Keywords

6.1 Introduction

The production of high-value biopharmaceuticals and other recombinant proteins for diagnosis or industrial application is mainly based on microbial and cell cultures in large bioreactors (De Jaeger et al. 2002; Tschofen et al. 2016; Kesik-Brodacka 2018). However, research has demonstrated the expression of hundreds of different recombinant proteins in plants over the last decades. Some have reached commercial production, confirming the viability and potential of this approach (Schillberg et al. 2019; Liu and Timko 2022). The advantages of using plants as an alternative to the other, more established systems based on microbial and cell culture are the reduced costs, the lower probability of infection by pathogens capable of infecting humans and other mammals, and the easy scale-up production (Fischer and Buyel 2020; Schillberg and Finnern 2021). Biomass production from plants is more sustainable than bioreactor-based production once it requires soil, fertilizers, and water, instead of expensive complex culture media (Obembe et al. 2011). Besides, the technology for cultivation is largely available and demands less skilled staff (Fischer and Buyel 2020; Liu and Timko 2022).

Different plant-based systems have been described for the production of recombinant proteins (see also chapter “Molecular Farming of Pharmaceutical Proteins in Different Crop Systems: A Way Forward”). These include the transient expression in leaves and the stable integration of the foreign gene for the expression in seeds, leaves, fruits, tubers, roots, aquatic plants, moss, hairy root culture, and cell suspension culture (Xu et al. 2018). Each of these systems presents advantages and specific applications. However, to be competitive, a plant-based system must ensure a high-level expression of the recombinant protein. To this end, several factors that can increase both gene expression and protein accumulation in the cell should be considered. These include the type of tissue or cell; selection of strong promoters and enhancer elements; and features of the target protein, subcellular targeting, and posttranslational modifications (Streatfield 2007; Ghag et al. 2021). Next, the downstream processes must be cost effective, and the final product must be safe, sufficiently pure, and biologically active (Wilken and Nikolov 2012).

This chapter discusses factors involved in producing recombinant proteins using seeds as a platform. The advantages, limitations, and challenges of seed-based production systems are commented, as well as biosafety issues and general aspects associated with the purification of heterologous proteins from seeds.

6.2 Seeds as Bioreactors for Producing and Storing Recombinant Protein

Expressing a recombinant protein in seeds has some unique advantages. During development, endosperm cells are committed to producing and storing proteins and other nutrients (oils, starch, carbohydrates, etc.) to nurture the developing plantlet after germination (Li and Berger 2012). Thus, the target protein expressed in the seed finds a cellular environment that favors protein accumulation (Robinson et al. 2005; Khan et al. 2012).

Product yield and protein quality are critical factors for any recombinant protein production system. Yield involves the efficiency of biosynthesis (i.e., transcription and translation) and the stability of the recombinant protein in the cell (Chen et al. 2020; Liu and Timko 2022). Quality may include the correct assembly and posttranslational modifications, which are involved in protein turnover processes, impacting yield, and are likely necessary for the protein to retain its biological function (Vitale and Boston 2008; Thomas and Walmsley 2015; Strasser et al. 2021). This balance between translation and turnover will affect protein accumulation and involves features of the recombinant protein, the subcellular location where it is directed, and the metabolic burden associated with transcription and translation of this protein (Thomas and Walmsley 2015).

In leaves, very high transcriptional and translational levels of a heterologous gene can be achieved, particularly in transient expression assays (Fischer et al. 1999; Gleba et al. 2004; Pogue et al. 2010). This high expression of a foreign gene, nonessential, may impose a metabolic imbalance on the cell and activate endogenous protection mechanisms that can limit the expression and accumulation of the heterologous protein or direct the cell to apoptosis (Thomas and Walmsley 2015). In seeds, the transcription level may not be as high as in leaves. Still, it can be steadily maintained throughout the developing endosperm and embryo, reducing the metabolic burden on these cells (Boothe et al. 2010). That does not imply that a seed-based production system will produce any recombinant protein. According to the type of recombinant proteins, some may be poorly expressed while others may not even be detected. There will also be differences in the target protein level between different transgenic lines expressing the same gene (Streatfield 2007; Hood et al. 2012). In any case, transgenic cereals and legume grains have been shown to accumulate high amounts of different types of recombinant proteins at consistent levels throughout different generations and batches (Hudson et al. 2014; Mirzaee et al. 2022). Seed production may take a couple of months. Still, a recombinant protein therein is steadily accumulated and stably maintained throughout seed development and can be stored for 4–6 years after the harvest, allowing scale-up production (Oakes et al. 2009; Boothe et al. 2010).

Once the recombinant protein is stably maintained in the cell, the final yield and purity will also depend on the extraction and purification processes (Menkhaus et al. 2004; Janson 2011; Wilken and Nikolov 2012). Optimizing each of these aspects is an effort to make molecular farming an increasingly attractive alternative for recombinant protein production.

6.3 Setting a Seed-Based Platform for Recombinant Protein Production

The production of recombinant proteins in seeds implies that the foreign candidate gene is stably integrated into the genome of a transgenic plant and that this gene is expressed in the seed and inherited by the progeny. Hence, a key factor in setting a seed-based platform for recombinant protein production is the ability to transform the candidate crop. For many species, the transformation protocol may be lengthy and cumbersome, even intractable, in some cases. However, the years-long plant transformation experience has resulted in relatively efficient protocols for species with high-protein content seeds, including cereals and grain legumes, such as rice, maize, barley, pea, and soybean.

To assure that the candidate gene is expressed in the seed, the usual approach includes a seed-specific promoter in the expression cassette used for transformation (Fig. 6.1). Several seed-specific promoters have been identified and tested, both for monocots and dicots (Furtado et al. 2009; Joshi et al. 2015; Xu et al. 2016; Mirzaee et al. 2022). Promoter sequences of storage proteins, for example, follow a tissue-specific pattern and are strictly regulated in time during embryogenesis (Chen et al. 1989). Endosperm-specific promoters that have been used for transgene expression in cereals include the rice globulin, prolamin, glutelin GluB-4, Gt13a maize zein, and barley D hordein (Kawakatsu and Takaiwa 2010). In legume seeds, some endosperm-specific promoters tested are the soybean β-conglycinin α′ subunit, the pea legumin, and the arcelin and phaseolin, from common bean (Chen et al. 1989; De Jaeger et al. 2002; Mirzaee et al. 2022).

Fig. 6.1
A schematic in 3 parts. a. A rectangular cell diagram. Labels include P S V, vacuole and apoplast in Golgi. b. A process flow of 3 and 6 reactions, respectively, in the E R and the Golgi. 3. 3 illustrations. They have a Y-shaped structure, formed with chains of diamonds, circles, and squares.

Schematic overview of N-glycosylation in plants. (a) Following transcription in the nucleus, the protein is translated and enters the ER, where the glycosylation process begins. Next, the protein moves to the Golgi where it is further processed. The glycosylated protein may be secreted to the PSV, PB, or apoplast. (b) Glycosylation process in the ER and Golgi corresponding enzymes. (c) Differences in glycosylation pattern of plant and mammal proteins, and the approach for glycol-engineering protein in a transgenic plant

Full size image

Other features that have also been shown to influence the expression levels in transgenic plants include the presence of introns, enhancer sequences, codon optimization, terminator sequence, and other 3′ flanking regions, such as scaffold matrix attachment regions (MARs) (Habibi et al. 2017; Webster et al. 2017; Diamos and Mason 2018).

Besides the promoter region, a cassette for protein expression in seeds includes a sequence for a signal peptide that will direct the protein to the endoplasmic reticulum (ER) and the secretory pathway (Fig. 6.1) (Arcalis et al. 2014). A signal peptide is a fragment of 20–30 amino acids present at the N- or C-terminal end of the target protein, which are recognized by specific complexes of RNA and proteins, called “signal recognition particles” (SRPs), that mediates the internalization of the target protein into the membranous organelles (Jolliffe et al. 2005; Robinson et al. 2005; Ashnest and Gendall 2018). Next, the signal peptide is cleaved, leaving the target protein at its intracellular destination (Bohnsack and Schleiff 2010). In the absence of signal peptides, the protein synthesized in free ribosomes accumulates in the cytoplasm, generally an unstable environment with high proteolytic activity (Obembe et al. 2011).

In seeds, storage proteins are directed to protein storage vacuoles (PSVs), which are derived from prevacuoles detached from the Golgi complex (Figs. 6.1 and 6.2). In Poaceae, which includes the cereals, the ER also forms protein bodies (PBs) that store mainly prolamin aggregates (Khan et al. 2012; Arcalis et al. 2014; Pedrazzini et al. 2016). PSVs are highly specialized vacuoles derived from the rough endoplasmic reticulum (Khan et al. 2012; Arcalis et al. 2014) and possibly from embryonic vacuoles (EVs), formed during seed development (Feeney et al. 2018). Their lumen has a pH close to neutral and practically no aminopeptidases, features that characterize them as a subcellular environment where protein degradation is minimal and an excellent target for addressing heterologous polypeptides (Takaiwa et al. 2007). In addition to providing a low oxidative environment, PSVs also harbor a high concentration of protease inhibitors, which increases their potential as a target for protein targeting aiming at increasing protein stability (Jolliffe et al. 2005; Oakes et al. 2009).

Fig. 6.2
2 photomicrographs. A. It has small clusters of tiny globules labeled O B and P S V on the left and right, respectively. B. O B is indicated within larger cellular bodies that are roughly ellipsoidal and P S V to its right within a plasma-like region.

Subcellular localization of recombinant FIX by immunocytochemistry in ultrathin sections of soybean cotyledons. (a) Immunogold-labeled FIX (white arrows) localizes to protein storage vacuoles (PSV) in transgenic soybean seeds. (b) Non-transgenic cotyledon. OB oil bodies. (Images: N.B. Cunha)

Full size image

In cereal grains and legume seeds, the families of storage proteins represent the major part of the total seed protein content. For example, the family of glutelins in rice comprises 80% of the seed protein content; the glutelins, in wheat, 40%; and the zein, in maize, 60% (Kawakatsu and Takaiwa 2010). In soybeans, the globulins, glycinin, and β-conglycinin account for up to 80% of the total protein in the seed (Hudson et al. 2014). The expression of these storage proteins along the development of the seed is highly regulated and might constrain the accumulation of a heterologous protein.

A rebalancing of the seed storage protein of soybean was tested by partially suppressing the α/α′ subunit of β-conglycinin, resulting in the increased accumulation of glycinin, along with heterologous green fluorescent protein (GFP) regulated by glycinin promoter and terminator (Schmidt and Herman 2008). In a similar approach, Kim et al. (2014) found that the increase in the recombinant methionine-rich 11 kDa δ-zein in soybean was dependent on a sulfur-rich medium supplementation. In maize, Hood et al. (2012) crossed transgenic lines expressing a recombinant cellulase with high protein elite genotypes and selected for lines with higher content of cellulase. These approaches demonstrate that seed-based platforms have many possibilities to further optimize the production, both in yield and quality of recombinant proteins.

6.4 Posttranslational Modification in Plants and Its Relevance in Molecular Farming

As eukaryotic organisms, plants have the metabolic pathways for posttranslational modifications of proteins—glycosylation, acetylation, and phosphorylation, among others. These modifications occur in the ER and Golgi and are relevant for molecular farming (Fig. 6.1). It is estimated that 50–70% of human proteins are glycosylated (Walsh and Jefferis 2006). Likewise, about 50% of the biopharmaceuticals currently produced are represented by glycosylated proteins (Mizukami et al. 2018; Montero-Morales and Steinkellner 2018). Also, glycosylation and other posttranslational modifications are involved in protein stability and turnover, potentially impacting the accumulation and, hence, the recombinant protein’s final yield (Thomas and Walmsley 2015; Varki 2017; Gupta and Shukla 2018).

The glycosylation pattern of proteins in plants, both N- and O-glycosylation, differs from that observed in insects, yeast, and animal cells. Glycosylated proteins from plants contain xylose, and arabinose residues (in O-glycans), which are not found in mammalian proteins. Moreover, plant N-glycans present α(1,3)-fucose, which is also present in mammalian cells but in β(1,6) linkage form. Plant glycans lack galactose and terminal sialic acids, which are present in mammalian glycoproteins (Fig. 6.1) (Strasser et al. 2021; Bohlender et al. 2022).

In many cases, these differences in the glycosylation pattern may not interfere with the biological activity or the functionality of the recombinant protein, particularly in non-pharma proteins. However, glycostructures can influence the pharmacokinetics, stability, and immunogenicity for biopharmaceuticals (Gupta and Shukla 2018; Bohlender et al. 2022). For example, the presence of nonhuman glycans, particularly the fucose and xylose residues, may cause allergies and immunogenic responses in humans (Montero-Morales and Steinkellner 2018), and non-sialylated glycoproteins are rapidly cleared from serum (Walsh and Jefferis 2006; Bohlender et al. 2022).

To circumvent these problems, Nicotiana benthamiana plants were glycol-engineered to present a more “humanlike” glycosylation pattern (Fig. 6.1). Transgenic N. benthamiana plants expressing β(1-4)-galactosyltransferase were successfully tested. The α(1,3)-fucosyltransferase and β(1,2)-xylosyltransferase genes were knocked out in the moss Physcomitrella, in transgenic Arabidopsis thaliana, and N. benthamiana, which produced N-glycans lacking xylose and fucose residues (Strasser et al. 2004). Next, N. benthamiana plants were modified to express the α-1,6-fucosyltransferase and the pathways involved in the biosynthesis, activation, transport, and transfer of Neu5Ac to terminal galactose of heterologous proteins (Fig. 6.1) (Castilho et al. 2010, 2011; Kallolimath et al. 2016).

Expression platforms that are able to produce proteins with extensively modified glycosylation patterns, as for N. benthamiana, for example, have not yet been developed for cereals or legume seeds. However, in one attempt to engineer the glycosylation pattern, Wang et al. (2017) expressed the human α-1,6-fucosyltransferase (FUT8) in rice, controlled by an endosperm-specific promoter. After crossing with a plant expressing recombinant human 1-antitrypsin, they confirmed the presence of α(1,6)-fucose and a reduction of β(1,3)-fucose both in the recombinant protein and in globulins. Vamvaka et al. (2016b) demonstrated that recombinant heavy chain of the HIV-neutralizing monoclonal antibody 2G12 expressed in rice seeds was predominantly non-glycosylated, potentially less immunogenic, and more potent in HIV-neutralization assays than the 2G12 antibodies produced in Nicotiana tabacum. Similar results were reported by Zhang et al. (2012), which found that approximately 70% of their rice-derived recombinant human a1-anti-α-trypsin was aglycosylated. Indeed, plants appear to tolerate alterations in the glycosylation pathways, not showing phenotypical alterations and being well suited for the production of glyco-engineered recombinant proteins. That flexibility was further demonstrated by the transient co-expression in N. benthamiana of specific glycosyltransferases allowing the production of the glycoproteins omega-1 and kappa-5 of Schistosoma mansoni containing the helminth-like glycosylation pattern (Wilbers et al. 2017; see also chapter “Tobacco Plants as a Versatile Host for the Expression of Glycoproteins”).

New available technologies, such as targeted genome editing, could be used to efficiently knock down β1,2-xylose and core α1,3-fucose residues, for example. In any case, developing a glycol-engineered seed-based platform is a promising yet challenging process (Buyel et al. 2021; see also Chaps. 3 and 4).

6.5 Comparison of Current Seed-Based Platforms

6.5.1 Cereals

Cereals are among the most cultivated and consumed crops worldwide, and most of them are considered a staple food for many countries. Currently, the main products obtained through the production of recombinant proteins in commercially available seed-based platforms are amylase, peroxidase, and cellobiohydrolase I (maize); growth factors and cytokines (barley); and a variety of enzymes (lactoferrin, albumin, transferrin, and lysozyme) and growth factors in rice (Fischer and Buyel 2020; Mirzaee et al. 2022).

Maize was the first seed-based platform used to produce an industrial reagent, avidin, by ProdiGene Inc. (USA). The product was indistinguishable from its counterpart from hen egg white and presented a high yield (2.3% of extractable protein from seed, on average) (Hood et al. 1997; Fischer and Buyel 2020; Moon et al. 2020).

Because maize is a cross-pollinating crop, working with transgenic maize is challenging and requires strict biosafety protocols to be followed. Despite that, maize has several advantages as a seed-based platform for the production of recombinant proteins, such as larger grain size, high yield, and lower production cost compared to other cereals. Moreover, maize has a higher endosperm proportion; a set of specific promoters for the seeds, which can be used to drive the expression of the transgene alone or in combinations; and easy genetic transformation, with established protocols (Watson and Ramstad 1987; Hood et al. 1997; Witcher et al. 1998). As a result, maize seeds have been used as a platform for the production of industrial reagents, such as enzymes and cosmetics, and also pharmaceuticals, such as antibodies to treat human and animal diseases (Rademacher et al. 2008; Egelkrout et al. 2020) and vaccines (Nahampun et al. 2015).

On the contrary, rice and barley are self-pollinating crops, making the risk of undesirable gene flow very low. The rice transformation system is effective in most varieties cultivated worldwide. However, some genotypes may be more suitable for producing a high level of recombinant proteins. The amount of proteins in rice seeds is 7–15% of the total seed weight (Takaiwa et al. 2007). The productivity of recombinant protein in relation to the total weight of the rice seed has reached high levels, from 1% of the total seed weight to 20% of dry seed weight (Vamvaka et al. 2016a). However, in some cases, the high content of recombinant proteins in rice seeds results in grains with an impaired phenotype, indicating that further research is required (Kusaba et al. 2003; Tada et al. 2003; Wakasa and Takaiwa 2013). Besides that, the ease of processing and scale-up production by well-established cultivation systems favor the choice of rice as a platform for producing recombinant proteins.

The rice seed-based platform has been explored for the production of biopharmaceuticals, such as vaccines, growth factors, and antiviral proteins (Takagi et al. 2005; Xie et al. 2008; Vamvaka et al. 2016a). As one of the main staple foods globally, it would be useful to exploit the rice seed-based platform to deliver pharmaceuticals as food. Some studies have shown that the recombinant protein produced in rice seeds remains active in the seeds, even after processing them as a fine powder or crude extracts, for oral administration to mice or macaques (Takagi et al. 2005; Xie et al. 2008; Nochi et al. 2009). In one study, the oral administration of rice seeds expressing immunogens, processed as a fine powder, effectively inhibited allergy-associated immune responses in mice (Takagi et al. 2005). Another study developed a rice-based vaccine that expresses the B subunit of cholera toxin (CT), initially tested in mice with positive results. Subsequently, the vaccine was orally administered to macaques and induced CT-neutralizing IgG antibodies, confirming its effectiveness against cholera in nonhuman primates. This vaccine (MucoRice-CTB) has recently passed phase 1 human clinical trials (Nochi et al. 2009; Yuki et al. 2021).

Another interesting example was the use of crude extracts of rice seeds expressing a microbicide against human immunodeficiency virus (HIV) in cytotoxicity and antiviral assays with human cells (Vamvaka et al. 2016a). Results showed that the crude extracts had stronger binding activity to HIV than the wild-type rice seeds, similar to the purified protein, and were not toxic to human cell lines. Also, the crude extracts expressing the microbicide presented the same oligosaccharide-dependent binding properties as the same recombinant protein expressed in Escherichia coli. Altogether, these results show that it is possible to administer rice seeds expressing biopharmaceuticals without processing or with only minimal processing (see also chapter “The Use of Rice Seed as Bioreactor”).

Barley is another cereal species that has been used as a seed-based platform for the production of recombinant proteins. The European regulatory agency (EFSA) has declared self-pollinating cereals such as wheat and barley as GRAS (generally recognized as safe) (Mirzaee et al. 2022; see also Chap. 14). In addition to being self-pollinating, barley has other characteristics of interest, such as its ability to regenerate, especially the cultivar Golden Promise, in which transformation amenability (TFA) alleles have been identified as responsible for its Agrobacterium transformation efficiency (Hisano and Sato 2016; Orman-Ligeza et al. 2020).

Other cereal seeds have been studied as a platform for producing heterologous proteins, however, on a smaller scale. For example, wheat seeds have been used to express TM-1 protein as an antigen to be used as an edible vaccine for chronic respiratory disease, a common disease in chickens, resulting in a significant level of protection (Shi et al. 2023).

6.5.2 Soybean

Among the plants that are candidates for seed-based stable accumulation of recombinant proteins, soybean plants, along with pea plants, present seeds with a high protein content (corresponding to approximately 40% of their weight) and, compared to other sources, represent a lower protein cost, due to their high seed yield (Mikschofsky and Broer 2012; Hudson et al. 2014; Vollmann 2016). Besides, recombinant proteins stored in soybean seeds were shown to remain stable and functional for long periods at room temperature (Oakes et al. 2009; Lobato Gómez et al. 2021). A study showed that the production of a functional subunit vaccine for Staphylococcal enterotoxin B in soybean seeds was stable over several soybean generations, and biochemically and immunologically similar to commercial recombinant forms (Hudson et al. 2014). The expression of cyanovirin-N, a lectin with antiviral activity, was demonstrated in soybean seeds at levels up to 10% of total soluble protein. Attempts to express this protein in transient assays in leaves of N. benthamiana were unsuccessful, demonstrating the potential of soybean as an alternative to express and accumulate this recombinant protein (O’Keefe et al. 2015).

From a regulatory point of view, it has a reduced risk of pollen contamination since soybean is largely self-pollinated (Paul and Ma 2011; Paul et al. 2011). Furthermore, soybean seeds accumulate proteins in PSV, resulting in optimal conditions for long-term storage of immunogenic and fully active recombinant proteins (Fig. 6.2) (Cunha et al. 2011).

Although soybean has not been commercially used for the production of recombinant proteins yet, numerous studies show the potential of this plant species for the production of pharmaceutical proteins, such as human growth hormone, proinsulin antiviral lectins, coagulation factor IX, antigens, antibodies, as well as non-pharma proteins (Moravec et al. 2007; Yamada et al. 2008; Cunha et al. 2011, 2014; Hudson et al. 2014; O’Keefe et al. 2015). Some of these studies have tested the potential of soybean expressing recombinant proteins when administered orally to mice, as seed extracts, with promising results (Moravec et al. 2007; Hudson et al. 2014). This indicates that soybean seeds could be formulated into edible products for oral delivery of pharmaceutical proteins (Adelakun et al. 2013).

Another important characteristic of this legume is its sensitivity to the photoperiod, expressed in the temporal modulation of the vegetative phase of its phenological cycle as a function of daily time and light intensity (O’Keefe et al. 2015). The production of seeds per plant can be greatly increased under controlled conditions in a greenhouse reaching up to 1000 seeds per plant. This increase in the production scale can be exploited by the molecular farming industry and is particularly relevant for biocontainment and more controlled cultivation conditions (Kantolic and Slafer 2007; see also chapter “Legume Seed: A Useful Platform for the Production of Medical Proteins/Peptide”).

6.6 Scale-Up of Seed-Based Production Systems

The commercial success of large-scale seed molecular agriculture depends on technology, economics, and public acceptance (see also Chap. 15). Factors important for the biopharmaceutical industry include the expected reduction in costs and the indirect effects on the biopharmaceutical market (Twyman et al. 2003). Seed-based platforms can be used to produce recombinant proteins at a significantly lower cost compared to other systems such as microbial fermentation and cell cultures (Giddings 2001; Hood et al. 2002; Twyman et al. 2003). It is estimated that the production costs of recombinant proteins in maize, for example, will be threefold higher than that for the production of maize for food use (Mison and Curling 2000). Still, the savings in operational expenses can generate a considerable reduction in terms of capital investment (about 75–80%) and manufacture (50–60%) compared to microbial fermentation and cell culture production (Buyel 2019). In addition, the production of proteins that require lower purity, such as industrial enzymes and oral vaccines, is significantly reduced compared to other products that require expensive purification processes (Nikolov and Hammes 2002; see also chapters “Molecular Farming of Industrial Enzymes: Products and Applications” and “Plant Molecular Farming for Vaccine Development”).

Field cultivation is the most reasonable option for the large-scale production of transgenic seeds. Although production in greenhouses may be feasible for transient transformations in short-cycle crops, generation time and space requirements would reduce the cost advantages of seed production (Boothe et al. 2010). The cost of production ($/g product) of the same recombinant protein produced in greenhouses is estimated as fivefold higher than that produced in an open-field system (Pogue et al. 2010). Besides, in terms of equipment for field production (from planting to harvesting), there is no difference between those used in conventional commodity crops. The existing technologies already meet this demand, enabling good scalability compared to other systems limited by the size of the culture reactor, for example. Furthermore, the long-term stability of the recombinant protein in the seeds allows the harvest to be decoupled from the purification process, generating greater flexibility and better stock management (Boothe et al. 2010).

Another important aspect of scale production is a high level of quality control to ensure high protein purity. Therefore, crop management must be carefully conducted, which is also necessary for biosafety reasons. Crops must provide high-quality seeds, which must present stable expression levels, and homogeneous products (e.g., glycosylation pattern and degradation levels) over generations and cultivation places to meet the established quality standards.

Finally, approximately 60–70% of the production cost of seed-based recombinant proteins is associated with downstream processing, so it is essential to develop techniques to improve it and make it increasingly efficient, reducing its cost (Dyr and Suttnar 1997). Therefore, understanding the conditions that affect the product’s quantity and quality is essential to meet the market standards and competitiveness. An overview of the methods currently used for protein purification is presented in the following section.

6.7 Basic Approaches for the Purification of Recombinant Proteins from Plant Cells

The choice of an expression system should consider the particularities of the target protein to be purified. Expression based on the bacteria Escherichia coli, for example, could be advantageous in terms of production time and costs (Rosano and Ceccarelli 2014; Lozano Terol et al. 2021). However, the lack of posttranslational modifications and contamination with bacterial endotoxins represent limitations for using this popular expression system (Sahdev et al. 2008). On the other hand, expression based on mammalian, yeast, and insect cells allows posttranslational modifications. However, these expression systems based on large bioreactors present a high upstream cost (Schillberg and Finnern 2021).

The production of biomass from plants requires soil and fertilizers, whereas bioreactor cultures demand complex media. That makes plant-based biomass production more sustainable than microbe and cell culture expression systems (Buyel et al. 2021), requiring fewer investments and demanding less specialized staff.

Downstream processes, independently of the expression system utilized, are critical in terms of cost and quality of the final product. It is estimated that downstream processing may represent 50–80% of the total costs, depending on the yield, recovery efficiency, and purity grade (Schillberg et al. 2019). Besides the costs, if purification of the recombinant protein is needed, one may be facing a technical challenge. Indeed, purification of a specific protein, either recombinant or endogenous, is generally not trivial, as each protein will have its specific physical characteristics. Plus, in the case of a recombinant protein, there will likely be no available protocol for purification from plant tissues. However, information on the purification method of the target protein from other expression systems may be helpful. Some factors do favor the downstream processing and purification of proteins from seeds since they are presented as a comparatively homogeneous starting material, have reduced water content, lack chlorophyll pigments and alkaloids, and confer high stability of the target protein stored in the seed storage vacuoles.

The source of the material to be processed needs to be handled carefully—plants need to be well cultivated so that the collected tissues are healthy. Next, some basic protein analyses are needed to confirm the presence of the heterologous protein and evaluate the expression level. These can be done from total extracts (i.e., tissues ground in extraction buffer), using well-established protocols for total protein quantification and detection, such as Western blot or ELISA. Finally, the amount of the recombinant protein can be expressed as a percentage of the total protein in the extract, or of the seed dry weight, as weight per mass of fresh or dry tissue (e.g., mg/g fresh leaves, μg/g dry seed weight).

The seed composition varies among species and requires specific extraction conditions regarding the contents of proteins, starch, oil, etc. For soybean seeds, for example, due to large amounts of oil (about 18–22% of seed weight), the material often requires homogenizing the seed meal with solvents, such as hexane. Furthermore, to ensure that the protein of interest maintains its stability and that its extraction process is efficient, it is essential that the extraction solutions present the appropriate pH and ionic strength (Robić et al. 2010). Hence, the extraction conditions consider pH, saline buffers, chaotropic agents/detergents, protease inhibitors, etc. (Janson 2011). Once extracted from the seeds, the target protein may be further purified by chromatography.

6.7.1 Non-affinity Absorption and Affinity Techniques to Purify Proteins

Proteins are made of amino acids (AAs) as basic building blocks assembled in a chain via amide bonds (peptide linkages). The 20 L-AAs found in proteins have four different ligand groups (an amino group, carboxyl group, hydrogen, and R-group). These groups and their interactions within the protein give unique biochemical characteristics and functions and influence their physiological and biological activities. The R-group of the 20 AA commonly found in proteins varies widely, especially their polarity at a biological pH (around pH 7.0), from polar and hydrophilic (water soluble) to nonpolar and hydrophobic (water insoluble) (Wu 2009). These physical properties can be exploited to aid the protein isolation and purification process.

In aqueous solutions, functional AA groups from folded proteins contribute significantly to the protein surface charge in a pH-dependent way. According to external pH, the overall charge may vary from positive (at low pH) to negative (at high pH). Therefore, separating a complex sample of proteins based on their surface charge helps purify a protein-rich sample with similar physicochemical characteristics in a reduced volume compared to the initial volume before separation (Bonner 2018). The purification may be optimized by using various techniques to exploit differences in the target protein’s charge and biospecificity. Several successful cases in the literature explore these methodologies for efficient protein separation from seeds. For example, Zhang et al. (2012) purified the human alpha-antitrypsin protein expressed in transgenic rice seed using different anion exchange columns. In another work, cellobiohydrolase I was expressed in transgenic corn seeds and purified with ammonium sulfate precipitation (a fractionation technique also used to isolate proteins from a complex sample) together with both cationic and anionic exchange chromatography, yielding 63% of pure protein (Hood et al. 2014).

It is a common strategy to engineer the recombinant protein with affinity tags to facilitate affinity-based purification procedures. The most common examples found in the literature are proteins containing polyhistidine tails (6×His or 10×His) (Valdez-Ortiz et al. 2005). The sequence of six (or ten) consecutive histidine residues is currently one of the most used strategies worldwide to purify recombinant proteins for biochemical and structural studies.

Unlike ion exchange chromatography, affinity chromatography does not explore the physicochemical characteristics of proteins. The affinity chromatography technique is based precisely on the unique biospecificity of the protein engineered for isolation from a complex sample of proteins. Biospecificity involves the interaction between two immiscible phases, that is, the reversible interaction between a ligand (which can be a small molecule, enzymes, among others) immobilized on a resin (known as the stationary phase) and the recombinant target protein inserted into a solute (mobile phase) (Janson 2011). Menkhaus et al. (2004) compared various techniques, such as precipitation with polyethyleneimine cationic polyelectrolyte (PEI), anion/cation exchange, diafiltration (molecular exclusion), and immobilized metal ion affinity chromatography (IMAC) for the purification of histidine-tagged β-glucuronidase from transgenic pea seeds. They observed an increased recovery of pure protein and higher enzyme activity when utilizing affinity chromatography (Menkhaus et al. 2004).

In some cases, proteins from the host can be present as contaminants in samples purified by affinity purification. To remove these contaminants, denaturation and refolding steps of proteins are sometimes necessary. Fujiwara et al. (2010) observed the need for two steps of affinity purification combined with a denaturation step with 6 M guanidine to remove protein contaminants from rice seed in the purification of IL-4 and -6 cytokines (Fujiwara et al. 2016). Similar results were previously obtained by Fujiwara et al. (2010) during the purification of human interleukin-10 (IL-10), also expressed in rice seeds. These results demonstrate the efficiency of combining different chromatographic and purification techniques to remove contaminants and consequently increase the yield and purity of the recombinant proteins obtained (Fujiwara et al. 2010).

6.7.2 Chromatography-Free Protein Purification

Another approach for recombinant protein purification is based on fusion proteins (FPs). The idea is to exploit the unique properties of the fusion partner, allowing an increase in stability and facilitating the purification process (Viana et al. 2013; Ki and Pack 2020). Examples of these fusion partners include the synthetic peptide elastin-like polypeptides (ELPs) (Ciofani et al. 2014), γ-zein (Torrent et al. 2009), and hydrophobin (Lahtinen et al. 2008).

Derived from its soluble precursor, tropoelastin, elastin has a hydrophobic motif composed of a repeated sequence of hydrophobic amino acids alanine (Ala) and valine (Val), in addition to the presence of other residues in significant amounts, such as glycine (Gly) and proline (Pro) (Partridge et al. 1955). At temperatures below 25 °C, the protein remains soluble; however, when the temperature is raised to 37 °C, a precipitated protein known as coacervate is observed. This process is fully reversible upon returning the protein to room temperature (Urry et al. 1969). Based on these properties, synthetic peptides were developed, known as the elastin-like polypeptides (ELPs), composed of the canonical sequence of the pentapeptide (Val-Pro-Gly-Xaa-Gly)n.

When investigating the strategy of using FP in conjunction with scalable purification processes, Phan et al. (2014) observed an enhancement in expression levels (about tenfold higher) of the avian influenza virus (H5N1) hemagglutinin subtype 5 (H5) protein fused with ELP at the C-terminus in transgenic tobacco seeds. The enhanced accumulation of HA, which is the major antigen of the influenza virus, by ELP-FP resulted in high concentrations of the ELPylated target protein in the aqueous crude extract. Further purification was facilitated by using optimized processes involving inverse transition cycling (ITC). In comparison with another strategy, also using FP (fungal hydrophobin I—HFBI), the same authors observed that only ELPylation was able to increase HA expression in seeds, resulting in high-purity protein (Phan et al. 2014), demonstrating the efficiency of using ELPylated proteins in the processes of expression and purification simply and inexpensively (Khan et al. 2012).

In contrast, in another study, Yang et al. (2021) showed that the γ-zein system, a member of the major prolamin storage family in maize, was more efficient for the accumulation of GFP in immature soybean seeds than the ELP system. In addition, the use of the γ-zein system provided a 3.9-fold increase in the accumulation of fused GFP in comparison with unfused GFP protein, demonstrating that the γ-zein system is a promising FP for future enhancement in the expression and purification of recombinant proteins in plants.

Although fusion proteins are generally non-immunogenic and biologically compatible, they may interfere with the activity of native proteins (Shamji et al. 2007; Viana et al. 2013). That implies the removal of the FP, which is done after purification by specific proteolytic enzymes that recognize cleavage sites placed at the junction of the target protein and the fusion partner. This additional step for recombinant protein purification may result in unspecific degradations of the target protein, reducing the final yield and impacting downstream costs (Tian and Sun 2011).

A promising alternative to proteolytic enzymes for the removal of FP is the use of inteins (Viana et al. 2013; Ki and Pack 2020). These proteins can catalyze their self-cleavage and, through amino acid substitution, can be regulated to cleave at either the N- or the C-terminus in response to reducing agents or changes in solution pH (Xu and Perler 1996; Perler 1998; Gillies et al. 2009). Therefore, the self-cleavage property of inteins can be applied to replace the traditional proteolytic cleavage. By fusing the intein (Eitag) with the ELP-FP, for example, the recombinant proteins can be purified by applying both ITC followed by autocatalysis by changing the pH of the solution. Tian and Sun (2011) explored the use of the ELP-intein system to increase the accumulation of the recombinant lectin fused with ELP in transgenic rice and tested the capacity of autocatalysis of intein after ELP extraction from seeds. Furthermore, the presence of Eitag + ELP did not alter the N-glycosylation patterns of the recombinant protein, demonstrating the potential application of the ELP-intein fusion system for the expression and purification of recombinant proteins in plants, especially in seeds.

6.8 A Brief Overview of Biosafety and Risk Assessment of Seed-Based Expression Systems

The technology for the production of recombinant proteins using these platforms is developing fast and focuses on two product lines: pharmaceuticals and non-pharmaceuticals. Despite intense research on developing biopharmaceutical production, most plant-based products currently available on the market belong to the non-pharma field, mainly because regulation processes are faster and less expensive for non-pharma products. These include products for the diagnosis, industrial reagents, and cosmetics, among others.

The biological safety assessment to produce recombinant proteins in seeds is an important issue. Biosafety involves several relevant issues, such as choice of plant platform, transgenic plants, field production, handling, harvesting, and transport. Therefore, one must consider plant biology from the perspective of productivity and how it impacts the environment, food security, and human health. Therefore, the best material based on technical aspects (e.g., seeds with better processing capacity, high protein content, stability, etc.) may not be the best choice if considering the regulatory issues of biosafety (Sparrow and Twyman 2009).

The use of seeds as a “bioreactor” has its risks of propagation in nature, contaminating non-transgenic plants, and being potentially hazardous to people and animals if used unintentionally as food and for insects and soil microorganisms (Lee et al. 2003). Most of the steps required to avoid mixing these seed-derived biopharmaceuticals in the food chain are relatively simple, such as meticulous planning and execution. The plants must be cultivated in an isolated area to avoid genetic and mechanical mixing of seeds containing biopharmaceuticals with those intended for food. Likewise, small-scale and large-scale field trials must be isolated from conventional practices with crops to avoid cross-pollination. Although these risks apply to all transgenic crops grown in the field, plants cultivated for molecular farming deserve special attention due to the nature of the recombinant proteins (i.e., biopharmaceuticals), consequently, with unpredictable, potentially hazardous outcomes (Basaran and Rodríguez-Cerezo 2008). Achieving an effective isolation level to avoid wind and insect pollination is challenging. If the plants are cultivated in confinement, the risks and threats to the environment would be reduced and would imply less strict regulatory issues.

Appropriate mitigation measures for recombinant protein-producing seeds will depend on several factors, including properties of the molecule, biology of the crop, and characteristics of the environment where it is being produced. Approaches of containment methods include identity preservation—using varieties that are visually distinct from traditional varieties (such as purple maize or black soybeans)—and use of marker genes, such as a fluorescent protein, barrier crops, and temporal barriers aiming to minimize undesirable crosses (Sparrow and Twyman 2009).

In general, regulatory guidelines for the production of recombinant proteins are similar across countries, but some specificities may apply. For example, in the United States, the production of biopharmaceuticals on transgenic plants is regulated by two agencies. The US Department of Agriculture (USDA), Animal and Plant Health Inspection Service (APHIS), focuses on the containment of these seed producers of protein. In contrast, the Food and Drug Administration (FDA) focuses on the manufacture of the drug or vaccine. APHIS reviews production license applications, assessing probable environmental impacts of these releases (Basaran and Rodríguez-Cerezo 2008). In the European Union, authorizations involve all member states and the European Commission (Breyer et al. 2009). In Brazil, GMO studies are only allowed in research institutions after authorization by the National Biosafety Technical Commission (CTNBio) (Mendonça-Hagler et al. 2008).

6.9 Conclusions and Perspectives

The demand for biopharmaceuticals continues to grow as new products are approved. To couple with this demand, a general trend for the production of recombinant proteins has been to increase yield and optimize upstream and downstream processes to reduce costs. In addition, bioequivalence is also very relevant for the biopharmaceutical industry, which pursues products that are as similar as possible to the original product.

As discussed in this chapter, several aspects of plant molecular farming are aligned with these demands. Producing recombinant protein in plants demands substantially lower costs for upstream processes as compared to the expression systems based on microorganisms and cell cultures. Concerning posttranslational modifications, efforts for glyco-engineering of plants have also achieved amazing progress. Although still restricted to model plants and transient assays using N. benthamiana, efforts toward a seed-based glyco-engineered platform are in progress. It is reasonable to consider its viability and availability before long.

As experience accumulates, methods for protein purification will become more efficient. These may compensate for limitations on the expression level, which can be very low, depending on the target protein. For seed-based expression, the final yield could be increased by applying more efficient extraction and purification, involving, for example, fusion proteins and fusion tags. Seed-based platforms, for their advantages in terms of protein content and long-term stability of the recombinant protein stored therein, offer great potential for new ideas to be implemented (see also Chap. 5). The progress witnessed in the last decades confirms the potential of molecular farming as an alternative system for expressing recombinant proteins and represents a field of opportunities.