Journal of Chemical Technology and Biotechnology J Chem Technol Biotechnol 83:286–293 (2008)

Processing of soybean (Glycine max)extracts in aqueous two-phase systemsas a first step for the potential recovery ofrecombinant proteinsOscar Aguilar and Marco Rito-Palomares∗Centro de Biotecnologıa, Departamento de Biotecnologıa e Ingenierıa de Alimentos, Tecnologico de Monterrey, Campus Monterrey, Ave.Eugenio Garza Sada 2501 Sur, Monterrey, NL 64849, Mexico

Abstract

BACKGROUND: The potential use of plants as production systems to establish bioprocesses has been establishedover the past decade. However, the lack of efficient initial concentration and separation procedures affect thegeneric acceptance of plants as economically viable systems. In this context the use of aqueous two-phase systems(ATPS) can provide strategies to facilitate the adoption of plants as a base for bioprocesses. Among the crops,soybeans (Glycine max) represent an attractive alternative since potentially they can produce high levels ofrecombinant protein. In this paper the processing of fractionated soybean extracts using ATPS is evaluated as afirst step to recover recombinant proteins expressed in plants, using β-glucuronidase (GUS; E.C. 3.2.1.31) as amodel protein.

RESULTS: The evaluation of the effect of system parameters provided the conditions under which the contaminantproteins from fractionated soybean extracts and GUS concentrated in opposite phases. A PEG 600/phosphatesystem comprising 14.5% (w/w) polyethylene-glycol (PEG), 17.5% (w/w) phosphate, a volume ratio (Vr) equal to1.0, and a system pH of 7.0 resulted in the potential 83% recovery of GUS from the complex mixture and anincrease in purity of 4.5-fold after ATPS.

CONCLUSIONS: The findings reported here demonstrate the potential of ATPS to process fractionated soybeanextract as a first step to isolate and purify a recombinant protein expressed in soybeans. The proposed approachcan simplify the way in which recombinant proteins expressed in plants can be recovered. 2007 Society of Chemical Industry

Keywords: aqueous two-phase systems; protein recovery; soybean extracts

INTRODUCTIONThere is considerable interest in the developmentof biotechnological processes that exploit the use ofplants as a host for producing recombinant proteins.The potential use of plants as bioreactors has beenestablished over the past decade. Transgenic plantsare a potentially inexpensive system for the large-scaleproduction of recombinant proteins for use in thepharmaceutical, agricultural and industrial sectors.1–3

The advantages of using transgenic plants includelow cost and flexibility in large-scale production, thepresence of natural storage organs such as seedsand tubers, and existing technology to harvestingand processing of plant material.2 However, thelack of efficient initial concentration and separationprocedures affect the potential generic acceptanceof plants as economically viable systems. One ofthe main drawbacks of the seed-based technologyis the inherent need for the plant to generate the

flower in order to produce the seed, raising bio-safetyconcerns regarding avoiding the release of pollen tothe environment and increasing the cost of productiondue to the need for containment and longer harvestingtimes.4 The selection of an adequate crop for theproduction of recombinant proteins is affected bynumerous aspects that contribute to the success andfinal cost of the products. The general alternativesinclude maize (Zea mays), rice (Oryza sativa), canola(Brassica sp.), tobacco (Nicotiana tabacum), peas(Pisum sativum) and soybean (Glycine max). Canola,maize and soy have been regarded as the most likelyproduction systems for commercial application ofplants as bioreactors.5 Maize is preferred over otherplant hosts due to its lower hydro-soluble proteincontent. Extensive research on the potential recoveryof recombinant proteins expressed in this particularcrop has been documented.6,7 In the particular caseof canola, compared to other crops, a greater fraction

∗ Correspondence to: Marco Rito-Palomares, Centro de Biotecnologıa, Departamento de Biotecnologıa e Ingenierıa de Alimentos, Tecnologico de MonterreyE-mail: mrito@itesm.mx(Received 9 March 2007; revised version received 10 August 2007; accepted 18 August 2007)Published online 11 December 2007; DOI: 10.1002/jctb.1805

2007 Society of Chemical Industry. J Chem Technol Biotechnol 0268–2575/2007/$30.00

Processing soybean extracts in aqueous two-phase systems

of its proteins are basic, thus recovery of anionicrecombinant proteins could be expected to be easier.Furthermore practical strategies for the recovery ofrecombinant proteins from canola have also beendocumented.8–10 Soybeans (Glycine max) representan attractive alternative since, potentially, they canproduce more recombinant protein that other crops.Soybeans contain approximately 40% of their weightas proteins compared to only 2% for potato tubers, forexample. Thus if the recombinant protein is expressedat an equal percentage of the total protein in these twocrops, 20 times more product could be producedfrom soybean. However, the potential recovery ofrecombinant proteins from soybeans has not beenwidely studied.

The use of seeds for heterologous protein expressionhas proven to be an efficient system for bothbiopharmaceutical and industrial proteins.11,12 Theuse of seeds as host organ for recombinant proteinproduction can be advantageous as they are naturalstorage organs with high expression levels, but thiscan be a challenge for downstream processing whenthe protein of interest has to be purified from theseed extract.8 Soybean seeds contain more proteinthan any other commercial crop (up to 48%)consisting of a mixture of proteins (α-, β-, and γ -conglycinin, glycinin and other globulins) ranging inmolecular weight from 140 to 300 kDa with differentphysicochemical properties allowing fractionationand differentiation.13–15 The most popular andsimplest method for fractionation is that described byThanh and Shibasaki13 using simultaneous isoelectricprecipitation to effectively isolate the two mainsoybean proteins 7S and 11S, named according totheir Svedberg sedimentation coefficient. Two mainglobulins 7S and 11S, account for about 70–80% ofthe total soybean proteins (TSP). Of these two, the7S fraction comprises about 35% of the TSP, wherethe majority of 7S proteins consist of the globulincalled β-conglycinin, a trimeric glycoprotein with amolecular weight of about 180 kDa that precipitatesat pH 4.8 and makes up about 85% of the 7Sfraction [13]. The 11S fraction comprises from 31to 52% of the TSP, with about 85% being the 11Sglobulin called glycinin. The currently accepted modelof glycinin is a hexamer with a molecular weight ofabout 360 kDa.14,15 Precipitation with variations isthe most commonly used method to recover soybeansproteins.16–18

In general, protein recovery from the starting plantmaterial includes extraction, clarification, proteincapture, purification and polishing. The overallproduction cost is mainly determined by the efficiencyof the initial capture and purification steps, wherefeed volumes are large until biomass solids andoils are removed and the protein is concentrated.5

Therefore, the establishment of efficient primaryrecovery procedures for the recovery of recombinantproteins from transgenic crop is needed. In thiscontext aqueous two-phase systems (ATPS) are an

attractive alternative to facilitate the adoption ofbioprocess based on plants as production systems.ATPS is a technique that has proved to have greatpotential for the recovery and purification of biologicalcompounds.19–23

Previous studies9,10 have shown the potential oftraditional downstream processing operations appliedto seed-produced recombinant proteins and the useof genetic engineering strategies to recover themfrom the bulk storage proteins in which the productof interest is immersed. The potential economicbenefits of substitution of costly unit operations, suchas chromatography, by ATPS without commitmentof the yield, have previously been addressed andthe same strategies can be applied to plant-madeproducts.7,24 Recently, the potential applicability ofATPS for integrated extractive partitioning applied tothe recovery of a model recombinant protein expressedin maize and tobacco has been demonstrated.7,25

However, the potential application of ATPS to processsoybean extracts as a route to the recovery ofrecombinant proteins has yet to be addressed.

In the present study, the potential application ofATPS to process soybean extracts was evaluated asa first step to establishing a practical strategy torecover recombinant proteins expressed in plants.Fractionated soybean protein extracts were obtainedby simple scalable methods and used as an exampleof a complex mixture of contaminants in which therecombinant proteins can be found. β-glucuronidase(GUS; E.C. 3.2.1.31) was chosen as a representativemodel protein to explore the potential use of ATPS toperform selective extraction of recombinant proteinsfrom soybeans. GUS has been previously used as amodel protein to be expressed in plant cells due toits high stability under a wide range of ionic forcesand its absence in higher plants.10,26 An artificialmixture formed by adding GUS to fractionatedsoybean extract served as an example to simulate thepresence of a recombinant protein. GUS is a tetramericacid hydrolase (68.2 kDa each) that catalyzes thecleavage of a wide variety of glycosides formed byβ-D-glucosiduronic acid and residues with hydroxylgroups, such as steroids, phenols, antibiotics, andother metabolites; as a result it is easy to monitor byenzymatic assays. This enzyme maintains its hydrolyticcapacity under very different environments, such asthose found in ATPS.8,10 A practical approach wasused to evaluate the effect of system parameterssuch as polyethylene glycol (PEG) and phosphateconcentration and nominal molecular weight of PEG,on the partition behaviour of proteins from thefractionated soybean extracts and GUS. This approachwas followed to establish the potential conditionsunder which the target protein (GUS) and thecontaminant proteins from the fractionated soybeanextract concentrated preferentially to opposite phases.This practical strategy can be useful as a starting pointto be applied in the recovery of recombinant proteinsexpressed in soybean.

J Chem Technol Biotechnol 83:286–293 (2008) 287DOI: 10.1002/jctb

O Aguilar, M Rito-Palomares

EXPERIMENTALProtein extraction from soybeansCommercial soybean seeds (Glycine max) wereprovided by the Agricultural Experimental Stationof Tecnologico de Monterrey. The seeds were brieflyrinsed with tap water to remove soil and other materialsand the whole grains were then ground to a fine powderusing a household grinder in 20 s intervals to avoidheating. Ground powder was passed through meshUS No. 100 to obtain flour. Soybean rich proteinextracts were obtained according to the procedurereported by Thanh and Shibasaki.13 Briefly, soybeanflour was suspended in 0.03 mol L−1 Tris-HCl buffercontaining 0.01 mol L−1 2-mercaptoethanol, pH 8 ata proportion of 1.0 g solids/20 cm3 buffer. The slurrywas stirred for 1 h and centrifuged at 10 000g, 20 minat room temperature to discard solids. The extract pHwas carefully adjusted drop-wise to 6.4 with 2 mol L−1

HCl and centrifuged at 10 000g for 20 min at 2–5 ◦C.The precipitate collected is referred to in this studyas the 11S fraction or 11S fractionated soybeanextract. The supernatant was further processed byisoelectric precipitation at pH 4.8, and the resultingprecipitate is referred to as the 7S fraction (or 7Sfractionated soybean extract). Both precipitates werere-suspended in 0.03 mol L−1 Tris-HCl buffer at pHto 7.5; any trace of precipitate, if present, was removedby centrifugation (5 min at 10 000g). These twofractionated extracts were further used for partitioningexperiments.

Aqueous two-phase experimentsAqueous two-phase systems (Table 1) were selectedbased upon previous experience19–21,24 to give a vol-ume ratio of 1.0 and a fixed weight of 2.0 g. Thestrategy behind the selection of the experimental sys-tems is well described elsewhere.19 The system tie-linelength (TLL), which represents the length of the linethat connects the compositions of the top and bottomphases in a phase diagram for a defined system, wascalculated as described previously.24 Predeterminedquantities of stock solutions of potassium phosphateand poly(ethylene glycol) (PEG; Sigma Chemicals, StLouis, MO, USA) of nominal molecular weights 600,1000, 1450 and 3350 g mol−1 were mixed with 0.2 gof fractionated soybean extracts (7S or 11S; with aprotein concentration of 32 and 9.0 mg cm−3, respec-tively) or 0.2 g of purified β-glucuronidase solutionwith 30 059 U cm−3 (Sigma Chemicals) to give thedesired PEG/salt composition (Table 1) with a finalweight of 2.0 g (the amount of fractionated soybeanextracts or the purified protein solution added to theATPS represent 10% w/w of the total system). Theamount of GUS in the stock solution (30 059 U cm−3)

was defined to ensure enzyme activity detection inthe ATPS. Additional, 2.0 g spiking experiments wereconducted, where the target protein (GUS) was addedto both fractionated extracts to simulate a proteinextract from a transgenic plant. ATPS were formu-lated in which the stock solutions of PEG and salt

were mixed with 0.2 g of spiked fractionated extracts(7S or 11S) containing a total protein concentrationof 33 or 10 mg cm−3, respectively, and 60 938 U cm−3

of β-glucuronidase. The stock solutions were gentlymixed for 15 min. Adjustment of the pH was madeby addition of 1 mol dm−3 orthophosphoric acid orpotassium hydroxide if needed. Complete phase sepa-ration was achieved by low-speed batch centrifugationat 1500g for 10 min. Visual estimates of the volumesof top and bottom phases were made in graduatedtubes. The volumes of the phases were then used toestimate the experimental volume ratio (Vr, defined asthe ratio between the volume of the top phase and thebottom phase). Samples were carefully extracted fromthe phases (top and bottom phase) and analyzed. Thetop and bottom phase recovery was estimated as theamount of target product present in the phase (volumeof the phase × product concentration in the phase)and expressed relative to the original amount loadedinto the system. Interface recovery was estimated asthe necessary amount of target product to completethe mass balance. Purification factor was estimated asthe ratio of the relative purity of GUS after and beforethe extraction stage. Relative purity was defined as theamount of GUS relative to that of the proteins fromthe fractionated soybean extracts. Results reported arethe average of three independent experiments.

Analytical techniquesThe enzymatic activity was estimated using theβ-glucuronidase microassay adapted from Feur-tado et al.27 Briefly, the modification included theuse of 2-amino-2-methylpropanediol (AMPD) as

Table 1. Systems selected for the evaluation of the partition

behaviour of the proteins from the fractionated soybean extracts

SystemMolecular weightof PEG (g mol−1)

PEG(% w/w)

Phosphate(% w/w)

TLL(% w/w)

1 600 14.5 17.5 32.02 15.5 18.0 37.13 15.8 19.5 41.54 17.0 20.5 45.25 1000 15.6 18.0 47.26 17.6 18.0 49.97 19.8 18.5 53.68 22.2 23.0 67.79 1450 13.7 13.1 27.110 15.7 13.9 34.411 18.6 15.2 41.912 21.0 16.0 47.813 3350 16.9 14.5 42.314 18.7 15.0 46.215 21.0 15.7 51.316 22.1 17.0 56.2

All 2.0 g systems comprising PEG of four different molecular weightswere selected to evaluate the impact of increasing TLL and molecularweight of PEG on the partition behaviour of the proteins from thefractionated soybean extracts. The volume ratio (estimated from blanksystems) and pH of the selected systems were kept constant andequal to 1.0 and 7.0, respectively. All systems were assembled andthe TLL estimated as described in the Experimental section.

288 J Chem Technol Biotechnol 83:286–293 (2008)DOI: 10.1002/jctb

Processing soybean extracts in aqueous two-phase systems

enzyme inhibitor. The assay mixture consisted of75 µL of McIlvine buffer pH 4.5, 15 µL of sub-strate (10 mmol L−1 4-nitrophenyl β-D-glucuronide(PNPG, Sigma-Aldrich)) in McIlvine buffer pH 5and 60 µL of enzyme extract-sample. Microtitre-assayplate was incubated at 37 ◦C for 15 min; the reac-tion was stopped by adding 75 µL of 0.2 mol L−1

AMPD (Sigma-Aldrich) and the yellow color pro-duced was measured at 405 nm in a microplate reader(BioTek Instruments Inc., Vermont, USA). Blankwells involved adding aliquots from the top or bot-tom phase of blank systems (ATPS without biologicalmaterial) for zero calibration. One unit of enzymeactivity was defined as the amount of enzyme pro-ducing 1.0pmol of 4-nitrophenol from 4-nitrophenylβ-D-glucuronide per minute at pH 4.5 and 37 ◦C.Protein concentration from the phases was determinedusing the method of Bradford.28

RESULTS AND DISCUSSIONProcessing of fractionated soybean extracts inATPSIn order to establish a purification strategy exploitingATPS for the recovery of target products froma particular mixture, the behaviour of the majorcontaminants in the extraction stage must becharacterised. In the particular case of the recoveryof a recombinant protein expressed in soybean,the process conditions under which the product ofinterest and the contaminants concentrate in oppositephase must be established. For the design of aparticular ATPS extraction stage, a strategy thatinvolves the characterisation of the behaviour of the

major contaminants proteins in selected ATPS canbe adopted. In this study two fractionated extractswere obtained from soybean exploiting precipitationsteps. The extraction of soybean proteins representsthe first step to develop a primary protein recoveryprocess. Thus the processing of the fractionatedsoybean extracts (identified as 7S and 11S fractions)in selected ATPS (Table 1) was initially attempted.In Table 2, the effect of system parameters (i.e.tie-line length and molecular weight of PEG) uponthe potential recovery of proteins from fractionatedsoybean extracts is illustrated. It is clear that proteinsfrom both fractionated soybean extracts (7S and 11S)exhibited bottom and interface preference in theATPS selected. In particular, proteins from the 7Sfractionated soybean extract concentrated relativelyequally in the interface and in the bottom phase.Increasing TLL within the same molecular weightPEG did not affect the protein behaviour (little or nochange was observed in partition coefficients of thesame molecular weight PEG); interface and bottomphase recovery remained relatively constant when TLLwas increased. However, increasing the molecularweight of PEG from 600 to 3350 g mol−1 caused aslight change in protein preference. Proteins from the7S fractionated soybean extract increased preferencefor the interface, particularly in PEG 3350 ATPS(systems 13 to 16 in Table 2). Such behaviour canbe attributed to a potential migration of the proteinsinitially concentrated in the top phase at ATPS withlow molecular weight PEG to the interface. Thischange in protein phase preference can be explainedby a reduction in the free available volume caused bythe increase in molecular weight of PEG in these types

Table 2. Effect of system tie-line lengths and molecular weight of PEG on the recovery of proteins from the fractionated soybean extracts in ATPS

Proteins from 7S fractionated soybean extract Proteins from 11S fractionated soybean extract

SystemTop phase

recovery (%)Bottom phaserecovery (%)

Interfacerecovery (%) Kp

Top phaserecovery (%)

Bottom phaserecovery (%)

Interfacerecovery (%) Kp

1 12 ± 1 42 ± 1 46 ± 1 0.28 ± 0.02 14 ± 1 67 ± 3 19 ± 3 0.22 ± 0.012 12 ± 1 44 ± 1 44 ± 1 0.27 ± 0.02 10 ± 0.1 61 ± 0.2 30 ± 0.2 0.16 ± 0.0023 11 ± 0.2 44 ± 1 45 ± 0.4 0.25 ± 0.01 10 ± 0.1 60 ± 1 30 ± 1 0.17 ± 0.0014 11 ± 1 37 ± 1 51 ± 1 0.30 ± 0.02 9 ± 1 46 ± 2 45 ± 3 0.20 ± 0.01

5 8 ± 0.1 42 ± 0.2 50 ± 0.2 0.20 ± 0.003 5 ± 0.2 60 ± 2 35 ± 3 0.09 ± 0.0016 5 ± 0.3 44 ± 1 51 ± 0.4 0.12 ± 0.01 4 ± 1 62 ± 2 35 ± 1 0.06 ± 0.027 5 ± 1 42 ± 0.3 53 ± 1 0.13 ± 0.02 4 ± 0.1 50 ± 0.4 46 ± 1 0.07 ± 0.0088 3 ± 0.1 24 ± 2 73 ± 2 0.14 ± 0.01 4 ± 1 8 ± 3 88 ± 4 0.51 ± 0.12

9 5 ± 1 43 ± 0.4 52 ± 0.4 0.12 ± 0.02 3 ± 0.2 72 ± 3 25 ± 3 0.05 ± 0.00110 5 ± 0.4 45 ± 1 50 ± 1 0.11 ± 0.01 5 ± 0.2 66 ± 1 28 ± 1 0.08 ± 0.0211 4 ± 1 45 ± 1 50 ± 2 0.09 ± 0.01 3 ± 1 68 ± 1 29 ± 1 0.04 ± 0.0112 5 ± 0.1 46 ± 1 49 ± 1 0.10 ± 0.002 4 ± 1 63 ± 3 33 ± 4 0.06 ± 0.01

13 3 ± 1 33 ± 1 64 ± 1 0.09 ± 0.02 3 ± 1 62 ± 2 35 ± 2 0.04 ± 0.0114 2 ± 0.3 37 ± 1 61 ± 1 0.05 ± 0.01 3 ± 0.4 62 ± 3 35 ± 3 0.05 ± 0.0115 2 ± 1 37 ± 0.2 61 ± 1 0.06 ± 0.01 1 ± 0.1 65 ± 1 34 ± 1 0.02 ± 0.00216 3 ± 0.4 37 ± 1 60 ± 1 0.07 ± 0.01 3 ± 1 68 ± 3 29 ± 3 0.04 ± 0.01

Compositions of the systems 1–16 are defined in Table 1. The concentration of fractionated extracts in all the ATPS was 10% (w/w). The top andbottom phase protein recovery is expressed relative to the initial amount of protein content in the soybean extracts loaded into the systems. Interfacerecovery was estimated as the necessary amount of protein to complete the mass balance. Protein partition coefficient (Kp) represents the ratio ofthe concentration of proteins in the phases.

J Chem Technol Biotechnol 83:286–293 (2008) 289DOI: 10.1002/jctb

O Aguilar, M Rito-Palomares

of ATPS.29,30 In the case of proteins from the 11Sfractionated soybean extract, although these proteinsalso exhibited interface and bottom phase preference,an increase in the protein concentrated in the bottomphase was observed. As a result a decrease in thepartition coefficients was observed when comparedwith those from the 7S fractionated soybean extractfor the same molecular weight PEG and TLL values(Table 2). This increase in bottom phase preference ofthe proteins from the 11S fractionated soybean extractin comparison to those from the 7S can be attributed tothe difference in the molecular size of the main proteinspresent in both fractionated extracts. The molecularweight of the majority of the proteins in the 11S(glycinin) and 7S (β-conglycinin) fractionated extractsis around 360 KDa and 180 KDa, respectively.13–16

It has been reported23,29 that proteins with highmolecular weight generally concentrate in the bottomphase. An increase in system TLL had no effect onthe partition behaviour of the proteins from the 11Sfractionated soybean extract. Increasing the molecularweight of PEG caused a slight increase in the proteinsconcentrated in the bottom phase. ATPS with lowmolecular weight PEG (i.e. 600 g mol−1) and shortTLL (i.e. 32% w/w; system 1 in Table 2) loadedwith 11S fractionated soybean extract concentratedthe majority of the proteins in the bottom phase (i.e.67%) and the rest in the top and interface (14% and18%, respectively). It is important to note that system8 exhibited a strange behaviour, since the majorityof the proteins, regardless of the fractionated soybeanextract used, concentrated at the interface, 73% and88% in ATPS loaded with 7S and 11S fractionatedsoybean extract, respectively. Such behaviour can beattributed to the fact that this system had the greatestTLL used (i.e. 67% w/w). A long tie-line length usuallyaffects protein solubility and causes proteins presenton the top and bottom phases to accumulate at theinterface.23,31

The results obtained from the processing offractionated soybean extracts in ATPS may suggestseveral recovery strategies for the potential purificationof a recombinant protein. It is clear that ATPS caneasily be implemented for the recovery of recombinantproteins expressed in soybean that exhibit top phasepreference (e.g. lysozyme and therapeutics proteinssuch as neuraminidase A and monoclonal antibodiesagainst HIV.7,25 In this case it can be anticipatedthat the contaminant proteins (obtained from eitherfraction 7S or 11S) and the target products willconcentrate in opposite phases. In the top phase theproduct of interest will be present with a reducedamount of contaminant proteins. In the case of targetrecombinant proteins that exhibit interface or bottomphase preference in ATPS, alternative strategiesexploiting precipitation stages and process parametersneed to be considered. Target protein can be recoveredfrom the 7S or 11S fraction based on the particularpartition behaviour of the protein of interest in ATPS.In order to further evaluate the potential application

of ATPS for the recovery of recombinant proteins thatcan be expressed in soybean, β-glucuronidase (GUS)was selected as model protein. GUS was selectedas a representative recombinant protein that can beexpressed in plants (i.e. pea seeds26) and monitored inthe presence of different solvents. As a first step, thepartition behaviour of the model protein was evaluatedin selected ATPS.

Partition of purified β-glucuronidase (GUS) inATPSThe partition behaviour (expressed as productrecovery) of glucuronidase in ATPS when increasingboth TLL and molecular weight of PEG is illustratedin Table 3. It is evident (based on mass balanceanalysis) that interface product accumulation accountsfor the majority of the protein loaded to the systems.In ATPS of low molecular weight PEG (i.e. 600 and1000 g mol−1; systems 1 to 8 in Table 3) GUS wasnot detected in the bottom phase. Enzyme activity wasdetected in the top phase, and the rest of the productloaded to the ATPS was considered to be accumulatedat the interface. The low product recovery from the topphase can be attributed to either product accumulationat the interface or loss of activity (or both). However,the presence of precipitate at the interface was visuallyobserved in the ATPS. Increasing TLL had nosignificant effect on product recovery. However, anincrease in the molecular weight of PEG from 600to 1000 g mol−1 caused GUS top phase recovery toslightly decrease. Further increase in molecular weightof PEG to 3350 g mol−1 resulted in an increase of GUSaccumulation at the interface. To further evaluate thepotential application of ATPS to recover GUS fromfractionated soybean extracts, four systems (1 to 4in Table 3) were selected. Systems in which GUSexhibited bottom phase preference (i.e. 9 and 10)were not considered due to the fact that additionalcontaminants, such as debris, tend to accumulate inthis phase together with proteins from the fractionatedsoybean extracts as seen in Table 2.

Recovery of GUS from fractionated soybeanextracts in ATPSThe potential recovery of GUS from fractionatedsoybean extract using ATPS was addressed usingPEG600–phosphate systems. In order to mimic thefeedstock derived from the potential production ofrecombinant GUS expressed in soybeans, artificialmixtures containing purified GUS and fractionatedextracts from soybeans were prepared and loadedinto the ATPS. The original soybean extract wasnot used in this study since it represents a verycomplex material that will challenge the efficiency androbustness of the extraction systems. Furthermore,the main reason for using fractionated extracts was tosimulate a process in which the target protein couldbe present in samples with different selective proteinprofiles. Isoelectric precipitation of spiked soybeanextract (data not shown) yields fractions of the total

290 J Chem Technol Biotechnol 83:286–293 (2008)DOI: 10.1002/jctb

Processing soybean extracts in aqueous two-phase systems

Table 3. Effect of system tie-line lengths and molecular weight of

PEG on the recovery of glucuronidase (GUS) in ATPS

SystemaTop phase

recovery (%)Bottom phaserecovery (%)

Interface recovery(%)

1 20±1 0.0 80±42 22±1 0.0 78±33 23±1 0.0 77±34 28±1 0.0 72±3

5 23±1 0.0 77±36 15±0.6 0.0 85±47 13±0.6 0.0 87±48 12±0.5 0.0 88±4

9 0.0 38±1.8 62±310 0.0 30±1.4 70±311 0.0 5.0±0.2 95±412 0.0 0.0 100.0

13 0.0 20±1 80±414 0.0 6±0.3 94±415 2±0.1 6±0.3 92±416 0.0 0.0 100.0

Compositions of the systems 1–16 are defined in Table 1. Theconcentration of the solution containing glucuronidase (30 059 Uml−1)

in all the ATPS was 10% w/w. The top and bottom phase proteinrecovery is expressed relative to the initial amount of GUS loadedinto the systems. Interface recovery was estimated as the necessaryamount of protein to complete the mass balance.

extract where GUS is widely distributed from pH3.5 to 6.5, the fraction precipitated at 4.8 being theone with the highest amount of GUS. The practicalrecovery and purification factor of GUS from ATPSloaded with 7S fractionated soybean extract is depictedin Table 4. A moderate top phase recovery for GUSwas obtained (30 to 33% in Table 4), which wasconsistent with the results derived when ATPS wereloaded with purified GUS (20 to 28%, Table 3).However, the purity of GUS from the artificial mixture(in the presence of 7S fractionated soybean extract)increased by up to three times after ATPS. Theincrease in purity can be explained by the poor topphase preference of the proteins from the 7S soybeanextract (less than 11%; Table 2). Potential recovery ofGUS from the bottom phase was not evaluated sinceGUS did not exhibit any preference for this particularphase and the contaminant proteins concentrated in

the bottom phase and interface. Alternatively, thepotential recovery of GUS from the interface can resultin a product recovery of up to 67% (system 3 and 4in Table 4). However, in this case the purity of GUSdid not benefit greatly (purification factor of 1.3–1.5)due to the accumulation of contaminant proteins atthe interface. Furthermore, the potential recovery ofGUS from the interface may be questionable sincethe activity of the enzyme needs to be evaluated.Additional solubilization stages may need to beconsidered. PEG 600/phosphate comprising 15.8%(w/w) PEG, 19.5% (w/w) phosphate, a volume ratioVr equal to 1.0, a system pH of 7.0 and TLL of 41.5%loaded with GUS and 7S fractionated soybean extractachieved a recovery of GUS from the top phase of 33%and 67% from the interface with an increase in purityof 3.0 and 1.5 times, respectively. These results suggestthe need for an alternative strategy. Furthermore, 7Sfractionated soybean extract is produced after twoprecipitation steps (see Experimental section) whichnecessarily increases the number of unit operationsneeded in the process. In contrast 11S fractionatedsoybean extract resulted from one single precipitationstep at pH 6.4. Thus it was decided to evaluatethe potential recovery of GUS from this type offractionated extract. Table 5 illustrates the practicalrecovery and purification factor of GUS from ATPSloaded with 11S fractionated soybean extract. In thisparticular case, the processing of the mixture of GUSand 11S fractionated soybean extract resulted in amoderate top phase recovery of GUS (i.e. 15% to28%) with maximum purification factor of 3.0, similarto that obtained from the processing of 7S fractionatedsoybean extract with added GUS. However, thepotential recovery of GUS from the interface was82% (system 1 in Table 5) and the increase in productpurity was 4.5 times after ATPS. Such results canbe explained by the fact that, for this particularATPS, proteins from 11S fractionated soybean extractexhibited a stronger preference (compare to that of theproteins from the 7S extract) for the bottom phase anda moderate preference for the interface, 67% and 18%,respectively (Table 2). However, it is important toconsider that the nature of the target product recoveredfrom the interface needs to be evaluated and additionalre-solubilization stages may be required.

Table 4. Recovery and purification factor of glucuronidase (GUS) from ATPS loaded with 7S fractionated soybean extract

Systems

Top phaseGUS

recovery (%)

Top phase GUSpurification

factor

Bottomphase GUSrecovery (%)

Bottom phase GUSpurification

factor

InterfaceGUS

recovery (%)

Interface GUSpurification

factor

1 30±1.5 2.5 8±0.4 0.20 62±3 1.42 30±1.5 2.5 4±0.2 0.08 66±3 1.53 33±1.5 3.0 0.0 0.0 67±3 1.54 33±1.5 3.0 0.0 0.0 67±3 1.3

Compositions of the systems 1–4 are defined in Table 1. The concentration of protein extract with glucuronidase (GUS) in all ATPS was 10% w/w(12 188 U per system). The top and bottom phase GUS recovery is expressed relative to the initial amount of protein loaded into the systems.Interface recovery was estimated as the necessary amount of proteins to complete the mass balance. Purification factor is the ratio between therelative purity of GUS after and before ATPS extraction. The relative purity of GUS in each phase is estimated as the amount of GUS relative to thatof the proteins from the soybean extract.

J Chem Technol Biotechnol 83:286–293 (2008) 291DOI: 10.1002/jctb

O Aguilar, M Rito-Palomares

Table 5. Recovery and purification factor of glucuronidase (GUS) from ATPS loaded with 11S fractionated soybean extract

System

Top phaseGUS

recovery (%)

Top phase GUSpurification

factor

Bottomphase GUSrecovery (%)

Bottom phaseGUS purification

factor

InterfaceGUS

recovery (%)

Interface GUSpurification

factor

1 15±0.7 1.0 3±0.1 0.04 82±4 4.52 20±1 2.1 4±0.2 0.06 76±3 2.63 23±1 2.2 0.0 0.0 77±3 2.64 28±1 3.0 0.0 0.0 72±3 1.6

Conditions as described in Table 4.

The general strategy proposed for the potentialrecovery of a GUS from fractionated soybean extractsis characterized by a two-stage process involving aprecipitation stage prior to ATPS extraction. A PEG600/phosphate system comprising 14.5% (w/w) PEG,17.5% (w/w) phosphate, a volume ratio Vr equalto 1.0, a system pH of 7.0 and TLL of 32% w/wresulted in recovery of 82% of the GUS protein fromthe interface. This strategy can result in a primaryrecovery process that increases protein purity up to4.5 times with respect to the purity of the initial crudeextract loaded to the ATPS. Potential processing ofthe interface of ATPS for the recovery of biologicalproducts has been addressed previously.23 However,particular conditions to process the interface for therecovery of GUS from soybean proteins need to beestablished using conventional techniques such asprecipitation or sequential ATPS extractions.

CONCLUSIONSThis paper reports a simplified strategy using aqueoustwo-phase systems to process fractionated soybeanextracts as a first step for the potential recovery ofrecombinant proteins, particularly GUS. It has beenshown that proteins from 7S and 11S fractionatedsoybean extracts concentrated at both the bottomphase and interface. Changes in phase preferencewere attributed to the nature of the proteins presentin the fractionated extracts and the effect of systemparameters on the partition behaviour of the proteins.The selected model protein, GUS, accumulatedpredominantly at the interface and exhibited a topand bottom phase preference in ATPS with low andhigh molecular weight PEG, respectively. Processingof the selected fractionated soybean extract with GUSby PEG600/phosphate ATPS resulted in an increasein the purity of the target protein. The operatingconditions established and derived from the proposedstrategy resulted in the potential recovery of GUSfrom the interface. These conditions concentrated thecontaminant proteins from the extracts preferentiallyto the top and bottom phase. Overall, the novelapproach proposed here represents a practical strategythat can simplify the way in which recombinantproteins expressed in plants can be recovered. Thisresearch described an approach that is necessary asa starting point to establish a practical protocol topotentially isolate and purify a recombinant protein

expressed in plants in general and in soybeans inparticular.

ACKNOWLEDGEMENTSThe authors wish to acknowledge the financial supportof Tecnologico de Monterrey, Biotechnology researchchair (Grant CAT005). The technical assistance ofIsabel Murillo is gratefully acknowledged.

REFERENCES1 Whitelam GC, Cockburn B, Gandecha AR and Owen MRL,

Heterologous protein production in transgenic plants.Biotechnol Gen Eng Rev 11:1–29 (1993).

2 Krebbers E, Bosch D and Vandekerckhove J, Prospects andprogress in the production of foreign proteins and peptidesin transgenic plants. in Plant Protein Engineering, ed. byShewry PR and Gutteridges S. Cambridge University Press,London, pp. 315–325 (1992).

3 Austin S, Bingham ET, Koegel RG, Matthews DE, Sha-han MN, Strab RJ, et al., An overview of a feasibility studyfor the production of industrial enzymes in transgenic alfalfa.Ann N Y Acad Sci 721:235–244 (1994).

4 Twyman RM, Stoger E, Schillberg S, Christou P and Fischer R,Molecular farming in plants: host systems and expressiontechnology. Trends Biotechnol 21:570–578 (2003).

5 Menkhaus TJ, Bai Y, Zhang C, Nikolov ZL and Glatz CE,Considerations for the recovery of recombinant proteins fromplants. Biotechnol Prog 20:1001–1014 (2004).

6 Azzoni AR, Kusnadi AR, Miranda EA and Nikolov ZL, Recom-binant aprotinin produced in transgenic corn seed: extrac-tion and purification studies. Biotechnol Bioeng 80:268–276(2002).

7 Gu Z and Glatz CE, Aqueous two-phase extraction for proteinrecovery from corn extracts. J Chromatogr B 845:38–50(2007).

8 Bai Y and Nikolov ZL, Effect of processing on the recovery ofrecombinant β-glucuronidase (rGUS) from transgenic canola.Biotechnol Prog 17:168–174 (2001).

9 Zhang C and Glatz CE, Process engineering strategy forrecombinant protein recovery from canola by cation exchangechromatography. Biotechnol Prog 15:12–18 (1999).

10 Zhang C, Love RT, Jilka JM and Glatz CE, Genetic engineeringstrategies for purification of recombinant proteins fromcanola by anion exchange chromatography: an example ofβ-glucuronidase. Biotechnol Prog 17:161–167 (2001).

11 Shyu DJH, Chou W-M, Yiu T-J, Lin CP and Tzen JTC,Cloning, functional expression, and characterization ofcystatin in sesame seed. J Agric Food Chem 52:1350–1356(2004).

12 Zeitlin L, Olmsted SS, Moench TR, Co MS, Martinell BJ,Paradkar VM, et al., A humanized monoclonal antibodyproduced in transgenic plants for immunoprotection of thevagina against genital herpes. Nature Biotechnol 16:1361–1364(1998).

292 J Chem Technol Biotechnol 83:286–293 (2008)DOI: 10.1002/jctb

Processing soybean extracts in aqueous two-phase systems

13 Thanh VH and Shibasaki K, Major proteins of soybean seeds.A straightforward fractionation and characterization. J AgricFood Chem 24:1117–1121 (1976).

14 Friedman M and Brandon DL, Nutritional and health benefitsof soy proteins. J Agric Food Chem 49:1069–1086 (2001).

15 Natarajan SS, Xu C, Bae H, Caperna TJ and Garret WM,Characterization of storage proteins in wild (Glycine soja)and cultivated (Glycine max) soybean seeds using proteomicanalysis. J Agric Food Chem 54:3114–3120 (2006).

16 Golubovic M, van Hateren SH, Ottens M, Witkamp G-J andvan der Wielen LAM, Novel method for the productionof pure glycinin from soybeans. J Agric Food Chem53:5265–5269 (2005).

17 Lakemond CMM, De Jongh HHJ, Gruppen H and Vora-gen AGJ, Differences in denaturation of genetic variants ofsoy glycinin. J Agric Food Chem 50:4275–4281 (2002).

18 Wu S, Murphy PA, Johnson LA, Fratzke MA and Reu-ber J, Pilot-plant fractionation of soybean glycinin and β-conglycinin. J Am Oil Chem Soc 76:285–293 (1999).

19 Rito-Palomares M, Practical application of aqueous two-phasepartition to process development for the recovery of biologicalproducts. J Chromatogr B 807:3–11 (2004).

20 Rito-Palomares M, Dale C and Lyddiatt A, Generic applicationof an aqueous two-phase process for protein recovery fromanimal blood. Process Biochem 35:665–673 (2000).

21 Rito-Palomares M and Lyddiatt A, Practical implementation ofaqueous two-phase processes for protein recovery from yeast.J Chem Technol Biotechnol 75:632–638 (2000).

22 He C, Li S, Liu H, Li K and Liu F, Extraction of testosteroneand epitestosterone in human urine using aqueous two-phasesystems of ionic liquid and salt. J Chromatogr A 1082:143–149(2005).

23 Benavides J, Mena J, Cisneros M, Ramirez OT, Palomares LAand Rito-Palomares M, Rotavirus like-particles primaryrecovery from the insect cell in aqueous two-phase systems. JChromatogr B 842:48–57 (2006).

24 Aguilar O, Albiter V, Serrano-Carreon L and Rito-Palomares M, Direct comparison between ion-exchange chro-matography and aqueous two-phase processes for the partialpurification of penicillin acylase produced by E. coli. J Chro-matogr B 835:77–83 (2006).

25 Platis D and Labrou NE, Development of an aqueous two-phasepartitioning system for fractionating therapeutic proteins fromtobacco extract. J Chromatogr A 1128:114–124 (2006).

26 Menkhaus TJ, Pate C, Krech A and Glatz CE, Recombinantprotein purification from pea. Biotechnol Bioeng 86:108–114(2004).

27 Feurtado JA, Banik M and Bewley JD, The cloning andcharacterization of α-galactosidase present during andfollowing germination of tomato (Lycopersicon esculentumMill.) seed. J Exp Botany 52:1239–1249 (2001).

28 Bradford MM, A rapid and sensitive method for the quantifica-tion of microgram quantities of protein utilizing the principleof protein–dye binding. Anal Biochem 72:248–254 (1976).

29 Benavides J and Rito-Palomares M, Potential aqueous two-phase processes for the primary recovery of colored proteinsfrom microbial origin. Eng Life Sci 5:259–266 (2005).

30 Rito-Palomares M and Hernandez M, Influence of system andprocess parameters on partitioning of cheese whey proteinsin aqueous two-phase systems. J Chromatogr B 711:81–90(1998).

31 Rito-Palomares M and Lyddiatt A, Process Integration usingaqueous two-phase partition for the recovery of intracellularproteins. Chem Eng J 87:313–319 (2002).

J Chem Technol Biotechnol 83:286–293 (2008) 293DOI: 10.1002/jctb