Characterization of green-tissue protein extract from alfalfa (Medicago sativa) exploiting a 3-D...

J. Sep. Sci. 2009, 32, 3223 –3231 O. Aguilar et al. 3223

Oscar Aguilar1

Charles E. Glatz2

Marco Rito-Palomares1

1Departamento de Biotecnolog�ae Ingenier�a de Alimentos,Centro de Biotecnolog�a,Tecnol�gico de Monterrey,Campus Monterrey, Monterrey,NL, M�xico

2Department of Chemical andBiological Engineering, IowaState University, Ames, IA, USA

Original Paper

Characterization of green-tissue protein extractfrom alfalfa (Medicago sativa) exploiting a 3-Dtechnique

There is a growing interest of pharmaceutical companies for plant-based productionsystems. To facilitate the general acceptance of plants as bioreactors, the establish-ment of efficient downstream operations is critical. It has been proposed that a bet-ter understanding of the properties of the contaminant proteins can benefit down-stream processing design and operation. The coupled application of 2-DE with aque-ous two-phase partitioning has been suggested as a practical 3-D method to charac-terize potential contaminant proteins from plant extracts. The application of thisnovel 3-D approach to a complex protein extract from alfalfa (Medicago sativa) contain-ing a model recombinant protein (human granulocyte colony stimulating factor(hG-CSF)) resulted in the quantification of 55 protein spots. The 3-D properties (Mr, pI,and Kp) obtained for 17 proteins comprising 69% of the alfalfa proteins, allowed theproposal of a prefractionation step as well as the identification of the target mole-cule (rG-CSF) from bulk of alfalfa proteins. The information obtained from this exper-imental approach was useful for the identification of the potential contaminant pro-teins that will occur in alfalfa when this plant is used as a host for recombinant pro-teins. Additionally, this method will assist in the design of adequate purificationstrategies for recombinant proteins expressed in alfalfa green tissue.

Keywords: Alfalfa protein / Aqueous two-phase systems / 2D-electrophoresis / G-CSF / Proteomics /

Received: March 24, 2009; revised: May 14, 2009; accepted: May 15, 2009

DOI 10.1002/jssc.200900184

1 Introduction

A wide number of pharmaceutical proteins have beenproduced in a variety of plant species (including tobacco,potato, rice, soybean alfalfa, tomato, and lettuce) reflect-ing the interest of biotechnology companies to benefitfrom the advantages of plant-based production systems[1, 2]. During the design of a recombinant protein pro-duction process, selection of the most adequate expres-sion system as well as an efficient extraction and purifi-cation strategy to maximize recovery of target protein,represent the major aspects to be considered. Down-stream processing costs typically contribute 80% of thetotal. Therefore, efficient and robust processing strat-

egies are essential [3]. In this context, the use of aqueoustwo-phase systems (ATPSs)-based strategies have resultedin the establishment of protocols for the recovery andpurification of biological compounds [4–7]. ATPSs havealso been used for the understanding of chemical proper-ties and behavior of proteins in solution [8].

It is clear that a better understanding of the propertiesof the contaminant proteins can benefit downstreamprocessing design and operation [9, 10]. Proteomic toolslike MS and 2-DE have become common techniques toaccurately detect and examine protein compositionfrom a variety of plant hosts. These techniques provideuseful information on the molecular properties of com-plex mixtures that can be exploited for the optimizationand better design of downstream strategies [11].

A 3-D technique for the molecular characterization ofcorn germ protein extracts was recently reported by Guand Glatz [9]. It was based on the coupled application ofaqueous two-phase partitioning to measure hydropho-bicity in terms of the partition coefficient of the proteins(Kp), and 2-DE to evaluate molecular weight (Mr) and pI ofindividual proteins [9]. The 3-D information obtained foreach protein (Mr, pI, and hydrophobicity) was used as a

Correspondence: Dr. Marco Rito-Palomares, Departamento deBiotecnolog�a e Ingenier�a de Alimentos, Centro de Biotecnolo-g�a, Tecnol�gico de Monterrey, Campus Monterrey, Ave. Euge-nio Garza Sada 2501 Sur, Monterrey, NL 64849, M�xicoE-mail: [email protected]: +52-81-8328-4136

Abbreviations: ATPS, aqueous two-phase system; LAC, a-lactalbu-min; LOD, limit of detection; LYS, lysozyme; rG-CSF, recombinantgranulocyte colony stimulating factor; RNA, ribonuclease A

i 2009 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.jss-journal.com

3224 O. Aguilar et al. J. Sep. Sci. 2009, 32, 3223 – 3231

basis for better visualization of the molecular propertiesof the host proteins from which a recombinant proteinmust be separated. However, in order to establish thegeneric application of this experimental approach, alter-native hosts need to be considered.

The aim of this work was to extend the application ofthis novel strategy to a new plant protein extract. Theapplication of the 3-D technique for the characterizationof alfalfa protein extracts containing an artificiallyadded recombinant model protein was evaluated.Human recombinant granulocyte colony stimulatingfactor (rG-CSF) was selected as an example of a recombi-nant product that could be readily produced in alfalfacells at low quantities. Colony-stimulating factors are gly-coproteins which act on hematopoietic cells by bindingto specific cell surface receptors and stimulating prolifer-ation, differentiation commitment, and some end-cellfunctional activation [12]. Previous reports have demon-strated the feasibility of recombinant human cytokineproduction in genetically modified plant cells [13, 14].One of the most relevant characteristics of proteins pro-duced in plants is the possibility of production of glycosy-lated forms of a recombinant protein. Any improvementin the function or life of the drug will have a remarkableimpact for patients with cancer or bone marrow trans-plantation [15]. The possibility of producing glycoformsof G-CSF in a commercially viable plant system has anenormous potential, considering the increase in the bio-logical activity of the molecule and the reduction in theproduction costs. However, such an experimental modelrepresents a real challenge for downstream processinggiven the high concentration of contaminant proteinsthat would be present.

In this research, several ATPS parameters were eval-uated to identify a system where partitioning is domi-nated by protein hydrophobicity, but that is also compat-ible with a complex green-tissue extract. By satisfyingthese criteria, a 3-D characterization technique of green-tissue proteins could be obtained by the coupled applica-tion of ATPS and 2-DE. Addition of a selected targeted pro-tein to the crude extract illustrates the generic applica-tion of this experimental approach for the recovery of arecombinant protein from plant extracts.

2 Materials and methods

2.1 Chemicals and reagents

PEG 3350, b-mercaptoethanol, DL-DTT, Tris, Bradfordreagent, and the selected model proteins: ribonuclease A(RNA), a-lactalbumin (LAC), BSA, and lysozyme (LYS) werepurchased from Sigma–Aldrich Chemicals (St Louis, MO,USA). The ReadyPrepm rehydration buffer, 11 cm Ready-Stripm IPG strips (pH 3–10), iodoacetamide, and PrecisionPlusm protein standard plugs were purchased from Bio-

Rad (Hercules, CA, USA). PMSF was purchased from Boeh-ringer Mannheim (Germany). Commercial rG-CSF (Biofil-granm) was obtained from BioSidus S.A. (Buenos Aires,Argentina). All the other chemicals were purchased fromSigma–Aldrich.

2.2 Plant materials

Commercial alfalfa seeds (Medicago sativa), kindly pro-vided by the Agricultural Experimental Station of Tecno-l�gico de Monterrey were field grown during 4 wk. Aerialparts (first stem and leaves) were harvested before flower-ing and immediately ground in liquid nitrogen withmortar and pestle, adding powdered glass to improvecell wall breaking. Powder stocks were stored at –868Cfor further use.

2.3 Alfalfa green-tissue protein extraction

Three different protocols (listed in Table 1) were eval-uated for protein extraction from powdered alfalfa greentissue at a proportion of 1.0 g solids/10 mL extraction buf-fer. Protocol A: green biomass was suspended in sodiumphosphate buffer (buffer A) [6] and stirred for 1 h withconstant pH monitoring, centrifuged (100006g, 30 min,room temperature; Centrifuge 5804R, Eppendorf, Ham-burg, Germany) and decanted to eliminate waste solids.The supernatant was filtered using 0.45 lm syringe filter(Corning, USA) and used immediately. Protocol B: pro-


Table 1. Efficiency of alfalfa protein extraction with threedifferent buffersa)

Extraction buffer Extracted protein,mg/g fresh alfalfab)

Protocol A20 mM Sodium phosphate 15.4 l 0.510 mM EDTApH 7

Protocol B50 mM Tris N HCl 14.6 l 110 mM MgSO4

0.1% b-Mercaptoethanol2 mM PhenylmethylsulfonylfluoridepH 7.8

Protocol C0.45 M Tris N HCl 27.1 l 20.45 M H3BO3

10 mM EDTApH 8

a) Alfalfa green-tissue ground on liquid N2 and extractedfor 1 h at 258C with constant pH monitoring with a 1:10plant material/buffer proportion.

b) Average of triplicate experiments. Protein concentrationmeasured by Bradford reaction.

J. Sep. Sci. 2009, 32, 3223 –3231 Other Techniques 3225

tein extraction with MgSO4-based buffer (buffer B), previ-ously reported for the preparation of Arabidopsis proteincrude extracts [16] was adapted for alfalfa green tissue.The slurry was stirred for 1 h and centrifuged at160006g for 30 min at room temperature and the super-natant filtered using 0.45 lm syringe filter. Protocol C:Tris-borate-EDTA (TBE) extraction buffer (buffer C) wasevaluated using the same biomass/buffer ratio. Theslurry was stirred for 1 h, and then centrifuged at120006g for 10 min at room temperature. The superna-tant was filtered using 0.45 lm syringe filter [17].

An artificial mixture of alfalfa protein containinghuman rG-CSF was prepared by adding an aliquot of thepurified human cytokine to the selected alfalfa proteinextract (derived from protocol A, B, or C) to have a finalconcentration of 63 lg of rG-CSF/mL extract. This concen-tration of protein was found to be within the range ofpreviously reported levels of this cytokine in plant cells[14].

Total protein determination for alfalfa extracts, phasesamples, and TCA precipitates was made using micro-plate Bradford reaction with BSA as standard [18]. Con-centration of model proteins was measured spectropho-tometrically at 280 nm using a microplate reader (Syn-ergy HT, BioTek Instruments, Vermont, USA). All proteindeterminations included calibration curves using propersolvents and blank ATPS for correction of any interfer-ence from phase-forming components.

2.4 Aqueous two-phase partitioning

ATPS were formulated according to the correspondingbinodal curves reported by Zaslavsky [19] and based onthe systems and methodology reported by Gu and Glatz[9] and Aguilar and Rito-Palomares [20] to give a fixedweight of 2.0 g for partitioning of alfalfa protein extractand for model proteins. PEG 3350–potassium phosphateand PEG 3350–sodium sulfate systems at pH 7 weretested in order to compare protein partitioning amongthem. Predetermined quantities of stock solutions ofPEG 3350, and sodium sulfate or potassium phosphatewere mixed with solid NaCl and protein sample to givethe following total composition: 15.7% w/w PEG 3350,8.9% w/w Na2SO4, 0–9% w/w NaCl for sulfate –ATPS and14.8% w/w PEG 3350, 10.3% w/w potassium phosphate0–9% w/w NaCl for phosphate–ATPS. NaCl effect onmodel protein partitioning was evaluated from 0 to 9%w/w with the same protein load. Sample was addedaccordingly to have 1.0 mg alfalfa protein per gram ofATPS. Partitioning of individual model proteins (LYS,BSA, LAC, and RNA) in the PEG –phosphate system wasperformed using the same concentration of each proteinper gram of ATPS (1.0 mg/g). All partitioning experimentswith alfalfa protein extracts and selected model proteinswere run in triplicate.

2.5 Sample preparation

To eliminate interferences from phase-forming com-pounds and increase protein concentration, TCA precipi-tation was performed on the top and bottom samplesbefore IEF according to the protocol reported by Gu andGlatz (2007) [9]. After precipitation, the protein pelletwas completely redissolved using 210 lL of rehydrationbuffer (8 M urea, 2% w/v CHAPS, 50 mM DTT, 0.2% v/vBioLyte from BioRad) and used for first-dimension IEF.Protein concentration was measured at this step dilutingsamples 1:3 to avoid highly concentrated urea interfer-ence with Bradford reagent. In the cases where proteinrecovery from precipitation was low, multiple replicateswere pooled at this step and considered for final proteinquantitation. All calibration equations for protein meas-urement were obtained using the proper solvents anddilutions for the correction of any interference.

2.6 2-DE

The first-dimension IEF was performed using 11.0 cm pH3–10 linear IPG strips (ReadyStrip, BioRad) in an EttanIPGphor3 apparatus (GE Healthcare). Strips were rehy-drated using 200 lL of sample to a maximum of 200 lg ofprotein per strip during 16 h at room temperature. IEFwas carried out for a total of 50 250 Vh. For the seconddimension, the focused IPG strips were equilibrated with6 M urea, pH 8.8, 75 mM Tris HCl, 2% w/v SDS, 29.3% v/vglycerol, 0.002% w/v bromophenol blue and 2% w/v DTTfor 15 min, and then acetylated for another 15 min usingthe same solution except replacing DTT with 2.5% w/viodoacetamide. Strips were placed onto 12.5% w/v linearpolyacrylamide gels prepared as described by Laemmli(1970) and electrophoresis was performed using a SE600Ruby electrophoresis unit (GE Healthcare) [21]. The gelswere visualized by staining with Coomassie Blue G-250,and scanned at 600 dpi resolution using a flat bed scan-ner in transmissive mode (Hewlett-Packard). Spot densi-tometry (area multiplied by the pixel intensity) was per-formed using PDQuest software (BioRad). The mass ofprotein for individual spots was calculated from the spotvolume relative to the total amount of protein loaded tothe gel. Protein concentrations from spot densities wereused to calculate partition coefficients (Kp, the ratiobetween top and bottom concentrations) of individualproteins. All the experiments were duplicated.

2.7 Protein assay

Total protein determination for alfalfa extracts, phasesamples, and TCA precipitates was made by using micro-plate Bradford reaction with BSA as standard (SynergyHT, BioTek Instruments) [18]. Concentration of modelproteins was measured at 280 nm in microplate reader.



All protein determinations included calibration curvesusing proper solvents and blank ATPS for correction ofany interference from phase-forming components.

3 Results and discussions

3.1 Protein extraction

In order to minimize protein degradation during grind-ing steps, liquid nitrogen was used to freeze–dry thestems and leaves immediately after harvesting. Threedifferent extraction methods were tested for the extrac-tion of alfalfa proteins from green tissue, the results areshown in Table 1. To avoid proteolysis during extractionat room temperature, all the extraction buffers includeda protease inhibitor in the formulation, either EDTA forsequestering metal ions needed for metalloproteasesactivity, or PMSF reported to be an irreversible inhibitorfor serine and cysteine proteases [22]. The amount of pro-tein extracted per gram of fresh alfalfa was found to besimilar between MgSO4- and sodium phosphate-basedbuffers, around 50% of the total protein reported foralfalfa leaves (30 mg/g fresh weight) [3]. However, TBE buf-fer dissolved nearly 90% of leaf proteins (27.1 g protein/gfresh alfalfa). Protocol C was used in subsequent parti-tion experiments and 3-D analysis, due to the betterresults obtained.

3.2 Aqueous two-phase partitioning experiments

Partitioning experiments performed using the sameATPS compositions previously reported by Gu and Glatz[5] for hydrophobic partitioning resulted in low partitioncoefficients (i.e., the extracted proteins are relativelyhydrophilic), as can be seen in Table 2 for the 3% NaCl sys-tems. As a result of this strongly biased partitioning, theresolving power for hydrophobicity differences for these

host proteins would be low. Total recovery of soluble pro-teins (sum of top and bottom phase recoveries) was alsolow (37%) because of accumulation of precipitated pro-teins at the interface.

The criteria to evaluate the choice of a particular ATPScomposition for 3-D characterization are defined by: (i) apartition coefficient (Kp) closest to 1.0, in order to obtainthe highest number of proteins in both phases; (ii) thehighest % of recovery, to keep most of the proteins solu-ble and not at the interface; and (iii) keep an intrinsichydrophobicity difference between phases to allow accu-rate hydrophobicity measurements.

In an attempt to better meet all criteria for the selec-tion of a particular ATPS for the 3-D characterization,ATPS with different NaCl concentrations were used.Table 2 shows that increasing NaCl from 0 to 3% in bothPEG 3350 –phosphate and PEG 3350–Na2SO4 systemscauses a decrease in Kp for total alfalfa protein. In the par-ticular case of PEG 3350 –phosphate systems, higherNaCl content A3% w/w caused an increase in Kp values.However, such effect on Kp can be explained with thelarge differences observed in recovery percentagesbetween top and bottom phases. The main observedeffect of NaCl was on the amount of protein that precipi-tates at the interface more than a significant shift on theprotein preference between the phases.

Although it has been reported that the addition ofNaCl to ATPS can shift protein partitioning, in this case,the objective was to improve alfalfa protein distributionbetween the two phases to yield a higher number of pro-tein spots in the gels. However, the addition of NaCl tothe ATPS containing alfalfa protein did not result in a sig-nificant improvement of protein distribution betweenthe phases or total protein recovery from the systems.For PEG 3350–phosphate system, the highest proteinrecovery was observed with 0% NaCl added. Although, inthe case of PEG 3350 –sulfate system, addition of 3% NaCl


Table 2. Effect of NaCl addition on the % recovery and Kp of alfalfa proteins in two different ATPSsa)

Systemb) w/w NaCl content w/w

0% 1.5% 3.0% 6.0% 9.0%

TLL 30% Top phase recovery (%) 8.1 l 0.9 0.8 l 0.6 2.9 l 0.9 5.6 l 3.6 4.6 l 2.7PEG 3350 14.8% Bottom phase recovery (%) 50.2 l 0.4 39.4 l 0.1 34.4 l 1.9 15.3 l 1.1 1.4 l 0.5Phosphate 10.3% Overall recovery 58.3 l 1.0 40.2 l 0.6 37.3 l 2.0 20.9 l 3.6 6.0 l 3.0

Kp 0.14 l 0.02 0.04 l 0.02 0.03 l 0.01 0.49 l 0.3 5.06 l 2.0TLL 32% Top phase recovery (%) 13.4 l 2.4 6.2 l 1.3 1.5 l 0.9 0.0 1.2 l 0.02PEG 3350 15.7% Bottom phase recovery (%) 16.0 l 2.4 15.2 l 1.3 35.9 l 0.9 12.1 l 1.9 8.2 l 2.8Na2SO4 8.9% Overall recovery 29.4 l 2.4 21.4 l 1.3 37.4 l 0.9 12.1 l 1.9 9.4 l 3.0

Kp 0.99 l 0.3 0.79 l 0.1 0.05 l 0.001 N.D. topc) 0.25 l 0.1

a) Partition coefficient (Kp) and % recovery data expressed as the average of triplicate experiments at 258C. % Interface precip-itation estimated to be the complement to meet 100% of the loaded protein.

b) pH of the systems and protein samples was previously adjusted to 7.0. Load of alfalfa protein was 1.0 mg protein/g ATPS.c) N.D. top, no protein detected in top phase.


resulted in highest total recovery, only 1.5% protein wasrecovered at the top phase. For both systems, the absenceof NaCl in the system resulted in better protein distribu-tion between the two phases.

The effect of NaCl added to ATPS has been documentedbefore [8, 23–25]. Such studies indicate that while someproteins are dramatically affected by high NaCl concen-trations, some others can remain unaffected. In the caseof complex protein mixtures, such as alfalfa extracts,some of the changes on Kp and top/bottom recoveriesobserved in Table 2 can be explained in terms of changesin the solubility of individual proteins that precipitate atthe interface and the different effect of NaCl on thehydrophobicity of the system depending on the type ofthe salt used [26].

Of the two hydrophobic systems tested, the PEG 3350 –potassium phosphate system with no added NaCl gavethe higher Kp (0.14) and highest % of protein recovery(58.3%). However, validation of hydrophobicity as theprinciple for partitioning in the absence of NaCl was nec-essary. The PEG 3350 –phosphate system with no NaClwas selected for the further validation of system hydro-phobicity to fulfill the previously defined criteria for theselection of the most adequate ATPS for partitioning.

3.3 Partition of model proteins in ATPS

The use of two-phase partitioning to estimate hydropho-bicity of proteins has been addressed before [8, 9, 26, 27].To test the ability of the particular ATPS selected to pro-vide a hydrophobicity measurement, four model pro-teins (LYS, BSA, RNA, and LAC) with known hydrophobic-ity values in terms of (NH4)2SO4 solubility. The parameter1/m* was reported by Hachem et al. [8] as a hydrophobicitymeasure. The m* parameter is the salt concentration atwhich a given protein starts to precipitate (given an ini-tial fixed concentration). The high linear correlation fac-tor obtained (see Fig. 1) between Kp measured in ATPSand the parameter 1/m* for the system composed by PEG3350 (14.8%) potassium phosphate (10.3%) without NaCladded, shows that it is a reliable method to measure thefunctional hydrophobicity of proteins. Some factors likesurface charge could also exert an influence on partition-ing besides surface hydrophobicity, however given thedifferent pI values for the model proteins used (BSA-5.6,LAC-4.8, RNA-9.6, and LYS-10.3) and the order of thehydrophobicity scale obtained, a significant chargeeffect would influence the logKp values, specially for theBSA–LAC order in the scale. No significant correlationwas found between partitioning of the four model pro-teins and the surface charge of the molecule in the PEG–phosphate system (Fig. 2). As previously reported [8],small changes in NaCl content could have different effecton Kp for different proteins making necessary a case by

case validation of the system hydrophobicity along withevery change on the system composition.

3.4 Application of 3-D method to green-tissuealfalfa extracts containing a recombinantprotein

To simulate a protein extract containing a recombinantprotein product human rG-CSF was combined with thealfalfa extract and loaded into the ATPS. This cytokinehas been produced in a variety of plant cells (i.e., tobacco,and tomato) concentrations ranging from 0.1 to 0.5% oftotal soluble protein (TSP) [13, 14], but was used here at alevel of 2% of TSP (20 lg of rG-CSF/mg alfalfa extract pro-tein). The level used has been reached for other recombi-nant proteins and is high enough for accurate identifica-tion in the characterization process.

After ATPS partitioning, 2-D gels from top and bottomphases were run under the same electrophoretic condi-tions, resulting in 55 protein spots (Table 3) over a widerange of molecular weight and pI, as can be seen in Fig. 3.Besides the quantified protein loss at the interface ofATPS (l40%), removal of phase-forming componentsusing TCA precipitation resulted in additional proteinlosses that were also quantified for each phase and wereconsidered for the calculation of yields for individualproteins. Approximately 30% of proteins were lost dur-ing top phase TCA precipitation, while for bottom phase43% of the proteins were lost. Only 17 spots detected on2-D gels provided matches between top and bottomphases. The number of protein matches is also limited bythe number of proteins detected over the LOD of the gels.


Figure 1. Correlation between surface hydrophobicitymeasured as logKp in ATPS with 1/m* parameter. Kp is thepartitioning coefficient of protein in PEG 3350 14.8% w/w,potassium phosphate 10.3% w/w system at pH 7, 258C andprotein loading of 1.0 mg/g ATPS for each protein, and m* isthe concentration of salt at the discontinuity point of the pro-tein in (NH4)2SO4 precipitation curve, reported by Hachem etal. [8] for selected model proteins using an initial concentra-tion of 2.0 mg/mL.


Besides this low number of matches, the overall partitioncoefficient estimated by spot densitometry was 0.13 l0.01 (the ratio of the total protein concentration (ppm) ofthe top and bottom phase from Table 3), and statisticallyequal to that measured by the Bradford method (0.14 l0.02; Table 2). Although these results demonstrate thatboth methods (2-DE gels and Bradford) can be readilyused to estimate Kp obtaining similar values, it is impor-tant to consider that a considerable amount of proteinswere not solubilized by the ATPS and remained at theinterface (l42%). The final result is a limited protein pro-file where only the proteins that can be readily dissolvedin both phases can be characterized by the 3-D tech-nique.

Despite the loss of proteins at the interface, the exten-sion of this 3-D strategy to the green-tissue proteins fromalfalfa resulted in molecular characterization of 17 pro-teins listed in Table 4. These protein spots were thosedetected in both top and bottom phases and theytogether comprised 69% of the total proteins detected on2-D gels. This means that 31% of the proteins showed


Figure 2. Correlation plots obtained from tridimensionalproperties of alfalfa green-tissue proteins (g) and selectedmodel proteins (f). (a) logKp versus pI, (b) logKp versus Mr.

Table 3. Total protein spots detected from 2-D gels afterATP partitioning of alfalfa protein extracts containing ahuman recombinant cytokine

Spotno.

Molecularweight(kDa)

pI Protein concentration(ppm)

Recoveredprotein,%a)

Top phase Bottom phase

1 17.71 3 1.82 2.81 0.32*2 54.74 3.1 – 25.40 1.783 57.68 3.15 – 2.43 0.174 57.96 3.2 – 45.96 3.225 22.83 3.2 0.37 – 0.036 18.43 3.27 0.32 0.028 25.02 3.6 1.85 – 0.139 25.12 3.81 3.52 – 0.25

10 13.36 3.95 2.61 9.11 0.82*11 26.54 3.97 16.96 5.78 1.59*12 16.36 4.05 0.11 – 0.0113 16.89 4.11 2.85 – 0.2014 29.1 4.18 1.31 – 0.0915 34.66 4.23 10.86 – 0.7616 18.76 4.3 2.69 3.02 0.40*17 16.81 4.49 13.90 – 0.9718 50.05 4.84 0.18 0.35 0.04*19 16.59 4.98 33.15 – 2.3220 49.88 5.03 0.92 – 0.0621 36.52 5.08 4.40 2.89 0.51*22 26.1 5.13 0.26 9.59 0.69*23 34.2 5.29 12.42 – 0.8724 26.08 5.3 1.22 8.75 0.70*25 33.75 5.5 12.40 – 0.8726 50.3 5.5 0.96 – 0.0727 26.38 5.51 0.31 – 0.0228 27.18 5.53 0.69 4.06 0.33*29 33.79 5.54 4.52 – 0.3230 33.36 5.58 2.24 – 0.1631 13.96 5.61 1.79 – 0.1332 18.09 5.61 0.10 – 0.0133 72.4 5.62 0.09 0.54 0.04*34 15.12 5.63 6.61 57.99 4.53*a)

35 33.46 5.64 4.61 – 0.3236 49.9 5.7 9.50 166.93 12.36*b)

37 25.88 5.73 1.27 8.24 0.67*38 18.44 5.74 0.35 – 0.0239 14.86 5.84 6.89 227.82 16.45*b)

40 26.49 5.84 0.36 35.96 2.55*41 46.3 5.87 – 5.23 0.3742 50.26 5.93 0.82 96.16 6.80*b)

43 53.16 6.07 – 82.29 5.7744 55.98 6.26 – 4.33 0.3045 14.18 6.45 4.30 283.87 20.19*b)

46 63.38 6.96 – 3.39 0.2447 25.53 6.98 – 27.32 1.9148 54.52 6.98 – 14.13 0.9949 55.57 7.38 – 19.87 1.3950 58.43 7.54 – 3.99 0.2851 14.61 7.55 – 39.78 2.7952 25.49 7.71 – 35.65 2.5053 55.36 7.93 – 5.70 0.4054 54.67 9.75 – 5.49 0.3855 58.2 10.0 – 12.47 0.87

Total 169.48 1257.62 100.0

All data are the average of two experiments using two-phase system:14.8% PEG 3350, 10.3% potassium phosphate, pH 7 and 258C. Loadof protein was 1.0 mg/g ATPS. Protein concentration is expressed inparts per million considering the average protein loss during TCAsteps for each phase. Proteins marked with * were selected for 3-Dcharacterization for being present at both phases.a) The molecular properties of the protein marked with a) corre-

sponded to those reported for human granulocyte colony stimu-lating factor. Recovered protein % was calculated as the totalamount of a particular protein divided by the total amount ofalfalfa proteins quantified by densitometry.

b) The molecular properties of the protein marked with b) corre-sponded to those reported for small and large subunits ofRubisco.


exclusive affinity for one of the two phases or the inter-face, and Kp could not be calculated in these cases. Figure3 illustrates the scatter plot obtained from 3-D propertiesof proteins from Table 4. Except for a couple of spots,most of the proteins showed bottom phase preference,confirmed by the negative values for logKp axis. An addi-tional correlation analysis of the 3-D data (Fig. 2) showedno evidence of molecular weight distribution (Mr) on thepartition of proteins (logKp) between the phases with anR2 a 0.1. No evident correlation was found either betweenpI and partitioning, with an R2 a 0.5, supporting hydro-phobicity as the main driving force for partitioning inthe system.

It is likely that the majority of the contaminant pro-teins from the green-tissue extracts are those relatedwith the photosynthetic system. Approximately 49% ofall the proteins visualized with Coomassie staining afterATPS (see in Table 3 spots 36, 39, and 45) can be attrib-uted to large and small subunits of ribulose-1,5-bis phos-phate carboxylase (Rubisco, E.C. 4.1.1.39) [28]. Experimen-tal molecular weight and pI for Rubisco subunitsobtained from 2-D gels of spiked samples were used forpositive spot identification through similarity with thesame proteins reported for M. sativa in the Swiss-Protdatabase. Rubisco from alfalfa has a reported molecularweight of l48–52 kDa and pI values of 5–5.5 [29]. Thistetrameric enzyme (and its subunits) accounts for 30 –50% of total protein from plant tissues [6]. The relativelyhigh concentration of these photosynthetic proteinsdemonstrates how important these proteins are forplants. However, the prominence of Rubisco subunits inspecific regions of the gel generally contributes to lower

quality of the 2-D gels, and prevents detection of lower ormoderate abundance proteins due mainly to their lowerconcentration and the limited range of detection ofstaining techniques used. In an attempt to improve thedetection of low abundant proteins, the presence ofhighly abundant proteins, such as photosynthetic pro-teins, was not considered. However, the amount of pro-tein that can be added to an IEF strip is a limiting factorthat must be considered. Additional experiments (notshowed here) were performed with an overload of alfalfaprotein in IPG strips. This strategy was followed to detectlow abundant proteins masked by the presence ofRubisco and its subunits. Gel analysis evidenced thatalthough a few low abundant proteins can be betterdetected and quantified, the total number of spots on 2-D gels did not increased dramatically as expected.

It has been previously reported that Rubisco precipi-tates at the interface of high molecular weight PEG –phosphate systems [6]. Figure 3b shows multiple spotscorresponding to the predicted Mr and pI values ofRubisco subunits. The spot 36 (49.9 kDa – pI 5.7) observedin Table 3 can be identified as Rubisco large subunit par-titioned in ATPS and showed clear preference for the bot-tom phase. The couple of spots 39 (14.86 kDa–pI 5.84)and 45 (14.18 kDa–pI 6.45) positively matched the molec-ular properties of the small subunit of approximately 14kDa. Different isoforms and degrees of phosphorylationhave been reported that shift its pI value appearing asmultiple spots with similar Mr [30]. The cluster observedin gel images (spots 36 and 42), apparently higher inintensity than the rest of the spots, is a result of the highconcentration of Rubisco subunits and its isoforms. How-


Table 4. 3-D properties and content of selected alfalfa proteins partitioned in ATPS

ATPSa) Spot no. Mr (kDa) pI logKp Subset of proteins,%b)

1 17.7 3.0 –0.189 l 0.1 0.5PEG 3350 (14.8% w/w) 10 13.4 4.0 –0.527 l 0.2 1.2Potassium phosphate 11 26.5 4.0 0.467 l 0.1 2.3(10.3% w/w) 16 18.8 4.3 –0.055 l 0.1 0.6

22 26.1 5.1 –1.562 l 0.01 1.0TLL 30% 18 50.1 4.8 –0.298 l 0.1 0.1Vr = 1.25 28 27.2 5.5 –0.766 l 0.01 0.5

33 72.4 5.6 –0.766 l 0.01 0.1Total count: 55 spots 21 36.5 5.1 0.181 l 0.1 0.717 Spots contain 69% 34 15.1c) 5.6 –0.936 l 0.1 6.6of total protein 24 26.1 5.3 –0.858 l 0.01 1.0

36 49.9 5.7 –1.245 l 0.01 17.939 14.9 5.8 –1.520 l 0.01 23.837 25.9 5.7 –0.813 l 0.01 1.042 50.3 5.9 –2.070 l 0.01 9.940 26.5 5.8 –1.997 l 0.01 3.745 14.2 6.5 –1.820 l 0.01 29.3

a) All data are the average of duplicate experiments run at pH 7 and 258C. Protein load was 1.0 mg/g ATPS.b) Calculated as the amount of a particular protein divided by the summation of the spots included in this table.c) Molecular properties corresponding to human rG-CSF.


ever, software analysis revealed the presence of multiplespots and protein streaking (not quantified) contributingto this apparently dominant protein. The set of spots cor-responding to the photosynthetic enzymes togetheraccounted for 81% of the subset of proteins detected andquantified in both phases (Table 4).

Regarding the reported molecular properties of themodel protein added to alfalfa protein extract, thehuman rG-CSF, this cytokine could be traced down toonly one spot with 15.1 kDa and pI 5.6 (spot 34 in Table4). Despite the relatively low amount of this protein (20lg of rG-CSF/mg alfalfa soluble protein), the protein spotcan be identified in the 3-D plot as the rG-CSF initiallyadded to alfalfa extract. The results reported here evi-denced the importance of the application of a 3-D charac-terization technique to green-tissue protein extractsfrom aerial parts of alfalfa containing a model recombi-nant protein. It was found that proteins extracted fromalfalfa green tissues tend to accumulate at the interfaceof hydrophobic ATPS, resulting in a limited protein pro-file from the 3-D characterization technique. The maindrawback that has to be overcome for the generic appli-cation of this technique to green-tissue extracts is thepresence of the main potential contaminant, Rubisco.Removal or depletion of this protein needs to be done

without compromising detection of a target protein orother low abundant proteins [31].

4 Conclusions

A 3-D characterization method was applied to alfalfagreen-tissue proteins, providing information on themolecular properties of a large number of host proteins.The method provided a convenient 3-D plot for the mainalfalfa proteins, characterized by both Mr and pI from 2-DE and a third dimension namely hydrophobicity,obtained from ATPS. The ATPS composed of PEG 3350(14.8%) and potassium phosphate (10.3%) resulted in 69%of total proteins partitioned in both phases from which3-D properties were calculated. The presence of a modelrecombinant protein in the extract such as rG-CSF as wellas the dominance of photosynthetic enzymes challengedthe potential application of this technique to a recombi-nant extract where a relatively low abundant proteincould be present. The use of 3-D mapping for analyzingprotein profiles allowed the identification of the molecu-lar properties from the main contaminant proteins.Such information will facilitate the establishment of pre-fractionation and purification conditions to processgreen-tissue extracts.


Figure 3. 2-D gels and 3-D scatter plot of green-tissue proteins from alfalfa using PEG 3350–phosphate system. ATPS: PEG3350 14.8% w/w, potassium phosphate 10.3% w/w at pH 7, 258C and 1.0 mg protein/g ATPS. (a) Top phase gel, (b) bottomphase gel, and (c) 3-D scatter plot of alfalfa proteins detected in both phases (see full data in Table 4). Spot volumes are propor-tional to the protein content. Marked spot corresponded to added G-CSF identified by its molecular properties.


Authors wish to acknowledge the financial support of Tecnol�gicode Monterrey, Biotechnology Research Chair (grant 020CAT161)and BioMaP REU Program 2007 for the technical assistance ofMattan Rojstaczer during this research.

The authors declared no conflict of interest.

5 References[1] Ma, J. K.-C., Drake, P. M. W., Christou, P., Nat. Rev. Genet. 2003, 4,

794 – 805.

[2] Stoger, E., Sack, M., Fischer, R., Christou, P., Curr. Opin. Biotechnol.2002, 13, 161 – 166.

[3] Abranches, R., Sylvain, M., Arcalis, E., Altmann, F., Fevereiro, P.,Stoger, E., J. Biotechnol. 2005, 120, 121 – 134.

[4] Azevedo, A. M., Rosa, P. A. J., Ferreira, I. F., Aires-Barros, M. R.,Trends Biotechnol. 2009, 27, 240 – 247.

[5] Gu, Z., Glatz, C. E., J. Chromatogr. B 2007, 845, 38 – 50.

[6] Platis, D., Labrou, N. E., J. Chromatogr. A 2006, 1128, 114 – 124.

[7] Benavides, J., Mena, J. A., Cisneros-Ruiz, M., Ram�rez, O. T., Palo-mares, L. A., Rito-Palomares, M., J. Chromatogr. B 2006, 842, 48 –57.

[8] Hachem, F., Andrews, B. A., Asenjo, J. A., Enzyme Microb. Technol.1996, 19, 507 – 517.

[9] Gu, Z., Glatz, C. E., Biotechnol. Bioeng. 2007, 97, 1158 – 1169.

[10] Asenjo, J. A., Andrews, B. A., J. Mol. Recognit. 2004, 17, 236 – 247.

[11] Natarajan, S. S., Xu, C., Hanhong, B., Caperna, T. J., Garret, W. M.,J. Agric. Food Chem. 2006, 54, 3114 – 3120.

[12] Neupogenm Data Sheet, Amgen Inc., Thousand Oaks, CA 2004.

[13] Kwon, T.-H., Kim, Y.-S., Lee, J.-H., Yang, M.-S., Biotechnol. Lett. 2003,25, 1571-1574.

[14] James, E. A., Wang, C., Wang, Z., Reeves, R., Han Shin, J., Magnu-son, N. S., Lee, J. M., Protein Expression Purif. 2000, 19, 131 – 138.

[15] Hoglund, M., Med. Oncol. 1998, 15, 229 – 233.

[16] Pirondini, A., Visioli, G., Malcevschi, A., Marmiroli, N., J. Chroma-togr. B 2006, 833, 91 – 100.

[17] Srere, P. A., in: Packer, L. (Ed.), Methods in Enzymology, Vol. 13, Aca-demic Press, London 1969, pp. 3 – 11.

[18] Bradford, M. M., Anal. Biochem. 1976, 72, 248 – 254.

[19] Zaslavsky, A., Aqueous Two-Phase Partitioning. Physical Chemistry andBioanalytical Applications, Marcel Dekker Inc., New York 1995.

[20] Aguilar, O., Rito-Palomares, M., J. Chem. Technol. Biotechnol. 2008,83, 286 – 293.

[21] Laemmli, U. K., Nature 1970, 227, 680 – 685.

[22] Berkelman, T., Stenstedt, T., 2-D Electrophoresis. Principles and Meth-ods, GE Healthcare, Buckinghamshire, UK 2004.

[23] Rosa, P. A. J., Azevedo, A. M., Aires-Barros, M. R., J. Chromatogr. A2007, 1141, 50 – 60.

[24] Azevedo, A. M., Rosa, P. A. J., Ferreira, I. F., Aires-Barros, M. R., J.Biotechnol. 2007, 132, 209 – 217.

[25] Cascone, O., Andrews, B. A., Asenjo, J. A., Enzyme Microb. Technol.1991, 13, 629 – 635.

[26] Andrews, B. A., Schmidt, A. S., Asenjo, J. A., Biotechnol. Bioeng.2005, 90, 380 – 390.

[27] Franco, T. T., Andrews, A. T., Asenjo, J. A., Biotechnol. Bioeng. 1996,49, 300 – 308.

[28] Watson, B. S., Asirvatham, V. S., Wang, L., Sumner, L. W., PlantPhysiol. 2003, 131, 1104 – 1123.

[29] Gasteiger, E., Hoogland, C., Gattiker, A., Duvaud, S., Wilkins, M.R., Appel, R. D., Bairoch, A., in: Walker, J. M. (Ed.), The ProteomicsProtocols Handbook, Humana Press, New York 2005, pp. 571 – 607.

[30] Incamps, A., H�ly-Joly, F., Chagvardieff, P., Rambourg, J. C.,Dedieu, A., Linares, E., Qu�m�neur, E., Biotechnol. Bioeng. 2005, 91,447 – 459.

[31] Garbis, S., Lubec, G., Fountoulakis, M., J. Chromatogr. A 2005, 1077,1 – 18.


Characterization of green-tissue protein extract from alfalfa (Medicago sativa) exploiting a 3-D...

Documents

Transcript of Characterization of green-tissue protein extract from alfalfa (Medicago sativa) exploiting a 3-D...