High throughput process development for the purification of rapeseed proteins napin and
cruciferin by ion exchange chromatography
Moreno-González, Mónica; Chuekitkumchorn, Pattra; Silva, Marcelo; Groenewoud, Roos; Ottens, Marcel
DOI
10.1016/j.fbp.2020.11.011
Publication date
2021
Document Version
Final published version
Published in
Food and Bioproducts Processing
Citation (APA)
Moreno-González, M., Chuekitkumchorn, P., Silva, M., Groenewoud, R., & Ottens, M. (2021). High
throughput process development for the purification of rapeseed proteins napin and cruciferin by ion
exchange chromatography. Food and Bioproducts Processing, 125, 228-241.
https://doi.org/10.1016/j.fbp.2020.11.011
Important note
To cite this publication, please use the final published version (if applicable).
Please check the document version above.
Copyright
Other than for strictly personal use, it is not permitted to download, forward or distribute the text or part of it, without the consent of the author(s) and/or copyright holder(s), unless the work is under an open content license such as Creative Commons. Takedown policy
Please contact us and provide details if you believe this document breaches copyrights. We will remove access to the work immediately and investigate your claim.
This work is downloaded from Delft University of Technology.
ContentslistsavailableatScienceDirect
Food
and
Bioproducts
Processing
jo u r n al ho m e p a g e :w w w . e l s e v i e r . c o m / l o c a t e / f b p
High
throughput
process
development
for
the
purification
of
rapeseed
proteins
napin
and
cruciferin
by
ion
exchange
chromatography
Mónica
Moreno-González,
Pattra
Chuekitkumchorn,
Marcelo
Silva,
Roos
Groenewoud,
Marcel
Ottens
∗DepartmentofBiotechnology,DelftUniversityofTechnology,vanderMaasweg9,2629HZDelft,theNetherlands
a
r
t
i
c
l
e
i
n
f
o
Articlehistory:
Received26May2020
Receivedinrevisedform9
November2020
Accepted22November2020
Availableonline30November2020
Keywords:
Canolamealproteins
Highthroughputexperimentation
Chromatography
IonExchange
Columnsimulation
a
b
s
t
r
a
c
t
Proteinsderivedfromplantresourcessuchasoilseedmeals,canolaandsunflower,are
con-sideredaviablealternativetoanimalproteinsforfoodconsumption.Thisworkpresentsa
rationalmethodology,usinghighthroughputexperimentation(HTE),fortheseparationof
cruciferinandnapin,thetwomajorproteinsofcanolameal,bychromatography.Eight
dif-ferentmixedmodeandionexchangeresinswereevaluatedatdifferentconditionswiththe
aimofcapturingnapinandidentifyingadsorption/desorptionbehavior,easeofdesorption
andselectivity.POROS50HSresultedasthemostpromisingresin.Theobtainedequilibrium
adsorptiondatafornapinandcruciferinwasusedinamechanisticchromatographymodel
andcomparedwithexperimentalresultsshowingaverygoodagreement.Themodelwas
usedtoidentifycolumnoperatingparametersthatleadto>98%yieldandpurityforboth
proteins.Subsequentlyaconceptualdownstreamprocessingwasproposed.
©2020TheAuthors.PublishedbyElsevierB.V.onbehalfofInstitutionofChemical
Engineers.ThisisanopenaccessarticleundertheCCBY-NC-NDlicense(http://
creativecommons.org/licenses/by-nc-nd/4.0/).
1.
Introduction
Globalproteindemandhasrisenasaresultofgrowing
pop-ulation and in order to satisfy this demand protein with
appropriatequalityneedstobegeneratedfromanimaland
plantsources(Wanasundaraetal.,2016).Valorizationof
agri-foodbyproducts,suchasoilseedmeal,couldhelptosatisfy
thedemandofthefuturefoodsupplyacrosstheworld.Canola
mealisacompetitiveplant-basedproteinsource,itcontains
∼40%proteincontent;∼12%ofcrudefiber,3%phenolic
com-poundsand3%ofphyticacidamongothers(Lomascoloetal.,
2012;Wanasundara,2011;Wanasundaraetal.,2017).Proteins
fromthismealhavepotentialapplicationsinbakeryproducts
asemulsifiers,indiaryand dressingproductsaseggwhite
proteinreplacement(Kodagodaetal.,1973),inbeveragesas
proteinsupplement,meatbinders,mayonnaiseandin
cheese-∗ Correspondingauthor.
E-mailaddress:M.Ottens@tudelft.nl(M.Ottens).
like products(Wanasundara et al., 2016). Moreover,canola
proteins are reported tohave antiamnestic, anti-oxidative,
antihypertensive, anorectic and antithrombotic properties
(Wanasundara,2011;Wanasundaraetal.,2016).
Napinandcruciferinarethetwomajorproteinsincanola
meal,theseare storageproteinswithdifferent
characteris-tics.Napin(2Salbumin)isabasicproteinwithanisoelectric
pointaround11andamolecularweightbetween12-14kDa.
HighlysolubleinwateratawidepHrange,stableathigh
tem-peratures (75◦C)(Jyothietal., 2007;Pereraet al.,2016)and
holdsfoamingproperties(AiderandBarbana,2011;Schmidt
etal.,2004;Wanasundaraetal.,2016a).Napinprotein
consti-tutesaround20%oftheproteincontentofcanolameal,while
cruciferinrepresent60%(Wanasundara,2011).Cruciferin(11S
globulin)hasanisoelectricpointofaround7.2anda
molec-ular weightbetween230−300kDa,muchhigherthan napin
protein. Thisproteinresemblesstructuralfeaturesofother
seedstorageproteins(Adachietal.,2003)andhaswell
orga-nizedstructurallevels(Wanasundaraetal.,2017).Incontrast
withnapin,cruciferinpossessesgelling,bindingand
emulsi-https://doi.org/10.1016/j.fbp.2020.11.011
0960-3085/©2020TheAuthors.PublishedbyElsevierB.V.onbehalfofInstitutionofChemicalEngineers.Thisisanopenaccessarticle
fyingpropertieswhichmakesitaninterestingproductforfood
applications(AiderandBarbana,2011;Wanasundara,2011).
Several methods have been described to successfully
extractproteinsfromdefattedoilseedmealsincludingsolvent
extraction,aqueousandalkalineextractionassistedwithsalt
andenzymes.Thereportedproteinyieldsvarybetween50%
and80% (Contreras etal., 2019; Fetzeretal., 2018; Pickardt et al., 2009; Vuorela et al., 2004). Despite the high protein
levelincanolamealextracts,thepresenceofglucosinolates,
phyticacidandphenolics,whicharealsoco-extracted,could
limitapplicationoftheproteinsinfoodproducts.Therefore,
effectiveseparationtechniquessuchasmembraneprocesses
(AkbariandWu,2015;Ghodsvalietal.,2005;XuandDiosady,
2002)arerequiredfortheseparationofsuchantinutritional
components(Wanasundaraetal.,2017).Eventhough
pheno-licsareconsideredantinutritionalcomponents,theirrecovery
mightpresentsomeeconomicpotential.Giventheir
antiox-idativeproperties,theyhavepotentialapplicationinthefields
of cosmetics, pharmaceutical and food products (
Moreno-Gonzálezetal.,2020).
Purificationoftheextractedproteinscanbeaccomplished
byseveralmethodssuchasisoelectricprecipitationfollowed
by membrane separation (Ghodsvali et al., 2005; Xu and
Diosady,2002)andproteinmicellarformation(Murrayetal.,
1980),beingisoelectricprecipitation(afteralkalineextraction)
themostapplied.However,extremealkalineconditionscan
havea negativeeffect on the functionalityof the proteins
due todenaturalization, loss ofessential amino acids and
lysinoalanineformation(Fetzeretal.,2018;Gerzhovaetal.,
2015;Houetal.,2017;Rommietal.,2015).Inaddition,high
pHmightcreateprotein-polyphenolscomplexes,whichmake
proteinproductsdarkandhavebitterflavor.Similarly,
pro-teinprecipitationmightreducethesolubilityofproductsand
promotesproteindenaturationcausedbyglobulinaggregates
(Bérotetal.,2005;RaabandSchwenke,1984).Moreover,
co-precipitationofbothproteinsmightoccurleadingtoprotein
mixturesratherthanindividualproteinisolates.Application
ofmore selective and milder conditions, such as aqueous
extractionandchromatographycouldbeappliedtokeep
pro-teinfunctionalityandimprovepurityofeachproteinproduct.
Studiesinpurificationofcanolaproteinsbychromatography
havebeenevaluated(Bérotetal.,2005;Pudeletal.,2015;Zhang
et al.,2007)using cationexchangeresins (CEX),
hydropho-bicinteraction (HIC) and size exclusion(SEC), showingthe
promisingpotentialofapplyingthistechnology.
Thisstudypresentsarationalstrategyfortheseparation
ofcruciferin and napin proteinsfrom canolameal extract
bypreparativeproteinchromatography.Proteinextractionis
assumedtobedoneatpH6andat0.3MNaCl.Thestrategy
involves theevaluation ofdifferentcation exchangeresins
(CaptoS,POROS50HS,CMSheparoseandMacroPrep50)and
mixedmoderesins(CaptoMMC,NuviacPrime,PPAHyperCel
andToyopearlMX-Trp-650M)atdifferentpHvaluesandsalt
concentrationstoidentifyadsorption/desorptionconditions
withthegoalofcapturingnapin.Resinscreeningwasdoneby
meansofhighthroughputexperimentation(HTE)
Addition-ally,resinselectionwasdonebyestablishingaresinselection
criteriabasedonnapincapacity,selectivity,easeofdesorption
and resinprice. Theobtainedequilibriuminformationwas
usedinacolumnadsorption/desorptionmodelwhichisthen
usedtosuggestaconceptualdownstreamprocessforthe
sep-arationofthetwomajorproteins(cruciferinandnapin)from
canolamealextract.
2.
Materials
and
methods
2.1. Chemicals
Forpreparationofthebuffersandsolutionsanalyticalgrade
chemicalswereused.Bis-tris(>98%),tris-HCl(>98%),
acetoni-trile(HPLCgrade),hydrochloricacidanalyticalreagent(37%),
trifluoroaceticacid(>99%)wereobtainedfromSigma-Aldrich,
theNetherlands.Sodiumchloride(>99%)waspurchasedfrom
J.T. Baker, Denmark, sodium hydroxide from Mallinckrodt
Baker, The Netherlands, and Ethanol:Emsure absolute for
analysiswasobtainedfromMerck,TheNetherlands.
Theusedproteinsare:napinisolate(>98%,ABIN1995012),
cruciferin isolate (>98%, ABIN1995013) and rapeseed
pro-tein mixture(57% napinand 43% cruciferin,ABIN1995014).
Theproductswere acquiredfromantibodies-online,GmbH,
Germany.As cruciferinisolatepresentedverylimited
solu-bilitypropertiesinwater,itwasdecidedtousetherapeseed
proteinmixturetoperformbinaryadsorptionexperiments.
Napinisolate,cruciferinisolateandrapeseedprotein
mix-ture were characterized by SDS-PAGE under non-reducing
conditions and reducing conditions (reducing agent TCEP
solution).SDS-PAGEwasperformedona4–12%Bis-TrisGel
(NuPAGETM Novex) at constant voltage (200V). The
non-reduced samplewas preparedwithNuPAGETM LDSsample
buffer. The reduced sample was incubated with LDS
sam-plebufferbeforeloadingongel.NuPAGEMark12TMwasused
asamolecularmarker. Theelectrophoresiswascarriedout
usingMESSDSasrunningbuffer.Afterrunning,thegelwas
stainedinGelCodeTM BlueSafeProteinStainanddestained
withMilli-Qwater.AllSDS-PAGEreagentswereobtainedfrom
ThermoScientific,TheNetherlands.
2.2. Resins
Threemixedmodecationresins:CaptoMMC(GEHealthcare,
Sweden), Nuvia cPrime(Bio Rad, USA) and Toyopearl
MX-Trp-650M(Tosoh,Japan);onemixedmodeanionresin,PPA
HyperCel(PallLifeSciences,France),twostrongcationresins,
Capto S (GE Healthcare, Sweden) and POROS 50HS
(Ther-moFisherScientific, TheNetherlands) andtwoweakcation
resins,CMSepharoseFastflow(GEHealthcare,Sweden)and
MacroPrep50CM(BioRad,USA),wereusedtoevaluatenapin
andcruciferinadsorption.Thecharacteristicsofallresinsare
showninTable1.
2.3. Buffersolutionandpreparations
For buffersatpH4,pH5,pH6andpH8,lacticacid,acetic
acid,bis-trisandtris-HClwereusedrespectively.Allbuffers
werepreparedbydissolvingtheamountofsaltcorresponding
to50mMinMilli-Qwater,adjustingthepHusing2MHClor
2MNaOH.Thesaltconcentrationwasadjustedto0.1M,0.3M,
0.7Mand1.0MbyaddingthecorrespondingamountofNaCl
beforecompletingbufferfinalvolume.Allbufferswerefiltered
previoustouseusingfilterswith0.45mporesize.
Napin stock solution was preparedby dissolving napin
protein(8g/L) inthe appropriatebuffer and filteredwitha
disposable0.22mPVDFfilter.Thisnapinstocksolutionwas
dilutedtodifferentnapinconcentrations(from1to8g/L)using
Milli-Qwater.
In binary component experiments, protein mixture
Table1–MixedmodeandIonexchangeresins.
Name/Type pKa Matrixcompositiona Liganddensitya
(mmol/L)a
CaptoMMC Mixedmodeweakcationexchanger ∼4.6b Highlycross-linkedagarose 80
NuviacPrime Mixedmodeweakcationexchanger ∼4.5b Macroporoushighlycross-linked
hydrophilicpolymer
65 ToyopearlMX-Trp-650MMixedmodeweakcationexchanger 2.9and9.4c Methacrylicpolymermatrix 110
PPAHyperCel Weakanionexchanger 8 Highporositycross-linked
cellulose
65
CaptoS Strongcationexchanger 1.2 Highlycross-linkedagarosewith
dextransurfaceextender
125
POROS50HS Strongcationexchanger 1.2 Cross-linked
polystyrene-divinylbenzene
104
CMSepharoseFastflow Weakcationexchanger 4.7 Cross-linkedagarose 80
MacroPrep50CM Weakcationexchanger 4.7 Methacrylatepolymerbased 210
a Providedbyresinsuppliers. b Basedon(ZhuandCarta,2016).
c BasedonthepKaofthetryptophanligand(NationalCenterforBiotechnologyInformation,2020).
Thesolutionwassonicatedatroomtemperaturefor30minin
ordertoincreasethesolubilityofcruciferin.Afterthat,
insol-ubilizedproteinwasremovedbyfilteringwithadisposable
0.22mPVDFfilter,thisstocksolutionwasdilutedtoevaluate
cruciferineffectonnapinadsorption.
2.4. Analyticalmethods
Insinglecomponentnapinexperiments,theconcentrationof
the proteinwasmeasured spectrophotometrically,
measur-ingthe absorbanceat280nmusingthe spectrophotometer
InfiniTePro200platereader(Tecan,Switzerland).The
mea-surementwas performed with100L of liquid solution in
a96-well half-area microplate (UV-STAR®, Greiner bio-one,
Germany)
To measure the concentration of both proteins, in the
proteinmixtureexperiments,reversephaseliquid
chromatog-raphy(RPC)wasapplied.TheanalysiswasdoneusinganUltra
High Performance Liquid Chromatography system (UHPLC
Ultimate3000)(ThermoScientific,USA)equippedwitha
Zor-bax 300 SB-C8 Rapid Resolution HD column (2.1×100mm,
1.8m)(Agilent,USA).Thecolumnwasequilibratedwith20%
acetonitrilesupplementedwith0.1%Trifluoroaceticacid(TFA)
at0.3mL/minkeepingcolumntemperatureat30◦C.The
sam-plewasinjectedandagradientofacetonitrilestartedfrom
20%to75%in7min,detectionwasdoneat280nm.Thenthe
columnwaswashedwith75%acetonitrilesupplementedwith
0.1%TFAfor5minbeforethenextinjection.Napin
calibra-tionlineswere evaluatedusingnapinproteinisolate.Since
cruciferinisolatedidnotshowtheexpectedcharacteristics,
asthe polypeptide profiledidnot show the corresponding
bands(Supplementarymaterial),astandardcurvecouldnot
beobtained.Therelativechangeofconcentrationwasusedto
evaluatecruciferin.Assumingthattheratiobetween
equilib-riumconcentrationandinitialconcentrationisproportional
topeakarearatio
CeCfeed =
Areapeakafteradsoprtion Areapeakoffeed
determinedby
UHPLC.
2.5. Batchadsorptionexperiments
Batch experiments were performed in order to determine
adsorptionequilibriumisothermsofnapinandcruciferinat
differentadsorption/desorptionconditionsondifferentmixed
modeandionexchangeresins,inordertoseparateboth
pro-teins.
2.5.1. Adsorptionequilibriumisotherms
Napinadsorptionequilibriumisothermswereevaluatedusing
aTecanEVOFreedom200roboticstation(Tecan,Switzerland)
equipped with an orbital mixer (Te-shake), an automated
vacuum system (Te-VacS),aplate reader(InfiniTe Pro200),
a roboticmanipulator (RoMa) arm(to movemicroplates to
thedifferentpositionsoftheroboticstation)andtwoliquid
handingarms(LiHaandMCA96).Theprocedureinvolvesthe
different stepsofthe chromatographyrununtiladsorption
(washing,equilibrationandadsorption).Aknownamountof
eachresin(15.6Lor23.4L)wasaddedtothewellsofa96
deep-wellfilterplate(catalognumber:MDRLN0410)from
Mil-lipore,USA.ResinswereaddedusingMediaScout®ResiQuot
resinloader(Atoll,Germany).Resinwerewashedtwotimes
withMilli-Q waterusing thevacuumsystem (Te-VacS)and
equilibratedwiththecorrespondingbufferfor20min.under
agitation(1200rpm).Equilibrationbufferwasremovedby
cen-trifugation with an Eppendorf centrifuge 5810 R (rotation
speed 4000rpmfor3min).Aftercentrifugation,resins were
contactedwith312Lofnapinsolutionsunderagitationuntil
equilibriumwasreached(2hat1200rpm).Onceequilibrium
was reached,the filterplate was centrifuged tocollect the
supernatant and equilibrium concentration was measured
spectrophotometrically (see2.4 Analytical methods).Napin
experimentswereperformedinduplicate.
EffectofpHandionicstrengthwereevaluatedfornapin
adsorptionbyperformingexperimentsat4differentpH(pH
4,5,6,8)keepingNaClconcentration(0.3M)constantandat
pH6varyingsaltconcentrationfrom0.1Mto1MNaCl.This
allowstoidentifydesorptionconditions.
Napinadsorptioncapacitywascalculatedaccordingtothe
massbalanceEq.(1)
qp,eq=
Cp,load∗Vload−Cp,eq(VH+Vload)
Vresin
(1)
Whereqp,eq istheadsorptioncapacity(mg/mLresin),Cp,loadis
theproteininitialconcentration(mg/mL),Vload isvolumeof
theliquidphase(mL)Cp,eqistheproteinequilibrium
concen-tration(mg/mL),VHistheholdupvolume(mL)andVresinisthe
Aftercentrifugation,someliquidcouldremainintheresin.
Thisliquidholdupwas determined usingthe method
sug-gested by Nfor et al. (2010). The resins were placed in a
deep-wellfilterplateandequilibratedwith350Lof1MNaCl.
After45minthefilterplatewascentrifugedwith5810R
cen-trifuge(Eppendorf,Germany) for3minat4000rpmandthe
flowthroughwascollectedtomeasureconductivity.Thiscycle
wasrepeateduntiltheconductivityofflowthroughwasequal
totheconductivityof1MNaClsolution. Consequently,the
equilibratedresinswerecontactedwith312LofMilli-Qwater
andincubatedovernightwithoutagitation.Aftercontacting,
thefilterplatewas centrifugedand theconductivityofthe
flow-through was measured.All the conditions were done
intriplicate.Theconductivity was measuredwitha
multi-parameteranalyzerC832(ConsortNV,Belgium).
Theholdupvolumewasmeasuredusingthesaltmass
bal-anceEq.(2)
CS,initial∗VH= CS,final∗ (VH+VW) (2)
Initialsaltconcentrationinallresinswas1MNaCl(CS,initial),
VHisholdupvolume(mL),CS,finalisthefinalsalt
concentra-tionaftercontactingwithMilli-QwaterandVW istheadded
volumeofMilli-Qwater.
Since cruciferin has low solubility in water, cruciferin
adsorptionisothermswereevaluatedusingtheprotein
mix-ture. A similar procedure to the one applied for napin
adsorptionisothermswasusedwiththeproteinmixturein
theliquidhandlingroboticstation.Theevaluatedconditions
were:pH6and 0.3MNaCl sincetheyare the same
condi-tionsintheproteinextract(adsorptioncondition).Cruciferin
experimentswereperformedinduplicate.
As the absolute cruciferin liquid concentration values
could notbemeasured,cruciferin adsorptioncapacitywas
determinedbydividingEq.(1)bythereferenceconcentration
(C0)(Eq.(3)),assumingthatCCp,eq
0 =
Areaatequilibirum Areaofreference
qp,eq=qCp,eq
0 =
Cp,load∗Vload−Cp,eq(VH+Vload)
VresinC0
(3)
Thereference cruciferin area peaks were the ones
cor-responding to the conditions of 0.3M NaCl and Napin
concentrationofaround6g/LforexperimentsatpH6(feed
condition).
2.5.2. Parameterestimation
Adsorptionisothermexperimentalresultsofnapinwere
fit-tedtoalinearisothermortoaLangmuirtypeisotherm(Eq.
(4)).Thistoidentifythe initialisotherm slope,which isan
indicationoftheaffinityofnapintotheresin.
qp,eq=
bp,iqmaxp,i Cp,i
1+bp,iCp,i
(4)
whereqmax
p,i isthemaximumadsorptioncapacity(mg/mLresin)
andbp,iistheLangmuirconstantalsoknownasequilibrium
constant (Carta and Jungbauer, 2010). Theinitial isotherm
slope,attheevaluatedconditions,wasusedtoevaluateresin
selection(see2.6Resinselection).
Themostsuitableresinequilibriumdata(POROS50HS)was
fittedtothemixedmodeisotherm,developedbyNforetal.
(2010).Thisisothermsisbasedonthethermodynamic frame-workofMollerupetal.(2008).Amoredetailexplanationofthis
isothermisfoundinsection2.8.1Mixedmodeisothermmodel
ofthispaper.Thefittedparametersofthisisothermmodel
were:1)thethermodynamicequilibriumconstant(Keq);2)the
stoichiometriccoefficientofsaltcounterion(v);3)the
param-eterthatdescribesthedifferencebetweenwater-proteinand
protein–protein interactions (Kp)and 4) the parameter that
describes the difference between water – protein and salt
–proteininteractions(Ks).Aspreviouslymentioned,binary
componentexperiments(napin+cruciferin)wereperformed
usingaproteinmixturepowder.Binarycomponent
adsorp-tionexperimentsshowedahighernapinadsorptioncapacity
thannapinsinglecomponentexperiments.Forisotherm
mod-ellingthis was adjusted byconsideringthe valueobtained
frombinarymixtureexperimentalresults,whichcorresponds
to44mg/mL.
DataregressionsweredoneusingMATLABR2017bandthe
functionlsqcurvefit.Inthemixedmodemodel,qpappearsin
both sidesofthe equation(Eq. (7)) Thenumericalsolution
ofthisequationwasfoundusingthefsolvefunctionof
MAT-LABanditwascombinedwiththeregressionusinglsqcurvefit
optimizer.Thus,theprocedureoftheparameterregressionis
applyinglsqcurvefitoptimizertominimizethesumofsquared
residualsbetweentheexperimentaldataandthenumerically
solvedadsorptiondatausingfsolvefunction.
Inaddition,asbinarymixtureexperimentpresentedavery
low change in cruciferin concentration. A linear isotherm
modelwasconsideredforthisproteinbyestimatingisotherm
slopebyfittingalinearcurvetotheexperimentaldata.
2.6. Resinselection
Thegeneratedequilibriumdatawasusedtoselectthemost
suitableresinfortheseparationofbothproteins.Thedesired
scenario is the adsorption of napin while cruciferin flows
through.Theselectioncriteriaweredefinedassuggestedby
Sevillanoetal.(2014),consideringnapinadsorptioncapacity
atfeedconditions(pH6and0.3MNaCl)andnapindesorption,
whichwasevaluatedusingtheinverseofthelowestisotherm
slopedeterminedattheevaluatedconditions.Thethird
crite-rion,selectivity,wasevaluatedusingtheratiobetweennapin
andcruciferinisothermslopesatadsorptionconditions.The
fourthcriterionwasthepriceoftheresinobtainedfromthe
differentsuppliers. Aweightbetween0and1wasgivento
eachcriterion,being0.5fornapincapacity,0.3fornapin
des-orption,0.1forselectivityand0.1forprice. Thereasonfor
givingalowweighttotheselectivitycriterionhadtodowith
theobservedpoorbindingofcruciferinontomostofthe
eval-uatedresins.
Eachcriterionwasnormalizedandresinscorewas
calcu-latedusingEq.(5).
Resinscore=
weight∗ criterionmaximumvalueofcriterion (5)
2.7. Columnadsorption/desorptionexperiments
The highestscoring resin,POROS 50 HS, was used to
per-formpulsecolumnexperimentsusingnapinandtheprotein
mixture(napin+cruciferin).Theexperimentswereperformed
inanAKTATMAvantsystem(GEHealthcareBio-SciencesAB,
Uppsala,Sweden)atroomtemperature(25◦C)operatedwith
Unicorn7.0software.Conductivity,pHandUVat280nm
sig-nalsweremonitoredduringtheexperiments.
AValiChrom11.3–100 POROS® 50 HS(Repligen,Sweden)
and the total bed heightwas 10cm, providing a total
col-umnvolumeof9.8mL.Extraparticleandtotalporositieswere
determinedbypulseinjectionsofbluedextranand1MNaCl
respectively.Theobtainedvalueswere0.83totalporosity(T)
and0.3forextraparticleporosity(b).
Fornapinand proteinmixtureexperiments,the column
was equilibrated with 50mM bis-tris pH6 with 0.3M NaCl
bufferfor5columnvolumes(CV)at2.5mL/min.Then1mL
ofnapinsolutionat6g/Lwasinjectedinthecolumn(10.1%
CV),columnwaswashedwiththeequilibrationbufferfor5CV
inordertoremovenon-bindingsubstratesoutofthecolumn.
Stepelutionwasdonebyapplying5CVofelutionbuffer,
Tris-HClpH8with0.7MNaCl.Aftereachexperiment,thecolumn
wasregeneratedwith1MNaClsolutionfor5CVas
recom-mendedbysupplier.
Forrapeseedproteinmixture,12g/Lofproteinmixturewas
dissolvedin50mMBis-TrispH6with0.3MNaClbuffer.This
correspondsto6.84g/Lofnapinand5.16g/Lofcruciferin.The
proteinmixture solution was then sonicatedfor 30min in
ordertoincreaseproteinsolubility.Thesolutionwasfiltered
with0.22mPVDFfilterbeforeinjectionincolumn.Afteruse,
thecolumnwasstoredin20%(v/v)ethanolassuggestedby
supplier.
2.8. Chromatographymodeling
2.8.1. Mixedmodeisothermmodel
ThemixedmodeadsorptionisothermmodelfromNforetal.
(2010) was appliedin this work. Themodel is a
combina-tionofHICandIEXmodelsfromMollerup(2007).Thismixed
modeisothermisbasedontheassumptionthataprotein(P)
bindsto namountofligand(L)byhydrophobicinteraction
andconcurrentlyexchangeswithvamountofsaltcounter-ion
(S)generatingaprotein-ligandcomplex(Ln+v)inafixed
stoi-chiometryasshowninEq.(6).Thestoichiometriccoefficient
ofsalt counterionisdefinedasthe ratiobetweenthe
pro-teinbindingchargedividedbythechargeofthesaltcounter
ion
v= zpzs
. Inthis work,sodium chloride wasused as a
saltcounterion.Thesinglecomponentadsorptionisotherm,
basedon reactioninEq.(6),resultsinthe isothermmodel
describedinEq.(7). P+vSL+nL ↔PLn+v+vS (6) qp cp = A∗
1−qqmaxp p v+n ∗˜p (7) whereA= ˜Keq()v+n 1 zscs v1 c n and˜p=exp (Kpcp+Kscs) (8)In Eq. (7), qp is the adsorbed protein concentration
(mg/mLresin), cp is the protein concentration in solution
(mg/mL),Keqisthermodynamicequilibriumconstantof
reac-tion(mL/mg),csistheconcentrationofsaltinliquidphase.c
ismolarityofthesolution(assumedtobewaterconcentration
(Mollerup,2007))andqmax
p ismaximumadsorptioncapacityof
protein(mg/mLresin),isliganddensityofthemixedmode
resinswhichareassumedtobeequalforHICandIEXligands
(mmol/L)andwasobtainedbasedonsupplierspecifications.
Theactivity coefficient ofthe protein ( ˜p) was determined
usingtheVan derWaalsequationofstateshowninEq.(8)
assuggestedbyMollerup(2006).Kpparameterdescribesthe
differencebetweenwater-proteinandprotein–protein
inter-actionswhileKsparameterdescribesthedifferencebetween
water– proteinand salt –proteininteractions (Nfor et al.,
2010).Inthe model,(1− qp
qmax
p )describesthe fractionoffree
ligands,whichis1ifnoproteinisboundanddecreases
asymp-toticallytozero.Thisisothermincorporatestheeffectofsalt
concentrationinthetermCs.
ThemixedmodeisothermdescribedinEq.(7)reducesto
theHICandIEXisothermmodelsderivedbyMollerup(2006)
whenv=0(electrostaticinteractionsnotpresent)andn=0(no
hydrophobicinteractions),respectively.
2.8.2. Columnchromatographymodel
Column chromatography was simulated based on the
transport-dispersive columnmodelwhich canbedescribed
percomponentasshowninEq.(9)
∂cp,i ∂t +
1−ε b εb ∂qp,i ∂t = −v ∂cp,i ∂x +DL,i ∂2cp,i ∂x2 (9)where,εbisthebedporosity,vistheinterstitialvelocityofthe
mobilephase(m/s),andDListheaxialdispersioncoefficient
(m2/s)
Masstransfercanbequantifiedbythe liquid-filmlinear
drivingforceapproximation.Thesolidstationaryphase
con-centrationterm ∂qp,i
∂t isdefinedasfollowsinEq.(10):
∂qp,i ∂t =kov,i
Cp,i−C∗p,eq,i (10)kov,iistheoverallmastranferscoefficient(1/s)andC∗p,eq.isthe
bulkequilibriumconcentrationwhichisobtainedusingthe
mixedmodeisotherm(Eq.(7)).Mixedmodeisotherm
param-eterswereusedtomodelnapinchromatographywhilealinear
isothermwasconsideredtomodelcruciferinchromatography.
The film masstransfer coefficient isdefined in Eq.(11)
(FelingerandGuiochon,2004). 1 kov,i= dp 6∗kf + d2 p 60∗εp∗Dp (11)
wheredpistheparticlediameter(m),kfisthefilmmass
trans-fer coefficient (m/s), εp isthe intraparticleporosity and Dp
is thepore diffusivity(m2/s). To determineadditional
rele-vantmodelparameters,correlationspresentedinTable2were
used.
ColumnboundaryconditionsaredescribedbyDanckwerts
fordispersivesystemsanditisassumedthatthecolumnis
notpreloadedwiththeproteinsCp(t=0)=0andqp(t=0)=0.
C (t,x=0)=Cinlet− DL uh · ∂C (t,x=0) ∂x (12) ∂C (t,x=L) ∂x =0 (13)
wherexistheaxialposition.Eq.(12)representstheboundary
attheinletofthecolumnandEq.(13)theboundarycondition
atthecolumnoutlet.Aspulseexperimentswereperformed,
theinjectionprofileismodelledasarectangularpulsewith
aconstantfeedconcentrationforagiventime,wheretpulse=
Vinj/˚v,where˚visthevolumetricflowrate(m3/s)andVinjis
theinjectionvolume.
Pulse Cinlet(t)=Cfeed,i at 0<t<tpulse (14)
Table2–Engineeringcorrelationsforcolumnmodelling.
Masstransferparameter Correlation Reference
Hydrodynamicradius StokesEinstein LeVanetal.(1999)
Freediffusivity Young Youngetal.(1980)
Filmmasstransfercoefficient Wilson&Geankoplis WilsonandGeankoplis(1966)
Poretortuosity Wakao&Smith WakaoandSmith(1962)
Porediffusivity Brenner&Gaydos BrennerandGaydos(1977)
Axialdispersioncoefficient Gunn Gunn(1987)
Pressuredrop Karman-Cozeny CartaandJungbauer(2010)
Eq.(9)isapartialdifferentialequation,dependentontime
andcolumnposition,thatcanbeapproximatedtoanordinary
differentialequationbyspatialdiscretization.Themethodof
lineswasusedtodiscretizeEq.(9)inspace.Thesetofordinary
differentialequationswassolvedinMATLABR2017busingthe
ODEsolverode15s.
3.
Results
and
discussion
3.1. Resinandproteinscharacteristics
Theusedproteinisolates,cruciferin,napinandprotein
mix-turewere analyzedbySDS-PAGE (Supplementarymaterial).
Napinandtheproteinmixtureshowedexpectedbandprofiles
thatwerecomparabletotheonesreportedelsewhere(Perera
etal.,2016).Cruciferindidnotshowntheexpectedprofileand
thesolubilityofthisproteinisolateinanyofthetested
condi-tionswasverylimited,thereforetheproteinmixturewasused
toevaluatecruciferinisotherms.
Toavoid an overestimation of the adsorbed phase
pro-teinconcentration, theresin liquid hold-upwasevaluated.
Theobtainedvalueswere9.3L,8.7L,7.0L,9.2L,10.0L,
9.8L,12.4Land8.5LforCaptoMMC,NuviacPrime,
Toy-opearlMX,PPAHyperCel,CaptoS,POROS50HS,CMSepharose
and MacroPrep50 CM respectively. These hold up volumes
correspond to ∼3−4% of the total volume applied in the
adsorptionexperiments(350L).
3.2. Napinadsorptionequilibriumisotherms
Napinadsorptionequilibriumisothermsweredeterminedfor
allresinsatfeedconditions,(0.3MNaClatpH6).Resultsare
shown inFig. 1, wherethe experimentalnapin adsorption
capacityisplotted againstthe equilibriumbulk
concentra-tion.Fromthefigure,onecanobservethatatfeedconditions,
thehighestadsorptioncapacityisobtainedusingresinCapto
MMC,followed byNuviacPrime,POROS HS, PPAHyperCel,
Toyopearl, Capto S, MacroPrep and CM Sepharose.Most of
theresinsshowedamaximumlimitintheamountadsorbed,
which can be described with a Langmuir isotherm model
(Eq.(4)).ResinsCaptoSandCMSepharosepresentedalinear
behavior.Theexperimentaldatawasfittedwithtwoisotherm
models-Langmuirorlineartype.
InadditiontotheresinisothermshapepresentedinFig.1,
theaffinityofnapintotheresinswasalsoobtainedby
deter-miningtheinitialisothermslopes.Theinitialisothermslopeis
anindicationoftheinteractionstrengthbetweentheprotein
andtheresin.Thelargertheisothermslope,thestrongerthe
interaction.Atfeedconditions,theisotherms’initialslopes,
foreachresin,canberankedfromlargesttosmallest:
Cap-toMMC,NuviacPrime,POROSHS, PPAHyperCel, Toyopearl,
Macro Prep, Capto S and CM Sepharose. Absolute values
(obtainedfromfitting)canbefoundinTable3.
Fig.1–Napinadsorptionisothermsonmixedmodeand ionexchangeresinsatpH6and0.3MNaCl.Symbols representexperimentalresults.Errorbarsresultedfrom duplicateexperiments.LinesrepresentLangmuirisotherm forCaptoMMC,NuviacPrime,POROSHS,Toyopearl,PPA HypercelandMacroPrepandlinearisothermforCaptoS andCMSepharose.
Thedifferencebetweenthenapinbindingstrengthonto
the different resins can be explainedbased on the pKaof
theresins(Table1),theisoelectricpointofnapinprotein-in
thiscasearound11-andtheproteinnetchargeatthe
eval-uated conditions.AtpH6,all resinsarenegativelycharged
and napin has a positive net charge. Therefore, attractive
electrostaticinteractionsarepossible.However,thereisalso
salt presentedinthe medium(NaCl) whichcould promote
hydrophobicinteractionsbetweennapinandthemixedmode
resins.Asaltconcentrationof0.3MNaCl–whichcanbe
con-sideredmedium-,provedtobealreadytoohighfortheweak
ion exchangers here evaluated (CM Sepharoseand
Macro-Perp)andevenforthestrongcationexchangerCaptoS.Asthe
preferablescenarioistheadsorptionofnapinatpH6and0.3M
NaCl,CMSepharose,Macro-PrepandCaptoSresinswere
dis-cardedassuitablecandidatesduetothepoorbindingofnapin
undersuchconditions.
TheeffectofionicstrengthandpHwasevaluatedforall
resinsinordertodeterminesuitabledesorptionconditions.
3.2.1. EffectofpHonnapinadsorption
Theionicstrength andpHmighthaveasignificant impact
onthe adsorptionofnapinonmixedmodecationicresins,
cationexchangeresinsandtheanionmixedmoderesinhere
evaluated.ExperimentswereperformedatdifferentpHvalues
keepingthesaltconcentrationconstantat0.3MNaCl,which
correspondtothesameionicstrengthasintheproteinextract.
NapinisothermresultsareshowninFig.2andFig.3.
From Fig. 2, one cannotice that for Capto MMC,Nuvia
Table3–NapinisothermslopeonmixedmoderesinandionexchangeresinatpH6andpH8anddifferentNaCl(Cs)
concentration.
Resin pH CS(M) Isothermslope
(mL/mLresin)
Resin pH CS(M) Isothermslope
(mL/mLresin) CaptoMMC 6 0.1 1185.4+245.2 Capto S 6 0.1 146.1+12 0.3 437.6+266.1 0.3 1.4+0.1 1.0 23.1+7.9 0.7 Notadsorbed 8 0.1 1114.4+248.8 8 0.1 25.2+3.8 0.3 99.3+25.6 0.3 0.2+0.0 0.7 17.8+6.8 0.7 Notadsorbed NuviacPrime 6 0.1 620.9+78.3 POROS50HS 6 0.1 453.8+76.2 0.3 67.5+17.23 0.3 52.5+14.9 1.0 1.3+0.01 0.7 0.7+0.9 8 0.1 276.6+57.5 8 0.1 108.9+22.5 0.3 20.2+3.5 0.3 6.0+1.8 0.7 2.1+0.2 0.7 Notadsorbed Toyopearl 6 0.1 97.4+13.6 CM Sepharose 6 0.1 8.5+3.0 0.3 5.9+5.02 0.3 0.3+0.0 1.0 0.4+0.2 0.7 Notadsorbed 8 0.1 52.4+9.1 8 0.1 13.5+1.9 0.3 1.8+0.3 0.3 0.2+0.0 0.7 0.5+0.1 0.7 Notadsorbed PPAHyperCel 6 0.1 1.9+0.1 MacroPrep 6 0.1 67.4+14.8 0.3 8.5+3.4 0.3 3.6+1.3 1.0 13.0+3.7 0.7 Notadsorbed 8 0.1 7.7+7.5 8 0.1 19.8+2.6 0.3 15.4+2.8 0.3 0.8+0.1 0.7 35.4+7.9 0.7 Notadsorbed
Fig.2–NapinadsorptionisothermsatdifferentpHkeepingsaltconcentrationat0.3MNaCla)CaptoMMCresin,b)Nuvia cPrimeresin,c)Toyopearlresinandd)PPAHyperCelresin.Symbolsrepresentexperimentalresults.Errorbarsresultedfrom duplicateexperiments.LinesrepresenttheLangmuirisotherm.
capacitywithincreasingpH.Thiscanbeexplainedbythe pro-tein’snetchargeandtheresinligand’spKa.Theweakcation exchangersarenegativelychargedatpHhigherthanthepKa, whiletheoppositeoccursforweakanionexchangers,which arepositivelychargedatpHlowerthanthepKa.Strongcation exchangersarepractically alwayschargedatanypH(Carta andJungbauer,2010).Theinteractionsbetweenmixedmode
resinsandproteinshavebeenexplainedbydifferentauthors
(Nforetal.,2010).TheinteractionisstrongestclosetothepKa
oftheligandanddecreasesifthepHisclosetothepIofthe
protein,whichisclearlyobservedhereforToyopearl,Capto
MMC,NuviacPrimeandPOROSHS.AtpH4,thenetcharge
ofnapinisaround12whileatpH8,thenetchargedecreases
significantlytoaround4.Thenetchargeofnapinat
Fig.3–EffectofionicstrengthonnapinadsorptiononCaptoMMC,POROSHSandPPAHyperCelatpH6andpH8.Symbols representexperimentalresults.Errorbarsresultedfromduplicateexperiments.Linesrepresentthemixedmodeisotherm thatthedatahasbeenfittedto.
sequence(obtainedfromtheRCSBProteinDataBank(PBD))
andEq.(16)(Supplementarymaterials).
znet=
i Nbasic,i 1+10pH−pKai − Nacidic,i 1+10pH−pKai (16)Whereznet isthe net chargeNbasic,i isthe numberofbasic
aminoacidsandNterminal,andNacidic,iisthenumberofbasic
aminoacidsandCterminal.
Theoppositebehavior isobservedwith theweak anion
mixed mode resin PPA HyperCel: an increase in
adsorp-tion capacitywithanincrease ofpH.Sincethe pKaofthe
PPA HyperCel resin ligand is around 8, the resin is
posi-tivelycharged.Becausenapinisalsopositivelychargedatall
theevaluatedpH values,thereisanelectrostaticrepulsion
betweenthe resin and the protein.The adsorptionis
con-sequentlyaresultofthehydrophobicinteractions between
napinandtheresinandtheelectrostaticrepulsion.ThepH
hasabiginfluenceontheadsorptionofnapininPPAresin,
astheisothermslopeandisothermshapechangewithpH:
thereisahighernapinaffinitytotheresinatahigherpH.
However,theadsorptioncapacityisstilllowerthanthe
max-imumobtainedatfeedconditionpH6and0.3MNaClwith
otherresins(e.g.40mg/mLbyCaptoMMC),whichdecreases
PPAresinscore.
Eventhough,thereisanotoriouschangeonnapin
adsorp-tionatdifferentpHvalues,theCaptoMMCresinistheleast
influenced,showingafavorableisothermevenatpH8.On
theotherhand,napinadsorptionstrengthonNuviacPrime,
POROS 50 HS and Toyopearl is significantly reduced. The
isothermslopesofallthe resinsatdifferentconditionsare
indicatedinTable3.
Theisothermsloperepresentstheaffinityoftheproteinto
theresin(thebindingstrength)whichisclearlymuchhigher
forCaptoMMCatalmostanypHvalue.However,thisisnot
necessary an advantage,as a higher pH might be needed
todesorbnapinwiththisresin,possiblyinterferingwithits
nativestructure,stablebetweenpH3–12(Wanasundara,2011),
Theseresultscanbeusedtoidentifydesorptionconditions
ofthebestperformingresin.
3.2.2. Effectofionicstrengthonnapinadsorption
Higher ionic strength is known to promote hydrophobic
interactionsduetotheso-called“salting-out”effect.Three
dif-ferentsaltconcentrations(ionicstrengths)weretestedattwo
differentpHconditions.Mostofthetestedresinsshowed
sim-ilartrends:adecreaseofnapinadsorptionstrengthwithan
increaseofionicstrength.Thiscanbeobservedbythelower
isothermslopevaluesandexperimentalcapacitiesobtained
athighersaltconcentrations(SeeFig.3andTable3).This
sug-geststhatnapinadsorptiononmixedmoderesinsismainly
controlledbyionicinteractions.
TheexceptionwasPPAresin,whichpresentedanincrease
innapin adsorptionwith increasing salt concentration. As
previouslymentioned,sinceforthisresinelectrostatic
inter-actionsarerepulsive,anincreaseinthebindingcapacityof
napinshouldbemainlyduetoanincreaseinthestrengthof
hydrophobicinteractions.Nosignificantimprovementinthe
adsorptionofnapinwas observedbetweenpH6and pH8,
atdifferentsaltconcentrations.Thismightsupportthefact
thatnapinadsorptiononPPAHyperCelischaracterizedonly
byhydrophobicinteractionsinthatrange.
Basedontheisotherm slope,onecannotice thatCapto
MMChasthelargestaffinityfornapin.However,thismight
complicatethedesorptionand recoveryoftheproteinafter
thecapturestep,asevenathighpHandsaltconcentrations
theisothermisfavorable.Asaconsequence,especiallyharsh
conditions(e.g.pHhigherthanthepIofthenapin)mightneed
tobeapplied.Thedeterminednapinisotherm slopesatall
testedconditionsareshowninTable3.
Fromtheobtainedresultsitisclearthattheeffectofionic
strengthismoresignificantthantheeffectofpH,especiallyfor
thePPAHyperCelresin.Whencomparingtheresultsforthat
resin(Fig.3eandf),theisothermsseemtooverlap.Thiswas
expected,astheadsorptionofnapinoccursmainlythrough
hydrophobicinteractions.Thiseffectcanbeevaluatedby
com-paringtheratiobetweentheisothermslopesattwodifferent
saltconcentration(ataspecificpH)andtheratiobetweenthe
isothermslopesattwodifferentpHconditions(ataspecific
saltconcentration).
Similar isotherm trendsto the onesdetermined inthis
workwereobservedintheresearchofNforetal.(2010).The
authorsevaluatedlysozyme(similarpIandmolecularweight
asnapin)adsorptiononCaptoMMCandPPAHyperCel.The
adsorptioncapacityoflysozymewasinthesameorderof
mag-nitudeastheoneobtainedfornapininthiswork.Inaddition,
similarisothermcurveswereobservedwhencomparingtwo
pHvaluesandonesaltconcentration.
Allthisinformationwasusedtoselectthemostsuitable
resinfornapincapture.
3.3. Proteinmixtureadsorptionexperiments
Batch adsorptionexperiments were performed using
solu-tionsofaproteinmixture(napin+cruciferin),toevaluateboth,
theeffect ofcruciferinonnapin’sadsorptionand to
deter-minecruciferin’sadsorptionisotherms.Atfeedconditions,it
isexpectedthatcruciferinpoorlybindstotheresins,asthepH
isclosetothepIofthisprotein(∼7).Acomparisonbetweenthe
expectednapinadsorptioncapacityandtheadsorption
capac-ityobtainedfromtheexperimentsusingtheproteinmixture
wasdoneandtheresultscanbefoundinFig.4.
Fig.4–Napinadsorptioncapacitycomparisonbetween expectedcapacityandexperimentallyobtainedfrom proteinmixtureexperimentsatpH6and0.3MNaCl. Expectedcapacitywascalculatedfromsinglecomponent isotherm.
Thepresenceofcruciferinseemstoenhancenapin
adsorp-tionatpH6and0.3MNaClforPOROSHSandCaptoSwhere
higher adsorption capacities than expected were obtained.
Thisbehaviormightbeexplainedbythechargeofthe
pro-teins.AtpH6,bothcruciferinand napinpossessapositive
netcharge,allowingcruciferintobindtotheresinsaswell.
However,cruciferinisconsideredalargeprotein,andin
solu-tion withnapin, the positivecharge oncruciferin’s surface
mighthaveresultedinastrongelectrostaticrepulsionofnapin
molecules.Bycreatingalessfavorablechemicalenvironment
intheliquidsolution,thepresenceofcruciferinmightthenbe
improvingtheadsorptionofnapinmoleculesontothetested
resins.
In all experiments (Supplementary material), cruciferin
showedaverysmallchangeinconcentrationafter
equilibra-tionwiththeresins,whichindicatedalowbindingstrength.
OnlyresinPPAHyperCelshowedarelativelyhighcruciferin
concentrationchange.Thecruciferin’sisothermslopevalues
were obtained by fitting a linear curve to the
experimen-tallyobtainedcapacities(asfunctionoftheequilibriumliquid
concentrations).Theobtainedvalueswere2.0+1.3,0.8+0.1,
2.3+1, 22.1+8.2, 1.6+0.5 and 0.9+0.7mL/mLresin for Capto
MMC,NuviacPrime,Toyopearl,PPAHyperCel,POROSHSand
Capto S respectively. Cruciferin’s isotherm slope is
signifi-cantly lower than the oneof napin for resin Capto MMC,
NuviacPrimeandPOROSHS,suggestingapreferential
bind-ingofnapintotheseresins.Napinisothermslopevalueswere
slightlyhigherforresinsCaptoSandToyopearlandcruciferin’s
isotherm slopewas higherthan napin’sisothermslopefor
resin PPA HyperCel. Asdiscussed before,for PPAHyperCel
resin,bindingismainlycharacterizedbyhydrophobic
inter-actions.Because,cruciferinisamorehydrophobicmolecule
thannapin,thismightexplainthehigheraffinitytowardsthis
resin.Alltheobtainedresultswereusedtoidentifythemost
suitableresinforthecaptureofnapin.
3.4. Resinselection
Themostsuitableresin fortheseparationofnapinan
cru-ciferin was selected using the previously definedselection
criteria (2.6 Resin selection). 1) Napin adsorption capacity
atfeedconditions(pH6with0.3MNaCl),2) bestdesorption
conditions(basedonthelowestisothermslopeobtainedfor
Table4–Valuesofeachcriterion(napincapacity,selectivity,resinpriceanddesorption)usedforresinselectionand resinscores.
Napincapacitya(mg/mL) Selectivityb(-) 1/slopedesorptionc(mL
resin/mL) Price/highestprice(-) Resinscore
CaptoMMC 40.0 219.9 0.1d 0.8 66%
NuviacPrime 34.3 81.3 0.8e 0.8 57%
Toyopearl 14.1 2.5 2.6f 1.0 37%
PPAHyperCel 26.6 0.4 0.5g 1.0 41%
POROSHS50 30.0 33.0 5.6h 0.5 75%
a Napincapacityat0.3MNaClandpH6.
b Ratiobetweennapinancrucferinisothermslopesat0.3MNaClandpH6. c Inverseofisothermslope:
d at0.7MNaClandpH8; e at1.0MNaClandpH6; f at1.0MNaClandpH6; g at0.1MNaClandpH6; h at0.7MNaClandpH8.
price.Adsorption(napincapacity)and desorptionwere the
two criteria with higher importance in the selection
pro-cess.Aspreviouslymentioned,resinsCaptoS,CMSheparose
andMacroPrepwherediscarded,astheyshowedsignificantly
lower napincapacity comparedtothe other resins atfeed
conditions.TheresultsareshowninTable4.
EventhoughresinCaptoMMCpossessesthehighestnapin
capacity(in allconditions tested)andselectivity,this resin
wasnotselectedasthemostpromising.Thereasonhasto
domainly withdesorption, possiblyrequiring a pHhigher
thannapin’spI,which,aspreviouslymentioned,canleadto
problemswiththestructuralstabilityofnapin.Therefore,
des-orbingatapHabovethepIisnotrecommended.Thebest
performing resin in terms of desorption performancewas
POROSHS, sinceitsresultsindicatethatdesorption canbe
performedathighsaltconcentrationwithoutincreasingthe
pHtoahigher valuethan napin’spI.Despite beingsecond
bestondesorptioncriteria.theToyopearlresinalsoshowed
amuchloweradsorptioncapacityfornapinwhencompared
toCaptoMMC andNuviacPrimeatfeedconditions.
There-fore,theToyopearlresinscoredthelowestintheadsorption
capacitycriterion.
PPAHyperCelandPOROSHSshowedcomparable
adsorp-tion capacities at feed condition. Nevertheless, from the
experiments performed with the protein mixture it was
clear that cruciferin also interacts with resin PPA
Hyper-Celhavinganevenlarger isothermslope than napin. This
wasobservedwiththeobtainedselectivityvaluewhichwas
lowerthanunity,indicatingthatatpH6and0.3MNaCl,
cru-ciferininteractsmorefavorablywithPPAHyperCelresinthan
napin.
Thelastevaluatedselectioncriterionwasresincost,with
resinPOROSHSbeingthe cheapestone,asthepricerange
isaroundhalfthepriceobtainedforthemixedmoderesins.
Eventhoughthiscriterionwasevaluatedwithalowerweight,
itwastakenintoaccount,asprocesseconomicsisasignificant
factorinprocessdevelopment.
3.5. Mixedmodeisothermparameterestimation
NapinadsorptiondataobtainedforPOROSHSresinwasused
inorder tofit theisotherm modeldescribed in 2.8.1Mixed
mode isotherm model. Since POROS HS is a strong cation
resin,therearenohydrophobicinteractionspresentedandthe
parameternfromEq.(7)wassettozero.Inordertoreducethe
complexityofsolvingthenon-linearsystem,themodelwas
linearizedbyapplyingthenaturallogarithminbothsidesof
theequationandtheregressionwasperformedasexplained
in2.5.2Parameter estimation.Theexperimentaldatafrom
thethreeevaluatedsaltconcentrationswerefittedto
deter-minethesetofparametersKeq,v,KpandKs,foreachpHvalue.
Themaximumadsorptioncapacity(qmax
p )wasobtainedfrom
adsorptionequilibriumexperimentsfromproteinmixtureand
theobtainedvaluewas44mg/mLresin.Theregressed
parame-tersareshowninTable5.
Thestoichiometriccoefficient(v) ofthe saltcounter ion
at pH6 was higher than at pH8, which was expected. For
thissystem,visequaltoproteinbindingcharge.AtpH
val-uesapproachingthepIoftheprotein(napinpI∼11),thenet
chargeoftheproteinapproacheszero(lower)andthe
electro-staticbindingstrengthbecomesweaker,resultinginalower
bindingcapacity.Kpistheparameterthatdescribesthe
differ-encebetweenwater-proteinandprotein–proteininteractions
(Mollerup,2006).TheregressedKpforbothpHvalues(6and
8)hadapositivevalue,suggestingthatinteractionsbetween
protein-water were stronger than protein-protein
interac-tions. Thisalso matches the observations obtainedin this
workandthedatafromliterature,indicatingahigh
solubil-ityofnapininwater.SimilartoKp,theparameterKsdescribes
thedifferencebetweenwater–proteinandsalt–protein
inter-actions (Mollerup, 2006). The obtained Ks values at both
evaluatedpHconditionswerearoundzero,implyingthatthe
strength ofprotein-saltinteractionsisnotsignificantly
dif-ferentthanprotein-waterinteractions.Byanalyzingthetwo
differentinteraction parameters-Kp andKs -,thestrength
ofwater-protein,protein-saltandprotein-proteininteractions
couldbeinferred.Inthissystem,giventhatKp>0andKs≈0,
protein-waterinteractionsweredominant.
ThepHvalueofmobilephasehasaninfluenceonthe
bind-ingchargeoftheproteinwhichconsequentlychangesthev
valueanditisexpectedtobelowerwhenthepHiscloserto
thepIoftheprotein.InordertoincludetheinfluenceofpHon
thestoichiometriccoefficient,approachesliketheonefrom
Pirrungetal.(2018)couldbeused.ThevforPOROSdecreases
withincreasingofpHwhichcorrespondstotheinfluenceof
pHonelectrostaticinteractions.
Alinearisothermmodelwasconsideredforcruciferin
Table5–IonexchangeisothermparametersofnapinproteininPOROSHSresinatpH6andpH8.
lnKeq v KP(mM-1) KS(mM-1) R2 SD
POROSHSpH6 6.6+0.3 1.9±0.4 2.3±2.7 −5×10-3+1×10-3 0.97 0.2
POROSHSpH8 5.8±0.2 1.2±0.2 0.8±1.0 −9×10-3+8x10-3 0.98 0.1
3.6. Columnchromatography
Twopulseexperimentswereperformedatsmallscale(∼10mL columnvolume)usingeitherasolutionofnapinora solu-tionofproteinmixture(napin+cruciferin).Theexperimental resultswere comparedwith the model output and can be foundinFig.5.Thedataobtainedfromthechromatography
stationwasnormalizedforeasiercomparisonwiththe
mod-elingresults.
AscanbeseeninFig.5,themodelisinexcellentagreement
withtheexperimentalresultsfornapinassinglecomponent
andalsowhennapinisinthepresenceofthesecondprotein,
cruciferin.Analyzing napinresults(Fig.5a),onecannotice
thatnodesorptionofnapinoccursafter5CVofwashingstep,
whichsuggestsastrongbindingattheseconditions(asshown
bythedeterminedisotherm).Elutionwithhighsalt
concen-tration(0.7MNaCl)and pH8showsaverysharp peak.This
wasexpectedasattheseconditions,napincapacityinPOROS
HSisclosetozero.HighsaltandpHconditionswerechosenin
ordertohavehighrecovery,whichcorrespondto100%based
onmassbalance(evaluatedbynumericalintegration).Note
thathighpHmightnotbeneededtofullydesorb this
pro-tein.Afterdesorption phase,the columnwaswashed with
aconcentratedsaltsolution(1MNaCl)untilnoabsorbance
at280nmwasdetected.Themaximumnapinconcentration
detectedwas6.8+0.1g/L.
Protein mixture chromatogram is presented in Fig. 5b.
Napinelutionprofileusingthisbinarymixturecorresponds
to the one obtained using single component napin. After
applyingtheelutionbuffer,againasharp peakisobserved
withapeakmaximumataround6.0CV(approximatelyone
columnvolumeaftertheswitchinconditions).Inthis
exper-iment, anextra peakisobserved duringthe washing step,
whichmightcorrespondtocruciferinprotein.Usingalinear
isothermmodelforcruciferinprotein(slope1.6mL/mL)and
performingthesimulation,onecanobservedapeakataround
1.8CV(washingstep).Comparingcruciferinsimulationresults
withexperimentalresults,itisclearthatcruciferinisotherm
slopewasoverestimated.Thisbecausetheexperimentalpeak
exitedthe columnearlier(∼1CV)andit issharperbut also
presentedtailing.Thistailingmightbecausedbythesizeof
cruciferin,whichisaround300kDa.Thedifferencebetween
theexperimentalandsimulatedresultsmightbeattributed
tocruciferinequilibriumparameter.Aspreviouslymentioned,
duringbinarymixtureexperimentscruciferindatapresented
verysmallchangeswhichweredifficulttoquantify.Besides
thesedifficulties, the model shows excellent agreement in
respecttonapinprotein.Themaximumnapinconcentration
obtainedinthechromatogramoftheproteinmixturepulse
experimentwas7.8+0.1g/L.Thisvalueishigherthantheone
obtainedintheexperimentwithpurenapinexperimentdue
totheslightlyhighernapinconcentrationapplied.
Using column experimental results from the protein
mixture, the slope of cruciferin was calculated to be
0.6mL/mLresin,whichconsequentlyincreasedtheselectivity
ofnapintoPOROSHSresinto87.0.
Thevalidatedmodelcanbethenusedforindustrialcolumn
designtoidentifycolumnsizingand operatingparameters.
Scaleupofchromatographyprocessesfromlaboratoryresults
isusuallydonebykeepingbedheightandvelocityconstant
andchangingcolumndiameter.However,thismakes
scalabil-ityinflexible,oftenresultingincolumnvolumesthatcannot
satisfythedesiredcapacityorthatdomarchthecapacities
foundatpilotand industrialscale,wherecompaniesoften
possessalreadyexistingequipment.Pre-determinedvolumes
(Stabyetal.,2017),withfixedcolumndiameterandadjustable
headers. The use of the previous model could reduce the
numberofexperimentsorinvestmentrequiredtosatisfy
pro-duction.
The following section describe the use of the previous
modeltodesignanionexchangecolumnfortheseparation
oftheseproteins.Themodelwasappliedtoahypothesized
casestudy consideringthe productionofrapeseedmealin
theNetherlands.Thiscasestudycanbeusedasbaseforlarge
scalepurificationofoilseedmeals.
3.7. Adsorptiveprocessdesignforthepurificationof napin,cruciferinandsinapicacidfromrapeseedmeal
Rapeseedproductionhassignificantlyincreasedinthe last
years, havinga globalproduction of75 million Mtin2018
where 25 million Mt were produced in Europe (Food and
AgricultureOrganizationoftheUnitedNations(FAO),2019).
Asmentionedbeforethemainproductobtainedfrom
rape-seedisoilforhumanconsumption, usuallyextracted from
theseedbymechanicalandsolventextraction(Fetzeretal.,
2018;Wanasundaraetal.,2017).Asbyproduct,rapeseedmeal
withahighproteincontent(∼40%drybasis),mainly
consist-ingofcruciferinandnapin(Contrerasetal.,2019;Lomascolo
et al., 2012; Wanasundara et al., 2017)isgenerated. In the
Netherlands,around3500MTofrapeseedmealwere
gener-atedin2018,correspondingtoaround1400MToftotalprotein.
Aconceptualdownstreamprocess(DSP)ofrapeseedmealfor
thepurificationofproteinsandpolyphenolsisshowninFig.6.
TheDSPincludes:anaqueousextractionassistedwithsalts
(Bérotetal.,2005)tosolubilizethedifferentcomponentsin
themeal,followedbyasmallmoleculeseparation(phyticacid,
polyphenolscompounds,carbohydrates)usingultrafiltration
with acut-off membraneof3kDa. Theprotein rich
reten-tateissenttoanionexchange(IEX)columnwherenapinis
capturedandcruciferinflowsthrough.Thepermeatecanbe
furtherprocessedtorecoversinapicacid(majorpolyphenolin
rapeseedmeal)usingadsorption(hydrophobicresins),as
rec-ommendedbyMoreno-Gonzálezetal.(2020)andSilvaetal.
(2018a) orliquid-liquid extraction(withorganic solventsor
ionicliquids)assuggestedbySilvaetal.(2018b).Thepurified
fractions of napin and cruciferin, obtained after the
chro-matographycolumn,arethensenttoanothermembraneunit
toremovethesaltsbeforebeingsenttothefinaldryingstage.
Following the adsorptionmethod forsinapic acid recovery
fromMoreno-Gonzálezetal.(2020),elutionandregeneration
mix-Fig.5–ModelpredictionvsExperimentaldataofNapin(a)andProteinmixture(b)onPOROS50HScolumn.Loading conditionpH6and0.3MNaCl,stepelutionatpH8and0.7MNaCl.Concentrationdatawasnormalized.
tureswhichcanberecoveredbydistillationandrecycledtothe
column.Toobtainsinapicacidcrystals,amultipleeffect
evap-orationcanbeusedtocrystalizethepolyphenol(Silvaetal.,
2018b).Theevaporatedsolventisdirectedtothedistillation
columnandsinapicacidcrystalsarethenfiltered,washedand
dried.
Consideringthattheannualproductionofrapeseedinthe
Netherlandsis5836tons(FoodandAgricultureOrganization
ofthe United Nations (FAO), 2019) and assuming that the
firststagesoftheprocess(aqueousextraction,centrifugation,
smallmoleculesseparation)accountanoverallproteinyield
of90%theinitialproteinextractflowratecanbeestimated.
Theionexchangecolumnforproteinpurificationcanbe
sizedusingthepreviousmodel.TheselectedresinPOROS50
HShasapolymericmatrix(rigidparticles)andaparticlesizeof
50mandbasedonsupplierspecification,thisresinpossesses
high mechanicalresistance (100bars). For astainless-steel
column,a maximumof 50barpressure drop was setfor a
columndiameterof1m(Schmidt-Traub etal., 2013).
Maxi-mumcolumnheightusingthisresincanbeestimatedusing
Karman-Cozenyequation,anditcorrespondsto1m.
Thecolumnmodelisthenusedtodeterminetheoperation
oftheionexchangecolumn,inspecifictoidentifytheloading
volume,whichisthiscaseistwocolumnvolumes.The
load-ingvolumeislowbecausetheconcentrationofeachproteinin
theproteinrichextractisrelativelyhigh(higherthan8g/L)and
theresingetstoequilibriumsaturationwithrelativelylow
vol-ume.Forthissimulation,anisothermslopeof0.6mL/mLresin
wasconsidered forcruciferin equilibrium.Themodel
indi-catesaresinutilizationof82%andanapinyieldof98%with
>99%purity.Intheflowthrough,cruciferinisrecovered(>99%
yieldand>98%purity).Theproductivitycalculatedwiththis
columnsize(1mdiameterand1mheight)is26.3gNapin/LR/h,
whichconsequentlyproduces52.1MT/a.Thisaccountsto12%
ofthetotalamountofnapinthancanberecoveredfromthe
rapeseedmealgeneratedintheNetherlandsusing oneion
exchangecolumn.
Itispossibletopurifytheannualproductionofrapeseed
mealbycalculatingthenumberofcolumnsneededto
oper-ateinparallel andthe number ofcyclesthat each column
can berun. It isimportant tokeep in mind that this is a
preliminaryevaluationandthatbreakthroughcurve
experi-mentsfornapinwillbestillneededtocorroboratetheselected
loading.Inaddition,forthisestimationasimilarcolumn
oper-ationintermsofdesorptionwasselected(5CV)whichcould
bereduced,asbasedonthecolumnexperiments,fullnapin
recoverymightbeachievedwithless volume.Operation of
theionexchangecolumncanalsobeadaptedtocontinuous
(Simulatedmovingbed)orsemi-continuous(CaptureSMBand
PeriodicCountercurrentChromatography,PCC)withmultiple
columnsconnectedinseriesoroperatinginparallel.In
addi-tion,economic feasibilityshould beassessedwith adetail
economicevaluationtoidentifyoverall capitalexpenditure
(CAPEX)andtheoveralloperationalcost(OPEX)and
perform-ingforinstanceacashflowanalysis.
4.
Conclusions
Highthroughputprocessdevelopment(HTPD)generates
reli-ableandsignificantinformationinashorttimeperiod.Inthe
caseofadsorptiveprocesses,HTPD,differentresinsand
condi-tionscouldbeassessedinparallelwhichallowsproperresin
choice.Thegeneratedinformationcanbeusedin
combina-tionwithmechanisticmodelstoevaluatetechnicalfeasibility
ofaprocessdesignwhichinvolvesprocessunderstandingand
contributestofastprocessdevelopment.
Declaration
of
interests
The authors declare that they have no known competing
financialinterestsorpersonalrelationshipsthatcouldhave
appearedtoinfluencetheworkreportedinthispaper.
Acknowledgments
ThisworkwassupportedbytheISPT(InstituteofSustainable
ProcessTechnology)undertheprojectCM-20-07inAdsorption
ofnon-volatilesfromfoodproducts.Thankstotheindustrial
partnersDSM,Unilever,FrieslandCampinaandRoyalCosun
fortheirvaluableinputthroughtheproject.Theauthorsare
alsothankfultoYiSongfromDelftUniversityofTechnology
foralltheprovidedlabsupport.
Appendix
A.
Supplementary
data
Supplementary material related to this article can be
found, in the online version, at doi:https://doi.org/10.
1016/j.fbp.2020.11.011.
References
Sevillano,D.M.,vanderWielen,L.A.M.,Hooshyar,N.,Ottens,M.,
2014.Resinselectionfortheseparationofcaffeinefromgreen teacatechins.FoodBioprod.Process.92,192–198.
Adachi,M.,Kanamori,J.,Masuda,T.,Yagasaki,K.,Kitamura,K.,
Mikami,B.,Utsumi,S.,2003.Crystalstructureofsoybean11S
globulin:glycininA3B4homohexamer.Proc.Natl.Acad.Sci. U.S.A.100,7395–7400.
Aider,M.,Barbana,C.,2011.Canolaproteins:composition,
extraction,functionalproperties,bioactivity,applicationsasa foodingredientandallergenicity–apracticalandcritical review.TrendsFoodSci.Technol.22,21–39.
Akbari,A.,Wu,J.,2015.Anintegratedmethodofisolatingnapin
andcruciferinfromdefattedcanolameal.LWT-FoodSci. Technol.64,308–315.
Bérot,S.,Compoint,J.P.,Larré,C.,Malabat,C.,Guéguen,J.,2005.
Largescalepurificationofrapeseedproteins(Brassicanapus L.).J.Chromatogr.B.818,35–42.
Brenner,H.,Gaydos,L.J.,1977.Theconstrainedbrownian
movementofsphericalparticlesincylindricalporesof comparableradius:modelsofthediffusiveandconvective transportofsolutemoleculesinmembranesandporous media.J.ColloidInterfaceSci.58,312–356.
Carta,G.,Jungbauer,A.,2010.ProteinChromatography:Process
DevelopmentandScale-up.JohnWiley&Sons.
Contreras,Md.M.,Lama-Mu ˜noz,A.,ManuelGutiérrez-Pérez,J.,
Espínola,F.,Moya,M.,Castro,E.,2019.Proteinextractionfrom
agri-foodresiduesforintegrationinbiorefinery:potential techniquesandcurrentstatus.Bioresour.Technol.280, 459–477.
Felinger,A.,Guiochon,G.,2004.Comparisonofthekinetic
modelsoflinearchromatography.Chromatographia60 (Suppl.1),S175–S180.
Fetzer,A.,Herfellner,T.,Stäbler,A.,Menner,M.,Eisner,P.,2018.
Influenceofprocessconditionsduringaqueousprotein extractionuponyieldfrompre-pressedandcold-pressed rapeseedpresscake.Ind.CropProd.112,236–246.
FoodandAgricultureOrganizationoftheUnitedNations(FAO),
2019.FAOSTATProductionQuantity(accessed24March2019)
http://www.fao.org/faostat/en/#data/QC.
Gerzhova,A.,Mondor,M.,Benali,M.,Aider,M.,2015.A
andconventionalextractionmethodontheextractability, compositionandphysicochemicalpropertiesofcanola proteinconcentratesandisolates.FoodBiosci.11,56–71.
Ghodsvali,A.,Khodaparast,M.H.H.,Vosoughi,M.,Diosady,L.L.,
2005.Preparationofcanolaproteinmaterialsusing
membranetechnologyandevaluationofmealsfunctional properties.FoodRes.Int.38,223–231.
Gunn,D.J.,1987.Axialandradialdispersioninfixedbeds.Chem.
Eng.Sci.42,363–373.
Hou,F.,Ding,W.,Qu,W.,Oladejo,A.O.,Xiong,F.,Zhang,W.,He,R.,
Ma,H.,2017.Alkalisolutionextractionofriceresidueprotein
isolates:influenceofalkaliconcentrationonprotein
functional,structuralpropertiesandlysinoalanineformation. FoodChem.218,207–215.
Jyothi,T.,Sinha,S.,Singh,S.A.,Surolia,A.,Rao,A.A.,2007.Napin
fromBrassicajuncea:thermodynamicandstructuralanalysis ofstability.BiochimicaetBiophysicaActa(BBA)-Proteinsand Proteomics1774,907–919.
Kodagoda,L.P.,Nakai,S.,Powrie,W.D.,1973.Somefunctional
propertiesofrapeseedproteinisolatesandconcentrates. CanadianInstituteofFoodSci.Technol.J6,266–269.
LeVan,D.M.,Carta,G.,Yon,C.M.,1999.Adsorptionandion
Exchange.In:Perry,R.H.,Green,D.W.(Eds.),Perry’SChemical Engineers’Handbook.,sevened.McGraw-Hill,NewYork,pp. 16–19.
Lomascolo,A.,Uzan-Boukhris,E.,Sigoillot,J.-C.,Fine,F.,2012.
Rapeseedandsunflowermeal:areviewonbiotechnology statusandchallenges.Appl.Microbiol.Biotechnol.95, 1105–1114.
Mollerup,J.M.,2006.Appliedthermodynamics:anewfrontierfor
biotechnology.FluidPhaseEquilibr.241,205–215.
Mollerup,J.M.,2007.Thethermodynamicprinciplesofligand
bindinginchromatographyandbiology.J.Biotechnol.132, 187–195.
Mollerup,J.M.,Hansen,T.B.,Kidal,S.,Staby,A.,2008.Qualityby
design—thermodynamicmodellingofchromatographic separationofproteins.J.Chromatogr.A1177,
200–206.
Moreno-González,M.,Girish,V.,Keulen,D.,Wijngaard,H.,
Lauteslager,X.,Ferreira,G.,Ottens,M.,2020.Recoveryof
sinapicacidfromcanola/rapeseedmealextractsby adsorption.FoodBioprod.Process.120,69–79.
Murray,D.E.,Terrence,M.J.,Barker,L.D.,Myers,C.D.,US4208323A,
USApatent1980.ProcessforIsolationofProteinsUsingFood
GradeSaltSolutionsatSpecifiedpHandIonicStrength.
NationalCenterforBiotechnologyInformation,2020.PubChem
Database.Tryptophan,CID=6305(accessedonJan.16,2020)
https://pubchem.ncbi.nlm.nih.gov/compound/Tryptophan.
Nfor,B.K.,Noverraz,M.,Chilamkurthi,S.,Verhaert,P.D.,vander
Wielen,L.A.,Ottens,M.,2010.High-throughputisotherm
determinationandthermodynamicmodelingofprotein adsorptiononmixedmodeadsorbents.J.Chromatogr.A1217, 6829–6850.
Perera,S.P.,McIntosh,T.C.,Wanasundara,J.P.,2016.Structural
Propertiesofcruciferinandnapinofbrassicanapus(canola) showdistinctresponsestochangesinphandtemperature. Plants.5,36.
Pickardt,C.,Neidhart,S.,Griesbach,C.,Dube,M.,Knauf,U.,
Kammerer,D.R.,Carle,R.,2009.Optimisationofmild-acidic
proteinextractionfromdefattedsunflower(Helianthusannuus L.)meal.FoodHydrocoll.23,1966–1973.
Pirrung,S.M.,ParrucadaCruz,D.,Hanke,A.T.,Berends,C.,Van
Beckhoven,R.F.W.C.,Eppink,M.H.M.,Ottens,M.,2018.
Chromatographicparameterdeterminationforcomplex biologicalfeedstocks.Biotechnol.Prog.34,1006–1018.
Pudel,F.,Tressel,R.P.,Düring,K.,2015.Productionandproperties
ofrapeseedalbumin.LipidTechnol.27,112–114.
Raab,B.,Schwenke,K.D.,1984.Simplifiedisolationprocedurefor
the12Sglobulinandthealbuminfractionfromrapeseed
(BrassicanapusL.).Nahrung28,863–866,
http://dx.doi.org/10.1002/food.19840280820.
Rommi,K.,Ercili-Cura,D.,Hakala,T.K.,Nordlund,E.,Poutanen,
K.,Lantto,R.,2015.Impactoftotalsolidcontentand
extractionphonenzyme-aidedrecoveryofproteinfrom defattedrapeseed(BrassicarapaL.)presscakeand
physicochemicalpropertiesoftheproteinfractions.J.Agric. FoodChem.63,2997–3003.
Schmidt,I.,Renard,D.,Rondeau,D.,Richomme,P.,Popineau,Y.,
Axelos,M.A.-V.,2004.Detailedphysicochemical
characterizationofthe2sstorageproteinfromrapeseed (BrassicanapusL.).J.Agric.FoodChem.52,5995–6001.
Schmidt-Traub,H.,Schulte,M.,Seidel-Morgenstern,A.,2013.
PreparativeChromatography,seconded.JohnWiley&Sons, Incorporated,Weinheim,Germany.
Silva,M.,Castellanos,L.,Ottens,M.,2018a.Captureand
purificationofpolyphenolsusingfunctionalizedhydrophobic resins.Ind.Eng.Chem.Res.57,5359–5369.
Silva,M.,García,J.C.,Ottens,M.,2018b.Polyphenolliquid-liquid
extractionprocessdevelopmentusingNRTL-SAC.Ind.Eng. Chem.Res.57,9210–9221.
Staby,A.,Rathore,A.S.,Ahuja,S.,2017.Preparative
ChromatographyforSeparationofProteins.JohnWiley& Sons,Incorporated,NewYork.
Vuorela,S.,Meyer,A.S.,Heinonen,M.,2004.Impactofisolation
methodontheantioxidantactivityofrapeseedmeal phenolics.J.Agric.FoodChem.52,8202–8207.
Wakao,N.,Smith,J.M.,1962.Diffusionincatalystpellets.Chem.
Eng.Sci.17,825–834.
Wanasundara,J.P.,2011.ProteinsofBrassicaceaeoilseedsand
theirpotentialasaplantproteinsource.Crit.Rev.FoodSci. Nutr.51,635–677.
Wanasundara,J.P.D.,McIntosh,T.C.,Perera,S.P.,
Withana-Gamage,T.S.,Mitra,P.,2016.Canola/rapeseed
protein-functionalityandnutrition.OCL23(4),D407.
Wanasundara,J.P.D.,Tan,S.,Alashi,A.M.,Pudel,F.,Blanchard,C.,
2017.Chapter18-proteinsfromcanola/rapeseed:current
status.In:Nadathur,S.R.,Wanasundara,J.P.D.,Scanlin,L. (Eds.),SustainableProteinSources.AcademicPress,San Diego,pp.285–304.
Wilson,E.J.,Geankoplis,C.J.,1966.Liquidmasstransferatvery
lowReynoldsnumbersinpackedbeds.Ind.Eng.Chem. Fundam5(1),9–14.
Xu,L.,Diosady,L.L.,2002.Removalofphenoliccompoundsinthe
productionofhigh-qualitycanolaproteinisolates.FoodRes. Int.35,23–30.
Young,M.E.,Carroad,P.A.,Bell,R.L.,1980.Estimationofdiffusion
coefficientsofproteins.Biotechnol.Bioeng.22,947–955.
Zhang,S.B.,Wang,Z.,Xu,S.Y.,2007.Downstreamprocessesfor
aqueousenzymaticextractionofrapeseedoilandprotein hydrolysates.J.Am.OilChem.Soc.84,693.
Zhu,M.,Carta,G.,2016.Proteinadsorptionequilibriumand
kineticsinmultimodalcationexchangeresins.Adsorption22, 165–179.