Method for producing a food grade protein product and / or a feed protein product from plant material

Abstract

A method for producing plant derived protein products is provided. The method includes extraction and purification of plant protein fractions, which may be used for food consumption and / or feed for animals. A plant protein juice is extracted from a green plant material, and used for production of a purified functional 'white' food protein product and / or a 'green' feed protein product.

Description

FIELD OF THE INVENTION The present invention relates to a method for producing plant derived protein. The method includes extraction and purification of plant protein fractions, which may be used for food consumption and / or feed for animals. More specifically, a green plant protein juice is extracted from a green plant material, and used for production of a purified white food protein product and / or a green feed protein product. BACKGROUND OF THE INVENTION The global human population is fast growing, and the consumption of food therefore also rises and there is a need for new ways to produce food products, food ingredients and food additives. One of the single most important compounds for this purpose is protein. Along with the need for more protein due to the expanding human population, there is a global demand for additional protein since more and more people prefer to eat animalbased products with the increasing income and urbanization in developing countries. Moreover, mainly in the western part of the world, there is a growing interest for plantbased food and food products. Therefore, there is a need for additional proteins, novel protein products, and further new ways of producing protein for human food consumption as well as producing proteins for animal feed, especially for monogastric animals (i.e. fish, poultry and pigs). The supply of protein for human food consumption and feed for monogastric animals with the right amino acid profile and at a price that is competitive to the world market is one of the major challenges for farmers and food producers. Some plants contain valuable proteins with a suitable amino acid composition that has high value for both humans and monogastric animals - examples of such plants are alfalfa (Medicago sativa), red clover (Trifolium pratense) and white clover (Trifolium repens), perennial ryegrass (Lolium perenne), and fescue grass species (Festuca sp.). All green plants with photosynthetic active leaves contain a substantial amount of protein in the leaves. In green leaves, the amount of protein based on wet weight basis, is between 1.2 - 8.2%. Of this, the protein ribulose-l,5-bisphosphat carboxylaseoxygenase (RuBisCO) constitute up to 50%. In the plant cells RuBisCO is located in the stroma of the chloroplast, and facilitates the binding of CO2, which is then converted to energy bearing molecules by photosynthesis. For human consumption, RuBisCO has a great amino acid profile, as it contains a large amount of many of the essential amino acids. In plants, the RuBisCO enzyme usually consists of two types of protein subunit, called the large chain (L, about 55,000 Da) and the small chain (S, about 13,000 Da). A total of eight large chains (= four dimers) and eight small chains assemble into a larger complex of about 540,000 Da. Extracts of legumes such as alfalfa and clover are generally recognized as safe (GRAS) by the US FDA and other equivalent national or regional regulatory authorities. Several methods have been used to recover leaf proteins, including pH-induced precipitation, organic solvents, as well as heat coagulation. The drawback of these methods is, that the functional properties of the protein (e.g. gelation, foaming, emulsifying, etc.) can be affected, which is e.g. seen when the protein is denatured. This, in turn, limits the potential applications of the protein as a functional food ingredient. Damborg et al. investigated crude protein yields for different plant materials for screw press extraction followed by a two-step heat precipitation. They found crude protein yields for the food protein fraction based on the crude protein content of the plant material of 1.0% for white clover, 1.0% for red clover, 1.2% for lucerne and 1.0% for perennial rygrass (Damborg et al., 2020). Other process set-ups have also been investigated, e.g. screw press extraction followed by heat precipitation (around 55 °C for 25 min), acid precipitation (pH 4.5) and re-solubilisation steps, where trials using lucerne resulted in a nitrogen based yield of around 1.5% (Nynas et al., 2021). Consequently, there is a need for an efficient, but milder, and sustainable process, which provides a high yield of functional proteins from the plant material. SUMMARY OF THE INVENTION The present invention relates to an improved and non-denaturing method of producing one or more protein products from plant material, and uses thereof, in particular for production of (white) food protein products and / or (green) feed protein products. The plant protein products obtained by the method of the present invention are advantageous by having a high content of proteins with exceptionally good functional properties, by retaining proteins in their native state. Nitrogen-fixing green plants with a high dry matter yield and a high protein content per hectare (HA), such as legumes, are a sustainable solution for protein production. By using a sustainable technology without the use of organic solvents, it is possible to obtain food proteins for human consumption as well as protein-rich concentrates as feed for monogastric animals. The present invention provides a method to be used within a so-called green biorefinery with the main goal of isolating functional proteins from plants, such as functional 3 proteins for use in food products, as well as produce a secondary animal feed protein product. In a first aspect, the present invention provides a method for producing a functional protein product and optionally a feed protein product from plant material, said method 5 comprising the steps: (i) providing a green plant material, and mixing said plant material with an aqueous solution; (ii) disintegrating and pressing the mixture from step (i) to obtain a pulp and a raw green juice; 10 (iii) optionally subjecting the raw green juice to a pre-filtration step to obtain a prefiltered green juice; (iv) subjecting the raw green juice or the pre-filtered green juice to a 1st stage crossflow filtration step to obtain a 1st stage retentate (retained green juice) and a 1st stage permeate (clarified juice); 15          wherein the nominal pore size of the 1st stage crossflow filtration is between 15- lOOnm, most preferably between 15-80 nm, such as between 40-80 nm, such as approx. 60 nm wherein the pressure applied in the 1st stage crossflow filtration is between 0.21.2 bar, preferably between 0.2-0.8 bar, most preferably between 0.2-0.6 bar, 20         and wherein the crossflow of the 1st stage crossflow filtration shall induce turbulent characteristics (i.e. Reynolds number > 2900 for a tube geometry); (v) subjecting the 1st stage permeate (clarified juice) to a 2nd stage crossflow filtration step to obtain a 2nd stage retentate and a 2nd stage permeate; 25          wherein the 2nd stage crossflow filtration has a MWCO between 1-100 kDa, such as between 5-20 kDa, such as approx. 10 kDa wherein the pressure applied in the 2nd stage crossflow filtration is between 1-10 bar, preferably between 1-6 bar, most preferably between 1-4 bar; (vi) subjecting the 2nd stage retentate to a diafiltration step to obtain a dia-concentrate 30 and a dia-permeate; (vii) recovering a functional protein product from the dia-concentrate; and (viii) optionally recovering a feed protein product from the 1st stage retentate (retained green juice). As disclosed herein, the characteristics of the 1st and 2nd stage crossflow filtration steps (e.g. types of membranes applied and the mode of their operation) are selected such that the 1st stage permeate and the 2nd stage retentate comprise RuBisCO protein. In one embodiment, the method of the present invention does not comprise heating above 60°C, such as not above 55°C, 50°C, 45°C, or 40°C, In one embodiment, the green plant material processed in the present invention to obtain a functional protein product and optionally a feed protein product, is selected from alfalfa, clover, lupines, grasses, crucifers, beets, chicory, carrot, radish, cassava, roadside crops and combinations hereof; preferably clover and / or grass; most preferably red clover (Trifolium prantese), white clover (Trifolium repens), perennial ryegrass (Lolium perenne), fescue grass species (Festuca sp.), and blends of one or more of these plant materials, such as clover grass. In one embodiment, the aqueous solution - which the plant material is mixed with - is a phosphate buffer having a pH in the range 6-8, said buffer comprising a reducing agent selected from sodium sulfite and mercaptoethanol and a stabilizing agent selected from EDTA, citrate, and other organic acids; preferably sodium sulfite as reducing agent and citrate as stabilizing agent. In one embodiment, the raw green juice obtained after pressing the plant material is subjected to the pre-filtration step, selected from (i) dead-end filtration, using a filter having a pore size in the range 1-50 pm and (ii) vibration sieve filtration using a sieve filter having a pore size in the range of 10-50 pm (e.g. steel screen). In one embodiment, the 1st stage crossflow filtration is operated using a ceramic membrane, at a crossflow velocity between 1-5 m / s and a pressure between 0.2-0.6 bar. In one embodiment, the 2nd stage crossflow filtration is operated using polymeric membrane and a pressure between 1-8 bar. In one embodiment, the diafiltration is operated using the same membrane as the 2nd stage crossflow filtration, and at a diafiltration factor of 2-10, preferably 2-8, most preferably 2-6. Preferably, the present invention provides a method for producing a functional protein product and optionally a feed protein product from plant material, said method comprising the steps: (i) providing clover and / or grass, and mixing said clover and / or grass with an aqueous solution, said aqueous solution comprising sodium sulfite as reducing agent and citrate as stabilizing agent; (ii) disintegrating and pressing the mixture from step (i) to obtain a pulp and a raw green juice; (iii) optionally subjecting the raw green juice to a pre-filtration step to obtain a prefiltered green juice; (iv) subjecting the raw green juice or the pre-filtered green juice to a 1st stage crossflow filtration step to obtain a 1st stage retentate (retained green juice) and a 1st stage permeate (clarified juice), wherein the 1st stage crossflow filtration is operated at a crossflow velocity between 2-3 m / s and a pressure between 0.3-0.6 bar, using a ceramic membrane having a nominal pore size of between 15-80 nm, and wherein the crossflow of the 1st stage crossflow filtration shall induce turbulent characteristics (Reynolds number > 2900 for a tube geometry); (v) subjecting the 1st stage permeate (clarified juice) to a 2nd stage crossflow filtration step to obtain a 2nd stage retentate and a 2nd stage permeate, wherein the pressure applied in the 2nd stage crossflow filtration is between 1-8 bar, using polymeric membrane having a MWCO between 5-20 kDa; (vi) subjecting the 2nd stage retentate to a diafiltration step to obtain a dia-concentrate and a dia-permeate, wherein the diafiltration step is operated using the same membrane as the 2nd stage crossflow filtration, and at a diafiltration factor of 2-6; (vii) recovering a functional protein product from the dia-concentrate; and (viii) optionally recovering a feed protein product from the 1st stage retentate (retained green juice). In a second aspect, the present invention provides a functional protein product obtainable by the method according to the present invention. Specifically, the functional protein product is recovered from the diaconcentrate in the process of the present invention - either as the direct product or by further processing of the diaconcentrate. In one embodiment, the functional protein product is characterized by being white or whitish in color and having neutral to slightly pleasant smell. In one embodiment, the present invention provides a soluble functional protein product obtainable by the method according to the present invention, wherein the functional protein product is fully soluble in an aqueous solution at pH 6-8, and preferably retains high solubility (>60%) over a broad pH range (pH 4-10) after drying. In one embodiment, the functional protein product has great foaming properties - i.e. having a foaming capacity (FC) >150%, and a foaming stability (FS) >60% after 10 minutes. In one embodiment, the functional protein product has good emulsifying properties -i.e. having an emulsification index (EAI) value >25 m2 / g and an emulsification stability index (ESI) value >35 minutes. In one embodiment, the functional protein product has great gelling properties - i.e. having a least gelling concentration (LGC) of <2% as a heat-set gel. In one preferred embodiment, the present invention provides a soluble functional protein product obtainable by the method according to the present invention, wherein the functional protein product is characterize by (i) being fully soluble in an aqueous solution at pH 6-8, (ii) having a foaming capacity (FC) >150%, (iii) having an emulsification index (EAI) >25 m2 / g, and (iii) having a least gelling concentration (LGC) of <2% as a heat-set gel. In a third aspect, the present invention provides a feed protein product obtainable by the method according to the present invention. Specifically, the feed protein product is recovered from the 1st stage retentate - i.e. the retained green juice, in the process of the present invention - either as the direct product or by further processing of the 1st stage retentate. DESCRIPTION OF THE INVENTION Brief description of the figures: Figure 1: An exemplary overview of a protein biorefinery concept according to the invention, including disintegration of green biomass by mixing with an aqueous liquid and shredding, production of raw green juice by pressing, and a two-stage crossflow filtration separation with subsequent diafiltration for separation and purification of a first stage retentate comprising proteins for feed application and a second stage dia-concentrate comprising protein for food application. Figure 2: An exemplary overview of a protein biorefinery concept according to the invention, including recirculation of process streams. Figure 3: Sketch of the two different process layouts which were evaluated on a demonstrator scale: (A) using bagfilters after the screw press, and (B) using heat treatment and centrifugation after the screw press. Figure 4: Flux and TMP during the first crossflow filtration stage (A) for Run 1, and (B) for Run 7. Figure 5: (A) SDS-PAGE gel for streams from Run 1. Lanes (from left to right): 1) Juice from screw press. 2) Heat-treated juice. 3) Juice after centrifugation. 4) Protein ladder (Bands from bottom up: 14.4; 18.4; 25.0; 35.0; 45.0; 66.2; 116.0 kDa). 5) Retentate from the 1st crossflow filtration. N.B this stream was diluted during stream-recovery. 6) Permeate from the 1st crossflow filtration. 7) Retentate from the 2nd crossflow filtration. 8) Permeate from the 2nd crossflow filtration. 9) Concentrate after diafiltration. 10) Permeate during diafiltration. (B) SDS-PAGE gel for streams from Run 7. Lanes (from left to right): 1) & 2) Empty. 3) Juice from screw press. 4) Empty. 5) Juice after 1 pm bag filter. 6) Empty. 7) Permeate from the 1st crossflow filtration. 8) Retentate from the 2nd crossflow filtration. 9) Permeate from the 2nd crossflow filtration. 10) Concentrate after diafiltration. 11) Empty. 12) Protein ladder (Bands from bottom up: 14.4; 18.4; 25.0; 35.0; 45.0; 66.2; 116.0 kDa). Figure 6: Crude protein concentrations in dry powders and plant material [mass%] for each run. Error bars indicates each of the two individual measurements used in the determination of the value. Only leaves from run 8 and forward were saved for crude protein analysis. Figure 7: Crude protein balance based on juice from the screw press for runs 6 to 17. Figure 8: Volcano plot illustrating protein-level depletion in (A) pre-filtered green juice vs. permeate from 1st stage filtration and (B) pre-filtered green juice vs. diafiltration concentrate following 2nd stage filtration step. 1st axis shows Iog2 fold-change of normalized LFQ intensities and 2nd axis shows adjusted p-value based on triplicate analysis. Depletion is defined as at least two-fold decrease (i.e. Iog2 fold change > 1) in protein abundance (normalized LFQ intensities) and significance level is set at 5% FDR (p-value > 0.05). Chlorophyll a-b binding proteins (Q6T705, Trifolium pratense, black arrow) and Photosystem II CP47 reaction center protein (K4PMQ7, Festuca pratensis, orange arrow) are shown as example proteins which are highly depleted by the filtration processes. Figure 9: Heatmap showing protein-level normalized LFQ intensities and hierarchical clustering for triplicates of pre-filtered green juice (PFGJ), 1st stage filtration permeate (clarified juice, CJ), and diafiltration concentrate (DC). (A) including all proteins, (B) only including proteins with differential abundance across the different samples. Figure 10: Visual appearance of the different freeze-dried protein fractions obtained from the process of the present invention. (A) Raw green juice. (B) 1st stage permeate. (C) Diafiltration concentrate. Figure 11: Schematic representation of the lab-scale emulation of prior art processes, where green biomass is initially pressed, and residual fibers are removed prior to heat treatment. Following centrifugation, the soluble fraction is precipitated by addition of acid to pH 4.2. As an alternative, heat treatment and a subsequent two-stage membrane filtration process is investigated as a combination of prior art and the process of the current invention. Figure 12: Reducing SDS-PAGE analysis of protein content in different fractions from the lab-scale experiment shown in Figure 11 solubilized in (A) Water and (B) 100 mM ammonium bicarbonate (pH 8.6) with 0.2 % (w / v) sodium dodecyl sulfate. Lane 1: Green biomass; lane 2: pulp; lane 3 raw green juice; lanes 4 & 5: retentate and permeate (green juice) following pre-filtration, respectively; lanes 6 & 7: precipitate and supernatant from heat treatment, respectively; lanes 8 & 9: retentate and permeate of 1st stage filtration of supernatant from heat treatment, lanes 10 & 11: retentate and permeate of 2nd stage filtration of supernatant from heat treatment, lanes 12 & 13 precipitate and supernatant from acid treatment of supernatant from heat treatment. The ~25 kDa chlorophyll a-b BP band is indicated with a dashed box while the desired RuBisCO subunits (large and small at ~50 kDa and ~15 kDa, respectively) are indicated in solid boxes. Figure 13: Reducing SDS-PAGE analysis of protein content in different fractions, obtained from the process of the present invention. Samples originate from the batch processed in run 11 of Example 3A. Lanes 1) Raw juice sample 2. 2) Raw juice sample 1. 3) Juice after SIL sample 1. 4) Juice after 50 pm bag filter sample 1. 5) Juice after 1 pm bag filter sample 1. 6) Juice after 1 pm bag filter sample 2. 7) Retentate from the 1st crossflow filtration. 8) Protein ladder (Bands from bottom up: 14.4; 18.4; 25.0; 35.0; 45.0; 66.2; 116.0 kDa). 9) Permeate from the 1st crossflow filtration sample 5. 10) Permeate from the 1st crossflow filtration mixed sample 1. 11) Permeate from the 1st crossflow filtration mixed sample 2. 12) Retentate from the 2nd crossflow filtration. 13) Concentrate after diafiltration. 14) Permeate from the 2nd crossflow filtration sample 2. The ~25 kDa chlorophyll a-b BP band is indicated with a dashed box while the desired RuBisCO subunits (large and small at ~50 kDa and ~15 kDa, respectively) are indicated in solid boxes. Figure 14: Histogram of selected proteins: the desired proteins (RuBisCO large and small subunits) and the undesired proteins, targeted for removal (here exemplified by "Chlorophyll a-b binding proteins" and "Photosystem proteins"). From lab scale experiments (Figure 11), the following samples are shown: unprocessed grass (UG), raw green juice (RGJ), precipitate (HP) and supernatant (HS) after heat treatment of RGJ, 1st stage retentate (HR1) of supernatant after heat treatment, 2nd stage retentate (HR2) of supernatant after heat treatment, and acid precipitate (AP) of supernatant after heat treatment. From pilot-scale production (batch corresponding to run 6, Example 3A), the following samples are shown: raw green juice (RGJ), 1st stage retentate (RI), 1st stage permeate (Pl), and diafiltration concentrate (DC). The food-grade protein product ('white protein') obtained from the process of the present invention (DC), shows no content of the undesired proteins. For comparison, the feed-grade protein product ('green protein') obtained from the process of the present invention (RI), shows high content of the undesired protein. Lastly, an example of prior art (AP) shows residual content of the undesired proteins. Figure 15: Visual appearance of representative samples selected for sensory analysis. A) Lyophilized dia-concentrate from run 6 as per Example 3A; B) Spray dried dia-concentrate from run 6 as per Example 3A; C) Lyophilized dia-concentrate as per Example 4; D) Prototype "feed-grade" protein isolate. Figure 16: Visual appearance of lyophilized process samples from the lab-scale emulation of a prior art process as per Example 5. A) Cryoground leaf material; B) Raw green juice; C) Pellet after heat treatment and centrifugation; D) Supernatant after heat treatment and centrifugation; E) Retentate from 1st stage filtration; F) Retentate from 2nd stage filtration; G) Pellet from heat treatment and acid precipitation. Abbreviations, terms, and definitions: The term "green biomass" or "green plant material" refers to chlorophyll-containing plants, such as leaves of plants. The green plant may be fresh plant material, but may also cover ensilaged or even dried plant material. The term "functional protein" or "native proteins" refers to a protein that is in its natural, biologically active, and correctly folded state as it exists within a living organism or in its natural environment. In this state, the protein typically exhibits its intended biological function(s) and interaction(s). Hence, in the present context of the plant proteins undergoing various processing steps, this term refers to proteins which have preserved at least one of the properties selected from protein activity, protein solubility, gelatinizing, water absorption, oil absorption, emulsifying, and foaming properties. The term "processing facility" refers to the facility where the harvested plant material is processed to extract and purify one or more protein(s) and protein product(s) as disclosed herein. The term "dry mechanical pretreatment" relates to mechanically processing the plant material in a dry process by cutting, chopping, crushing, milling, or a similar process, such that the plant material is reduced in size. The term "wet mechanical pretreatment" is in the present context a mechanical treatment of the plant material in aqueous phase, resulting in an efficient opening of the plant cells. This may be performed using a blender, mill, macerator, pulper, or refiner. The term "aqueous solution" refers to a water-based solution. In the present context, the aqueous solution comprises components that act as pH buffer and / or other compounds that help stabilize the proteins extracted from the plant material. The term "pulp" refers to the moist solid residue produced by pressing (e.g. screw pressing, juicing or extruder) green plant material. The term "raw green juice" refers to the liquid stream produced by pressing (e.g. screw pressing) green plant material. The term "green juice" generally refers to the feed stream going into the 1st crossflow filtration stage. This stream may be a raw green juice, or may have been treated by prefiltration or other preprocessing means. The term "1st stage retentate" refers to the retentate obtained from the 1st crossflow filtration stage. The term "1st stage permeate" refers to the permeate obtained from the 1st crossflow filtration stage. The term "2nd stage retentate" refers to the retentate obtained from the 2nd crossflow filtration stage. The term "2nd stage permeate" refers to the permeate obtained from the 2nd crossflow filtration stage. The term "dia-concentrate" refers to the concentrate obtained after diafiltration of the retentate from the 2nd crossflow filtration stage. The term "dia-permeate" refers to the permeate obtained from the diafiltration of the retentate from the 2nd crossflow filtration stage. The term "Transmembrane pressure" (TMP) refers to the pressure difference between feed and permeate side during membrane operations. The term "Crossflow velocity" (CFV) refers to the velocity of the flow tangential to the membrane surface on the feed side of the membrane during crossflow membrane operations. The term "Molecular weight cut-off" (MWCO) refers to the lowest molecular weight where 90% of a substance of that molecular weight is retained by the membrane. The term "green protein" refers to a process stream or product comprising protein, and additionally comprising green color pigment. In the process of the present invention, a "green protein product" is obtainable from the first stage retentate. The term "white protein" refers to a process stream or product mainly comprising protein, mostly devoid of green color pigment. In the process of the present invention, a "white protein product" is obtainable from the second stage dia-concentrate. Detailed description of the invention: The present invention concerns a method for producing plant derived protein. The method includes extraction and purification of plant protein fractions, which may be used for food consumption and / or feed for animals. More specifically, a green plant protein juice is extracted from a green plant material, and used for production of a purified white food protein product and / or a green feed protein product. The present inventors have surprisingly found that using the method of the present invention (1) a considerably higher yield of food grade protein is obtained compared prior art technologies - see Example 3A; (2) as a function of the process conditions applied, the quality of the product is improved compared to prior art technologies, e.g. taste (removal of 'bitter taste' proteins), color (removal of 'green' proteins) - see Example 4 and 5; and other improved functional properties incl. foaming and emulsifying properties - see Example 6. Several prior art publications disclose extraction of protein from plant material. However, none of the prior art suggests the combination of a two-stage ultrafiltration and diafiltration to obtain a functional food protein product devoid of undesired bad-taste compounds, as well as a feed protein product, as disclosed herein. The prior art does not disclose that heat treatment of plant material may be substituted by such two-stage ultrafiltration, optionally combined with a pre-filtration, as disclosed herein. Specifically, the present invention provides a process comprising two membrane filtration steps, wherein the first filtration step is characterized by a smaller pore size compared to a standard microfiltration membrane filtration step, and the process does not have a heat treatment and centrifugation step. The present invention provides the benefit of obtaining a native, non-degraded, functional protein product, while at the same time getting rid of undesired color and taste compounds from the final 'white' food protein product. It would not immediately be expected that a smaller pore size would facilitate an improved process in regards to process performance and food grade protein yield, compared to more open membranes. On the contrary it might be expected that the smaller pore size would result in additional fouling buildup, as more components will be retained by the membrane, thereby reducing both flux and protein selectivity. A smaller pore size also generally negatively impacts flux due to an increased flow resistance through smaller pores. Surprisingly, the present invention finds an improved process performance by utilizing the two-stage ultrafiltration filtration process, as further disclosed herein. Without wishing to be bound by theory, it is suspected that - the membrane separation applied in the present invention avoids precipitation and coagulation of proteins, resulting in improving techno-functionality and yields of the food grade proteins, compared to using heat and / or acid precipitation; - the use of a small pore size membrane in the first stage filtration mitigates fouling (the pores are small and some of the larger fouling components cannot internally foul the membrane); - the selective removal of "green" colorants ("grassy sensory attributes") is due to 1st stage filtration removal of residual chloroplasts and other particles, as well as chlorophylbinding proteins bound to residual cellular matrix components or bound in larger complexes which also include the chlorophyll itself (because things are retained native under physiological conditions). The chlorophyll is likely responsible for color, but is also expected it to influence taste and smell; - the stabilizing buffer present during the protein extraction increases protein quality and yields; and - quick processing and exposure to the stabilizing buffer is important for protein quality. I. Method of preparing one or more protein product(s) from plant material As disclosed herein, the present inventors found that there is a large potential for recovering functional proteins with valuable properties from plants, and further additionally utilize other lower value proteins from the plant. The present invention concerns a method for providing a functional protein product for food application and / or a feed protein product. The invention comprises several steps that result in high overall protein yields. In the first step, freshly harvested plant material is disintegrated and pressed in an aqueous stabilizing solution to yield a liquid fraction and a press cake. This step facilitates a high release of soluble proteins into the aqueous solution while at the same time ensures stabilization of the extracted proteins. The press cake may be further subjected to further pressing steps in order to release more proteins from the press cake. The final press cake may be used to produce feed pellets for cattle, dairy cows, horses, poultry, or pets, or may be used as a substrate in biogas production, or in a fermentation process, such as bioethanol production. The food and feed proteins - which are the main focus of the present invention - are obtained from the liquid fraction after pressing, by further processing using different filtration steps, including an optional pre-filtration, followed by a stage 1 and a stage 2 filtration, and finally a diafiltration step, as detailed below. A feed protein product is obtained from the retentate from the stage 1 filtration step, while a food protein product is obtained from the concentrate from the diafiltration step. The desired plant protein RuBisCO is smaller than the pore size used in the stage 1 filtration, hence the majority of the RuBisCO will be in the permeate, which is collected to be used as feed in the subsequent stage 2 filtration. The stage 1 filtration therefore acts to remove larger species (as well as RuBisCO closely bound to and / or adsorbed onto larger species including (but not limited to) residual cellular material). Less RuBisCO is 'lost' compared to the undesired constituent in the plant juice. Afterwards the stage 2 filtration is used in combination with diafiltration to concentrate RuBisCO and to wash out smaller species, thereby increasing the purity of the final food protein product. As RuBisCO is larger than the pore size used in the stage 2 filtration, the retentate from this process contains the desired protein enriched for RuBisCO, which may be further concentrated and dried to the final protein powder. The method is distinctive in that the specific filtration steps provide a high yield of native, non-degenerated, high value food protein for human consumption and feed protein for monogastric animals. In a first aspect the present invention concerns a method for producing a functional protein product and optionally a feed protein product, said method comprising: (i) mixing a green plant material with an aqueous solution; (ii) disintegrating and pressing mixture from (i) to obtain a press cake and a raw green juice; (iii) subjecting the raw green juice to a 1st stage crossflow filtration step to obtain a 1st stage retentate (retained green juice) and a 1st stage permeate (clarified juice); (iv) subjecting the 1st stage permeate (clear juice) to a 2nd stage crossflow filtration step to obtain a 2nd stage retentate and a 2nd stage permeate; (v) subjecting the 2nd stage retentate to a diafiltration step to obtain a dia-concentrate and dia-permeate; (vi) recovering a functional protein product from the dia-concentrate; and (vii) optionally recovering a feed protein product from the 1st stage retentate (retained green juice). The process parameters of the 1st stage crossflow filtration step and 2nd stage crossflow filtration step are chosen such that the 1st stage permeate and the 2nd stage retentate will comprise RuBisCO protein. In one preferred embodiment, the aqueous solution helps stabilize components extracted from the plant material, prevents denaturation of proteins, and reduces undesired protein modifications including (but not limited to) oxidation and cross-linking. In one embodiment, the pressed raw green juice is subjected to a prefiltration step prior to the 1st stage crossflow filtration, to obtain a prefiltered green juice which is then subjected to the 1st stage cross flow filtration. In one embodiment, the 1st stage retentate (retained green juice) is further processed into a feed protein product. Several different separation processes including microfiltration and heat treatment has previously been proposed to fractionate juices produced from green biomass and retain residual particulate matter along with e.g., chlorophyll and other compounds associated with color, smell and taste, while producing a white protein rich fraction. The present inventors have, however, found that a direct use of a two-stage ultrafiltration method, can advantageously be used for this separation, wherein the first stage membrane filtration is characterized by a smaller pore size compared to a standard microfiltration step, and without a previous heat treatment or decanter centrifugation step. This retains the favorable functional properties for the fractionated proteins (e.g. high solubility, foaming, gelling), while limiting production rate decline and avoiding the need for more extensive pre-treatment, which is traditionally needed to remove foulants prior to conventional microfiltration membrane operations. The 1st stage membrane filtration efficiently removes particulate matter along with residual chloroplasts, cell debris organelle-bound and / or large macromolecular complexes including (but not limited to) bound chlorophyll as well as potentially viable contaminant microorganisms endogenous to the plant material. The 2nd stage membrane filtration with an even smaller molecular weight cut-off compared to the 1st stage filtration, preferably combined with a diafiltration step, will further purify and concentrate the white protein fraction, thus removing lower molecular weight compounds such as salts from the aqueous solution as well as undesired small molecules endogenous to the plant material. Preferably, the process of the present invention does not involve any steps that lead to precipitation of the food grade proteins. As disclosed herein, food proteins result from two consecutive filtration steps of pressed green grass juice, where the permeate of the first filtration step is applied in the second filtration, and where the retentate of the second filtration step is diafiltered, resulting in the food protein product, as disclosed in greater details herein. It is this process stream, that shall not involve any steps that lead to precipitation of proteins. Precipitation of proteins could for example occur due to heat treatment (e.g. temperatures above 50 degree Celcius) and / or a pH change (e.g. pH below 5). Hence, in one embodiment, the process stream leading to food grade protein does not involve heating above 50 degree Celsius. Hence, in one embodiment, the process stream leading to food grade protein does not involve reducing the pH to below pH 5. I.i Plant material All kinds of fresh plant materials from various plant species and genera may be useful in the method of the present invention. However, it is presently preferred to employ plant materials that are generally considered as safe for humans (GRAS). The plant material may be organic or conventional plant material. As one of the objects of the present invention is to provide plant material derived proteins, it is preferred to use plant materials with a high content of proteins. Protein content is measured by the Kjeldahl, Dumas, or elemental analysis methods (e.g. CHNO(S)), where the content of nitrogen is representative for the amount of crude protein in the same, i.e. crude protein is N*6.25. Hence, in a useful embodiment of the present invention the plant material has a protein content of at least 0.1% (w / w) based on fresh (non-dried) plant material, such as least 0.2% (w / w), such as 0.5% (w / w), such as at least 1% (w / w), such as at least 2% (w / w), such as at least 3% (w / w), such as at least 4 % (w / w), such as at least 5% (w / w), such as at least 6% (w / w) based on fresh (non-dried) plant material. In further embodiments, the plant material has a protein content in the range of 0.1-6% (w / w) based on fresh (non-dried) plant material, such as in the range of 0.2-5% (w / w), such as in the range of 0.3-4% (w / w), such as in the range of 0.4-3% (w / w), such as in the range of 0.5-2% (w / w), such as in the range of 1.5-4% (w / w), such as in the range of 1-3% (w / w), such as in the range of 2-5% (w / w), such as in the range of 2-3% (w / w), such as in the range of 0.1-2.5% (w / w), such as in the range of 1-2% (w / w) based on fresh (non-dried) plant material. In another useful embodiment of the present invention the plant material has a dry matter (DM) protein content of at least 2% (w / w) - i.e. calculated based on dried plant material, such as at least 4% DM (w / w), such as at least 6% DM (w / w), such as at least 8% DM (w / w), such as at least 10% DM (w / w), such as at least 12% DM (w / w), such as at least 14% DM (w / w), such as at least 16% DM (w / w), such as at least 18% DM (w / w), such as at least 20% DM (w / w), such as at least 22% DM (w / w), such as at least 24% DM (w / w), such as at least 26% DM (w / w), such as at least 28% DM (w / w), such as at least 30% DM (w / w), such as at least 32% DM (w / w), such as at least 34% DM (w / w), such as at least 36% DM (w / w), such as at least 38% DM (w / w), such as at least 40% DM (w / w). In further embodiments, the plant material has a protein content in the range of 2-40% DM (w / w), such as in the range of 10-40% DM (w / w), such as in the range of 15-40% DM (w / w), such as in the range of 20-38% DM (w / w), such as in the range of 25-35% DM (w / w), such as in the range of 30-35% DM (w / w), such as in the range of 5-35% DM (w / w), such as in the range of 10-30% DM (w / w), such as in the range of 12-28% DM (w / w), such as in the range of 15-25% DM (w / w), such as in the range of 16-24% DM (w / w), such as in the range of 18-22% DM (w / w), such as around 20% DM (w / w). It is further useful that the dry matter content of the plant material is at least 10%. Hence, in a useful embodiment of the present invention the plant material has a dry matter content of at least 8% (w / w), such as at least 9% (w / w), such as at least 10% (w / w), such as at least 11% (w / w), such as at least 12 % (w / w), such as at least 13% 16 (w / w), such as at least 14% (w / w), such as at least 15% (w / w), such as at least 16% (w / w), such as at least 17% (w / w), such as at least 18% (w / w), such as at least 19% (w / w), such as at least 20% (w / w), such as at least 21% (w / w), such as at least 22% (w / w). In further embodiments, the plant material has a dry matter content in the range of 8-30% (w / w), such as in the range of 10-28% (w / w), such as in the range of 12-26% (w / w), such as in the range of 14-24% (w / w), such as in the range of 15-20% (w / w. It may be preferred to use perennial plants with high biomass yield and / or low cost plant materials in order to reach an improved process economy. The green parts of the plants are preferred for the claimed process. In one preferred embodiment, the plant material is a legume. In one embodiment, the plant material provided and used in the method of preparing one or more protein products is selected from alfalfa, clover, and lupines. Other plant species and families such as grasses, crucifers, beets, chicory, carrot, radish, cassava, roadside crops and / or combinations hereof may also be utilized in the present method of preparing one or more protein products. Preferably, the plant material is clover and / or grass. Most preferably, the plant material is selected from red clover (Trifolium prantese), white clover (Trifolium repens), perennial ryegrass (Lolium perenne), fescue grass species (Festuca sp.), and blends of one or more of these plant materials, such as clover grass, which refers to a blend of one or more clover species and one or more grass species. As evidenced in the examples section, a functional (white) protein product can be obtained from such plant material using the method as disclosed herein. I.ii Harvest, transport and storage of plant material The green plant material is harvested (e.g. by cutting) and then transported to a processing facility for obtaining one or more protein products from the plant material. In the time from harvest and until transport to the processing facility, the green plant material may be left in swaths followed by transport to the processing facility. Preferably, the plant material shall be transported to the processing facility as fast as possible after harvest, and also be further processed to obtain the proteins product(s) as fast as possible after harvest. The harvested plant material may optionally be stored before use. It is preferred that the plant material after harvest and transport to the processing facility is stored in a cold place for as short as possible time. In a preferred embodiment, the storage temperature of the green plant material until use is preferably below 20°C, but higher than 0°C, in order to avoid spoilage. In one embodiment the storage temperature of the plant material is preferably below 20°C, such as below 18°C, such as below 15°C, such as below 10°C, such as below 8°C, such as below 6°C, such as below 5°C, such as below 4°C, such as below 3°C, such as below 2°C, such as below 1°C, but higher than 0°C. In a further preferred embodiment the time between harvest of the plant material and further processing of the harvested plant material to obtain the proteins product(s) is less than 24 hours, such as less than 20 hours, such as less than 14 hours, such as less than 12 hours, such as less than 10 hours, such as less than 8 hours, such as less than 6 hours, such as less than 4 hours, such as less than 2 hours. Prior to the following disintegration of the plant material, as disclosed below, the plant may be washed, preferably in cold water, to remove impurities and to lower the temperature of the plant material. To avoid spoilage it may be preferred that the plant material has a temperature at or below 10°C, such as below 9°C, e.g. below 8°C, such as below 7°C, e.g. below 6°C, such as below 5°C before disintegration. Preferably the plant material has a temperature at or below 10°C before disintegration. Thus, in an embodiment the plant material is washed and preferably cooled to a temperature at or below 10°C, such as at or below 9°C, such as at or below 8°C, such as at or below 7°C, such as at or below 6°C, such as at or below 5°C; such as in the range from 2-10°C, such as in the range from 5-10°C. I. Hi Disintegration and pressing of the plant material Disintegration refers to the process or action of breaking down or breaking apart into smaller components, resulting in the loss of the original form or structure. Specifically, for the present invention disintegration shall result in the opening of the plant material to facilitate extraction of plant proteins. The harvested plant material is optionally reduced in size before further processing, such as reduced in size by mechanical dry treatment of the plant material. The plant material may be reduced in size by an initial mechanical dry treatment to obtain lengths suitable for the subsequent processing (make handling easier), such as resulting in plant material having an average length of between 1-10 cm, such as between 1-8 cm, such as between 1-6 cm, such as between 2-5 cm, such as between 2-4 cm. In one embodiment, the dry mechanical treatment comprises cutting, chopping, crushing, shredding, milling, grinding, and / or combinations thereof. Plant material can be disintegrated into smaller pieces by chipping or shredding. Milling involves the use of mechanical forces, such as impact, compression, and shear, to disintegrate the biomass particles. Grinding involves using grinders or pulverizers to disintegrate the plant material into finer particles. In one preferred embodiment, the harvested plant material is mixed with an aqueous solution and disintegrated by a wet mechanical treatment, resulting in disintegration of the plant cells. In one embodiment, the wet mechanical treatment comprises shredding, milling, grinding, and / or combinations thereof. In one embodiment, the harvested plant material is first disintegrated by a mechanical dry treatment, and then mixed with an aqueous solution, and further disintegrated by a wet mechanical treatment. A dry mechanical treatment is preferably done immediately before a wet mechanical treatment and pressing, as cutting the plant materials may result in activation of enzymes - e.g. proteolytic and oxidative enzymes - which are undesired. In another embodiment, the harvested whole plant material is mixed with an aqueous solution and then disintegrated by a wet mechanical treatment. Stabilization of the proteins and prevention of disintegration of the protein part during extraction and purification process is very important. Therefore, in one embodiment, an aqueous solution comprising stabilizing agents for preventing disintegration and / or undesired modifications of the proteins is used, said solution preferably further having buffering properties to keep the pH within a preferred range. In another embodiment, an aqueous solution having buffering properties is used to keep the pH within a preferred range, and further stabilizing agents may be added to prevent disintegration and / or undesired modifications of the proteins. In one embodiment, sodium sulfite is used as reducing agent in the aqueous solution. Preferably, the aqueous buffering solution comprises 2-25 mM sodium sulfite such as 10-20 mM, more preferably 5-20 mM, such as 5-15 mM, most preferably 5-15 mM, such as 10-15 mM. In another embodiment, mercaptoethanol is used as reducing agent in the aqueous solution. In one embodiment, sodium citrate tribasic dehydrate is used as stabilizing agent in the aqueous solution. Preferably, the aqueous buffering solution comprises 5-40 mM sodium citrate tribasic dehydrate such as 5-35 mM, more preferably 10-40 mM, such as 20-40 mM, most preferably 25-40 mM, such as 25-35 mM. In another embodiment, EDTA is used as reducing agent in the aqueous solution. Preferably the pH of the grass juice after addition of the aqueous solution is in the range pH 6-8 to help prevent denaturing of the proteins. In one embodiment, the pH of the grass juice after addition of the aqueous solution is in the range of 4-10, such as in the range of 4.5-9.5, such as in the range of 5-9, preferably in the range 5.5-8.5, more preferably in the range 6-8. As recognized by a person skilled in the art, the pH buffer capacity of an aqueous solution may be adjusted by selecting different buffer systems. Preferably, a phosphate buffer system is used, comprising potassium phosphate and sodium phosphate to ensure good buffering capacity to help prevent denaturing of proteins. The aqueous solution in which the plant material is soaked and / or mixed, preferably has buffering properties. In one embodiment, the aqueous solution is a phosphate buffer, such as a phosphate buffer comprising potassium phosphate (e.g. potassium phosphate monobasic) and sodium phosphate (e.g. di-sodium hydrogen phosphate dihydrate). In one embodiment, the aqueous solution further comprises a reducing agent, such as a reducing agent selected from sodium sulfite and mercaptoethanol and / or a stabilizing agent, such as a stabilizing agent selected from EDTA, citrate (e.g. sodium citrate tribasic dehydrate), and other organic acids in order to improve stabilization of the proteins. In a preferred embodiment, the aqueous solution is in the range pH 6-8 by a phosphate buffer, and further comprises sodium sulfite as reducing agent, and sodium citrate tribasic dihydrate as stabilizing agent. Preferably, the aqueous solution is in the range pH 6-8 by a phosphate buffer, and comprises 10-40 mM sodium citrate tribasic dehydrate (CeHsNasO? (x2H2O)), and 5-20 mM sodium sulphite (Na2SOs). Preferably, the aqueous solution has a pH in the range 6-8, and comprises 10-100 mM di-sodium hydrogen phosphate dehydrate (Na2HPO4 (x2H2O)), 2-20 mM potassium phosphate monobasic (KH2PO4), 10-40 mM sodium citrate tribasic dehydrate (CeHsNasO? (X2H2O)), and 5-20 mM sodium sulphite (Na2SOs). Most preferably, the aqueous solution comprises 22 mM di-sodium hydrogen phosphate dehydrate (Na2HPO4 (X2H2O)), 7 mM potassium phosphate monobasic (KH2PO4), 40 mM g / L sodium citrate tribasic dehydrate (CeHsNasO? (X2H2O)), and 12 mM sodium sulphite (Na2SOs). The plant material is mixed, soaked, and / or dispersed, or by other means brought in contact with the aqueous solution. It should preferably be ensured that the plant material is thoroughly mixed and / or soaked, such that the aqueous buffer solution helps prevent degradation of proteins from the plant material. In one embodiment, the ratio of aqueous buffer solution (kg) to plant material (kg dry weight) is max 10:1 - i.e. no more than 10 kg buffer I kg DM plant material. In one embodiment, the ratio of aqueous buffer solution (kg) to plant material (kg dry weight) is between 1-10 kg buffer / kg DM plant material, such as between 2-10 kg buffer / kg DM plant material, preferably between 3-10 kg buffer / kg DM plant material. A wet mechanical treatment is performed to open the plant cells for efficient extraction of the proteins. Specifically, the wet mechanical treatment refers to the plant material being soaked, dispersed, or mixed with the aqueous buffer solution while being mechanically treated, such as by shredding, milling and / or refining the plant material while in contact with the aqueous solution. In one embodiment, the plant material is shredded while in contact with the aqueous solution, before being pressed, to facilitate efficient pressing of the plant material, as further disclosed herein. The plant material may be shredded by use of a blender. In one embodiment, the plant material is milled while in contact with the aqueous solution, before being pressed, including e.g. conical refiners or a disc type refiners, operated at ambient or atmospheric pressure: so-called "atmospheric refining". In one embodiment, the plant material is refined while in contact with the aqueous solution, before being pressed, to shear or crush the plant material suspended in the aqueous solution, such as by using e.g. a toothed colloid mill. A suitable efficient mechanical wet process is one, which efficiently increases the overall surface area in order to enable degradation of the material into a satisfactory level and opens up the cells thus facilitating an efficient release of plant juice containing carbohydrates and proteins, while not being too destructive - i.e. while ensuring the integrity of the extracted proteins, in order to obtain functional protein after extraction. After wet mechanical processing, preferably such as after shredding, the plant juice is separated from the remaining plant solids. This separation may be performed by pressing, centrifugation, decantation, or filtration. In a preferred embodiment, the plant material is pressed to extract a plant juice from the solids. The pressing and separation of the plant material result in two fractions: a liquid (a raw green juice) and a moist solid residue (pulp). The plant press cake comprises fibers of cellulose, hemicelluloses, pectin and lignin as well as residuals of what is in the plant juice. The plant juice mostly comprises small fibers, cell debris, chloroplasts, proteins, as well as organic acids, amino acids, peptides, sugars, and salts in suspension. In a preferred embodiment, the pressing process is carried out using a screw press, juicer or an extruder. A screw press is a mechanical device designed to extract liquids from solids, separate different components of a mixture, or compress materials using the principle of a rotating screw or auger within a cylindrical or conical enclosure. The pressing step - e.g. the screw press - may be operated as a one-step or a two-step pressing process, and / or a twin screw press could be of interest. The screws press may be operated with a counter pressure of between 1-15 bars, such as between 2-12 bars, preferably between 2-10 bars, such as between 4-10 bars, most preferably between 4-8 bars; at a speed of between 1-50 rpm, such as between 2-45 rpm, preferably between 2-40 rpm, such as between 5-30 rpm, most preferably between 5-20 rpm. The liquid (raw green juice) obtained from the pressing preferably has a protein content (protein per dry weight) of at least 10% (w / w), such as at least 15% (w / w), such as at least 20% (w / w), such as at least 25% (w / w), such as at least 30% (w / w), such as at least 35% (w / w); or a protein content in the range of 10-40% (w / w), a protein content in the range of 10-30% (w / w), a protein content in the range of 10-20% (w / w), a protein content in the range of 20-40% (w / w), a protein content in the range of 30-40% (w / w), such as in the range from 12-38% (w / w), such as in the range from 14-36% (w / w), such as in the range from 15-35% (w / w), such as in the range from 18-32% (w / w), such as in the range from 20-30% (w / w). During or after the above described disintegration and pressing of the plant material, enzymes may be added in order to obtain an at least partial degradation of other plant cell components, such as hydrolysis of pectin, cellulose and other carbohydrates, resulting in enhanced release of bound protein. In one embodiment, at least one enzyme is added, selected from a group consisting of cellulase, kitinase, -fructosidase, -glucanase, hemicellulase, xylanase, invertase, glactosidase, polygalacturonase, xylosidase and arabinosidase. In useful embodiments, two or more enzymes, such as three or more enzymes, four or more enzymes, or five enzymes or more enzymes, are added. Sufficient pressing results in a press cake having a dry matter content of at least 18% w / w, such as at least 19% w / w, at least 20% w / w, such as at least 22% w / w, such as at least 24% w / w, such as at least 26% w / w, such as at least 28% w / w, such as at least 30% w / w, such as at least 32% w / w, such as at least 34% w / w, such as at least 36% w / w, such as at least 38% w / w; or a dry matter content in the range 20-40%, such as in the range 25-40%, such as in the range 30-40%, such as in the range 32-38%. The press cake may be use as an animal feed (e.g. fodder pellets or silage) or in a fermentation process, such as a biogas or bioethanol production. The following sections concern the further processing for purifying the plant juice, to obtain different protein products. Specifically, the plant juice may be split into a fraction suitable for food use and a fraction suitable for feed use. The purification relies on different filtration process steps, operated as specified herein. Liv Pre-filtration (Police filter) An optional pre-filtration of the first plant juice may be performed, primarily in order to remove sand, larger particulate matter and other debris. One important effect of this step is to minimize fouling in the subsequent filtration process. Fouling is undesired, as it may alter the selectivity and reduce the flux, and thereby lower the productivity of the system. A more pure green juice is obtained from the pre-filtration step. Pre-filtration may be performed as a single or multistage pre-filtration - i.e. one, two, three, four or more separate pre-filtrations may be performed. As a first pre-filtration step the plant juice may be drained / filtered through a filter with relatively large openings, such as a strainer type filter. In one embodiment, the pre-filtration is performed using a filter having a pore size in the range of 1-50 pm, such as in the range of 5-50 pm, such as in the range of 10-40 pm, such as in the range of 15-35 pm, such as in the range of 10-20 pm. The material of the pre-filter may be selected from a steel screen, a polypropylene screen, and other standard screens. Preferably the pre-filter is in the form of a vibration sieve. The pre-filtration can be performed either by dead-end filtration (e.g. utilizing bag-filters with a pore size in the l-50pm), or preferably by the using a more open filter (pore size in the range of 10-50pm) such as the use of a steel screen in a vibration sieve operation. In one embodiment, the pre-filtration step is selected from (i) dead-end filtration mode, using a filter having a pore size in the range 1-50 pm (e.g. polymeric filter), and (ii) a vibration sieve filtration using a sieve filter having a pore size in the range of 10-50pm (e.g. steel screen). The pre-filtration may be performed simply using gravity, and / or by adding pressure. As an alternative to pre-filtration, simple sedimentation and / or centrifugation may be applied to ensure that sand, larger particulate matter and other debris is removed prior to the Stage 1 filtration. It is particularly preferred that sand is removed as it may negatively affect the operation of the following filtration steps. I. v Stage 1 filtration The pressed plant juice (raw green juice) or the pressed plant juice which has been prefiltered (prefiltered green juice) is subjected to a stage 1 filtration. The stage 1 crossflow filtration is characterized by the use of a membrane having a nominal pore size between 15-100nm, such as having a nominal pose size between 15-90nm, between 15-80nm, between 15-70nm, between 15-60nm, between 15-50nm, between 15-40nm, between 15-30nm, between 15-20nm, such as having a nominal pose size between 20-90nm, between 20-80nm, between 20-70nm, between 20-60nm, between 20-50nm, between 20-40nm, between 20-30nm; such as having a nominal pose size between 90-100nm, between 80-100nm, between 70-100nm, between 60-lOOnm, between 50-100nm, between 40-100nm, between 30-100nm, between 20-lOOnm; such as having a nominal pose size between 90-99nm, between 80-99nm, between 70-99nm, between 60-99nm, between 50-99nm, between 40-99nm, between 30-99nm, between 20-99nm; such as having a nominal pose size between 80-90nm, between 70-90nm, between 60-90nm, between 50-90nm, between 40-90nm, between 30-90nm, between 20-90nm, between 15-90nm; such as having a nominal pose size between 70-80nm, between 60-80nm, between 50-80nm, between 40-80nm, between 30-80nm, between 20-80nm, between 15-80nm such as having a nominal pose size between 20-90nm, between 30-80nm, between 40-70nm, between 50-60nm, such as having a nominal pose size between 30-70nm, between 40-60nm. Preferably the stage 1 crossflow filtration is characterized by the use of a membrane having a nominal pore size between 15-100nm, more preferably between 15-90nm, even more preferably between 15-80nm, most preferably between 20-80nm. In one embodiment, the stage 1 crossflow filtration is characterized by the use of a membrane having a nominal pore size of at least 15nm, but below lOOnm, such as a nominal pore size of at least 20nm, but below lOOnm, such as a nominal pore size of at least 25nm, but below lOOnm, such as a nominal pore size of at least 30nm, but below lOOnm.In one embodiment, the pressure applied in the 1st stage crossflow filtration is between 0.1-1.2 bar, such as between 0.1-1.1 bar, such as between 0.1-1.0 bar, such as between 0.1-0.9 bar, such as between 0.1-0.8 bar, such as between 0.1-0.7 bar, such as between 0.1-0.6 bar, such as between 0.1-0.5 bar, such as between 0.1-0.4 bar. In one embodiment, the pressure applied in the 1st stage crossflow filtration is between 0.2-1.2 bar, such as between 0.2-1.1 bar, such as between 0.2-1.0 bar, such as between 0.2-0.9 bar, such as between 0.2-0.8 bar, such as between 0.2-0.7 bar, such as between 0.2-0.6 bar, such as between 0.2-0.5 bar, such as between 0.2-0.4 bar, such as between 0.3-0.8 bar, such as between 0.3-0.7 bar, such as between 0.3-0.6 bar, such as between 0.30.5 bar. Preferably, the pressure applied in the 1st stage crossflow filtration is between 0.2-1.2 bar, more preferably between 0.2-0.8 bar, most preferably between 0.2-0.6 bar. Turbulence is characterized using Reynolds number (Re). The Reynolds number is defined as Re= (u L) / v = (p u L) / p , where p is the density of the fluid (kg / m3), u is the flow speed (m / s), L is a characteristic length (m), p is the dynamic viscosity of the fluid (Pa-s or N-s / m2 or kg / (m-s)), and v is the kinematic viscosity of the fluid (m2 / s). As recognized by a person skilled in the art, if considering a tube geometry, flow is generally considered to be turbulent when Re > 2900. Preferably, the crossflow in the 1st stage crossflow filtration shall induce turbulence. In one embodiment, the Reynolds number (Re) for the crossflow in the 1st stage crossflow filtration >2900, such as between 2900-15000, preferably between 4000-12000, such as between 5500-11000, most preferably between 5800-10000. As recognized by a person skilled in the art, turbulence can be introduced by the movement of the feed solution or the membrane, in terms of high recirculation flow (crossflow), membrane rotation (dynamic filtration) and / or membrane vibration. Turbulence can also be enhanced by manipulating the geometry of the membrane. In one embodiment, the turbulence is created by a crossflow with a velocity between 1-5 m / s, such as between 1.5-4 m / s, preferably between 2-3.5 m / s, most preferably between 2-3 m / s. In one preferred embodiment, the stage 1 crossflow filtration is operated in constant flux mode. As recognized by a person skilled in the art, the constant flux may be achieved by altering the TMP during membrane operation. In one embodiment, the type of filter applied in the stage 1 crossflow filtration is selected from an inorganic membrane and a polymeric membrane; preferably an inorganic membrane, most preferably a ceramic membrane. In a preferred embodiment, the stage 1 crossflow filtration is characterized by the use of a ceramic membrane having a nominal pore size 15-100nm, wherein the pressure applied in the 1st stage crossflow filtration is between 0.2-1.2 bar, and wherein the crossflow shall induce turbulence. Preferably the crossflow velocity is between 1-5 m / s. In a most preferred embodiment, the stage 1 crossflow filtration is characterized by the use of a ceramic membrane having a nominal pore size 15-80nm, wherein the pressure applied in the 1st stage crossflow filtration is between 0.2-0.6 bar, and wherein the crossflow shall induce turbulence. Preferably the crossflow velocity is between 1-5 m / s. The stage 1 filtration is preferably characterized as being an ultrafiltration step, not a microfiltration step - as characterized by the stage 1 crossflow filtration using a membrane having a nominal pore size between 15-100 nm, preferably less than 100 nm, such as between 15-80 nm as disclosed herein. In one preferred embodiment, the process of the present invention does not comprise any crossflow membrane filtration steps using a microfiltration membrane. The stage 1 filtration aims to retain residual particulate matter along with e.g. choloroplasts, bound chlorophyll and "green" colorants (grassy sensory attributes), which are not desirable in a product aimed for food production (foul tasting proteins). The chlorophyll is likely responsible for color, but is also expected to influence taste and smell. In the Stage 1 filtration, larger species (including particles) are separated and concentrated, forming a product suitable for feed applications. The permeate from this step (containing protein targeted for food applications) is then filtered and concentrated in the stage 2 filtration, as disclosed below. The process parameters of the 1st stage crossflow filtration step are chosen such that the 1st stage permeate will comprise RubisCO protein - i.e. the process parameters are chosen such that the RuBisCO protein can pass through the filter of the 1st stage filtration step. Preferably, the process parameters of the 1st stage crossflow filtration step are chosen such that the 1st stage permeate will comprise native RuBisCO protein - i.e. the process parameters are chosen such that the RuBisCO protein can pass through the filter of the 1st stage filtration step and such that the RuBisCO protein is not denatured. The stage 1 filtration process is operated in crossflow mode (tangential flow). An advantages of crossflow filtration is, that the filtration can be operated as a continuous process. Another significant advantage is that the tangential flow reduces fouling cake formation on the filter surface, and a higher flux can therefore be maintained. The stage 1 crossflow filtration results in a retentate (i.e. a 'retained green juice' stream for feed application, comprising proteins) and a permeate (i.e. 'clear juice') which goes through a Stage 2 filtration, as disclosed below. As evidenced in Example 4 and 5, this stage 1 filtration efficiently facilitates that proteins, which are unwanted in food products, end up in the green juice retentate, which may be recovered and used as a feed product. This includes substances which cause a foul taste. Meanwhile, the 1st stage permeate comprises much fewer - is essentially free of - such unwanted substances. The proteins of the 1st stage retentate (i.e. proteins for feed application) may optionally be precipitated by heat treatment as a further treatment, such as by raising the temperature to a temperature between 60°C and 90°C, such as between 70°C and 88°C, preferably between 80°C and 85°C. In another embodiment, the proteins of the 1st stage retentate are optionally recovered by acid precipitation as a further treatment. In one embodiment the acid precipitation is performed by the addition of at least one acid, such as an organic acid or inorganic acid or combinations thereof. Examples of useful protein precipitating acids are CH3CHOHCOOH, CH3COOH, HCOOH, H2SO4, HNO3, and H3PO4. In order to obtain a satisfactory precipitation it is preferred that the pH of the juice is adjusted to a pH in the range from 2 to 5, such as in the range from 2.5 to 4.5, such as in the range from 3 to 4, such as in the range from 3.5 to 4. The precipitation may be followed by a centrifugation step, such as decanter centrifugation, in order to obtain a 'green protein concentrate'. This concentrate may have a high degree of denatured proteins due to the heat treatment. The product may be dried, in order to lower the amount of water in the protein concentrate; a step that is obviously dependent on the customer's wishes and needs. The purified green protein product is for example suitable as animal feed, such as fodder pellets. I. vi Stage 2 filtration The permeate from the stage 1 crossflow filtration is subjected to a stage 2 filtration. In one embodiment, the stage 2 filtration is an ultrafiltration. The stage 2 filtration is performed using a tighter membrane compared to the stage 1 filtration. The stage 2 crossflow filtration is characterized by the use of a membrane having a molecular weight cut off (MWCO) between 1-100 kDa, such as between 1-90 kDa, such as between 1-80 kDa, such as between 1-70 kDa, such as between 1-60 kDa, such as between 1-50 kDa, such as between 1-40 kDa, such as between 1-30 kDa, such as between 1-20 kDa, such as between 5-20 kDa, such as having a 10 kDa MWCO, such as having a 100 kDa MWCO. In one embodiment, the membranes has a MWCO between 1-500 kDa, such as between 1-450 kDa, such as between 1-400 kDa, such as between 1-350 kDa, such as between 1-300 kDa, such as between 1-250 kDa, such as between 1-200 kDa, such as between 1-150 kDa. In one embodiment, the membranes has a MWCO between 10-150 kDa, such as between 10-100 kDa, such as between 10-90 kDa, such as between 10-80 kDa, such as between 10-70 kDa, such as between 10-60 kDa, such as between 10-50 kDa, such as between 10-40 kDa, such as between 10-30 kDa, such as between 10-20 kDa. In one embodiment, the 2nd stage crossflow membrane filtration has a MWCO between 1-500 kDa, preferably between 1-200 kDa, such as between 5-20 kDa, such as between 50-150 kDa, more preferably between 5-150 kDa or between 10-100 kDa. In one embodiment, the pressure applied in the 2nd stage crossflow filtration is between 1-10 bar, such as between 1-9 bar, such as between 1-8 bar, such as between 1-7 bar, such as between 1-6 bar, such as between 1-5 bar, such as between 1-4 bar. Preferably, the pressure applied in the 2nd stage crossflow filtration is between 1-10 bar, more preferably between 1-6 bar, most preferably between 1-4 bar. In one preferred embodiment, the stage 2 crossflow filtration is operated in constant flux mode. In one embodiment, the type of filter applied in the stage 2 crossflow filtration is selected from a polymeric membrane and an inorganic membrane (e.g. ceramic membrane); preferably a polymeric membrane. In a preferred embodiment, the stage 2 crossflow filtration is characterized by the use of a polymeric membrane having a MWCO between 5-20 kDa, wherein the pressure applied in the 2nd stage crossflow filtration is between 1-4 bar. In a most preferred embodiment, the stage 2 crossflow filtration is characterized by the use of a polymeric membrane having a MWCO between 1-200 kDa, wherein the pressure applied in the 2nd stage crossflow filtration is between 1-4 bar. The stage 2 filtration membrane allows for transmission of low molecular weight compounds while retaining larger species, including proteins. The retained retentate is then washed in a diafiltration step, preferably using the same membrane, to further remove low molecular weight compounds from the final concentrate. The stage 2 crossflow filtration results in a permeate (which may be recirculated, as disclosed herein) and a retentate which goes through a diafiltration step, as disclosed below. The process parameters of the 2nd stage crossflow filtration step are chosen such that the 2nd stage retentate will comprise RuBisCO protein - i.e. the process parameters are chosen such that the RuBisCO protein should not pass through the filter of the 2nd stage filtration step. Preferably, the process parameters of the 2nd stage crossflow filtration step are chosen such that the 2nd stage retentate will comprise native RuBisCO protein - i.e. the process parameters are chosen such that the RuBisCO protein should not pass through the filter of the 2nd stage filtration step and such that the RuBisCO protein is not denatured. The stage 2 crossflow filtration permeate may potentially be used in buffer recirculation (as further disclosed herein), as a biofertilizer, as for recovery of minerals and / or small molecule bioactives through e.g. reverse osmosis or other applications. I. vii Diafiltration The retentate from the stage 2 crossflow filtration is subjected to a diafiltration step. The diafiltration step is preferably performed using the same membrane and operating conditions as the for the stage 2 crossflow filtration, and adding water (washing). A diafiltration factor of between 2-15 is used, depending on required purity of the protein product. In one embodiment, the diafactor of the diafiltraiton step is between 2-15, such as between 2-12, such as between 2-10, such as between 2-8, such as between 2-6. Preferably the diafactor is between 2-10. The diafiltration factor (diafactor) is often also referred to as the number of diavolumes. In one embodiment, the diafiltration is operated using the same membrane as the 2nd stage crossflow membrane filtration, and at a diafiltration factor of 2-10, preferably 28, most preferably 2-6, wherein the pressure applied for the diafiltration is between 110 bar, preferably between 1-8 bar, more preferably between 1-6 bar, most preferably between 1-4 bar. The diafiltration step results in a dia-concentrate ('white protein' for food application) and a permeate (which may be recirculated, as discussed herein). In one embodiment, the concentrate resulting from the diafiltration comprises at least 40% (w / w) crude protein based on total DM weight, such as at least 45%, such as at least 50%, such as at least 55%, such as at least 60%, such as at least 65%, such as at least 70%; such as at least 75%, such as at least 80%, such as at least 85%, such as at least 90%, such as at least 95%. Specifically, the concentrate mainly comprises functional proteins (i.e. native, non-denatured proteins). The concentrate may further be dried to obtain a dried functional protein product, which may be used as a high value protein product for food consumption. The dia-permeate may potentially be used in buffer recirculation (as further disclosed herein), as a biofertilizer, as for recovery of minerals and / or small molecule bioactives through e.g. reverse osmosis or other applications. I.iix Further processing The 'white' plant protein concentrate obtained after diafiltration may be further processed. In one embodiment, the 'white' plant protein concentrate may be concentrated by removal of aqueous liquid, such as by evaporation, lyophylization, and / or drying; e.g. evaporation at low pressure and low temperature, such as using a rotary (vacuum) evaporator, a plate evaporator, a thin-film evaporator, a falling film evaporator, a forced circulation evaporator, or similar equipment; e.g. drying using a spray dryer, a hydrocyclone, a fluidized bed dryer, a rotary drum, a pulse combustion dryer or similar equipment. The resulting dried and / or evaporated product may be directly formulated into a food protein product or be further separated into food protein ingredients. For example, the food protein product may be further refined to enrich more for RuBisCO. Preferably, the final product is a powder product. Such powder may be obtained after drying / evaporation by spraying. The 'green' plant protein concentrate (green juice) may also be further processed. It may be further purified or directly formulated into a feed protein product. In one embodiment, the 'green' plant protein concentrate may be concentrated by removal of aqueous liquid, such as by evaporation and / or drying; e.g. evaporation at low pressure and low temperature, such as using a rotary (vacuum) evaporator, a plate evaporator, a thin-film evaporator, a falling film evaporator, a forced circulation evaporator, or similar equipment; e.g. drying using a spray dryer, a hydrocyclone, a fluidized bed dryer, a rotary drum, a pulse combustion dryer or similar equipment. The resulting dried and / or evaporated product may be directly formulated into a feed protein product or be further separated into feed protein ingredients. Preferably, the final product is a pellet product. I.ix Recirculation of process steams One or more of the process stream(s) may be recirculated within the process to improve process economy; mostly to ensure reuse of the aqueous solution and / or decrease water demands. An example of such recirculation of streams is illustrated in Figure 2. To ensure efficient stabilization of components - e.g. proteins - extracted from the plant material, already from the time of disruption of the plant material, the plant material is suspended (i.e. submerged) in a stabilizing aqueous solution, as disclosed herein. This requires a minimum amount of stabilizing solution in relation to the amount of plant material, which limits the concentration of extracted components in the produced plant juice. To allow for higher concentrations of extracted components, while still achieving sufficient contact between the stabilizing components of the aqueous buffer solution and the plant material, the plant juice obtained from the pressing stage (comprising stabilizing components) may be fully or partly recirculated to act as the stabilizing suspension medium for fresh plant material. This allows for more components from the plant material to be extracted into the plant juice, while minimizing the amount of stabilizing aqueous solution. In one embodiment, the raw green juice after pressing, or a fraction thereof, is used for a second round of extraction of plant material, by mixing said juice with fresh plant material and further mechanically wet treat this fresh plant material, as disclosed herein. This reuse of the press juice for a new round of extraction may be done once, twice, three times, four times, five times, or even more. In one embodiment, the present invention concerns a method for producing a functional protein product and optionally a feed protein product, wherein one or more process stream(s) is / are recirculated within the process, said method comprising: (i) mixing a first batch of green biomass with an aqueous solution; (ii) disintegrating and pressing the mixture from (i) to obtain a first pulp and a first raw green juice; and using the first raw green juice or a fraction thereof for a second round of extraction of plant material, by mixing said first raw green juice or a fraction thereof with a second batch of green biomass, and disintegrating and pressing to obtain a second pulp and a second raw green juice; optionally using the second raw green juice or a fraction thereof for a third round of extraction of plant material, by mixing said second raw green juice or a fraction thereof with a third batch of green biomass, and disintegrating and pressing to obtain a third pulp and a third raw green juice; (iii) subjecting the first raw plant juice or a fraction thereof, the second raw plant juice or a fraction thereof, and / or the third raw plant juice or a fraction thereof, to a 1st stage crossflow filtration step to obtain a 1st stage retentate (retained green juice) and a 1st stage permeate (clear juice); (iv) subjecting the 1st stage permeate (clear juice) to a 2nd stage crossflow filtration step to obtain a 2nd stage retentate and a 2nd stage permeate; (v) subjecting the 2nd stage retentate to a diafiltration step to obtain a dia-concentrate and dia-permeate; (vi) recovering a functional protein product from the dia-concentrate; and (vii) optionally recovering a feed protein product from the 1st stage retentate (retained green juice). The stage 2 filtration permeate contains smaller molecular species including nutrients extracted from the plant material as well as the low molecular weight stabilizing components originating from the aqueous buffer solution. For the purpose of reusing the stabilizing components and to accumulate nutrients in the permeate of the stage 2 filtration (i.e. allowing for recovery of more concentrated streams), a partial recirculation of this second stage ultrafiltration permeate may be considered. In one embodiment, the stage 2 filtration permeate, or a fraction thereof, is used for a second round of extraction of plant material, by mixing said permeate with fresh plant material and further mechanically wet treat this fresh plant material, as disclosed herein. In one embodiment, the present invention concerns a method for producing a functional protein product and optionally a feed protein product, wherein one or more process stream(s) is / are recirculated within the process, said method comprising: (i) mixing a first batch of green biomass with an aqueous solution; (ii) disintegrating and pressing the mixture from (i) to obtain a press cake and a raw green juice; (iii) subjecting the raw plant juice or a fraction thereof to a 1st stage crossflow filtration step to obtain a 1st stage retentate (retained green juice) and a 1st stage permeate (clear juice); (iv) subjecting the 1st stage permeate (clear juice) to a 2nd stage crossflow filtration step to obtain a 2nd stage retentate and a 2nd stage permeate; (v) subjecting the 2nd stage retentate to a diafiltration step to obtain a dia-concentrate and dia-permeate; (vi) recovering a functional protein product from the dia-concentrate; and (vii) optionally recovering a feed protein product from the 1st stage retentate (retained green juice), wherein the 2nd stage permeate or a fraction thereof is used for a second round of extraction of plant material, by mixing said 2nd stage permeate or a fraction thereof with a second batch of green plant material, further processing the second batch as disclosed in steps (ii), (iii), (iv), (v), (vi), and (vii). The dia-permeate may also contain smaller molecular species including nutrients extracted from the plant material as well as the low molecular weight stabilizing components originating from the aqueous buffer solution. For the purpose of reusing the stabilizing components and to accumulate nutrients in the permeate of the diafiltration, a partial recirculation of this dia-permeate may be considered. In one embodiment, the dia-permeate, or a fraction thereof, is used for a second round of extraction of plant material, by mixing said permeate with fresh plant material and further mechanically wet treat this fresh plant material, as disclosed herein. The potentially recircled streams disclosed above, such the green juice after pressing, the stage 2 filtration permeate, and the diafiltration permeate may be used in combination or as separate recircled streams, such as two of them combined or all three combined. I.ix Temperature Temperature is of significance to the present invention, as high temperatures lead to denatured proteins (e.g. proteins altered in the native structure). Proteins isolated after precipitation by heat and / or acid are denatured. Accordingly, such proteins cannot be used for purposes where the properties of native (i.e. functional) proteins are needed. For example, temperatures above 60°C can lead to denatured proteins. Hence, in a most preferred embodiment all processing steps leading to the functional protein product are carried out at temperatures below 60°C, such as below 58°C, such as below 56°C, such as below 54°C, such as below 52°C, such as below 50°C, such as below 48°C, such as below 46°C, such as below 44°C, such as below 42°C, such as below 40°C. In a preferred embodiment the disintegration, pressing, filtration, and / or any further processing steps are kept at a temperature between 5-55°C, such as between 5-50°C, such as between 5-45°C, such as between 5-40°C, such as between 5-35°C. In a preferred embodiment the disintegration, pressing, filtration, and / or any further processing steps are kept at a temperature between 5-30°C, such as between 6-29°C, such as between 7-28°C, such as between 8-27°C, such as between 9-26°C, such as between 10-25°C, such as between 11-24°C, such as between 12-23°C, such as between 13-22°C, such as between 14-21°C, such as between 15-20°C, such as between 16-19°C, such as between 17-18°C. In one preferred embodiment, the method of the present invention does not comprise heating above 60°C, such as not above 58°C, such as not above 56°C, such as not above 54°C, such as not above 52°C, such as not above 50°C, such as not above 48°C, such as not above 46°C, such as not above 44°C, such as not above 42°C, such as not above 40°C, such as not above 38°C, such as not above 36°C, such as not above 34°C, such as not above 32°C, such as not above 30°C. II. Food (white) and feed (green) protein products In one aspect, the invention relates to protein products, such as protein concentrates, obtainable by the method according to the present invention. II. i Functional (white) protein product The present invention provides a functional (white) protein product. In one embodiment, the functional protein product is a food protein product or a food protein additive. In one embodiment, the present invention provides a functional (white) protein product obtainable by the method according to the present invention disclosed herein. The functional white protein product is obtainable from the method disclosed herein, either directly obtainable as the dia-concentrate, or obtainable from the dia-concentrate via further processing, as disclosed herein. The functional (white) protein product obtainable by the method of the present invention is a composition comprising one or more functional proteins. In one embodiment, the functional protein product comprises at least 40% (w / w) crude protein based on total DM weight, such as at least 45%, such as at least 50%, such as at least 55%, such as at least 60%, such as at least 65%, such as at least 70%; such as at least 75%, such as at least 80%, such as at least 85%, such as at least 90%, such as at least 95%. Specifically, the concentrate mostly comprises functional proteins (i.e. native, non-denatured, non-degraded proteins). It is preferred that the function of the proteins in the final 'white' product obtained from the process of the present invention is preserved during the various process steps so that the proteins in the functional protein product have preserved at least 50% or up to 100% of their activity, compared to the natural proteins. In one embodiment, at least 50%, such as at least 55, 60, 65, 70, 75, 80, 85, or at least 90% of the proteins in the functional protein product concentrate are functional proteins (i.e. native, nondenatured proteins). In one preferred embodiment, the one or more functional protein(s) is / are water-soluble protein(s) retained in the native form. In one embodiment, the functional proteins comprise RuBisCO protein. In one embodiment, the functional proteins comprise RuBisCO protein as the major protein constituent. In one embodiment, the concentrate resulting from the diafiltration comprises at least 10% (w / w) RuBisCO protein based on crude protein, such as at least 15%, such as at least 20%, such as at least 25%, such as at least 30%, such as at least 35%, such as at least 40%; such as at least 45%, such as at least 50%. In one preferred embodiment, the functional protein product is essentially devoid of one or more of chlorophyll, chloroplastic residues, chlorophyll-binding proteins, and photosystem proteins. In one embodiment, the concentrate resulting from the diafiltration comprises less than 20% (mol / mol) chlorophyll-binding and photosystem protein based on relative iBAQ quantification by mass spectrometry, such as less than 15%, such as less than 10%, such as less than 5%, such as less than 4%, such as less than 3%, such as less than 2%; such as less than 1%. The functional protein product is termed a 'white' functional protein product because the color of this protein product is preferably white or whitish, preferably essentially free of green color. Such colorless products are favorable in 'white' products. Further, another favorable property of the functional protein product is that it is essentially free of the typical grassy odor and taste. The functional protein product obtainable by the method according to the present invention may further be characterized by its favorable functional properties, including foaming, gelling, and solubility. In one embodiment, the functional protein product has great foaming capacity (FC) and foaming stability (FS). Specifically, in one embodiment, the functional protein product is characterized by having FC >150%, and FS >60% after 10 minutes. Preferably, the functional protein product has FC > 150%, more preferably FC > 200%, most preferably FC > 250%. Preferably, the protein product has FS > 60%, more preferably FC > 75%, most preferably FC > 90%. In one embodiment, the functional protein product has great emulsifying properties. Specifically, in one embodiment, the functional protein product is characterized by having an emulsification index (EAI) value >25 m2 / g and emulsification stability index (ESI) >35 minutes. Preferably, the functional protein product has EAI > 25 m2 / g, more preferably EAI > 50 m2 / g, most preferably EAI > 75 m2 / g. Preferably, the functional protein product has ESI > 35 min, more preferably ESI > 70 min, most preferably ESI > 100 min. In one preferred embodiment, the functional product obtainable by the present method is characterized by having FC >150%, FS >60% after 10 minutes, an emulsification index EAI>25 m2 / g and ESI >35 minutes. In one embodiment, the functional protein product has great gelling properties - i.e. having a least gelling concentration (LGC) of < 5% (w / v) as a heat-set gel. Preferably, the functional protein product has LGC < 5%, more preferably LGC < 3%, such as LGC between 1-1.5%, such as < 2%, most preferably LGC < 1%. In one embodiment, the functional protein product is fully soluble in an aqueous solution pH 6-8. In one embodiment, the functional protein product is fully soluble in an aqueous phosphate buffer solution pH 6-8. In one embodiment, the functional protein product is fully soluble in an aqueous phosphate buffer solution pH 6-10. Further, the functional protein product retains high solubility after drying over a broad pH range. Specifically, the dried product has a water solubility in the range from pH 4 - 10 of > 60%, preferably > 75%, most preferably > 90%. More preferably, the dried product has a water solubility in the range from pH 6 - 10 of > 60%, preferably > 75%, most preferably > 90%. Most preferably, the dried product has a water solubility in the range from pH 6 - 13 of > 60%, preferably > 75%, most preferably > 90%. Additionally, the functional protein product obtainable by the method according to the present invention may further be characterized by its smell and taste properties. In one embodiment, the functional protein product is characterized by having neutral to slightly pleasant smell. In one embodiment, the functional protein product is characterized by having a "malty" taste with association to bread, pistachio, caramel, and / or beer, and a "creamy" and slightly "salty" sensation, with limited bitterness. In one preferred embodiment, the functional product obtainable by the present method is characterized by being white or whitish in color, having a neutral smell, and that the functional protein(s) are water-soluble protein(s). In one most preferred embodiment, the functional product obtainable by the present method is characterized by being white or whitish in color, having a neutral smell, being fully soluble in an aqueous solution pH 6-8, and having FC >150%, FS >60% after 10 minutes, EAI >25 m2 / g, ESI >35 minutes, and LGC < 2%. II. ii Feed (green) protein product In one embodiment, the present invention provides a (green) feed product obtainable by the method according to the present invention disclosed herein. The feed protein product is obtainable from the method disclosed herein, either directly obtainable as the 1st stage retentate, or obtainable from the 1st state retentate via further processing, as disclosed herein. The feed (green) protein product obtainable by the method of the present invention is a composition comprising one or more plant proteins. In one embodiment, the feed protein product comprises at least 20% (w / w) crude protein based on total DM weight, such as at least 25%, such as at least 30%, such as at least 35%, or at least 40% (w / w) crude protein based on total DM weight, such as at least 45%, such as at least 50%, such as at least 55%, such as at least 60%, such as at least 65%, such as at least 70%; such as at least 75%, such as at least 80%, such as at least 85%, such as at least 90%, such as at least 95%. The feed product is named a 'green' feed protein product because the color of this protein product is most often green or greenish. III. Potential uses of the protein products In one aspect, the invention relates to the use of the protein obtainable by the method of the present invention. The present invention concerns production of proteins for food or feed use. Furthermore, the present invention may relate to a biorefinery concept for producing a protein concentrate to be used for different purposes other than feed and food, e.g. medical and industrial purposes. The functional protein product obtainable by the method according to the present invention may be used as a food additive - e.g. to provide water absorption, fat absorption, emulsifying, gelling, or foaming properties. The functional protein product may additionally be used as a food additive in order to increase the nutritional value of the food product. Moreover, the functional protein product obtainable by the method of the present invention may be used as a vegetable protein. The 'green' protein product obtainable by the method according to the present invention may be used as an animal feed. Due to its contents, the green protein concentrate may also be subjected to further processing to obtain at least one high value product to be used as an animal feed additive (e.g. fodder pellets). IV. Methods of analyzing / characterizing the protein products Methods for characterizing the protein products of the present invention are provided herein. Protein content may be determined as follows: The protein content may be measured as N (nitrogen) times 6.25 (often named crude or raw protein) by methods such as Kjeldahl, Dumas or elemental analysis (e.g. CHNO(S)). In the present context the terms "protein content", "raw protein" and "crude protein" are used herein interchangeably. Proteins can be characterized by molecular techniques e.g. X-ray crystallography, Nuclear Magnetic Resonance, Cryo-electron microscopy, Circular dichroism, isothermal titration calorimetry, liquid chromatography, mass spectrometry, gel electrophoresis, or combinations hereof. The qualitative protein composition of the functional protein product may be assessed by a range of analytical methods for identification of RuBisCO, as recognized by a person skilled in the art. One such method could be reducing SDS-PAGE, wherein proteins are separated by size and RuBisCO may be identified as distinct and intense bands around 50-55 kDa and 10-15 kDa, which correlates with the size of the large and small subunit of the RuBisCO protein, respectively, as further described herein. To increase confidence of identification, more specific methods may be applied such as by Western Blotting, wherein RuBisCO-specific antibodies may be used to specifically visualize RuBisCO from SDS-PAGE analysis. Preferably, the applied antibody is raised against linear epitopes of the large and small subunits of RuBisCO and represent domains conserved across intra-and inter-species RuBisCO isoforms. More preferably, RuBisCO may be identified by highly specific analytical methods such as by mass spectrometry-based proteomics, such as by bottom-up proteomics, wherein the large and small subunit of intra- and interspecies RuBisCO isoforms may be distinguishable. The quantitative protein composition of the functional protein product may be assessed by a range of analytical methods, as recognized by a person skilled in the art. One such method could be densiometry wherein band intensities from SDS-PAGE analysis may be used to estimate the relative quantity (mass / mass) of each band. Preferably, quantitative composition may be determined using mass spectrometry-based proteomics, such as by relative iBAQ quantification, wherein the relative abundance (mol / mol) of individual and / or grouped proteins may be determined. In such analysis, the molar abundance of RuBisCO would correspond to the sum of riBAQ for all identified RuBisCO subunits and isoforms. The nativity of proteins of the functional protein product may be assessed by a range of analytical methods, as recognized by a person skilled in the art. One such method could be native-PAGE, where proteins are separated by size but under non-reducing and nondenaturing conditions, whereby the native hexadecameric RuBisCO complex would be maintained. Preferably, the nativity of the functional protein product may be assessed by means of thermal analysis such as by differential scanning caliometry, wherein the temperature of thermal denaturation of proteins may be determined. The absence of a thermal denaturation event will indicate loss of nativity. More preferably, the nativity of RuBisCO may be validated by chromatographic methods such as by size exclusion chromatography, wherein the RuBisCO complex would be retained less and elute faster than its constituents (large and small subunits, respectively). Verification of protein identity may subsequently be accomplished through qualitative analysis such as by the methods described herein. If the protein for instance is an enzyme, the activity can be measured as either the consumption of substrate or production of product over time. A large number of different methods of measuring the concentrations of substrates and products exist in the art and many enzymes can be assayed in several different ways, such as but not limited to initial rate expression, progress curve experiments, transient kinetics experiments and / or relaxation experiments. Additional means for characterizing the protein product is disclosed in the Examples, such as foaming properties (Example 6.2.1), emulsification properties (Example 6.2.2), solubility (Example 6.2.3), and sensory evaluation (Example 6.2.4). It should be noted that embodiments and features described in the context of one of the aspects of the present invention also apply to the other aspects of the invention. V. Numbered Embodiments of the invention Numbered Embodiment 1. A method for producing a functional protein product and optionally a feed protein product from plant material, said method comprising the steps: (i) Providing a green plant material, and mixing said plant material with an aqueous solution; (ii) disintegrating and pressing the mixture from step (i) to obtain a pulp and a raw green juice; (iii) optionally subjecting the raw green juice to a pre-filtration step to obtain a prefiltered green juice; (iv) subjecting the raw green juice or the pre-filtered green juice to a 1st stage crossflow membrane filtration step to obtain a 1st stage retentate (retained green juice) and a 1st stage permeate (clarified juice), wherein the nominal pore size of the 1st stage crossflow membrane filtration is between 15-100 nm, wherein the pressure applied in the 1st stage crossflow membrane filtration is between 0.2-1.2 bar, preferably between 0.2-0.8 bar, more preferably between 0.3-0.6 bar, and wherein the crossflow of the 1st stage crossflow filtration shall induce turbulent characteristics (Reynolds number > 2900 for a tube geometry); (v) subjecting the 1st stage permeate (clarified juice) to a 2nd stage crossflow membrane filtration step to obtain a 2nd stage retentate and a 2nd stage permeate, wherein the 2nd stage crossflow membrane filtration has a MWCO between 1-100 kDa, preferably between 5-20 kDa, most preferably approx. 10 kDa, and wherein the pressure applied in the 2nd stage crossflow membrane filtration is between 1-10 bar, preferably between 1-6 bar, more preferably between 1-4 bar; (vi) subjecting the 2nd stage retentate to a diafiltration step to obtain a dia-concentrate and a dia-permeate; (vii) recovering a functional protein product from the dia-concentrate; and (viii) optionally recovering a feed protein product from the 1st stage retentate (retained green juice). Numbered Embodiment 2. The method according to Numbered Embodiment 1, wherein the method does not comprise heating above 60°C, such as not above 55°C, 50°C, 45°C, or 40°C, Numbered Embodiment 3. The method according to Numbered Embodiment 1 or 2, wherein the green plant material is selected from alfalfa, clover, lupines, grasses, crucifers, beets, chicory, carrot, radish, cassava, roadside crops and combinations hereof; preferably clover and / or grass; most preferably red clover (Trifolium prantese), white clover (Trifolium repens), perennial ryegrass (Lolium perenne), fescue grass species (Festuca sp.), and blends of one or more of these plant materials, such as clover grass. Numbered Embodiment 4. The method according to any one of Numbered Embodiments 1-3, wherein the aqueous solution is a phosphate buffer having a pH in the range 6-8, said buffer comprising a reducing agent selected from sodium sulfite and mercaptoethanol, and a stabilizing agent selected from EDTA, citrate, and other organic acids. Numbered Embodiment 5. The method according to any one of Numbered Embodiments 1-4, wherein the raw green juice is subjected to the pre-filtration step, and wherein the pre-filtration step is selected from (i) dead-end filtration, using a filter having a pore size in the range 1-50 pm (e.g. polymetic filter) and (ii) vibration sieve filtration using a sieve filter having a pore size in the range of 10-50 pm (e.g. steel screen). Numbered Embodiment 6. The method according to any one of Numbered Embodiments 1-5, wherein the 1st stage crossflow membrane filtration is operated using a ceramic membrane, at a crossflow velocity between 1-5 m / s and a pressure between 0.2-0.6 bar. Numbered Embodiment 7. The method according to any one of Numbered Embodiments 1-6, wherein the 2nd stage crossflow membrane filtration is operated using polymeric membrane and a pressure between 1-8 bar. Numbered Embodiment 8. The method according to any one of Numbered Embodiments 1-7, wherein the diafiltration is operated using the same membrane as the 2nd stage crossflow membrane filtration, and at a diafiltration factor of 2-10, preferably 2-8, most preferably 2-6. Numbered Embodiment 9. A functional protein product obtainable by the method according to any one of Numbered Embodiments 1-8, preferably wherein the functional 5 protein product is soluble in an aqueous solution pH 6-8, and / or preferably wherein the functional protein product has (i) a foaming capacity (FC) >150%, and / or a foaming stability (FS) >60% after 10 minutes, (ii) an emulsification index (EAI) value >25 m2 / g and / or an emulsification stability index (ESI) value >35 minutes, and / or (iii) a least gelling concentration (LGC) of <2% as a heat-set gel. 10 Numbered Embodiment 10. A feed protein product obtainable by the method according to any one of Numbered Embodiments 1-8. EXAMPLES Example 1: Screening of different membranes at lab scale 15 These experiments investigate the influence of the choice of membrane type / material on the first stage filtration. Several different membranes were investigated during this screening. This included: • a Whatman flat-sheet Regenerated Cellulose Membrane (RC58) with 0.2pm nominal pore size 20        • two different Alfa Laval flat-sheet membranes with 0.2pm nominal pore size, both having a support layer of polypropylene, and their respective active layers are polysulphone (MFG2) and a fluoro polymer (MFP2) • two different silicon carbide ceramic membranes from Liqtech with nominal pore size of approximately 0.3pm: a multichannel (LT) tubular membrane 25          (COM0250305 UF-03), and a monotube (LTM) tubular membrane (COM0250305xxx-17). For membrane modules, an Alfa Laval LabStak M20 module and a Sterlitech® Sepa CF cell have been utilized. The Liqtech membranes were mounted in custom membrane modules supplied by Liqtech. 30    1.1 Preparation of Grass Juice Typically, the grass was harvested from 7-8 am, juiced from 8-9 am, centrifuged from 9-10.30 am. The buffer was prepared with 86.5 mM Na2HPO4, 14.5 mM KH2PO4, 5 mM EDTA and 12.5 mM Na2SO3 and stored overnight in a refrigerator. The grass juice was prepared by harvesting grass and pressing it in an Angel juicer. During pressing, buffer 35 was added to the grass juice in the ratio of 2:1. When screw pressing 1 kg of grass, roughly 0.5 L of grass liquid was obtained, so, including the buffer, 2.5 L of grass juice was produced. The pH of the grass juice was measured to be in the range of 6.9-7.2. The grass juice is then centrifuged for 30 minutes at 10.000 G and at 4°C. The grass juice is then transferred to a filtration setup. 1.2 Influence of the membrane type on protein transmission The selectivity and permeability of the membranes were investigated. The selectivity was evaluated based on a visual determination of the permeate color and based on the protein content determined by SDS-PAGE. From the initial experiments it became clear that the choice of membrane has significant importance for the process, especially in regard to the protein transmission. In general, it was observed that the initial protein transmission appears largest for the RC and LT / LTM membranes, with lower initial protein transmission observed for the MFG membrane and a negligible protein transmission for the MFP membrane. A common observation seen for all these experiments is, that the color corresponds with the RuBisCO content. It was observed that greener permeate samples had a higher RuBisCO content. This indicates that no sufficient separation between green and white proteins was achieved using these membranes under the investigated operating conditions. 1.3 Fouling K significant degree of fouling was seen for all these experiments. Fouling gives rise to multiple problems: first of all, fouling will decrease the flux over time and thereby the productivity of the system. Additionally, fouling of the membrane results in a lower, or in some cases close to zero, protein transmission after the initial phase. This results in the protein being retained and thereby concentrated in the retentate, which is contrary to the goal of the 1st stage filtration step in this biorefining process. The experiments also indicated, that the CFV (crossflow velocity) was able to significantly impact the flux, likely by removing part of the fouling layer, since the experiments showed that the flux increased significantly more than expected by an increase in TMP (transmembrane pressure) alone (in this set-up TMP increased as CFV was increased). Furthermore, it was observed in the MFG-experiment, that the protein transmission could be affected during the filtration, likely due to changes in the membrane fouling as a result of the changes in CFV. The same correlation was not found for the two LT-experiments. Based on these initial experiments, it is speculated a high CFV in combination with a low TMP would be beneficial for the process, as the low TMP should ensure less fouling and cake compression, and the high CFV should ensure less fouling by removing the formed cake layer. However, in the lab scale setup, it was only possible to apply high CFV when the TMP is also high. 1.5 Concluding remarks Multiple different membranes were tested for the filtration of grass juice. The ceramic SiC membrane showed the highest consistent protein transmission compared to the other types of membranes. The permeate did, however, display varying intensities of green color. This indicates that larger species (including green protein, not suitable for food production) are permeated. Both the flux and protein transmission were observed to decline over time for the majority of the conducted experiments, most likely due to the formation of fouling. It was found that fouling currently constitutes a significant problem in the filtration process. Besides lowering the productivity, the protein transmission is also significantly decreased due to fouling. In the following Examples, the fouling issues as well as the issues related to 'green' protein permeation were addressed by testing membranes with a smaller pore size at pilot scale, including ceramic membranes. Example 3A: Production of food and feed protein at pilot scale The aim of this study was to develop a pilot scale demonstrator, capable of processing kilograms of fresh leaves, that can produce protein containing streams suitable for food and feed applications. Yields as well as product quality were characterized and related to operational conditions and process layout. 3.1 Plant material Leaves were harvested from three different fields, all located in Jylland, Denmark. Field number 1 (location: 57°02'16.5"N 9°58'42.4"E) was planted using a commercial blend (seed mix 35) containing e.g., white clover and lucerne, in addition red clover was also present in this field. More specifically, Field number 1 contained 13% white clover and 87% ryegrass of different cultivars, and red clover was also observed in this field. Plant material from this field was used in the majority of the test runs. Field number 2 (location: 56°49'07.0"N 9°38'47.9"E) and 3 (location: 56°37'55.3"N 9°30'23.4"E) contained locally developed mixes, rich in clover and ryegrass. More specifically, Field 2 was planted using a custom blend (DLF, Denmark) containing 5% white clover, 58% ryegrass, 25% X Festulolium, and 12% Timothy; while Field 3 contained a locally developed mix, rich in red clover and ryegrass. The plant material was harvested between the middle of June and middle of October. For the experiments, each of the individual batches were harvest in the morning, about 1 hour before shredding and pressing. A manual bush trimmer tool (Einhell, Agillo PXC Solo) was used to harvest the leaves. Care was taken to avoid the presence of soil in the recovered leaves fraction by harvesting the plants around 5 cm above soil level. 3.2 Experimental set-up Two different process layouts were evaluated on a demonstrator scale, as illustrated in Figure 3. For both process options, fresh leaves were first shredded by a blender while being suspended in an aqueous buffer solution. Specifically, the aqueous solution comprised 4.00 g / L di-Sodium hydrogen phosphate dehydrate, 1.00 g / L Potassium phosphate monobasic, 11.70 g / L Sodium citrate tribasic dehydrate, and 1.58 g / L Sodium sulphite. One experiment with an aqueous solution with a lower buffering capacity was also performed. In this case the concentration of potassium phosphate and di-sodium hydrogen phosphate was halved, the concentration of the rest of the stabilizing components as well as the preparation procedure remained the same, however. The ratio between fresh leaves and buffer solution was kept at 1:2 in respect to mass for all experiments. Specifically, the plant material was weighed and divided into 5 kg fractions. Each 5 kg fraction was successively added to 10 kg of stabilizing solution (with a temperature around 20 °C) while being shredded by a blender (Model HMI012, Hamilton Beach Commercial) set at maximum effect. Each 5 kg fraction was treated for at least 5 minutes with the blender. Thereafter the leaf suspension was pressed in a screw press where the raw green juice was recovered. Specifically, the screws press used (a Vincent KP6) was operated with a counter pressure of 4 bars at a speed setting of "1" (corresponding to about 3.3 rpm) for all the experiments. The two processes differ in the treatment of the pressed plant juice before it is fractionated using membrane processes. For the first process option (runs 1-2 in Table 2), a mild heat-treatment step followed by centrifugation was used to remove particles / suspended matter before the first crossflow filtration step, to limit fouling (Figure 3B). Specifically, the heat treatment was performed by pumping the plant juice from the screw press through a coiled tube submerged in a 72 °C heat bath. The pump speed (and thus the residence time) was regulated to keep the outlet temperature of the juice between 63 - 65 °C. As fouling from the juice was deposited on the tube wall (decreasing the heat transport from the bath to the juice), the residence time for the heat treatment was increased from about 70 to 120 s. Following the heat-treatment, the juice was quickly cooled to around 20 °C by transporting it through a coiled tube submerged in an ice bath. The heat-treated juice was centrifuged at 10 000 g for 15 min using an Avanti JXN-26 centrifuge from Beckman Coulter. After completed centrifugation the supernatant was recovered through manual decantation; care was taken to ensure that no visual particles were included in this fraction. For the second process option (runs 3-16 in Table 2), a pre-filtration using bag filters was used for the purpose of removing foulants (Figure 3A). Specifically, a range of polypropylene bag filters from Spectrum®, were used. Before the use of bag filters, the plant juice from the screw press was filtered through a strainer (with lxl mm square openings). The bag-filtration was performed manually, pouring the raw green juice into the filter, and initially draining it using gravity. As retained material increased the flow resistance, additional pressure was applied by manually squeezing the bag filter, and as even more materials was deposited in the filter it was replaced by a fresh filter. For run 17 the raw green juice from the screw press was only filtered through a strainer (with lxl mm square openings). Following the heat treatment and centrifugation (first process option) or the pre-filtration (second process option), both process options thereafter utilized a two-step membrane approach. Here larger species (including particles) as well as smaller species associated to the large species are separated and concentrated in an initial crossflow filtration step, forming a product suitable for feed applications. This first step aims to retain residual particulate matter along with e.g. chlorophyll which are not desirable in a product aimed for food production. The permeate from this step (containing protein targeted for food applications) was then filtered and concentrated in a second membrane step where a 10 kDa molecular weight cut-off ultrafiltration membrane was used. This membrane allowed for transmission of low molecular weight compounds while retaining larger species. The retained retentate was then washed in a diafiltration step using the same 10 kDa ultrafiltration membrane, further removing low molecular weight compounds. 3.3 Membrane specifications The membranes utilised are listed in Table 1. Nominal pore size in this table is reported based on the suppliers' specifications, except for the CoMem® membrane where no specification was found. The mean flow pore size of this membrane was instead determined using capillary flow porometry (PX1000, Porometer). The two ceramic membranes from LiqTech as well as the polymeric Synder v0.2 were all investigated as membrane media for the first filtration stage, while ST Sanitary was used in the second filtration stage. Both LiqTech membranes are tubular membranes where each membrane has 30 round channels with an inner diameter of 3 mm. The membranes from Synder are spiral-wound elements, with spacers reported in Table 1. Table 1. Membrane specifications Membr ane Manufac turer Description Membrane area Pore size / MWCO CoMem ® LiqTech Asymmetric silicon carbide ceramic membrane 0.34 m2 0.16 pm HTM LiqTech Asymmetric silicon carbide (SiC) and zirconia (ZrO?) ceramic 0.34 m2 60 nm V0.2 Synder Polyvinylidene difluoride (PVDF) membrane (spacer: 46 mm, V0.2 2540F) 2.04 m2 0.2 pm ST Sanitary Synder Polyethersulfone (PES) membrane (spacer: 31 mm diamond, ST 2540F) 3 m2 10 kDa Two different commercially available pilot units were utilized in the demonstrator process, one suitable for mounting "HTM"and "CoMem" ceramic membranes (a Labbrain unit from LiqTech) and the other one designed to house "VO.2" and "ST Sanitary" polymeric membranes (a SW25 System from MMS membrane systems). Both systems were equipped with pressure transducers to determine transmembrane pressures (TMP), flow meters for cross flow and permeate flux determinations as well as temperature probes. Each of the two pilot units were also capable of automatic logging of process parameters and data from each of these two rigs were recorded every two seconds during operations. Diafiltrations were performed using the same membrane as used during the second filtration stage and the TMP was kept the same if nothing else is stated. Between each run, the membranes were cleaned using CIP (cleaning in place) procedures. The following commercial cleaning chemicals were used during the cleaning: Ro dan 264, Ro dan acid and Hypochlor DES from Novadan as well as Ultrasil (125) from Ecolab. 3.4 Experiments and operational conditions A total of 17 runs were performed under various conditions, see Table 2 and Table 3. Run 1 and 2 were performed using a mild heat-treatment step followed by centrifugation before the first crossflow filtration step (process layout version 1) while subsequent runs were instead prefiltered using bag filters (process layout version 2). The TMP during the first filtration step was kept at 0.3-0.5 bar. The temperature of any of the different runs was never above 35°C, during the first filtration. The volumetric crossflow during the second filtration step was set at ~2.9 m3 / h for all experiments. The temperature of any of the different runs was never above 35°C, during the second filtration. The product streams (the retentates from the first and second crossflow filtrations) were divided into multiple one-liter portions and frozen at -18°C as soon as they had been recovered. Samples from the product streams were later freeze-dried to produce a powder. Table 2 Run Source Harvest time Batch size (kg) Leaf dry weight (kg) Aqueous solution0 Shredding 1 Field 1 13-Jun 22 16 3.7 Std. Yes 2 Field 2 20-Jun 22 20 3.2 Std. Yes 3 Field 1A 27-Jun 22 10 - Std. Yes 4 Field 2 28-Jun 22 10 1.8 Std. Yes 5 Field 2 29-Jun 22 10 1.7 Std. Yes 6 Field 2 4-Jul 22 20 3.6 Std. Yes 7 Field 1 2-Aug 22 14.1 2.7 Std. Yes 8 Field 3 10-Aug 22 15 2.5 Std. Yes 9 Field 1 16-Aug 22 15 2.1 Std. Yes 10 Field 1 23-Aug 22 15 2.9 Std. Yes 11 Field 1 29-Aug 22 15 2.3 Std. Yes 12 Field 1 6-Sep 22 15 3.5 Std. No 13 Field 1 12-13 Sep 22 45B 7.4 Std. Yes 14 Field 1 20-Sep 22 15 2.6 Red. Yes 15 Field 1 26-Sep 22 15 2.3 Std. Yes 16 Field 1 4-Oct 22 15 2.1 Std. Yes 17 Field 1 11-Oct 22 15 2.5 Std. Yes A Leaves needed to be collected at the edge of the field and the mixture might be slightly different. B Juice was produced and processed up to and including the first crossflow filtration stage during two consecutive days. The juice from these two days was then mixed (after storing at 5°C) and processed in the second stage crossflow filtration step 5 and the following diafiltration. c Std. = standard buffer; Red. = reduced concentration of buffer components, specifically the concentration of potassium phosphate and disodium hydrogen phosphate was halved, while the concentration of the rest of the stabilizing components remained the same. Table 3 1st stage filtration 2nd stage filtration Run PrefiltrationA or heat / centri. membrane crossflow8 membrane TMP Dia-factor 1 Heat / centri. HTM 3 m / s ST Sanitary 1 bar 5.6 2 Heat / centri. HTM 3 m / s ST Sanitary 1 bar 5.7 3 BF 50 and 10 pm CoMem® 3 m / s N / A N / A N / A 4 BF 50, 10, 5, and 1 pm CoMem® 3 m / s ST Sanitary 1 barB ~10c 5 BF 50, 10, and 1 pm HTM 3 m / s ST Sanitary 0.9 bar 8.5 6 BF 50, 10, and 1 pm HTM 3 m / s ST Sanitary 2 bar 9.6 7 BF 50, 10, and 1 pm HTM 3 m / s ST Sanitary 1 bar 9.5 8 BF 50 and 1 pm HTM 3 m / s ST Sanitary 2 bar 9.8 9 BF 50 and 1 pm HTM 3 m / s ST Sanitary 2 bar 9.7 10 BF 50 and 1 pm V0.2 2.9 m3 / hD ST Sanitary 2 bar 9.5 11 BF 50 and 1 pm HTM 2 m / s ST Sanitary 2 bar 9.5 12 BF 50 and 1 pm HTM 3 m / s ST Sanitary 2 bar 9.6 13 BF 50 and 1 pm HTM 3 m / s ST Sanitary 2 bar 9.5 14 BF 50 and 1 pm HTM 3 m / s ST Sanitary 2 bar 9.5 15 BF 50 and 1 pm HTM 3 m / s ST Sanitary 3 bar 9.6 16 BF 50 and 1 pm HTM 3 m / s ST Sanitary 2 bar 18.6 17 Only strainer HTM 3 m / s ST Sanitary 2 bar 9.6 A BF = bag-filters. B The Reynolds number for flow in a tube geometry will be approximately 6700 and 10000 at a crossflow velocity of 2 m / s and 3 m / s respectively. c The pressure during diafiltration was adjusted from 1 to 0.5 bar halfway during the diafiltration as the dia-water feed pump reached maximum capacity. This also led to an additional uncertainty when determining the diafactor. D Crossflow is reported in m3 / h for spiral wound membranes. 3.x Analytical methods Dry matter determinations: The dry matter of process and product streams were determined by drying samples at 105 °C. The samples were dried to full dryness, overnight, and total dry matter was calculated based on initial and dry weight. The dry matter of the pure buffer solution was determined using the same methodology. A double or triple sample was dried for each stream and the dry matters reported are the average of these samples. pH and conductivity determinations: The pH and conductivity of selected process streams were also determined using calibrated electrodes from Mettler-Toledo (Inlab® Expert Pro-ISM and InLab® 731-ISM respectively). SDS-PAGE: The presence of protein in the different process streams were visualized with SDS-PAGE using gradient gels (4 to 20%). For each stream 10 parts of sample was mixed with 5 parts of water and 7 parts of reducing buffer and heated to 95 °C for 10 min. Out of this mixture, 20 pL was then loaded to each well in the pre-casted SDS-PAGE gel. A constant voltage of 160 V was applied for 40-45 minutes to separate the proteins. Coomassie blue was used to stain the gels. Nitrogen content: The nitrogen content of food and feed product streams as well as selected process streams were analyzed using a FlashSmart™ Elemental Analyzer from Thermo Fisher Scientific. Crude protein (CP) amounts were estimated based on measured nitrogen amounts using a conversion factor of 6.25. 3.5 Results - yield of extracted plant material for each process step Estimation of the yield of extracted components in the juice for each process step was performed using the stream sizes and dry matter measurements for the juice streams together with the dry matter of the fresh leaves. 3.5.1 Mixing, shredding and screw press While there were variations between the individual runs, an average of around 34% (±5.7 percent units) of the total dry material in the leaves was extracted into the green juice after the screw press stage. The single run using a buffer with lower buffering effect (run 14) resulted in an extraction of 35% of the total dry material of the leaves. This was close to the average of all the runs, indicating that no significant effect on the extraction efficiency with regards to amount could be observed in this single experiment. It is further noted that run 12, performed with no shredding, displayed a lower degree of extraction (26%) compared to the other runs. 3.5.2 Heat-treatment / centrifugation or bag pre-filtration It was found that a considerably larger amount of material is removed by the heat-treatment / centrifugal step (compared to using bag filters). The vast majority of the removed material was retained in the centrifugation step, where a green-colored fraction with grass sensory attributes was recovered. For run 1 and 2, 24% and 40% of the extracted material was removed (i.e., 76% and 60% of the material entering this step remains in the stream feed into the crossflow filtration stages); while for runs 3-16, only 3 to 10% of the material was removed. This indicates that a relatively small amount of the extracted material is present in the form of particles large enough to be separated by a 1 pm filter. For run 17, where only a strainer and no bag filters are used, only about 1% of the extracted material was removed. 3.5.3 First stage crossflow filtration Extensive fouling / gel layer formation is generally known to occur during crossflow filtration of streams containing extracted plant material, even after centrifugation pretreatment (Zhang et al., 2015). To mitigate such fouling (which may result in loss of both permeate flux capacity as well as protein transmission) during the first crossflow filtration stage, operational conditions with crossflow velocities from 2 to 3 m / s (Re approximately 6700 and 10000 respectively, thereby inducing turbulent flow (Re > 2900)) and transmembrane pressures from 0.3 to 0.5 bar were chosen. Examples of the development of the permeate flux (including TMP-data) for run 1 and run 7 can be found in Figure 4. Comparable fluxes were achieved for both process options (heat-treatment / centrifugation vs. bag pre-filtration), with a significant flux decline in the start of the filtration for both process options. The first crossflow stage separates mainly larger species as well as undissolved particles and concentrates these in the retentate. For the runs with heat-treatment and centrifugation steps (run 1 and 2), most of the green color had been removed already in the upstream centrifugation step. For the bag filtered streams (runs 3-16, as well as run 17), the retentate of the 1st stage crossflow filtration was observed to maintain a green color as well as grassy sensory attributes. For both process configurations the permeate out from the first filtration stage had lost its green color and grassy scent. For the processes using bag pre-filtration between 51 to 58% of the extracted leaf material was permeated through the first crossflow filtration stage, while the residual thus would be retained in the retentate or deposited on / in the membrane. For the heat-precipitated / centrifuged process, less material was retained in the first cross flow filtration stage (e.g. for run 2 almost 80% of the extracted leaf material was permeated). This is not surprising as more material was removed in the upstream heat-precipitated / centrifugal step, and thus less undissolved material would be present in the feed into the first filtration stage. 3.5.4 Second stage crossflow filtration and diafiltration stage An initial flux decline was also observed in the second filtration step, when the juice is fed to the ultrafiltration unit with a lower MWCO membrane compared to the first crossflow filtration step. After the initial flux decline, the filtration (where the retentate is circulated and gradually concentrated) is characterized by relatively stable flux behavior. Conductivity measurements of the removed permeate during diafiltration confirmed the reduction of the buffer and stabilizing components during diafiltration. This was expected as the buffer and stabilization components are considerably smaller than the 10 kDa cut-off of the membrane. In addition, some extracted and / or degraded leaf components (e.g. smaller sugar molecules and degraded protein) would also be expected to pass through the 10 kDa membrane. Mass balance and mass yield estimations were also performed over the second filtration stage where the ultra- and diafiltration was conducted. It was found that a considerable amount of the remaining material was removed, leaving between 8 to 14% of the extracted material entering this stage in the final, dia-washed, concentrate. 3.6 Results - protein content and protein yields of the process streams 3.6.1 SDS-PAGE analysis of protein content The protein content of process streams from run 1 and 7 were compared by SDS-PAGE analysis, Figure 5. For several of the process streams distinct bands were seen at around 55 kDa and 15 kDa, which correlates well to the size of the large and smaller subunit of the desirable RuBisCO (Ribulose 1,5-bisphosphate carboxylase / oxygenase) protein, respectively. For run 1 (comprising heat treatment), a significant amount of the protein was removed in the centrifugation step. For both run 1 and run 7, no protein bands could be observed in the permeates from the 2nd crossflow filtration, indicating that the 10 kDa membrane was efficient in retaining the protein constituents. 3.6.2 Crude protein yield per total dry matter The crude protein concentration as a percentage of the total dry matter was estimated using total nitrogen analysis. Figure 6 shows the results for the leaves fed into the screw press, the raw green juice from the screw press, as well as the two main product streams of the process: 'white protein' (diafiltrated concentrate) and 'green protein' (from retentate from the first crossflow filtration stage). Applying the mixing, shredding, and screw press as described herein together with the stabilizing solution, a crude protein concertation of around 20% (based on the total dry mass) was achieved in the raw green juice from the screw press for the majority of the runs (Figure 6). For the 2nd stage filtration dia-filtrated concentrate ('white' protein product), protein concentrations between 50% up to 70% could be achieved using process version 2 (i.e. bag filter treatment), while lower concentrations were observed for the two runs with process version 1 (i.e. heat-treatment / centrifugation steps). For the retentate from the first crossflow filtration ('green' protein product), the bag filtration process version yielded retentates with crude protein concentrations typically between 30 to 40%. For further comparison of the different process configurations, the crude protein content (based on the total dry mass) for the 1st stage permeate is reported in Table 4 (NB for this stream the buffer and stabilizing components contributes to the total dry matter). A higher value indicates a higher relative protein content compared to the other compounds in the permeate. The data surprisingly indicates that for the tighter HTM membrane (smaller pore size), a higher relative protein content is found in the permeate. Without being bound to theory this suggests that a tighter membrane (in the ultrafiltration range) may alter fouling behavior to allow for higher protein permeation. This is further supported by the yield over the 1st stage filtration (crude protein content in the 1st stage permeate per crude protein content in the 1st stage feed) reported in Table 4, where a higher crude protein yield is achieved for the tighter HTM membrane, further indicating that the tighter membrane has a better protein permeation. Table 4. Crude protein content and yield for 1st stage crossflow filtration permeate Experiment Process details Crude protein content (wt% of total dry matter) Crude protein yield Process version 1 (Heat treatment; June) Run 1&2 Process with heat pretreatment 7.9% (7.2%-8.6%) - Process version 2 (CoMem®; June) Run 4 Process with MF ceramic membrane in the first stage (no heat pretreatment) 9.3%*- ** 25.8%*, ** Process version 2 (VO.2; Aug.) Run 10 Process with MF polymeric membrane in the first stage (no heat pretreatment) 8.4%* 28.3%* Process version 2 (HTM; June to Aug.) Run 5-9 and 11 Process with UF ceramic membrane in the first stage (no heat pretreatment) 12.1% (10.3%-14.2%) 38.0% (31.9%-40.3%) Process version 2 (HTM; Sept, to Oct.) Run 12-17 Process with UF ceramic membrane in the first stage (no heat pretreatment) 11.0% (9.7%-12.7%) 38.4% (31.0%-47.9%) *Only single run, ** Sampled during run 3.6.2 Crude protein yield per crude protein content in the plant material The crude protein yields for different process streams, based on the crude protein content in the leaves used for each respective run (with available data), can be found in Table 5. Table 5: Crude protein content in different process streams per crude protein content in the dry leaves Run Raw green juice from screw press 1st Crossflow filtration retentate 2nd Crossflow filtration recovered 8 51% 27% 5.5% 9 53% 33% 3.4% 10 40% 27% 1.9% 11 59% 34% 5.1% 12 35% 16% 4.6% 13 53% 29% 4.7% 14 55% 28% 6.3% 15 48% 25% 5.9% 16 51% 32% 5.4% 17 43% 28% 5.0% The lowest amount of extracted crude protein in the juice produced by the screw press can be found for the run where no shredding of the plant material was performed, i.e. run 12 (see Table 5). If the process options are evaluated by the protein yield of the dia-filtered concentrate, then run 10 displayed the lowest yield (1.9%, see Table 5). This run used a different membrane in the first crossflow filtration stage (a polymeric membrane with a larger pore size compared to the other runs), which suggests that membrane selection may be an important factor for the efficiency of the process. Generally, for process version 2 where a ceramic membrane with a smaller pore size was used, yields around 5% was shown to be achievable for a dia-filtered concentrate stream based on the crude protein content of the leaves, which is a significant improvement. Finally, a crude protein balance based on the extracted juice from the screw press which shows in which stream the crude protein exits the demonstrator process can be found in Figure 7. Due to the difficulty of measuring the retentate volume directly, this volume (which was needed in these calculations) was estimated by applying a nitrogen balance over the first crossflow filtration stage. Of the crude protein extracted into the juice about 45 to 65% is collected in the first cross flow filtration retentate, see Figure 7. About 5 to 10% are captured during the bag filtration and about 20 to 30% is washed out during the second crossflow filtration step, leaving around 10% in the dia-filtered concentrate. 3.6.3 Crude protein yield in 'food stream' per crude protein content in the grass juice The crude protein yields for the final 'food stream' product - i.e. product obtained after diafiltration, based on the crude protein content in the green juice for each respective run (with available data), can be found in Table 6. Table 6: 'Food stream' protein yield Run Crude protein in 'food stream' per crude protein in grass juice 1 5.0% 2 4.4% 6 14.0% 7 12.7% 8 10.9% 9 6.6% 10 5.0% 11 8.5% 12 12.5% 13 9.3% 14 12.3% 15 13.4% 16 10.5% 17 12.1% Run 1 and 2 applied heat treatment of the gras juice before the first stage filtration step, leading to significantly lower protein yield in the final 'food stream' product than the other runs. Furthermore, as mentioned previously, run 10 used a polymeric membrane in the first stage filtration with a larger pore size, differentiating it from the other runs, andleading to significantly lower protein yield in the final 'food stream' product than the other runs, which suggests that membrane material and pore size may be an important factor for process yield. Example 3B: Production of food and feed protein at production scale The aim of this study was to run the process at production scale, capable of processing even greater amounts of fresh leaves, producing protein containing streams suitable for food and feed applications. Overall process layout was as illustrated in Figure 3A. 3.7 Plant material Leaves were harvested from three different fields, all located in Jylland, Denmark and planted with commercial grass blends. For field number 4, the grass blend (Seed Mix 035, DSV) contained 13% white clover and 87% ryegrass of different cultivars. Field number 5 (Agrowgrass 350, Nordic Seed) contained 3% red clover, 10% white clover and 87% ryegrass of different cultivars and field number 6 (Eco grass protein, DSV) contained 16% red clover, 10% white clover, 20% ryegrass, 20% X Festulolium, 25% tall fescue and 7% Timothy, as well as a small amount of herbs. For the experiments, each of the individual batches were harvest and fine cut by commercial agricultural machinery in the morning, about 1-2 hours before shredding and pressing. 3.8 Experimental set-up The experimental setup generally follows the procedure described in Example 3A for process version 2 (Figure 3A), but with scaled up equipment, leading to a few differences. Additionally, recirculation of the grass juice was also explored in these experiments. Briefly, fresh grass was added to the aqueous buffer solution (21.9% w / w di-Sodium hydrogen phosphate dehydrate, 5.5% w / w Potassium phosphate monobasic, 64% w / w Sodium citrate tribasic dehydrate, and 8.6% w / w Sodium sulphite) in a mixer tank and then shredded by running the suspension through a deflaker (GLD 200 B Labyrinta 2023, Cellwood Machinery). The shredded grass was then pressed in a screw press (Scud 6000 Extra 2022, Celikel), producing the raw grass juice. For experiments without grass juice recirculation, the fresh grass was divided into two roughly equal batches. Each batch was added to 500 L of the aqueous buffer solution (containing half the total amount of buffer components each) and had multiple passes through the deflaker before pressing. For experiments with recirculation, fresh grass was added to 500 L of aqueous buffer solution (containing the total amount of buffer components). As grass juice was produced from the screw press, it was then continuously recirculated back to the mixing tank while fresh grass was added. In both cases, the raw grass juice, after pressing was complete, was then prefiltered using a vibration sieve (A40S-1-68CIP-32 2020, AMKCO) with a 32 pm pore size. Two filtration units built by MMS membrane systems were utilized for the two crossflow membrane filtration stages. The filtration unit for the first stage filtration was built with the option to mount different membrane geometries. Two different membrane types were tested for the first stage filtration: Ceramic HTM membranes (membrane area = 22.11 m2) and polymeric V0.2 membranes SW6338 / 48 (membrane area = 32.7 m2). For the second filtration stage and diafiltration, polymeric ST-2BD-6338 membranes (membrane area = 40.88 m2) were used. All three membranes were also investigated in Example 3A. For additional membrane specifications not mentioned here, refer to table 1. Different process conditions were also tested in these experiments, an overview of which can be seen in Table 7. No experiments were performed with any sort of heat treatment (i.e. neither process option 1 (Figure 3B) or otherwise). The TMP during the second filtration step was approximately 1.5 bar and for the diafiltration approximately 1.25 bar. The volumetric crossflow during the second filtration step and diafiltration was set at 34 m3 / h for all experiments. The temperature of any of the different runs was never above 35°C. The product streams (the retentates from the first and second crossflow filtrations) were divided into multiple containers and frozen at -21°C as soon as they had been recovered. Table 7A. Process specifications Run# Source Harvest date Batch size (kg) Leaf dry weight (kg) Total buffer components (kg) Grass juice recirculation (Y / N) 2_A Field 4 11-Jul 23 368.5 73.0 18.280 N 2_B Field 5 05-Sep 23 600 58.8 21.936 Y 2_C Field 6 03-Oct 23 600 56.4 21.936 Y 2_D Field 6 12-Oct23 600 75.6 21.936 Y Table 7B. Process specifications (continued) Run# 1st stage filtration 2nd stage filtration Membrane Pore size (nm) Pressure (bar) Crossflow velocity* Flux (at VCR 3) (LMH) Membrane Dia factor 2_A HTM 60 0.3 2.6 m / s 40c ST Sanitary 8.9 2_B HTM 60 0.3 2.6 m / s 22 ST Sanitary 8.7 2_C V0.2 200 0.3 23 m3 / hB ~5.5 ST Sanitary 8.7 2_D HTM 60 0.8 2.6 m / s 20 ST Sanitary 10.0 A The Reynolds number for flow in a tube geometry will be approximately 8700 at a crossflow velocity of 2.6 m / s. B Crossflow is reported in m3 / h for spiral wound membranes. Crossflow flow was only -60% of the recommended which may affect flux 5 and selectivity. c Recirculation of both retentate and permeate for 1 h in the 1st filtration stage (Flux given for this time). 3.9 Results - Crude protein yield in 'food stream'per crude protein content in the grass juice The crude protein yields for the final 'food stream' product - i.e. product obtained after 10 diafiltration, based on the crude protein content in the green juice for each respective run, can be found in Table 8. Calculations are based on stream sizes, dry matter measurements and crude protein content, determined according to Example 3A. Table 8: 'Food stream' protein yield Run Crude protein in 'food stream' per crude protein in grass juice 2_A 12.8% 2_B 13.0% 2_C 4.8% 2_D 1.8% Process conditions for Run 2_A and 2_B resembled optimal conditions found in Example 3A, leading to high protein yield in the final 'food stream' product. Run 2_B used a membrane in the first stage filtration having a larger pore size and a different membrane material compared to the other runs, leading to significantly lower protein yield in the final 'food stream' product than the other runs. Flux was also significantly lower, showing that process performance is negatively affected as well. Run 2_B used higher pressure in the first stage filtration, leading to lower protein yield. Example 3C: Investigation of membrane MWCO in the second stage filtration and diafiltration The aim of this study was to investigate the effect of MWCO during the second stage membrane filtration and the diafiltration. 3.10 Plant material Leaves were harvested from two different fields, both located in Jylland, Denmark and planted with commercial grass blends. Field number 4 (see section 3.7) was planted with Seed Mix 035 (DSV), while Field 7 was planted with a conventional version of the same blend (Seed Mix 35, DSV). For the experiments, each of the individual batches were harvest by commercial agricultural machinery in the morning, about 1-2 hours before fine cutting, shredding and pressing. 3.11 Experimental set-up The experimental setup generally follows the procedure described in Example 3B, with the following changes: The screw press was rebuild with a new gear and new retention mechanism in order to increase pressing efficiency and a Mono Series A muncher was installed in order to fine cut the leaves immediately before the mixer tank and subsequent shredding, instead of fine cutting on the fields. In addition to the membrane equipment utilized in Example 3B, a smaller additional membrane unit was built, in order to run the second stage filtration in parallel, thereby testing two different membranes on the same feed (Permeate from the first filtration stage). This membrane equipment had a 3.8-inch membrane housing (M3.8-1-PN40, Alfa Laval), a LKH-10 / 163 pump (Alfa LavaL DMS 4.0 kW) and was equipped with manual pressure gauges and a SM7000 and a SM8420 flowmeter, both from IFM, for measuring permeate flow and crossflow respectively. The membrane used for this equipment was a 100 kDa GR40PP membrane (GR40PP-3838 / 48, Alfa Laval), with a membrane area of 4.2 m2. This membrane was then compared to the membrane used in Example 3.B, a 10 kDa ST-2BD-6338 membrane from Synder. This was done in two experiments, where in both cases 10.968 kg of buffer components were added and grass juice recirculation was used. For the first experiment, W3_A, 300 kg fresh leaves (39.0 kg dry weight) were harvest from Field 7 on the 24th of September 2024, and for the second experiment, W3_B, 200 kg (31.6 kg dry weight) were harvest from Field 4 on the 1st of October 2024. The first stage filtration was carried out under the same conditions for both, using the HTM membrane (60 nm pore size), 0.3 barTMP and 2.6 m / s crossflow velocity (Re ~ 8700). The permeate from the first stage filtration was then divided into 2 fractions, according to the deadvolume size of the two second stage filtration units, to reach as comparable concentration ratios as possible in the two units. The TMP during the second filtration stage was 1.5 bar in both cases, while the TMP during the diafiltration was 1.25 bar and 1.5 bar for the 10 kDa and 100 kDa membranes respectively. 3.12 Results - Crude protein and dry matter contents in the diafiltrated concentrates The dry matter and crude protein content in the final diafiltrated streams from the two experiments and each of the two membranes can be seen in Table 9. The final volume concentration ratio (VCR) and the final flux during the second filtration stage, as well as the diafiltration factor can also be seen. Results are obtained as describes in Example 3A. Table 9: Dry matter and crude protein content in diafiltrated concentrate streams and process results for the different second stage membranes Run MWCO Dry matter content Crude Protein content Second stage final VCR Flux at final VCR [LMH] Diafactor W3_A 10 kDa 1.13% ~51% 12.1 ~32 10 100 kDa 1.11% ~63% 11.0 ~70 9.0 W3_B 10 kDa 0.89% 74.8% 10.8 ~43 9.5 100 kDa 0.78% 76.1% 11.7 ~117 10.6 The dry matter content can be seen to be relatively consistent in each experiment, regardless of the membrane chosen. However, a larger difference is seen for the crude protein content in W3_A, where the crude protein is higher for the 100 kDa membrane. This indicates that the larger membrane allows for some components to permeate, that would otherwise be retained by the 10 kDa membrane, thereby increasing the purity of the final 'food protein' concentrate. For W3_B, the crude protein content is also higher for the larger membrane, but the difference is not nearly as pronounced as for W3_A. Additionally, it can be observed that the larger MWCO of the 100 kDa membrane has a significant effect on the flux, which increases by more than double at approximately equivalent VCRs. The results therefore indicate, that increasing the MWCO during the second stage filtration and diafiltration potentially has a positive impact on protein purity, while significantly benefitting the filtration flux. Example 4: Identification of proteins 4.1 Objectives This example describes the exploration of proteins found in plant juice from clover grass. It outlines the proteins which are removed as a result of the membrane process of the present invention. The investigation was carried out using shotgun bottom-up proteomics analysis by tandem mass spectrometry of full technical triplicates. The investigation provides qualitative and quantitative insight into the proteins and their fate during processing. 4.2 Method Clover grass was harvested and pressed, and then further processed according to the present invention using a 1st stage filtration step and a 2nd stage filtration step followed by diafiltration; the process conditions were identical to the standard conditions for the second process option in Example 3A, using the UF ceramic membrane (HTM) in the first stage, bag filter pre-filtration, the standard buffer solution and with shredding. The experiment was carried out in early October 2022. Representative samples of the prefiltered green juice (PFGJ), the 1st stage filtration permeate (clarified juice, CJ), and the diafiltration concentrate (DC) were taken and immediately cooled and subsequently lyophilized to dryness. Dry triplicate samples were prepared for shotgun bottom-up proteomics analysis by liquid chromatography and tandem mass spectrometry (LC-MS / MS) as described below. 4.2.1 Sample preparation for LC-MS / MS Samples were prepared for proteomics analysis using the "iST Kit for Plant Tissues (PreOmics, Germany)" according to manufacturer recommendations. Briefly, a dry sample corresponding to ~80 pg protein (based on N*6.25 in accordance with Example 3A) was dissolved in 100 pL "Lyse" and heated to 95°C for 10 minutes at 1000 rpm in a heating block to facilitate dissolution, disulfide reduction, and cysteine alkylation. As samples were fully soluble and had already been subjected to pressing and pre-filtration, whereby cell debris and residual particular matter was removed, shearing was omitted based on previous investigations. After cooling, proteins were digested with Trypsin / LysC for 3 hours at 37°C and 500 rpm before terminating the reaction by adding "Stop". Samples were centrifuged (16000 ref, 1 min) before transferring to the cleanup cartridge and subjected to a three-step washing procedure (Wash 0 + 1 + 2) and subsequently eluted into new tubes. Eluates were lyophilized to dryness and resuspended in "LC Load" and diluted to a total concentration of ~0.5 mg / mL based on absorbance (Nanodrop A280, 1 Abs Unit = 1 mg / mL). 4.2.2 Bottom-up proteomics analysis by LC-MS / MS Protein digests were analyzed on an LC-ESI-MS / MS consisting of an EASY-nLC system (Thermo Scientific) coupled to a Q Exactive HF mass spectrometer (Thermo Scientific) with a Nanospray FLex ion source (Thermo Scientific). Approximately lpg protein digest was loaded on a reverse phase Acclaim PEPMAP NANOTRAP column (C18, 100 A, 100 pm. x 2 cm, (Thermo Scientific)) in solvent A (0.1% FA) followed by separation on a reverse phase ACCLAIM PEPMAP RSLC analytical column (C18, 100 A, 75 um x 50 cm (Thermo Scientific)). Peptides were eluted by constant flow at 300 nL / min during a 60 min ramped gradient from 5 to 100% of solvent B (0.1% FA in 80% ACN, Fischer Scientific). MS was operated in positive ion and data dependent top-20 mode, were the (up to) 20 most intense MSI precursors were selected for higher energy C-trap dissociation (HCD) fragmentation at 28 eV using a window of isolation of 1.2 m / z. Survey scans were obtained at a resolution of 60.000 at 200 m / z and HCD spectra were obtained at 15.000 at 200 m / z. Maximum ion injection time was set to 50 for MS and 45 for MS / MS scans. The underfill ratio was set to 3.5% and a dynamic exclusion of 30 sec was applied. During acquisition, "peptide match" and "exclude isotopes" were enabled. 4.2.3 LC-MS / MS data analysis Data was analyzed in MaxQuant (vl.6.0.43) for identification and quantification of proteins using standard setting: Two missed clevages and up to five modifications (fixed: Cys carbamidomethylation; variable: Met oxidation and N-term alkylation) allowed, a seven amino acid minimal length was applied, a false discovery rate (FDR) fixed at 1% on both peptide- and protein-level, reverted sequences were used as decoys to control FDR, common contaminants were included in searches. Data was searched against a custom protein database including all proteins found in UniProt for the Trifolium (taxon id:3898), Lolium (taxon id:4520), and Festuca (taxon id:4605) genera (downloaded May 7th 2022). Quantification was performed including both unique and razor peptides using MaxLFQ (for enrichment analysis between samples) and iBAQ (for in-sample protein distribution by relative molar abundance). In the differential analysis, a protein was considered differentially abundant if at least two-fold difference (i.e. Iog2 fold change > 1) in protein abundance (normalized LFQ intensities) and significance level is set at 5% FDR. (p-value > 0.05) is found. For relative molar abundance, the iBAQ intensity for each protein group was divided by the sum of iBAQ intensities for all proteins groups within an individual sample replicate, following filtering of contaminants and false positive identifications. 4.3 Results Across the nine analyzed samples (three sample triplicates), 1483 proteins were quantitatively identified in at least one sample. In the three sample types, at least 800 proteins were identified in at least two of three replicates. In the analysis of samples from the purification process according to the present invention, it was found by differential proteomics, that 192 proteins were significantly depleted at 5% FDR following the 1st stage filtration step, compared to the pre-filtered green juice (Figure 8A). Chlorophyll a-b binding protein (Q6T705, Trifolium pratense) and Photosystem II CP47 reaction center protein (K4PMQ7, Festuca pratensis) are examples of proteins which were highly depleted by the filtration processes (black and orange arrow in Figure 8A, respectively). In the pre-filtered green juice, the differentially abundant proteins (across compared samples as determined by normalized LFQ intensities) constitute ~32% of the total protein (molar fraction as determined by riBAQ); whereas in the permeate of the 1st stage filtration step, they constitute ~2.5% (by riBAQ). Among these, particularly three groups stand out: • Chlorophyll a-b binding proteins (Group A, 15 proteins) • Photosystem-related proteins (Group B, 16 proteins) • Histones (Group C, 4 proteins) Prior to the 1st stage filtration step, the three groups constitute 17.1% (by riBAQ) of the total protein (8.2%, 8.0% and 1.9% for group A, B, and C, respectively); while following 1st stage filtration, they are reduced to constitute 0.21% (by riBAQ) of the total protein (0.03%, 0.04%, and 0.14% for group A, B, and C, respectively). Group C (Histones) are involved with DNA folding / storage and may potentially be considered as indicators of intact cells. Significant depletion indicates efficient removal of residual cells in the 1st stage filtration step. Groups A and B share the property of being related to light harvest (energy production), are associated with chloroplasts, and both have the ability to bind high amounts of chlorophyll. Chlorophyll is the dominant green pigment in leaves and considered unappealing in a protein product. In addition to color, chlorophyll extracts have previously been reported to be bitter. Following the 2nd stage filtration step and subsequent diafiltration, the number of significantly depleted proteins was 145 compared to the pre-filtered green juice (Figure 8B). Chlorophyll a-b binding protein (Q6T705, Trifolium pratense) and Photosystem II CP47 reaction center protein (K4PMQ7, Festuca pratensis) remained highly depleted (black and orange arrow in Figure 8B, respectively). Depleted proteins constitute 1.2% of the total protein compared to 30.3% in the pre-filtered green juice. The three aforementioned groups depleted during 1st stage filtration remain depleted, constituting 0.11% (by riBAQ) of the total protein (0.01%, 0.04%, and 0.06% for group A, B, and C, respectively). The difference between the pre-filtered green juice and the filtration fractions is also illustrated in a heatmap, including all proteins (Figure 9A) and the proteins found depleted following 1st stage filtration (Figure 9B). Both examples show clear differentiation in protein composition, indicating selectivity across the membranes. The visual color difference of the protein fractions is illustrated in Figure 10. The prefiltered green juice is green, while the green color has been depleted from the 1st stage permeate and the diafiltration concentrate. Removal of green color correlates well with removal of chloroplastic (and chlorophyll-affiliated) proteins. In addition to groups A and B as discussed above, this was further substantiated by removal of e.g. the chloroplastic "Outer envelope pore protein 16" (Uniprot AC# A0A392PBQ3), which constitutes 0.25% of the total protein in the pre-filtered green juice and was not detected following filtration in either of the fractions. Removal of the green color also correlates with removal of bitterness of taste while retaining protein nativity and thus obtaining a highly soluble and functional product (see Example 6) 4.4 Conclusion The chlorophyll-binding proteins and photosystem proteins were efficiently removed by the filtration steps of the present invention (from >16% of the total protein to <0.1%), correlating with the removal of the green color and reducing the bitterness of the final product substantially. Example 5: Protein-level - comparison of present invention to prior art 5.1 Objectives This example describes the quantitative, protein-level composition of a product obtained using the membrane process of the present invention in comparison to a prior art approach. The example investigates a broad selection of process samples from both approaches and was carried out using shotgun bottom-up proteomics analysis by tandem mass spectrometry. The investigation provides qualitative and quantitative insight into the proteins and their fate during processing and a direct comparison of the method of the current invention to the prior art. 5.2 Methods To emulate prior art, a lab-scale process was designed (Figure 11). The process included screw pressing of clover grass (from University of Copenhagen), police filtering to remove particulate matter, heat precipitation at 60 °C, and centrifugation at 2000 g for 10 minutes. The supernatant was subsequently split into two parallel processes, where the first represent a two-stage membrane filtration process (200 nm PES and 10 kDa RC), while the second represent prior art by including acid precipitation (pH 4.2) and subsequent centrifugation. From this process, representative samples were taken at each step for analysis by SDS-PAGE and the following samples were selected for analysis by LC-MS / MS: Unprocessed grass (UG), raw green juice (RGJ), precipitate following heat treatment (HP), supernatant following heat treatment (HS), retentate from 1st stage filtration of HS (HR1), retentate from 2nd stage filtration of HS (HR2), sediment after centrifugation of acid precipitate from HS (AP). All samples were lyophilized and resolubilized prior to analysis. As representative of the process of the present invention, a representative sample from Example 3A (Run 6) was used. Process samples submitted for analysis were prepared as described herein. From this run, representative samples were taken at each step for analysis by SDS-PAGE and the following samples were selected for analysis by LC-MS / MS: Raw green juice (RGJ), 1st stage crossflow filtration retentate (RI), 2nd stage crossflow filtration permeate (Pl), and diafiltration concentrate (DC). Samples were analyzed prior to drying. SDS-PAGE analysis was performed as described in Example 3A, where lyophilized samples from the emulated prior art process were solubilized in either deionized water or 100 mM ammonium bicarbonate (pH 8.6) with 0.2 % (w / v) sodium dodecyl sulfate. LC-MS / MS sample preparation and analysis as well as downstream data processing was performed as described in Example 4. 5.3 Results SDS-PAGE analysis was performed on sub-sample from different processing steps in the emulation of prior art (Figure 12) and the process of the current invention (Figure 13). Moreover, the sub-samples from the emulated prior art process were resolubilized in both water (Figure 12A) and detergent-containing buffer (Figure 12B) to illustrate solubility. The ~25 kDa chlorophyll a-b BP band is indicated with a dashed box while the desired RuBisCO subunits (large and small at ~50 kDa and ~15 kDa, respectively) are indicated in solid boxes. The analysis illustrates that while heat treatment is capable of producing a soluble supernatant fraction enriched in desired protein (Figure 12A, lane 7), downstream membrane filtration still shows presence of the undesired protein (Figure 12A, lane 10) whereas application of acid precipitation (Figure 12A, lane 12) renders the product insoluble in water. By addition of detergent to the process streams to facilitate resolubilization, it is further substantiated that although the majority of the undesired protein band at ~25 kDa is depleted during heat treatment (Figure 12B, lane 6 and 7), the step results in a massive loss of protein by precipitation in addition to the residual undesired protein and solubility issues. In contrast, the process of the current invention shows no indication of undesired protein after the 1st stage crossflow filtration (Figure 13, lanes 9-13), as undesired protein, visible in upstream samples (Figure 13, lanes 1-6), is retained in the 1st stage retentate (Figure 13, lane 7). The outcome of the process of the present invention was also demonstrated by identifying the amount of desired vs undesired proteins (as per Example 4) in different fractions from different steps of the processes using bottom-up proteomics by LC-MS / MS. Specifically, RuBisCO subunits (both large and small) are desired proteins, while Chlorophyll a-b binding proteins and Photosystem proteins are undesired proteins due to their association with the green pigment chlorophyll. In Figure 14, the content of RuBisCO subunits, Chlorophyll a-b binding proteins and Photosystem proteins (relative within each sample) are shown for different fractions obtained from the purification process according to the present invention and the process emulating prior art. It was found that the process of the present invention is superior for removal of the specific unwanted protein groups. While only minor amounts (~0.5% by riBAQ) of undesired protein was found in the ultrafiltration retentate subsequent to heat treatment (HR2), residual protein was identified quantitative, in agreement with observations from SDS-PAGE analysis. Using a process fully emulating prior art, i.e. heat treatment and subsequent acid precipitation, a larger proportion of undesired protein was identified (AP). In contrast, the undesired protein was fully removed (not identified) following 1st stage crossflow filtration of the current invention (Pl) and remained fully depleted (not identified) in final diafiltration concentrate (DC). Moreover, the analysis also shows a relative enrichment of RuBisCO protein from constituting ~35% of the total protein (by riBAQ) in the raw green juice (RGJ) in both the lab-scale emulated process and the representative sample from the process of the current invention, to constituting 60% of the total protein (by riBAQ) in the diafiltration concentrate (DC). This is an even higher content of the desired RuBisCO protein than obtained using the prior art (AP), where RuBisCO constitutes 55% of the total protein (by riBAQ). 5.4 Conclusion The example corroborates the finding from Example 4 that chlorophyll-binding proteins and photosystem proteins are efficiently removed by the process of the present invention. While these proteins are also largely removed by prior art, they still remain in minor amounts, which impacts color and flavor of the end product. Moreover, prior art also renders the protein isolate insoluble in water, whereas the process of the current invention is based on gentle processing of aqueous solution, thereby producing a watersoluble product. Furthermore, the process of the present invention can concentrate the desired RuBisCO protein to a higher relative protein content than prior art, thereby improving the nutritional value of the product due to the favorable amino acid composition of RuBisCO. Example 6: Properties of the protein product(s) 6.1 Objectives This example investigates the functional properties of the products obtained by application of the process of the current invention. For this purpose, the foaming and emulsification properties of the lyophilized dia-concentrate from all runs, as per Example 3A, were screened using standardized coarse methodology. The lyophilized dia-concentrate from Example 4 was investigated for pH-dependent solubility. Furthermore, sensory properties (i.e. color, smell, and taste) were evaluated on selected samples. 6.2 Methods Dia-concentrates from all runs, as per Example 3A, were lyophilized prior to screening for foaming and emulsification properties. The dia-concentrate, as per Example 4, was lyophilized prior to investigation of pH-dependent solubility. Taste and smell were evaluated on dia-concentrates as per Example 4 and Example 5 (both lyophilized and spray dried dia-concentrate of "Run 6" as per Example 3A) as well as a prototype of a "feed-grade" protein extract. The prototype was produced from the 1st stage crossflow filtration retentate, which was subsequently heat treated and sedimented in a decanter centrifuge. Color was assessed on process samples as per Example 4 and representative samples used for tasting. 6.2.1 Foaming properties Foaming properties were evaluated by determination of foaming capacity (FC) and foaming stability (FS). Lyophilized dia-concentrates were dispersed in 15 mL water at 2 mg / mL (protein concentration based on protein content and dry matter as per Example 3A) in a 50 mL graduated cylinder and a foam was generated by homogenization using an UltraTurrax at 16000 rpm for 1 min. FC was determined as foam volume relative to initial volume as: FC = V_foam I VJnitial * 100%, where VJnitial corresponds to the initial 15 mL volume, while V_foam is the foam volume immediately after foam generation. FS was evaluated at 10 min and 30 min subsequent to foam generation and determined as: FS_x = V_foam_x / V_foam * 100%, where FS_x is the foam stability after x minutes and V_foam_x is the foam volume after x minutes. 6.2.2 Emulsification properties Emulsification properties were evaluated by determination of the emulsification activity index (EAI) and emulsification stability index (ESI). Lyophilized dia-concentrates were dispersed in deionized water at 4 mg / mL (protein concentration based on protein content and dry matter as per Example 3A). Solutions were mixed with rapeseed oil to a final oil content of 23.5% (w / w) and emulsified using a two-stage protocol. Primary homogenization was performed using a Polytron PT 1200e homogenizer at 16000 rpm for 30 seconds followed by secondary homogenization using Microson XL2000 sonicator (Pl probe) at 75% amplitude for two 30 second passes with 1 minute break. Immediately after emulsification, a sample aliquot was taken from the bottom of the container and diluted 100-fold in 0.1% aqueous sodium dodecyl sulfate (SDS). The diluted emulsion was gently inverted and the absorbance was measured at 500 nm. Sampling was repeated after 10 minutes. EAI was determined as: EAI (mA2 / g) = (A_0 * D * 2 * 2.303) / (0 * c * 10,000), where A_0 is the absorbance (1 cm light path) at 500 nm of the sample taken immediately after emulsification, D is the dilution factor in 0.1% SDS (i.e. 100), 0 is the oil fraction (i.e. 0.235), and c is the protein concentration used for emulsification (i.e. 0.004 g / mL). ESI was determined as: ESI (min) = (A_0 / (A_0 - A_10)) * 10 min, where A_0 is the absorbance (1 cm light path) at 500 nm of the sample taken immediately after emulsification and A_10 is the absorbance of the sample taken 10 minutes after emulsification. 6.2.3 Solubility Lyophilized dia-concentrate, as per Example 4, was solubilized in phosphate buffer with a pH adjusted in the range from 2-12 at 2 mg / mL (protein concentration as per protein content per dry matter) by vortexing for 30 seconds and agitation (roller mixer) for 30 minutes. Subsequently, the solution was centrifuged at 7500 ref for 15 minutes and the supernatant taken for characterization. Nitrogen content was determined as described in Example 3A. Solubility was determined as solubilized nitrogen relative to added nitrogen as: Solubility = N_supernatant / N_sample. Aliquots (10 pL corresponding to 20 pg protein if fully solubilized) were analyzed by SDS-PAGE, as described in Example 3A. 6.2.4 Sensory evaluation Color was evaluated by visual inspection and documented by photography. Sensory properties (smell and taste) were evaluated by a small (n = 11), untrained panel. Smell and taste were evaluated on dry product as well as redispersed product in tap water in a concentration range of 1 mg / L to 10 g / L (DM basis). Aqueous samples were evaluated from low to high concentration with thorough pellet washing with water between samplings. Along with the samples of the current invention, reference samples including bovine casein, pea protein, and brewer's spent grains were included. Panelists were asked to rate their experience of the different samples in terms of pleasantness of smell and taste from unpleasant (1) to pleasant (5) as well as bitterness from not bitter (1) to bitter (3). Furthermore, panelists were asked to note down any notes / associations obtained during the evaluation. 6.2.5 Gelling Gelling properties may be investigated by determination of the least gelling concentration for thermo-set gels. In brief, the functional protein product is solubilized in a range of different concentrations (e.g. 0.1-10% (w / v)) and subsequently thermoset by heating (e.g. 85°C for 1 hour). Subsequently, thermo-set gels are cooled down (e.g. 4°C overnight) before being inspected. A set gel is defined as a gel that can withstand tube inversion without running / falling down the sides of the tube (e.g. a 15 ml Griner tube), and the lowest concentration providing a set gel is defined as the least gelling concentration (LGC). 6.3 Results 6.3.1 Foaming properties Foaming capacity (FC) and foaming stability (FS) after 10 minutes and 30 minutes was determined for all runs as per Example 3A (Table 10). With the exception of "run 3" (where no dia-concentrate was produced), all lyophilized dia-concentrates showed excellent FC (>150%) and good short-term stability (>60% after 10 minutes). Longer term stability (30 minutes) differed more substantially between runs ranging from 24% to 93%. Compared to the lab-scale emulation of prior art (heat treatment and acid precipitation as per Example 4) which showed an FC of 33%, a FC of 25% after 10 minutes, and a fully collapsed foam after 30 minutes, the "food-grade" dia-concentrates are largely superior. The observed foaming properties of the "food-grade" product of the present invention, which are better than many other relevant plant-based proteins (e.g. Ma et al 2022), makes it a high potential ingredient for food applications. 6.3.2 Emulsifying properties The emulsification index (EAI) and emulsification stability index (ESI) was determined for all runs as per Example 3A (Table 10). Compared to observed foaming properties, emulsifying properties were found to be more variable. Nevertheless, the dia-5 concentrate from several runs indeed showed good emulsifying properties with EAI > 60 mA2 / g and ESI > 100 minutes which surpass literature reports of other relevant plant-based protein (e.g. Ma et al 2022) and commonly used animal-based proteins such as sodium caseinate. This was an interesting and surprising observation, as native proteins have often been reported to display poor emulsifying properties compared to 10 e.g. their denatured counterparts due to the requirement for structural reorganization at the oil / water interphase. Preliminary trials with enzymatic hydrolysis indicates improved emulsification. Table 10 Foaming and emulsifying properties Run FC (%) FS (%) at 10 min FS (%) at 30 min EAI (m2 / g) ESI (min) 1 267 85 63 79.9 50.8 2 200 97 83 97.1 17.6 3 - - - - - 4 33 60 40 32.6 34.5 5 300 89 78 12.4 101.1 6 200 80 33 26.6 20.2 7 293 73 45 20.7 115.8 8 180 93 93 62.2 100.1 9 180 93 56 27.6 40.5 10 267 88 80 39.5 98.9 11 233 77 43 35.9 47.1 12 167 60 24 82.6 104.6 13 153 75 35 76.7 171.5 14 267 85 68 23.7 97.6 15 233 83 43 48.3 40.7 16 233 86 57 60.0 44.9 17 167 80 52 69.2 73.2 6.3.3 Solubility 15 The pH-dependent solubility of the lyophilized dia-concentrate (as per Example 4) showed good solubility across a wide pH span. Based on solubilized nitrogen, a solubility >60% was observed in the range from pH 4 to pH 10 with the highest solubility observed at pH 8. At more acidic conditions (pH 2) a solubility of 39% was observed whereas more alkaline conditions (pH 12) resulted in a solubility of 58%. SDS-PAGE analysis revealed no substantial bias in protein solubilization based on applied pH value, albeit band intensity of RuBisCO large subunit did decrease at acidic pH. As the dia-concentrate during production is a fully soluble aqueous solution, the protein product will prior to drying be fully water soluble. Resolubilization of the dried product is affected by both the applied drying and solubilization process, which can be further optimized. During sensory evaluation (see Example 6.3.4), it was observed that spray dried product was substantially harder to solubilize than lyophilized product. This is suspected to relate to a partial denaturation of the protein at the drying conditions applied. Preliminary data suggests that a substantially improved solubility of spray dried product is possible by decreasing inlet and nozzle temperatures during drying. 6.3.4 Sensory evaluation 6.3.4.1 Color The color of the dried product (Figure 15, A-C) was found to be light with pale, brownish hints. The color of the feed-grade prototype (Figure 15, D) was, in contrast, found to be dark with a greenish and brownish appearance. These observations correlate well with observations during processing where the feedstock for 1st stage cross-flow filtration (Figure 10, A) was found to be intensely green whereas the dia-concentrate (Figure 10, C) was bright with pale, brownish hints. The feedstock for 2nd stage cross-flow filtration (Figure 10, B) was brighter in color due to the high salt content from the stabilizing solution, which were removed during diafiltration. During the lab-scale emulation of prior art (as per Example 5) it was found that while a combination of heat treatment and acid precipitation produced a whiter product with less brownish hints, the product instead contained pale, greenish hints (Figure 16, G). Moreover, the drying method was observed to influence the color, where spray drying produced a lighter product. It is suspected that this relates to a smaller particle size and finer powder obtained with spray drying. 6.3.4.2 Smell The smell of the dia-concentrates (Figure 15, A-C) was evaluated by the panel as neutral to slightly pleasant (average score = 3.4). The panelists noted that the dry and redispersed product from the process of the current invention was slightly sweet and presented "earthy" notes with hints of chocolate and licorice. No panelists reported "grassy" smell sensations. In contrast, the prototype of the feed-grade product (Figure 15, D) presented "grassy" notes and a more intense "earthy" and "hay-like" smell. In general, dried products were reported to have a more pleasant smell in dry form compared to redispersed form at high concentration (~0.5 points increase). These observations align with smell observations during process development, where the "grassy" notes were eliminated following the 1st stage cross-flow filtration. In contrasts, in was observed that the product emulating the prior art (Figure 16, G) still retained slight "grassy" notes as well as "acidic" notes, substantiating the requirements for further washing of a product produced using such a process. 6.3.4.3 Taste The taste of the dia-concentrates was evaluated by the panel as neutral in pleasantness (average score = 3.1). The panelists reported that dry dia-concentrates had a "malty" taste with association to bread, pistachio, caramel, beer and a "creamy" and slightly "salty" sensation. The dry dia-concentrates were generally reported to have very limited bitterness (average score = 1.3). The feed-grade prototype was considered substantially less pleasantly tasting (average score = 2.3) with a clear "grassy" taste as well as association to seaweed, ocean, green tea, herbs, and disintegrated plant material with more moderate bitterness (average score = 1.6). As found for smell, the dried dia-concentrates generally received a higher score than redispersed samples at high concentration (~0.5 points increase). At lower concentrations (< 1 g / L), the taste of the dia-concentrates was generally rated as neutral in pleasantness (average score = 2.8). Due to the low solubility of the feed-grade prototype, the evaluation was found unreliable in relation to concentration effects for redispersed samples although a decreased pleasantness was reported with increasing concentration (based on amount of suspended and sedimented solids). 6.4 Conclusions The product from the process of the current invention displays excellent foaming properties with a capacity of > 150% and a stability > 60%, when the specified membranes are used and regardless of other process parameters. Foaming properties are superior to literature reports of other plant-based protein products. Emulsifying properties of the product appears to be less pronounced and somewhat variable, although process parameters may be found to produce a product with promising emulsifying properties. Nevertheless, downstream processing, through e.g. enzymatic hydrolysis, has repeatedly been reported to enhance functional properties (in particular emulsification) of plant proteins (consistent with preliminary trials), which would allow for tailoring of functionality based on desired application. The gentle process of the present invention retains proteins in their native state, hereby securing high solubility over a wide pH range and retaining functionality of the protein while removing color. Sensory evaluation indicates that the process of the current invention allows for the production of a protein product with a neutral perception of taste and smell, which associates particularly to malt in taste. Moreover, the feed-grade protein produced from the process of the current invention generally has a more negative perception and clear associations with grass. Ultimately, functional and sensory evaluation indicated that the process of the current invention effectively removes undesired color, taste, and smell characteristics meanwhile producing a soluble, protein-rich product with very promising functional properties. This product has great potential for application as a functional ingredient in foods. REFERENCES Damborg, V. K., Jensen, S. K., Weisbjerg, M. R., Adamsen, A. P., & Stodkilde, L. (2020). Screw-pressed fractions from green forages as animal feed: Chemical composition and mass balances. Animal Feed Science and Technology, 261, 114401. Nynas, A. L., Newson, W. R., & Johansson, E. (2021). Protein fractionation of green leaves as an underutilized food source—protein yield and the effect of process parameters. Foods, 10(11), 2533. Ma, Kai Kai, et al. "Functional performance of plant proteins." Foods 11.4 (2022): 594. https: / / doi.org / 10.3390 / foodsll040594 Zhang, W., Grimi, N., Jaffrin, M. Y., & Ding, L. (2015). Leaf protein concentration of alfalfa juice by membrane technology. Journal of membrane science, 489, 183-193.

Claims

1. A method for producing a functional protein product and optionally a feed protein product from plant material, said method comprising the steps:(i) providing a green plant material, and mixing said plant material with an aqueous 5 solution;(ii) disintegrating and pressing the mixture from step (i) to obtain a pulp and a raw green juice;(iii) optionally subjection the raw green juice to a pre-filtration step to obtain a prefiltered green juice;10 (iv) subjecting the raw green juice or the pre-filtered green juice to a 1st stage crossflow membrane filtration step to obtain a 1st stage retentate and a 1st stage permeate,wherein the nominal pore size of the 1st stage crossflow membrane filtration is between 15-80 nm, preferably between 40-80 nm, such as approx. 60 nm,wherein the pressure applied in the 1st stage crossflow membrane filtration is 15          between 0.1-0.6, preferably 0.2-0.6 bar, andwherein the crossflow of the 1st stage crossflow filtration shall induce turbulence;(v) subjecting the 1st stage permeate to a 2nd stage crossflow membrane filtration step to obtain a 2nd stage retentate and a 2nd stage permeate,wherein the 2nd stage crossflow membrane filtration has a MWCO between 1-500 20          kDa, preferably between 1-200 kDa, more preferably 1-100 kDa, most preferablybetween 5-20 kDa, such as approx. 10 kDa, andwherein the pressure applied in the 2nd stage crossflow membrane filtration is between 1-10 bar, preferably between 1-6 bar, more preferably between 1-4 bar;(vi) subjecting the 2nd stage retentate to a diafiltration step to obtain a dia-concentrate 25 and a dia-permeate;(vii) recovering a functional protein product from the dia-concentrate; and(viii) optionally recovering a feed protein product from the 1st stage retentate.

2. The method according to claim 1, wherein the 1st stage permeate and the 2nd stage 30 retentate comprise RuBisCO protein.

3. The method according to claim 1, wherein the method does not comprise heating above 60°C, such as not above 55°C, 50°C, 45°C, or 40°C.

4. The method according to claim 1 or 2, wherein the green plant material is selected from alfalfa, clover, lupines, grasses, crucifers, beets, chicory, carrot, radish, cassava, roadside crops and combinations hereof.

5. The method according to any one of claims 1-4, wherein the green plant material is clover and / or grass; most preferably red clover (Trifolium prantese), white clover (Trifolium repens'), perennial ryegrass (Lolium perenne), fescue grass species (Festuca sp.), and blends of one or more of these plant materials, such as clover grass.

6. The method according to any one of claims 1-5, wherein the aqueous solution is a phosphate buffer having a pH in the range 6-8, said buffer comprising a reducing agent selected from sodium sulfite and mercaptoethanol, and a stabilizing agent selected from EDTA, citrate, and other organic acids.

7. The method according to any one of claims 1-6, wherein the raw green juice is subjected to the pre-filtration step, and wherein the pre-filtration step is selected from (i) dead-end filtration, using a filter having a pore size in the range 1-50 pm (e.g. polymetic filter) and (ii) vibration sieve filtration using a sieve filter having a pore size in the range of 10-50 pm (e.g. steel screen).

8. The method according to any one of claims 1-7, wherein the 1st stage crossflow membrane filtration is operated using a ceramic membrane, at a crossflow velocity between 1-5 m / s and a pressure between 0.2-0.6 bar.

9. The method according to any one of claims 1-8, wherein the nominal pore size of the 1st stage crossflow membrane filtration is between 40-80 nm, such as approx. 60 nm.

10. The method according to any one of claims 1-9, wherein the 2nd stage crossflow membrane filtration is operated using polymeric membrane and a pressure between 18 bar.

11. The method according to any one of claims 1-10, wherein the 2nd stage crossflow membrane filtration has a MWCO between 1-200 kDa, such as between 1-100 kDa.

12. The method according to any one of claims 1-11, wherein the 2nd stage crossflow membrane filtration has a MWCO between 5-20 kDa, such as approx. 10 kDa.

13. The method according to any one of claims 1-12, wherein the diafiltration is operated using the same membrane as the 2nd stage crossflow membrane filtration, and at a diafiltration factor of 2-10, preferably 2-8, most preferably 2-6.

14. The method according to claim 13, wherein the pressure applied for the diafiltration is between 1-8 bar.

15. The method according to any one of claims 1-14, wherein turbulence is characterized by a Reynolds number > 2900 for a tube geometry.

16. A functional protein product obtainable by the method according to any one of claims 1-15.

17. The functional protein product according to claim 16, wherein the functional protein product is soluble in an aqueous solution pH 6-8.5    18. The functional protein product according to claim 16 or 17, wherein the functionalprotein product has(i) a foaming capacity (FC) >150%, and / or a foaming stability (FS) >60% after 10 minutes,(ii) an emulsification index (EAI) value >25 m2 / g and / or an emulsification stability index 10 (ESI) value >35 minutes, and / or(iii) a least gelling concentration (LGC) of <2% as a heat-set gel.

19. The functional protein product according to any one of claims 16-18, comprising RuBisCO protein.

20. A feed protein product obtainable by the method according to any one of claims 115    15.

Images