Nicotiana hairy roots for recombinant protein expression, where to start? A systematic review

Hairy roots are a plant-tissue culture raised by Rhizobium rhizogenes infection (formerly known as Agrobacterium rhizogenes). Nowadays, these roots have been gaining more space in biotechnology due to their benefits for the recombinant expression of valuables proteins; it includes simplified downstream processing, protein rhizosecretion, and scalability in bioreactors. However, due to methodological inconsistency among reports, the tissue platform is still a promising technology. In the current paper, we propose the first step to overcome this issue through a systematic review of studies that employ Nicotiana hairy roots for recombinant expression. We conducted a qualitative synthesis of 36 out of 387 publications initially selected. Following the PRISMA procedure, all papers were assessed for exclusion and inclusion criteria. Multiple points of root culture were explored, including transformation methods, root growth curve, external additives, and scale-up with bioreactors to determine which approaches performed best and what is still required to achieve a robust protocol. The information presented here may help researchers who want to work with hairy roots in their laboratories trace a successful path to appraisal the literature status.


Introduction
Plant molecular farming (PMF) emerged in the 1990s as a fast, cheap, and easy-to-handle biotech approach to produce recombinant proteins (RP) [1,2]. However, PMF has problems such as the non-human pattern of glycosylation, low-rate recombinant production, and complex protein purification process, which open the way to the second wave of plant-based hosts that overcome these problems (reviewed in [3]). The hydroponic hairy roots culture is one of these newly employed platforms of recombinant expression, usually explored to simplify downstream processing (DSP) of foreign proteins from plants.
The hairy root occurs when the soil-borne prokaryote Rhizobium rhizogenes (previously recognized as Agrobacterium rhizogenes) triggers the highly-branched root upon plant infection [4]. The bacteria cause this phenotype through the transfer of a DNA fragment carrying root locus (rolA, rolB, rolC, and rolD) and opine genes from its rootinducing plasmid to the nucleus of plant cells [5,6]. Over the R. rhizogenes establishment, the rol genes work synergistically to regulate neoplastic roots emergence by controlling the plant's phytohormones (e.g., IAA, auxins, and cytokinins). Although their initial biotechnological focus was on secondary metabolites production, later the hairy root culture became a tool for recombinant protein expression.
Heterologous proteins have led to huge advances in biomedical and industrial biotechnology. Since 2002, the U.S. Food and Drug Administration (FDA) has approved over 300 biopharmaceutical products, and this number increases with the expansion of the modern medical industry. The RP produced in hairy root so far includes monoclonal antibodies 1 3 [7,8], vaccines [9][10][11], hormones [12], enzymes [13,14], and proteins difficult to produced by bacterial hosts (e.g., complex proteins and antimicrobial peptides; [15,16]).
One of the major advantages of HR is the possibility to release folded and biologically active RP into the culture medium, a process known as rhizosecretion [17]. The protein recovering from culture medium ensures a faster DSP, reducing the bench-work and counteracts one of the major planthost drawbacks, which is the particle contamination generated during protein extraction [18]. That issue contributes to about 80% of the total PMF expenses [19][20][21]. The root system has the benefits of speed, high scalability, and low risk of contamination with human pathogens. Furthermore, several reports show the root-based recombinant expression as more productive than the regular leaf system [22][23][24][25][26].
Indeed, hairy root culture has been emerging as a powerful platform for bioproduction. However, a remaining issue is the lack of best practices to use. Where to start, and what are the best steps to follow? Researchers employ different strategies to obtain RP from roots, such as a variety of transformation, signal peptides, culture establishment, and protein recovery methods, with each experimental design leading to distinct outcomes, without following a clear path. Some work sought to optimize root culture implementation [11,27,28]. Although, no research has investigated the whole hairy root bench-line, highlighting the need for a review assessment.
Herein, we aim to provide a thorough systematic review of the use of Nicotiana sp. hairy root for bioproduction, focusing on the reported methodologies to summarize the common practices in its culture implementation. Furthermore, we also highlight the strengths and bottlenecks of hairy root technology.

Material and methods
To perform this systematic review, we defined the eligibility criteria (Table 1) and the preliminary keywords following the PICO method [29]. Using the R package litsearch, we created the final search code from these words submitted to the Scopus database [30,31]. Scopus was used in these refinement steps due to its higher response rate than the other two databases chosen to conduct the collecting of papers (Web of Science and PubMed).
We collected data following the steps proposed by the Preferred Reporting Items for Systematic Reviews and Meta-Analyzes (PRISMA) methodology [32,33] and with the support of the covidence web tool (www. covid ence. org/). Database searches included only original peer-reviewed scientific articles, published until January 2022. As mentioned in the fourth exclusion criterion (Table 1), this work did not cover reporter proteins because it often overperforms yields related to therapeutic/industrial ones. For example, a study expressing red fluorescent protein achieved the best RP level to date in Nicotiana hairy root [8]. We removed this source of bias to make this study as straightforward as possible.
We also used our screening results with an inter-rater reliability κ = 0.798 ± 0.08 (p < 0.05) to measure the selection bias, indicating moderate to good inter-author agreement according to Cohen's Kappa statistical index [34]. In addition, we found 98.35% agreement between the authors MOS and MMA during the paper's selection. All incompatible decisions were re-evaluated among the authors. These values also highlight the consistency between our selection's previously set inclusion and exclusion criteria. After reading all the papers included in the qualitative synthesis, we retrieved the following information from the papers: (i) the RP and its function; (ii) its biomass/secreted productivity; (iii) culture day that the hairy root had the highest productivity; (iv) percentage of secreted RP; (v) transformation method; (vi) Rhizobium strain used; (vii) time it took to emerge the firsts roots; (viii) culture maintain method; (ix) culture conditions; (x) culture medium additives; (xi) plant species used; (xii) plant age; and (xiii) the configuration of the expression construct. We contacted the authors by email if some information was missing in their papers. The R package juicR [35] was used for data collection and ggplot2 for graphs design [36].

Results and discussion
Our systematic search resulted in 387 hairy root studies found (Table 1). Based on the eligibility criteria, 32 papers have advanced to the next step, in which we performed a full assessment of the articles, resulting in 48 removals. Finally, we added two articles by cross-referencing (i.e., finding studies by reading the initial phase papers' references) and ended up with 36 reports to perform the systematic review, comprising 30 distinct recombinant proteins (Table 2). Among them, 20% are antibodies, 6.6% are vaccines, and 26.6% are human proteins that are the most frequent in the hairy roots expression system, followed by antimicrobial peptides (13.3%) and industrial proteins (6.6%).
Only two Nicotiana species were used for hairy root implementation in articles, N. tabacum (mainly cv. Xanthi) and N. benthamiana ( Supplementary Fig. S1). Even though species diversity is limited, the recombinant productions reported in literature differ significantly from each other. Considering the amount of RP obtained among the papers in total soluble protein (TSP), dry weight (DW), and fresh weight (FW) of the tissue, the median values were 1.4% TSP, 800 μg/g DW, and 13.7 μg/g FW. In addition, the minimum and maximum values were ~ 0-4.4% TSP, 70-1690 μg/g DW, and 18-100 μg/g FW (the number of papers contributing to each measurement was 9, 8, and 16 for TSP, DW, and FW, respectively; Table 2). Regarding the rhizosecreted yield, the average was 3.17 μg/mL, with the highest and lowest values being 16.3 μg/mL and 0.022 μg/mL, respectively. This subject will be discussed in Sect. "Rhizosecretion performance".
During the establishment of hairy roots in the bench line, many features may affect the recombinant performance, such as gene transformation and long-term root tissue maintenance. Considering that, this paper is divided into short and long-term methodological considerations for hairy roots deployment.

Transformation strategy
Genetic transformation occurs with the transfer of exogenous genes into the host DNA. There are several methods for this in plants, such as viral vectors, protoplasts (microinjection or electroporation), biolistics, and genetic transfer by Rhizobium bacterial strains (reviewed in [59,60]). Regardless of the gene insertion protocol, the growth of hairy roots requires that the rol genes be transferred into the plant at some point. Thus, there are three related ways for establishing the recombinant roots (Figs. 1, 2): the indirect way, where a wild-type R. rhizogenes transfers the rol genes to a transgenic plant; the direct form, in which an R. rhizogenes containing the plasmid vector mediates the delivery of the T-DNA (including the rol genes) along with transgene sequence; and finally, a form independent of R. rhizogenes, in which the rol genes are carried over by one of the previously mentioned transformation methods. The choice of one of these strategies seems to affect all subsequent steps in the establishment of hairy roots, including plasmid components, biomass accumulation, and RP extraction, being crucial the planning of bench work starts at this point.
The indirect form provides better RP yields and root growth in all studies where transformation methods were compared [28,61]. However, this is the most time-consuming approach since it depends on an existing transgenic plant. In previous studies, indirect production was found to result in 1.7-fold RP secretion and 1.84-fold RP biomass over direct form [47]. On the other hand, the direct way saves time and money but may suffer from clustering expression cells due to chimeric transformation. In any case, the choice of method must meet the researcher's needs, as each one has its advantages and limitations. In most papers, the indirect protocol is preferred, in part because it is often convenient for the researcher to have access to established transgenic lines [7,40,54].
As part of the transformation, the next step is to design an expression cassette. The plasmid preparation focuses on defining the RP sequence and its components, such as the signal peptide, amino acid tag, and promoter (supplementary Fig. 1). Each constituent can affect recombinant expression. Strong promoters may push gene transcription up, tags improve purification efficiency and/or protein solubility, and signal peptide guides the protein to the right place for workable purification.
Many studies have focussed on using peptides to direct the recombinant proteins to specific compartments within Varasteh-Shams et al. [28] plant cells, such as the apoplastic space or endoplasmic reticulum (ER). In earlier studies, signal peptides triggered changes in biomass growth, medium culture properties, and RP yield [12,40]. Authors often choose between two of those peptides for Nicotiana hairy root: the KDEL, a C-terminal retention signal for ER, and the N-terminal calreticulin, which promotes RP secretion. KDEL causes an increase in RP accumulation in endoplasmic reticulum (ER) lumen, whereas calreticulin promotes it in the culture medium (i.e., rhizosecreted RP boost). A signal sequence addition also increases hairy root productivity. Comparisons performed between ER retention signal γ-zein exhibited a slight productive advantage compared to apoplastic space [26]. The better yield with ER signal could be explained by size limitation of secretion route owing to cell wall pores averaging 50 Å (although in some reports the secreted proteins exceeding this limit [11,38,62]). Furthermore, the longer time in the ER allows more precise folding of proteins and the proteasefree environment reduces protein degradation risk. Nevertheless, the research group's objectives will determine the most appropriate secretory approach, but because hairy root culture excels at DSP ease up, efforts to increase rhizosecretion should always be considered. The promoter plays a critical role in triggering the expression of target genes, determining the tissue, and setting the right timing for expression. The recombinant expression levels are also determined by the terminator sequence, which plays a key role in transcript termination, mRNA stability, and polyadenylation. Most of the accessed works use the constitutive Cauliflower Mosaic Virus 35 s (pCaMV 35 s), usually adding multiple line-up copies to increase transcriptional activity. Other promoters employed include tissue-specific (mas2' and (Aocs)3AmasPmas; [38,46]), inducible (NtQPT2; [48]), and constitutive (TEV; [23]). However, none of them ensures better results than pCaMV 35 s, except for the heat shock protein 18.2 promoter, which yields 1.4-fold higher than pCaMV 35 s in the production of the recombinant sweetener protein miraculin [16]. In these trials, the heat shock 18.2 terminator was also used instead of the commonly employed NOSt. Such replacement of regular promoters and terminators has already been done in other plant expression hosts, which provided more stability to the mRNA and were less prone to transgenic sequence silencing [63].

Transformation procedures
A well-executed transformation results in several transgenic lines and elite events (i.e., a high-yield plant usually propagated in trials) is likely to occur with a successful procedure [8,54]. Obtaining an elite event for RP production via HR according to strategy on Fig. 2B, reduces labor and saves FW fresh weight, DW dry weight, TSP total soluble protein money, limiting most analyses to a smaller group. And for that, the root induction rate can be increased by adjusting the R. rhizogenes strain, bacterial volume, acetosyringone concentration (AS), inoculation time, duration of co-cultivation, and type and size of explants.
Induction of hairy roots is initiated by infecting plant tissue with R. rhizogenes. Then, the bacteria can transfer rol genes into the cell's nucleus (as well as the foreign gene in direct transformation). The most used approach among articles to promote bacteria-plant interaction is to inoculate a 4-5 week old and 1 cm 2 leaf disc into a bacterial solution with OD 600 = 1, allowing R. rhizogenes to enter the intercellular space through the leaf wounds [51]. Another method is to incise the midrib of a plant with a scalpel previously exposed to R. rhizogenes colonies [7,54,61]. In the root induction protocol, liquid co-culture (i.e., the immersion of tissues in bacterial culture medium) and solid co-culture (i.e. when leaves are kept in a solid culture medium without antibiotic selection) have been performed at different times, with the most commonly applied periods being 30 min and 48 h, respectively [28] found that a 10-min in liquid and a 72-h in solid co-culture provided the most favorable results in a comparative approach (employing a bacterial solution at OD 600 = 0.5). However, longer treatment times of liquid co-cultivation were never tested, even though it performed well for hairy root induction in other species, such as Marking nut and Dang shen [65,67].
Furthermore, some exogenous compounds, such as acetosyringone (AS) that are commonly used during gene transfer. Plants use that natural wound-induced phenol as a weapon against bacterial infections [68]. In response, the pathogenic Rhizobium sp. has evolved to recognize it and activate vir genes responsible for coding proteins that transfer the T-DNA [69]. AS is the major external reagent used along plant transformation to induce hairy roots in plants by R. rhizogenes [22,70]. However, few papers herein examined mention its use in trials. In these reports, either preculture [49], liquid [11,39], or solid co-culture [37,45] were application points of AS. The most common concerns about hairy root transformation are addressed in the paragraphs above. Even though there is a widely employed methodology that is followed by several papers, there remain gaps in the understanding of how to transform using R. rhizogenes effectively. To design a robust approach, empirical data is still required. Some issues must be answered, such as determining the most successful R. rhizogenes strain, the optimal period for co-culture, and the appropriate AS application strategy. Clearing these points in the future would make performing an elite event simpler.

Hairy root growth curve
After hairy root induction, maintaining the root culture in vitro is critical for high performance of recombinant protein. Recombinant hairy roots are highly stable hosts that can be cultivated for a long time, producing RP over 19 months in some cases [7]. Understanding its developmental stages helps to control recombinant yield and improve outcomes. Root growth generally involves three phases (Fig. 3a): (i) the initial phase, in which hairy root adjust its metabolism by biochemical and gene changing fitting then to the culture conditions; afterwards in (ii) the mid-log phase, the hairy root use the resources in the culture medium for growing; and finally (iii) the stationary phase, that ends the cycle when roots expansion stabilizes due to spatial or nutritional constraints [55]. In addition, some authors describe a death phase as a later point in the root cycle when tissue necrosis and phenolic emission occur [16]. It is common for the higher RP accumulation to occur between the mid-log and stationary phase, usually between the fourth and sixth week of culture. Because recombinant levels typically decay after that point, it is important to harvest the proteins-RP from biomass and medium-and subculture the roots at that moment ( Fig. 3b; [40,42,53,58]).
Nevertheless, Nicotiana roots respond differently to culture conditions in terms of growth and recombinant expression. The compiled papers reported the peak of recombinant yield occurring between the sixth and 36th of culture day. An equally wide variation might be noted in biomass accumulation (Fig. 3b). Culture conditions and medium composition are critical factors that, if carefully handled, may control these drastic variations. Among the main drivers are the aeration system [71], pH, light, medium volume, agitation, temperature, conductivity, osmolarity, sugar content, carbon source, micro-and macronutrients, plant growth regulators, vitamins, and medium type.
The choice of an appropriate medium is critical to the success of plant tissue culture. Murashige and Skoog (MS; [72]), B5 [73], and Woody Plant (WPM; [74]) are the most used medium in the papers. All reports agreed that Gamborg's B5 is better for development and RP production in Nicotiana hairy roots [7,27,28]. This can be explained by the high nitrate/ammonium ratio (13 mM) since nitrogen uptake is known to be related to dry matter growth and energy setups [75]. Otherwise, another reason for the growth advantages provided by the B5 medium addition is the presence of thiamine, which promotes root growth [76]. Fig. 3 Line plot of biomass accumulation over 30 days of culture for each selected paper. In A this information is expressed in fresh weight (FW), with the background curve as a hypothetical marker of hairy root growth. Similarly, the second graph B illustrates the growth lines, without a time limit, in terms of FW and dry weight (DW). Lastly, C displays the highest growth rates achieved for Nicotiana roots (removed from the first graph for better visualization) Nevertheless, the highest ever observed biomass rate was achieved using MS, it reached 0.45 g of fresh weight (FW) per mL (Fig. 3c; [8]).
It is also important to look for other mediums than those previously mentioned. K3 and Schenk-Hildebrandt nutritional solutions had better results than Gamborg's B5 for tobacco growth in early studies [49]. Hoagland medium was also once used for adventitious tobacco roots propagation, but there is a gap in knowledge about its performance compared to other compounds [13]. It is also worth noting that there are unexplored culture mediums and mixtures for tobacco root culture (e.g., root culture White's medium, Gresshott and Doy basal salt, Linsmaier and Skoog tobacco tissue culture medium, and MS1/2-B5 mixture).
On the other hand, the maintenance conditions are usually overlooked and can affect the production of recombinant products if not appropriately handled. Factors such as pH, which often varies over time [7,41,44,55], can lead to aggregation and precipitation of rhizosecreted RP, preventing a well-functioning DPS. Acid-base balance is also closely related to salt solubility and tissue nutrition uptake, which should always be around 5.8 [77]. In addition, the authors disagree on other parameters, such as light and agitation. The agitation is essential to the wide availability of nutrients for roots, though is a typically non-empirical choice made by authors who operate a shake flask culture, with 100 rpm being the most used rotational parameter (60 rpm speed is also common in many reports). However, according to comparisons between shake speeds, 90 rpm was the most beneficial setting for root growth in flask [27]. Regarding lighting, some papers keep the culture on exposure, while others keep it in the dark. It has been suggested that long photoperiods may favor root growth [78]. However, due to conflicting evidence [79], it is necessary to compile these conditions and compare them specifically for Nicotiana species.

Rhizosecretion performance
Rhizosecretion is the most cost-effective way to collect foreign proteins, whereby the target proteins are harvested directly from the liquid medium, and the laborious methods of plant biomass extraction are avoided. However, the low production is its main setback, and in some cases, RPs are barely detected in the root culture medium [50]. In the compiled papers, the rhizosecreted-to-total RP ratio varied from 1 to 57%. That value is directly linked with the transformation method, species, root growth stage, and expression level. To increase this ratio, it is critical to have attention not only to these influencing aspects, but also to those that reduce the amount of protein in the medium. The most disturbing factors reported are proteases, containment surface adsorption, RP length, imbalanced pH, and mishandling of signal peptides. Plant proteases often act on RP on the plant molecular farm system during protein extraction, lowering yield and generating contaminants fragments that are difficult to be removed from the final product (e.g., [80][81][82][83]).
Proteases are found in all organisms and are essential for life. They cleave peptides to fine-tune metabolism and remove damaged, misfolded, or potentially dangerous proteins. The core proteome of N. benthamiana contains around 1240 predicted proteases [84,85], some of which acts on the secretory pathway, affecting rhizosecreted RP. Aspartic (family 1), cysteine (family 1), metallopeptidase (family 1), and serine (family 1, 8, 10, and 33) proteases are the main protein found in secreted media [86,87], although only serine subtilisin-like proteases and CND41 (chloroplast nucleoid DNA binding protein) were reported in the tobacco root exudate so far [8,88,89].
The target protein is exposed to a broader range of proteases triggered by Rhizobium sp. infection, throughout hairy root establishment culture [84]. With aging, the amount of protease release in the medium increases, and it is pointed out as the main limiting factor of long-time RP production (over 30 days of culture) [44,88]. The onset of protease accumulation differs between articles, found on days 6, 20, 34, and 42 [41,44,54,58]. Although, all reports agree that protease peak occurs at the end of the mid-log phase of biomass growth, being directly related to the vegetative growth phases. Hairy root employers should therefore monitor the hairy-root growing in order to control protein degradation, since this is the most significant factor that restricts RP production in secretion.
One of the protease-handling approaches used for Nicotiana sp. hairy root was the co-expression of a soybean Bowman-Birk serine protease inhibitor (I12 family) linked with calreticulin sequence. The co-expression resulted in a 2-to 2.5-fold increase in the recombinant antibodies accumulation in previous work [38]. Nevertheless, the authors notice that constitutive expression of the inhibitor interferes with plant growth performance, reducing lifetime reproductive capability and individual size. Clearly, this scenario reflects the demand for finding and testing inhibitors that may protect the target protein while interfering as little as possible with the plant's metabolic function.
In general, knowledge of RP's features and cleavage sites can aid in the selection of the most effective inhibitor(s) to counteract the proteases. Externally supplied chemicals are a means to reduce proteolytic activity in the hairy root system without the need for gene integration. In the papers compiled, additive supplementation is the most used method for managing rhizosecretion quality. An example was the addition of bacitracin into culture medium in a previous study, an antibiotic polypeptide mixture known for its broad-spectrum protease inhibition [42]. The use of bacitracin resulted in faster root growth, but no decrease in protein degradation was observed. It is possible that the concentration used in studies so far was insufficient to effectively prevent proteolysis [90]. However, other external additives approaches have been remarkably efficient, increasing protein levels by up to 28-fold (Figs. 3, 4; [8]). This is a promising direction to optimize the hairy root to the field of recombinant expression in biotechnology.
A broad spectrum of biological effects is attributed to the compounds used as additives, such as growth stimulators, osmotic regulators, nutrients, and protein protection factors. In addition, medium additives may be used solo or in combination. The solo mode ensures a better-monitored DPS, less pleiotropic effects, and greater dose-dependence control. Alternatively, the second approach usually provides higher yields since it combines agents that protect or increase RP production in distinct ways [8]. The most common additive used in Nicotiana hairy roots is polyvinylpyrrolidone (PVP), a water-soluble polymer that acts as a protein stabilizer and reduces adsorption to vessel surfaces. Proteins may be protected by PVP due to its ability to form complexes and colloids. It has excellent properties for PMF, such as metabolic inertness, low cost, and easy separation along the DPS, unlike other polymers, such as bovine serum albumin [91]. In addition, PVPs do not affect root growth or RP yield in biomass [25,27,40,58]. In its best performances, PVP showed a 43% increase in the ratio of RP rhizosecretion. TSP −1 and held protein degradation at minimum rates during 46 days of culture [7,44].
Among solo applications, the most significant increases in rhizosecreted RP volume were promoted by 2,4-dichlorophenoxyacetic acid (2,4-D) and 1-naphthaleneacetic acid (NAA). Both are synthetic auxins that act on tissue division, differentiation, and elongation, promoting lateral and adventitious rooting [92,93]. Notably, 2,4-D-containing medium has also been used to culture recombinant hairy Fig. 4 Scheme of rhizosecretion fold increase using solo additives in Nicotiana hairy root. PVP was employed at different molecular weights by each author. ✝Although the paper reported the best concentration as 0,1% (w/v), Wongsamuth and Doran (1997) adopted Gamborg's B5 medium, which already has 0.25% (w/v) KNO3. *Reports that obtained no differences between RP rhizosecretion with and without additives 1 3 roots in other species, including turnip and Arabidopsis, resulting in increased RP secretion in both species [94,95]. However, treatment with auxins at high doses may alter cell wall composition and permeability, which could induce morphological malformations [8]. This may explain why [11] found opposite results when 2,4-D was used as an additive for Nicotiana root growth, which did not affect rhizosecretion compared to the control treatment. The optimal auxin concentrations for tobacco hairy root culture have yet to be solved. And due to such adverse effects, growth regulators should be used cautiously to avoid impairing the expression system. In this sense, Lonoce et al. [8] endeavored to find the optimal auxin concentrations for scFV2B8-FcΔXF protein expression in tobacco HR. The highest RP accumulation was reached by adding 2,4-D and NAA both at 0.1 mg/L to elicit the protein production 28 days after HR induction, yielding 16.3 mg/L and 8.5 mg/L of antibodies on culture medium, respectively. This result was notable since in adjusted conditions in a few months it is possible to produce grams of antibody.
Meanwhile, the combination of additives led to the most efficient recombinant secretion results ever achieved for Nicotiana species. Among the highlights are the largest rhizosecretion yield [8], major booster in recombinant secretion [37], and greater rhizosecreted-to-total RP ratio obtained in Nicotiana roots [27]. All these results were obtained from different mixtures of KNO 3 , NAA, and PVP solutions. So, that comprises a nitrogen source, an auxin, and a stabilizing agent, respectively. Nevertheless, the mix of additives entails a challenge: setting the concentration of each component. This decision's complexity stems from the potential interactions between chemicals and the damage they may cause to plant tissue. To overcome this challenge, [27] performed a statistical experimental design that individually considered the optimal results for using NAA, KNO 3 , and PVP in tobacco root culture. Using this method, they could predict the most effective configuration when the additives were together. Furthermore, many ingredients still need to be evaluated as medium incrementation ingredients, including 2,4-D (used only in solo), albumin, and polyethylene glycol, showing an exciting direction for future research.
Another way to maximize the amount of RP obtained by secretion is to grow roots in fully submerged bioreactors. An example is a tobacco hairy root growing in a Life-Reac-torTM plastic bioreactor expressing GFP and reaching a 20% RP rhizosecreted/total ratio, with a protein concentration of 810 μg/L in a 1.5-L container [61]. Details about bioreactors will be covered in the following topic.

Culturing hairy root in bioreactors for molecular farming
A bioreactor consists of a sterile vessel containing a culture whose physicochemical parameters (pH, dissolved gases, aeration, and temperature) are monitored online or offline. Although bioreactors often do not excel in yields, absolute production far surpasses conventional systems, supporting liters of culture medium. Besides, the vessel contains an agitation system to facilitate mass/energy transfer in the culture and channels for the in/outflow of oxygen, culture media, antifoaming, buffer, and samples.
Different bioreactors have been used for the plant cell suspensions culture (stirred tank, bubble-column, air-lift, wave). However, the hairy root bioreactor adaptability is an outstanding drawback due to its complex growth (reviewed in [96]). The machine's basic functions are hindered by the intertwined nature of roots, such as agitation system, homogeneous transfer of energy, nutrients, and metabolites. Moreover, root hairs are delicate structures, highly sensitive to stress and changes in temperature, pH, oxygen, and other nutrient levels. Therefore, the traditional bioreactor culture is not suitable, and the process cannot be maintained.
A few traditional bioreactors have been modified for hairy root, such as liquid-phase reactors, where hairy root are entirely immersed in culture media (e.g., stirred tank, bubble column, air-lift reactors); gas-phase reactors, where the hairy root are exposed to air or a mixture air/liquid medium (e.g., trickle bed, droplet phase, nutrient mist reactors), and hybrid reactors, in which a combination of liquid-and gasphase reactors is used (e.g., bubble column and trickle bed, bubble column and nutrient mist, bubble column and spray bioreactors). The most suitable system for hairy root seems to be air-lift bioreactors that provide aeration and agitation by releasing compressed air from the vessel base; and mist reactors that spray roots throughout the day with a light coat of nutrient media, also including a rich supply of oxygen.
However, most hairy root bioreactors papers are related to secondary metabolites [96][97][98]. And so far, only a few studies have applied a bioreactor to recombinant tobacco roots. Such imbalance with the protein focus might be due to the recent host adoption for RP production and the additional protease problem.
The generation of a full-length murine IgG1 monoclonal antibody by tobacco growing in a 2-L QuickFit was one of the first reports of RP produced by a hairy root in a bioreactor [7]. After 30 days of growth, the air-driven bioreactor produced 9.0 g/L of biomass, while flask cultivation obtained 11.1 g/L. The former, however, had a higher RP output. After 14 days, antibody production in the shaking flask culture dropped significantly, reaching 1.7 times lower than in the bioreactor, probably caused by higher biomass degradation in flask culture after the root growth ceased.
Other use of Nicotiana hairy root bioreactor includes the secretory version of monoclonal antibody M12 [27], cultured in a 100 mL prototype bioreactor (Medicel Explorer Cultivation Unit, Medicel, Espoo, Finland), aerated with a sinter sparger (1.5 or 2.5 vvm, i.e., volume of gas flow per liquid volume per minute) and mixed either by airflow alone or with an additional magnetic stirrer. Compared with the performance of hairy roots cultured in 250 mL Erlenmeyer flasks, growth did not differ significantly when airflow humidity was introduced in the reactor. On the other hand, aeration variation was ineffective for growth but influenced M12 yield, resulting in 3-times higher antibody amount.
Lastly, the mammalian immunomodulator interleukin-12 was made into two different bioreactor systems, in a mist and airlift bioreactor [55]. Both were equipped with 2-L chambers (plastic bag for mist), immersed in B5 medium, and inoculated with 16 g of fresh roots. In airlift, roots were not uniformly distributed, arranging into a dense ring around the vessel wall and surrounding the central column of aeration bubbles. Also, pockets of dark phenolic-like roots appeared, suggesting nutrient starvation in those regions. Nevertheless, the biomass growth was slightly favored. For RP making, interleukin-12 intracellular concentration in the mist reactor was 4.7 µg/g FW while 2.9 µg/g FW was obtained in the airlift reactor, likely due to oxygen limitation. The concentration of interleukin-12 in the media was 434, 196, and 308 µg/L in shake flasks, airlift reactor, and mist reactor, respectively.
Despite its great progress, no valuable heterologous protein has been commercially produced in hairy roots at a large scale yet. The low yields are one of the main reasons for that non-adoption trend, but also their requirements compared with microbial or mammal cultures, which grew faster and homogeneously. Therefore, it is pointed out the need to explore new strategies and bioreactor designs to enhance productivity of foreign proteins in hairy roots.

Future perspectives
Years ago, plants were discovered as an alternative for RP production. Today, many green cultures are being explored for recombinant protein expression. Among them, roots provide DSP benefits and easier maintenance. This systematic review covers the second-wave plant platform for recombinant expression, hairy root culture in its bench-work background and applications. The content compiled herein may assist researchers in grasping the status of the literature, providing a starting point for hairy root implementation, and suggesting a path for future research.
Recombinant production varies widely among approaches. Considering the recent use of hairy roots in recombinant expression, several authors have wondered which parameter is most suitable for establishing these roots in some instances. Several researchers are looking to find these variables, such as the liquid medium that roots respond best to, the co-culture period that induces the higher biomass of hairy roots, and the additives applications that enhance rhizosecretion. Nevertheless, some questions remain. Addressing them might consolidate the technological pathway for optimizing roots as recombinant hosts, for example, in determining which bacterial strain is better able to transform Nicotiana, which flask size is most conducive to root growth, how light regime affects production of RP, which combination of additives most effectively stimulates rhizosecretion, how to suit bioreactors for better support the roots, etc.
However, before addressing these open questions, it is worth considering some of the weaknesses of many hairy root reports. That is the failure to communicate clearly critical data, which should be included in the articles. For example, the yields of all recombinant clones are rarely reported with statistical content (standard error, standard deviation, sample description, and hypothesis testing). Another shortcoming is the different means by which RP yields are estimated; most papers report the data only in order of TSP, DW, or FW. However, these measures are not comparable. Ideally, in this case, the community should either agree on the choice of the most appropriate measure or the data should be estimated for all three metrics. Due to the lack of such details, meta-analytical investigations of the Nicotiana hairy roots literature were unviable (seeking to answer questions such as "hairy root vs. leaf-based RP expression", "therapeutic protein vs. reporter protein in hairy root", and "additives in hairy root secretion content"). Thus, the clear presentation of the data is critical for corresponding studies to elucidate the methodology for using hairy roots. Therefore, the way forward appears to be to strengthen the consistency of articles and to investigate methodological steps that are still unclear, such as those described previously. Although, hairy roots are a powerful biotechnological tool and have advantages over conventional green platforms. Reports of proteins produced by hairy roots and regenerated shoots revealed that the roots produced more for small and large proteins [9,52]. In hydroponic cultures, roots provide more RP than leaves [46]. Recombinant hairy root hosting is highly effective, as its genetic stability provides a longerterm expression, and its ability to exude protein simplifies DSP. This allows RP to be produced over a long period of time without the need for laborious protein purification methods. Finally, the preservation of elite events showing higher production of HR may be useful in biobanks for industry accession and aiming to avoid genetic modification over long-time of culture [99] and for considering the cryopreservation aspects such as biochemical, genetic, and morphological alterations during the cryopreservation process is recommended a fully review in Popova et al. [100]. Thus, here we demonstrate that Nicotiana's hairy root is a good system to consider for recombinant protein expression.