract de-hulled mustard seed, and showed that the extracted crude oil contained 15% of non-lipid dry matter with 12.8 g of carbohydrates and 2.2 g of nitrogenous matter per 100 g of dry crude oil. It could be deduced that the extraction yield of non-lipid components was around 6% of the initial mass of dry flour in the tested conditions. Furthermore, studies have shown that the extraction efficiency of ethanol and isopropanol is hugely dependent on solvent water content and temperature. Thus, increasing water content in the solvent enhances the extraction of nonlipid components but reduces the efficiency of oil extraction. Fauduet et al. (1995) studied the influence of ethanol-water mixtures on the extraction of glucosinolates from rapeseed meal and showed that increasing water content in the solvent increased glucosinolates extraction yield: 94%, 40% and 14% of glucosinolates were removed by pure water, ethanol at 80 wt.% and ethanol at 97 wt.% alcohol, respectively. However, the use of pure water led to a 23% loss of the initial mass of proteins. Similarly, van Megen (1983) reported that decreasing alcohol content in ethanol (from 85 wt.% to 65 wt.% alcohol) increased the extraction yields of glucosi-nolates (from 49 to 92% at 60°C) and dry matter (from 21 to 27% at 60°C).
Studies to date have mainly tested hydroalcoholic solvents with high water content and dealt with pre-defatted matter to evaluate the ability of solvents to concentrate proteins in the meal and/or detoxify the meal. The aim of this study was to find a non-hexane solvent with good oil extraction capacity without losing non-lipid extraction efficiency. This study investigated the influence of solvent type, i.e. ethanol and isopropanol at various water contents, on oil extraction and rapeseed meal quality.

Materials
Rapeseed was purchased from Alliance Occitane (Toulouse, France). Table 1 reports the rapeseed composition.
Hexane, ethanol and isopropanol were purchased from Quaron (Cestas, France) and were "technical" grade.

Extraction procedure
The process flow diagram in Figure 1 presents the rapeseed extraction procedure. The seeds were first crushed by a Silverson L5M-A high-shear laboratory mixer in the solvent phase (4 min at 8400 rpm and 15°C). The extraction experiments were then carried out in an agitated Nutsche filter (POPE Scientific Inc., Saukville, WI) of total capacity 2.78 L, equipped with a double jacket for temperature control, an agitator for mechanical stirring, and a stainless steel 20 mm mesh (5 layers) on the bottom of the filter to separate the miscella from the solid after the desired extraction time. Fourstage cross-current extraction was performed by immersion of crushing seeds in preheated solvent (solvent/seed ratio: 1500 g/ 100 g per stage; 10 min per stage, stirring at 42 rpm and 50°C). Miscella was recovered by filtration after each washing step, and the miscella samples were used to measure the extracted matter or to analyze the proportion of lipid and non-lipid fractions in the extracted dry matter. Finally, the meal was exposed to ambient air to remove the solvent. Isopropanol at 84.2 wt.% or 87.8 wt.% alcohol, ethanol at 92.0 wt.% or 95.6 wt.% alcohol, and hexane (as reference) were tested as extraction solvents.

Analytical measurements
Samples of miscella were desolventized in an oven (103.5°C, 13 h) to measure dry matter content. Lipid and nonlipid contents of each miscella sample were determined by liquid-liquid extraction in hexane then water (1:1:0.4 wt/wt/wt miscella/hexane/water).
Total oil content of seeds and meals was determined according to NF v03-908. Total oil content of the meal is the sum of the extractable and non-extractable oil content. Extractable oil content was determined after Soxhlet extraction of the meal in hexane. Non-extractable oil content is the part of the residual oil content of the meal that is not readily extractable due to a seed grinding problem. It was quantified by a second grinding and extraction step run on the meal in hexane according to NF v03-908.
Moisture, protein and glucosinolate content of seeds and meals were determined according to ISO 771, ISO 5983-2 and ISO 10633-1, respectively. Protein solubility was determined by precipitation in potash according to ISO 14244. Protein and glucosinolate contents of seeds and meals were expressed on de-oiled dry matter basis using the following equation: where C DDM i is concentration of component i (protein or glucosinolates) in the meal expressed on de-oiled dry matter basis (per mass of DDM), C DM i is the concentration of the component i expressed on a dry matter basis (per mass of DM) and determined by analysis, and C DM L is total oil content in the meal determined by analysis (g/100 g of DM).

Extraction yield
Cumulated lipid and non-lipid dry matter extraction yields (Y k ) were determined using the following equation: where m seed is mass of seeds (g), m mis is mass of miscella recovered at each washing step (g), C seed k and C mis k are content of component k (lipids or non-lipid dry matter) in the seeds and the miscella (%), respectively.

Statistical analysis
Experiments and analyses were repeated twice. To evaluate the effect of extraction conditions, data was submitted to analysis of variance (ANOVA, in Excel) with p < 0.05.  solvents was mainly observed in the first wash: lipid solubility was reduced by the presence of water in alcohol, and hexane remained the most efficient solvent in terms of oil extraction. Lipid extraction yield thus reached 91% with hexane and 83-84% with azeotrope alcohols (95.6 wt.% EtOH and 87.8 wt.% i-PrOH) at the first extraction step. A 3.6 wt.% increase in water content reduced lipid extraction yield to 46% and 66% for 92 wt.% EtOH and 84.2 wt.% i-PrOH, respectively. Consequently, as oil solubility decreased with alcohol content, the process required a higher quantity of solvent, and so a larger number of washings was required in order to extract the same amount of oil. The negative impact of the water content in alcohol on oil solubility was previously demonstrated by Arnold (1956 and1957) for both ethanol and isopropanol and by Sinichi and Diosady (2014) for isopropanol. Nevertheless, as shown in Figure 2a, final yield, i.e. (89.5 ± 3.7) wt.%, was not significantly impacted by the water content in the alcohol/water mixture (p = 0.298).

Crude oil extraction yield
Furthermore, the variation in water content had less influence on lipid extraction by isopropanol than by ethanol. This result is in agreement with Arnold (1956, 1957) who demonstrated that the miscibility of vegetable oils was greater in isopropanol than in ethanol at a given water content. Figure 2b reports non-lipid dry matter extraction yield. Alcohols extracted some non-lipid components as previously reported (Berot and Briffaud, 1983;Sinichi and Diosady, 2014). The extracted non-lipid dry matter represented 11 to 15 wt.% of the initial defatted matter depending on water content. As observed, the higher the water content, the higher the extraction of non-lipid components. The maximum values of non-lipid dry matter yield were achieved after two washing steps. Note that non-lipid components were not extracted with hexane.
After alcoholic extraction and conventional distillation of the miscella, the crude oil contained 75-87% lipids and 12-25% non-lipid compounds depending on solvent tested. These non-lipid compounds are impurities and will complicate oil recovery value. Recent results suggested that applying an oil recovery method by cooling the alcohol miscella instead of distillation leads to recovery of a non-lipid-compound-free oil, as the non-lipid compounds are retained in the solvent-rich phase (Oliveira et al., 2012;Citeau et al., 2018). Table 2 reports the composition of the meal after 4 extraction stages. Protein and glucosinolate concentrations were expressed on a de-oiled dry matter basis to compare results independently of the variation in residual oil content, particularly the oil content that was nonextractable due to a seed grinding problem.

Rapeseed meal quality
The initial rapeseed contained 38 g of proteins per 100 g of de-oiled dry matter. The use of alcohol increased the protein concentration of the meal to around 42-44 g/100 g without significant difference between alcohol types and water content (p = 0.087) but significantly higher than after hexane extraction (p = 1.64 Â 10 À4 ). Thus, the extraction of additional non-lipid compounds (estimated as 11-15 wt.% of the initial defatted matter) increased meal protein concentration by 11-16% compared to the hexane extraction meal. Berot and Briffaud (1983) demonstrated that protein concentration could reach 60 wt.% after ethanol or isopropanol extraction when using dehulled rapeseed. However, dehulling was not applied in the current study.
As shown in Table 2, alcohols removed glucosinolates from the meal, and the higher the water content, the lower the residual glucosinolates concentration, in agreement with Fauduet et al. (1995) and Van Megen (1983). Isopropanol 84.2 wt.% eliminated the most glucosinolates from the oilseed, decreasing glucosinolate concentration by 49-73% compared to meals extracted by the other alcohols.
Table 2 also shows that the use of alcohols reduced protein solubility. Note that the impact was greater with higher-watercontent solvent, in agreement with Sawada et al. (2014) who demonstrated that the protein solubility of soybean was decreased by the presence of water in ethanol and the loss of solubility was accentuated by increasing temperature. A loss of solubility reflects changes in protein conformation and a partial denaturation of the protein that could impact its functionality.

Conclusion
This study investigated the influence of solvent type, i.e. ethanol and isopropanol, at two water concentrations, on rapeseed extraction efficiency and rapeseed meal quality. Alcohol-based solvents extracted 11-15 wt.% non-lipid matters along with the oil, thus increasing protein concentration to 42-43 g per 100 g of meal and decreasing glucosinolates concentration to 7-19 g per 100 g of meal depending on type of alcohol. In comparison, protein and glucosinolate concentrations were 38 g and 25 g per 100 g of meal, respectively, after extraction with hexane. Lipid extraction was less affected by the variation of water content in isopropanol than in ethanol. In the tested conditions, the increase of water content did not affect the final protein concentration of the meal, but strongly affected protein denaturation and removal of glucosinolates. In particular, isopropanol with the highest water content eliminated the most glucosinolates from the oilseed, decreasing glucosinolate concentration by 49-73% compared to meals extracted by the other alcohols.