Journal of Chromatography A, 1376 (2015) 7483

Contents lists available at ScienceDirect

Journal of Chromatography A

jo ur nal ho me pag e: www.elsev ier .com/ locate /chroma

Fabricating electrospun cellulose nanobre adsorion-exc

Stewart . Bra Department o C1H b Innovations T , ScienOxford, Didcot OX11 0QX, UKc School of Science and Technology, Nottingham Trent University, Nottingham, NG1 4BU, UK

a r t i c l e i n f o

Article history:Received 24 JuReceived in reAccepted 4 DeAvailable onlin

Keywords:ElectrospinninConvective maDiethylaminoeCarboxylateTEMPO-media

a b s t r a c t

Protein separation is an integral step in biopharmaceutical manufacture with diffusion-limited packed

1. Introdu

The conprescriptionestimated t

CorresponCollege LondoTel.: +44 0 20

E-mail add

http://dx.doi.o0021-9673/ ly 2014vised form 2 December 2014cember 2014e 12 December 2014

gss transferthyl

ted oxidation

bed chromatography remaining the default choice for industry. Rapid bind-elute separation using con-vective mass transfer media offers advantages in productivity by operating at high owrates. Electrospunnanobre adsorbents are a non-woven bre matrix of high surface area and porosity previously inves-tigated as a bioseparation medium. The effects of compression and bed layers, and subsequent heattreatment after electrospinning cellulose acetate nanobres were investigated using diethylaminoethyl(DEAE) or carboxylate (COO) functionalisations. Transbed pressures were measured and compared bycompression load, COO adsorbents were 30%, 70% and 90% higher than DEAE for compressions 1, 5 and10 MPa, respectively, which was attributed to the swelling effect of hydrophilic COO groups. Dynamicbinding capacities (DBCs) at 10% breakthrough were measured between 2000 and 12,000 CV/h (2 s and0.3 s residence times) under normal binding conditions, and DBCs increased with reactant concentrationfrom 4 to 12 mg BSA/mL for DEAE and from 10 to 21 mg lysozyme/mL for COO adsorbents. Comparingcapacities of compression loads applied after electrospinning showed that the lowest load tested, 1 MPa,yielded the highest DBCs for DEAE and COO adsorbents at 20 mg BSA/mL and 27 mg lysozyme/mL, respec-tively. At 1 MPa, DBCs were the highest for the lowest owrate tested but stabilised for owrates above2000 CV/h. For compression loads of 5 MPa and 10 MPa, adsorbents recorded lower DBCs than 1 MPa asa result of nanobre packing and reduced surface area. Increasing the number of bed layers from 4 to 12showed decreasing DBCs for both adsorbents. Tensile strengths were recorded to indicate the mechanicalrobustness of the adsorbent and be related to packing the nanobre adsorbents in large scale congu-rations such as pleated cartridges. Compared with an uncompressed adsorbent, compressions of 1, 5and 10 MPa showed increases of 30%, 110% and 110%, respectively, for both functionalisations. The datapresented show that capacity and mechanical strength can be balanced through compression after elec-trospinning and is particular to different functionalisations. This trade-off is critical to the development ofnanobre adsorbents into different packing congurations for application and scale up in bioseparation.

2014 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY license(http://creativecommons.org/licenses/by/3.0/).

ction

tribution of biotechnology products to the global and over-the-counter pharmaceutical markets wereo be worth $118 billion in 2011 with increased focus

ding author at: Department of Biochemical Engineering, Universityn, Bernard Katz Building, Gordon Street, London WC1H 0AH, UK.7679 9580; fax: +44 0 20 7209 0703.ress: d.bracewell@ucl.ac.uk (D.G. Bracewell).

in the therapy areas of oncology, anti-diabetes and vaccines [1].Some individual products are reaching annual sales of over $1billion [2]. As the market moves towards developing more com-plex biomolecules such as fusion proteins and antibody fragments,purication stages in downstream processing are becoming moreexpensive. The advancement of cell line engineering in upstreamprocessing, including transfection methods and media develop-ment, in upstream processing have realised increased product titresover the past two decades [3]. However, downstream processinghas yet to achieve a dramatic improvement in process ef-ciency partly due to limitations in widely used packed-bed resins

rg/10.1016/j.chroma.2014.12.0102014 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/3.0/).hange chromatography

R. Dodsa,b, Oliver Hardicka, Bob Stevensc, Daniel Gf Biochemical Engineering, University College London, Bernard Katz Building, London Wechnology Access Centre Micro and Nanotechnology, Rutherford Appleton Laboratorybents for

acewell a,

0AH, UKce and Technology Facilities Council, Harwell

S.R. Dods et al. / J. Chromatogr. A 1376 (2015) 7483 75

including diffusive mass transfer, achievable ow rates and scale-up volumes. Protein bioseparation media using convective masstransfer such as porous membranes and monoliths have receivedincreased attention because they avoid this diffusion limitation andhave a higher capture efciency and reduced buffer use to improveoverall productivity [4]. In the last 30 years, rigid porous monolithshave also been introduced and developed. The single solid contin-uous matrix has no interstitial voids and can also vastly improveproductivity by operating at much higher owrates than packed-bed chromatography [5]. Current advantages in industry have beenrealised in the polishing stage of monoclonal antibody purica-tion using owthrough mode where a membrane column bindsimpurities and allows the target to pass through [6].

Nanobre electrospinning involves passing a viscous polymersolution through a microneedle charged at a high voltage (>5 kV)to deposit a continuous bre strand to a grounded collector andform a non-woven mat with a bre diameter of less than 1 m[7]. Electrospun nanobres have been investigated for a mul-titude of applications including tissue engineering [8], catalysisand sensors [9,10], ltration [11] and composites [12]. Celluloseis a commonly used material in membrane chromatography andltration for being chemically resistant, cheap and has good non-specic binding properties [4]. However, cellulose raises manychallenges in electrospinning because it is difcult to dissolve andthe solvent systems required can lead to non-uniform nanobredeposition lulose derivby regenerabre depos(viscosity), ditions havacetate nancal strengthFig. 1 showdifferent mresin and anadsorbent. porosity wi

Chemicahydroxyl grhave been bioseparatimers includlaminoethyfabricated b

ethyl chloride hydrochloride (DAECH) to show improved sepa-ration productivity compared with porous membranes [19,20].Alcohol groups on cellulose have been controllably oxidised tocarboxylate (COO) groups using (2,2,6,6-tetramethylpiperidin-1-yl)oxyl (TEMPO) as a catalyst and sodium hypochlorite as theoxidant [21,22]. The main application of TEMPO-mediated oxida-tion is in the preparation of nanocellulose from wood pulp wherethe ionic repulsion of COO groups helps force cellulose bres apartduring processing, reducing the mechanical energy required [23].The use of TEMPO-mediated oxidised electrospun cellulose nano-bre has been used before to bind metal ions [22] and viruses[24]. The physical and chemical methods applied in fabricatingelectrospun cellulose nanobre adsorbents affect bioseparationperformance and controlling parameters is important to fabricatinga reproducible material. Compressing nanobre sheets combinedwith annealing via heat treatment is used to further improvemechanical properties than heat treatment alone. A robust nano-bre adsorbent is essential in scaled up packed bed congurationssuch as pleated cartridges, as seen in membrane chromatogra-phy. However, chemical modications applied to nanobres mayadversely affect morphology and structure.

2. Materials and methods

2.1. Fabricating cellulose nanobre adsorbents

shor cons

i.d. withity se

(Mrn watio as sp(200 age sneed00 m2, whus stquarn tw

die.lic p

Fig. 1. Scanni nd S cand heat treat ked-bediameters and[13]. As such, electrospinning readily dissolvable cel-atives such as cellulose acetate are preferred followedtion to cellulose via hydroxide treatment. For uniformition of cellulose acetate, controlling polymer solutionow rate and voltage as well as environmental con-e been shown to be critical [14]. Annealing celluloseobres with heat is a common step to improve mechani-

by creating spot welds at bre strand overlap points.s scanning electron microscopy (SEM) images of theorphologies for a cast porous membrane, packed-bed

annealed electrospun regenerated cellulose nanobreA nanobre adsorbent balances a high surface area andth the benets of convective mass transfer.l modications of chromatographic media usingoups on the support for application in bioseparationresearched [15]. Electrospun nanobre adsorbents inon have been reported for cellulose [16] and other poly-ing polysulfone [17] and polyacrylonitrile [18]. Diethy-l (DEAE) cellulose electrospun nanobres have beeny Williamson ether synthesis using 2-(diethylamino)

To mat of0.5 mmin linehumidacetatesolutioat a rtion wdrum tion stof the sheet 630 g/mprevio[19]. Sbetweeas thehydrau

ng electron microscopy images comparing protein purication media. (a) Sartobied regenerated cellulose nanobre adsorbent. (c) Fractogel EMD TMAE HiCap pac

approximately 0.1 m pore diameter.ten the electrospinning time and produce nanobreistent bed height, four microneedles (100 mm length;) were used and the collector was moved side-to-side

the needle array. The operating voltage was 30 kV, thet to 70% and temperature to 25 C. A 20 wt.% cellulose

= 29,000, 40% acetyl groups, Sigma-Aldrich, Dorset, UK)as prepared in acetone:DMF:ethanol (SigmaAldrich)of 2:2:1 as previously described [14,19]. The solu-un at 2.5 mL/h for 10 h. The collector was a rotatingmm dia.; 300 mm length) set at 60 rpm on a transla-et at 300 mm x-axis displacement (150 mm either sidele array centre) at a rate of ve loops per minute. Am 180 mm was produced equating to approximatelyich was comparable to the nanobre mat used in our

udy but a reduction in spinning time from 36 h to 10 hes (80 mm 80 mm) were cut, layered and placed ino 100 mm 100 mm square aluminium blocks to act

Compression was performed for 2 min in a manualress (Specac, Kent, UK) under different loads of 1000,

ellulose membrane (Sartorius Stedim, Epsom, UK). (b) Compressedd resin (EMD Millipore, Darmstadt, Germany) with 4090 m bead

76 S.R. Dods et al. / J. Chromatogr. A 1376 (2015) 7483

5000 and 10,000 kg as indicated on the gauge, corresponding to0.98, 4.9 and 9.8 MPa, respectively, for brevity, we used roundedup values; 1, 5 and 10 MPa, respectively. To study the effects of themain reactant concentration, 8 layers were compressed at a load of5 MPa. To scompressedcompressedin a preheafor 30 min. by deacetylovernight. Tdeacetylatio

2.2. DEAE m

Cellulosform DEAE100 mL deioof DAECH atdamaging trepeat reaccellulose (2was stirred(90 C) 0.5 Mdissolve anwas rinsed

2.3. TEMPO

A COO-cmodied frous mixturSigma-AldrNaOH. The nused to dropfor the thrsodium hypitored to enof the C6 hwas 10 minEthanol (10and stirred pure waterchlorite (Siacid) was p[26].

2.4. Morph

SEM imaWorld BV, Eof 10 keV. ric softwarFourier traATR) was ua Thermo Sough, UK). 4000500 cmeasured wstrength m10 MPa, weinto a tensohighest reco

2.5. Equilibrium adsorption capacities

To assess the nanobre equilibrium binding capacity the matswere cut into 25-mm discs. Discs were incubated in 0.02.0 mg/mL

protV-63ter bAE c

10 mlyso. Thndinbres . Thr

cons (CaxKdCein bearis

bestightaki, perimlculae adroteients

BCs a

dynasicent ausly ution

anddsorpell eq

betw of 2O adessedent amg/m

DEA wit0 cmffer

The equid voteine loae in m anes wAE a, 0.1sorbL, 4 lessels. Toeaseordelter htudy the effects of physical modication, 8 layers were at loads of 1, 5 and 10 MPa and 4, 8 and 12 layers were at 5 MPa. The nanobre sheet was immediately placedted oven (NR30F, Carbolite, Shefeld, UK) set at 213 CCellulose acetate mats were regenerated to celluloseation using 0.1 M sodium hydroxide in 2:1 H2O:EtOHhe addition of ethanol was essential to ensure completen.

odication

e adsorbents can be reacted directly with DAECH to ligands via alkylation [19,20]. A reaction solution ofnised water was employed with varying concentrations

50 and 200 mmol/g cellulose stirred at 250 rpm to avoidhe nanobre mat. To improve DEAE functionalisation, ation was performed using 200 mmol DAECH per gram 200 mmol) as previously reported [25]. The reaction

for 15 min at 250 rpm. Then the mat was treated in hot NaOH solution for 10 min to complete the reaction and

y unwanted reactants. The DEAE-cellulose adsorbentin copious amounts of water.

-mediated oxidation

ellulose adsorbent was produced following a procedureom that previously reported [21]. A 100 mL aque-e of TEMPO (0.002 g; Sigma-Aldrich) and NaBr (0.02 g;ich) was adjusted to a pH of 10.5 using aqueous 0.1 Manobre mat was stirred for 5 min. A syringe pump waswise add sodium hypochlorite (NaClO; Sigma-Aldrich)ee concentrations investigated; 5, 10 and 20 mmolochlorite (NaClO) per gram cellulose. The pH was mon-sure pH remained above 10.5 to encourage oxidationydroxyl on the cellulose. The time taken to add NaClO

and the mixture was allowed to stir for a further 5 min. mL, Sigma-Aldrich) was added to quench the reactionfor 10 min. The mat was washed thoroughly with ultra-. To oxidise any remaining aldehyde groups, sodiumgma-Aldrich) treatment (0.45 g in 45 mL in 1 M aceticerformed for 48 h in the dark to as previously described

ological, chemical and tensile strength analyses

ging was performed using a Phenom G2 Pro (Phenom-indhoven, The Netherlands) at an accelerating voltageImages were captured and analysed with Firbomet-e (Phenom-World BV) to estimate bre diameter.nsform infra-red attenuated total reectance (FTIR-sed to characterise the chemical group changes oncientic Nicolet iS10 FT-IR Spectrometer (Loughbor-Spectra were recorded from dry samples in the rangem1 by an accumulation of 50 scans. A background wasith 10 scans prior to each sampling. For ultimate tensile

easurements, compressed adsorbent samples, 1, 5 andre cut into 15 mm 10 mm (L W) strips and placedmeter. The strips were stretched at 1 mm/min and therded force before breaking was used.

model(Jasco ing, afthe DEtein instudy, pH 5.5tive binanotestingproteintrationQ = Qmof protThe linline ofbed heKawasThe exwas caand thered padsorb

2.6. D

TheAKTA Bsurempreviodistribbuffersrium awas wvariedtrationand COcompradsorband 5 for theformedat a 61tion bubuffer.lowingthe vothe promust bvolum0.014 cvolumFor DE10 MPaCOO ad0.20 mnon-prcontroto incrthe recing a ein solutions and UV absorbance readings at 280 nm0, Essex, UK) were taken at each step of before bind-inding (16 h), wash (1 h) and elution (1 h). For testingellulose nanobre, BSA was used as the model pro-M Tris buffer at pH 8.0. For the COO-cellulose bindingzyme was used in 20 mM sodium acetate buffer ate elution buffers used were the same as the respec-g buffer and containing 0.5 M NaCl. After elution, thewere regenerated in 0.1 M aqueous NaOH for repeatee tests were performed and the adsorbed equilibriumcentrations (Q) and liquid phase equilibrium concen-) were averaged. The Langmuir adsorption isotherm/Kd + C was used, where Qmax is the maximum capacityound, and Kd is the equilibrium dissociation constant.ed form of Langmuir isotherm was plotted and from a

t the Qmax and Kd values were estimated. The wet was measured with a digital micrometer (Mitutoyo,Japan; 0.001 mm resolution) to calculate the volume.

ent was performed three times. Elution performanceted as a ratio of the protein concentration after elutionsorbed equilibrium protein concentration. The recov-n concentration was 75% for DEAE and 90% for COO.

nd transbed pressures

amic binding capacity (DBC) was measured using an (GE Healthcare, Uppsala, Sweden) system with UV mea-t 280 nm. A custom-made 25-mm PEEK lter holder wasdesigned using frit spacers to ensure full radial ow

across an adsorbent at very high owrates [19]. The model proteins were the same as used in the equilib-tion study. The nanobre adsorbent in the lter holderuilibrated prior to binding. The binding owrates wereeen 10 and 610 cm/h. BSA or lysozyme at a concen-

mg/mL in a 2-mL sample loop was injected for DEAEsorbents testing in most cases. For 8-layer adsorbents

at 1 MPa all of the protein injected was bound to thend the protein concentration was increased to 3 mg/mLL of BSA and lysozyme to provide a maximum DBCE and COO adsorbents, respectively. Elution was per-h a 30% mix of 1 M NaCl in respective binding buffer/h and the adsorbent was further cleaned with elu-at 610 cm/h, followed by re-equilibration with bindingDBC was calculated at 10% breakthrough using the fol-ation DBC10% = ((V10% V0) + CLoad)/VBed where V0 islume of the entire system, CLoad is the concentration of

solution loaded, and V10% is the volume of sample thatded before achieving 10% breakthrough. VBed is the bedmillilitre measured from a bed height range betweend 0.07 cm with a digital micrometer (Mitutoyo). Bed

ere taken as an average of the three adsorbents tested.dsorbents: 1 MPa, 0.35 mL; 5 MPa (8 layers), 0.16 mL;6 mL; 4 layers, 0.07 mL; and 12 layers, 0.21 mL. Forents: 1 MPa, 0.34 mL; 5 MPa (8 layers), 0.19 mL; 10 MPa,ayers, 0.09 mL and 12 layers, 0.21 mL. Blank tests usingd unmodied cellulose adsorbents were performed as

measure transbed pressure, the AKTA was programmed in owrate in steps up to 50 mL/min and presented asd back pressure minus the system back pressure includ-older containing no adsorbent.

S.R. Dods et al. / J. Chromatogr. A 1376 (2015) 7483 77

3. Results and discussion

3.1. Surface morphology

SEM sergies and brepresentatricate DEAEmaterial prdle and hadopen with (Fig. 2a). Thand the samhere at an awere compduce regencompact mdiameters rc(ii) show by repeat tr20 mmol NDEAE adsobre strandbeen repories [19,20]. above protmatrix uponbents also sin the unmooxidation hpare nanocpolymer chwood pulp assists in degrade wothe DEAE angroups on tsion forcingnoticeable. nanobre mwas no lonDAECH for considered in the morpcation.

3.2. FTIR-AT

FTIR-ATRgroups on ththe differenFig. 3. Celluclear from t1740 cm1,(OC O); 12the 33003showed no low concenthe cellulos[19]. Fig. 31731 cm1,salt (COONaC O peak hThe applicawith detect

3.3. Tensile strength

Previous investigations into non-compressed nanobre adsor-bents suffered damage during chemical modication from mixing

us remogg anful ts wratedand % inessed

imping pnd 1on e

diffe streere tand C

proion hegradpulscatiod thampleutescked

ansb

nsbeadsool D

Fig. g comte wiessioessurositon-cdsorEAE b). C

to ethyg effangey theous

shee beessurecorice asent obee of ed vessuhad c

stre capves as an important tool for investigating morpholo-re diameters of electrospun nanobres. Fig. 2 showsive SEM images and modication steps taken to fab-

and COO adsorbents. The cellulose acetate startingior to any compression or baking was fragile to han-

a cotton-wool like texture. The morphology appearedlarge black spaces between the straight bre strandse electrospinning conditions were previously reportede small range in nanobre diameters was measured

verage of 0.5 m [14]. Eight layers of cellulose acetateressed at 5 MPa, oven-baked and deacetylated to pro-erated cellulose. The general appearance was a moreatrix with strands retaining a linear appearance andemained in the same range (Fig. 2b). Figs. 2c(i) andbre matrices following chemical modications to DEAEeatment of DAECH (2 200 mmol) and COO groups ataClO concentration per gram adsorbent, respectively.rbents showed a slightly distorted appearance withs losing some of their linear character, which has notted for singly treated DEAE nanobres in other stud-More than two repeats of DAECH treatment using theocol led to a complete degradation of the nanobre

drying, becoming a hard opaque material. COO adsor-how a loss of linear character and distortion not seendied regenerated cellulose matrix. TEMPO-mediatedas use in the nanobrillation of wood pulp to pre-ellulose particles and brils. By charging the celluloseains with COO groups the process of homogenisingconsumes less energy because ionic repulsion forcesbril separation [23]. The ionic repulsion is used tood brils into nanobrils 34 nm in diameter [27]. Ind COO nanobre adsorbents here, the effect of charged

he nanobre strand in solution may lead to ionic repul- nanorils apart that make up the strand becomingUpon drying for SEM analysis the appearance of theatrix was only slightly distorted because the surface

ger charged. The chemical conditions of 2 200 mmolDEAE and 20 mmol NaClO for COO adsorbents werethe highest possible before any considerable changehology rendered the adsorbent unsuitable for appli-

R

was employed to investigate the changes in chemicale surface of cellulose nanobre adsorbents throughoutt modication and representative spectra are shown inlose acetate deacetylation to regenerated cellulose washe replacement of acetate peaks (ester carbonyl (C O);

carbon-methyl (CCH3); 1365 cm1 and ester linkage21 cm1) by a broad and larger alcohol (OH) peak in500 cm1 region. DEAE modication (2 200 mmol)new peaks because the weak stretching and relativelytration of the tertiary amine bonds were masked bye peaks, regardless how high a modication is usedb shows that COO-cellulose created a new peak at

corresponding to the carbonyl group of the carboxylate). As the amount of oxidant NaClO was increased, theeight increased, indicating an increase in COO groups.tion of FTIR-ATR was convenient to investigate groupsable peaks.

and thand hospinninbe useSampleregeneDEAE and 30compr10 MParecord130% adeviatitisticaltensileples wDEAE ing anyoxidatbre dionic remodiimpliethis sacontriband pa

3.4. Tr

Trausing 200 mmbents. varyinowracomprbed prand pothan nCOO athan D(Fig. 5shownboxymswellinThe chsised b

Poras atincreasbed pr5 MPa ing twadsorbof nanincreassmall bbed prwhich tensilereducequired a less vigorous approach, hindering the amounteneity of functionalisation. Compressing after electro-d followed by heat treatment was therefore found too reduce failures and produce consistent adsorbents.ere compared with non-compressed and heat treated

cellulose indicated as the No Press sample (Fig. 4).COO adsorbents compressed at 1 MPa showed a 40%crease in tensile strengths compared with the non-

sample, respectively. Increasing compression to 5 androved tensile strength over non-compressed further,ercentage differences of 85% and 105% for DEAE and20% for COO, respectively. However, the large standardrrors for 5 and 10 MPa compressions suggest no sta-rence between them and may indicate a maximum ofngth achieved for these cellulose adsorbents. The sam-ested when dry and the weak ion-exchange groups ofOO would be in their neutral form and not exhibit-

nounced effect from ionic repulsion. TEMPO-mediatedas been used in nanocellulose production, assistingation during mechanical processing in solution throughion [27]. The changes in morphology between chemicalns were negligible and similarities in tensile strengthst chemical modication has little or no effect, at least in

size. Improving the mechanical strength of adsorbents to creating a robust material capable of being handled

into large scale pleated or spiral-wound congurations.

ed pressures of varying bed layers and compressions

d pressures were recorded for increasing owratesrbents prepared at chemical modications of 2AECH for DEAE and 20 mmol NaClO for COO adsor-

5a shows similarly increasing transbed pressures forpressions of 8-layer DEAE adsorbents with increasing

th 10 MPa showing the highest of the three. Increasingn during fabrication was expected to increase trans-res because the nanobre matrix was more packedy was reduced. DEAE transbed pressures were higherompressed DEAE adsorbents previously reported [19].bents show considerably higher transbed pressuresand differences between increasing compression loadsOO groups are hydrophilic and have been previouslycause a higher back pressure when comparing car-l - with DEAE modied cellulose beads, suggesting thisect contributes to increases in transbed pressure [28].

in matrix packing was clear in COO adsorbents empha- hydrophilic nature of COO groups.membranes, like nanobre adsorbents, are producedets and designing a media with multiple layers tod height is one method to increase bed volume. Trans-res of 4-, 8- and 12-layer adsorbents compressed atded noticeable differences with COO adsorbents show-

high pressures than DEAE (Fig. 5c and d). The 12-layerbed volume was similar to 8-layer, but the packingr matrix much higher, which was evidenced in thetransbed pressure. The 4-layer adsorbents were of sucholumes that hardly any pressure was recorded. Trans-res provided an insight to the packing of nanobres,learly increased with increasing compression from thength results. However, higher transbed pressures mayacity because of the channelling effect seen in porous

78 S.R. Dods et al. / J. Chromatogr. A 1376 (2015) 7483

Fig. 2. Fabricamodication. Carboxylate (C

membranesarea of an a

3.5. Equilibmodication

Reactanusing 8-laytion of electrospun cellulose nanobre adsorbents and representative scanning electron(b) Regenerated cellulose adsorbent after 5 MPa compression, heat treatment and deaceOO) cellulose adsorbent.

where the proteins are not accessing all the surfacedsorbent.

rium absorption capacities of varying chemicals

t concentrations were varied to investigate capacitieser adsorbents compressed at 5 MPa (Fig. 6 and Table 1).

DEAE modiand 200 mmThe Qmax aof Langmucentration of the Kd v200 mmol atrolled usin microscopy images. (a) Cellulose acetate nanobre mat before anytylation. (c)(i) Diethylaminoethyl (DEAE) cellulose adsorbent. (c)(ii)

cations were varied with DAECH concentrations of 50ol/g cellulose and a repeated treatment (2 200 mmol).

nd Kd values were evaluated using the linearised formir isotherm. However, lacking data in the low con-region of the isotherm detracted from the reliabilityalues. Capacity was increased from the 50 mmol todsorbent, showing that functionalisation can be con-g DAECH amount. The 200 mmol concentration Qmax of

S.R. Dods et al. / J. Chromatogr. A 1376 (2015) 7483 79

Fig. 3. FTIR-AT s showCellulose acet t. (b)(C O) from ca

Fig. 4. TensileNo Press show

13 mg BSA/muncompresof Zhang etviously shocapacity hethat of singlincrease in NaClO (Fig.ber of COO20 mmol Qmpreviously rresins, whicThe swellincult to accocontribute thave been

Table 1Equilibrium bithe maximum

Sample

Reactant con

Qmax (mg/mKd (mg/mL) R spectra of diethylaminoethyl (DEAE) and carboxylate (COO) cellulose adsorbentate (CA) starting material, regenerated cellulose (RC) and DEAE cellulose adsorbenrboxylate salt group (COONa) with increasing concentrations of NaClO applied. strengths of diethylaminoethyl (DEAE) and carboxylate (COO) cellulose adsorbents incrn was eight layers of uncompressed regenerated cellulose. Error bars indicate SD.

L was similar to that we previously recorded for ansed nanobre adsorbent [19] but was lower than that

al. [20]. Repeating the DAECH treatment has been pre-wn to increase adsorbent capacity and the equilibriumre was increased to 27.4 mg BSA/mL, which was twicee treatment [25]. COO nanobre adsorbents showed anQmax for increasing concentrations of oxidising reagent,

6b and Table 1) and agrees with the increased num- groups suggested in the FTIR spectra (Fig. 3b). Theax of 47.5 mg lysozyme/mL was comparable to someeported values for commercially available packed-bedh typically have exceptionally high surface areas [29].g effect noted in the transbed pressure tests was dif-unt for under equilibrium binding conditions and mayo increasing capacity. Other COO nanobre adsorbentsstudied for electrospun polyacrylonitrile where Chiu

et al. achieof a similatechniquesbinding capnanobre ain this studof polymer ues suppormedium.

3.6. DBCs o

An AKTAmate the Dsame adsorshown to e

nding study of the varying chemical modications used to fabricate diethylaminoethyl (D capacity of protein bound, Qmax, and dissociation constant, Kd , estimated using a Langmu

DEAE cellulose

c. 50 mmol 200 mmol 2 200 mmol L) 7.5 13.0 27.4

0.077 0.044 0.11 ing the change in chemical groups during adsorbent fabrication. (a) COO modication shows an increasing height of the carbonyl peakeased with increasing compressions applied during fabrication. The

ved an equilibrium binding capacity using lysozymerly high value as in this study [30]. Polymer grafting

have shown advantages in vastly improving proteinacities [31]. An equilibrium capacity range using a

dsorbent was shown to be 425 times higher than thaty, also using lysozyme and dependent on the amountgrafted [16]. These studies reporting high capacity val-t using the high surface of nanobres as an adsorbent

f varying chemical modications

system and custom lter holder were used to esti-BCs at 10% breakthrough at varying owrates of thebents studied above (Fig. 7). The residence times arexemplify how little time is required for convective

EAE) and carboxylate (COO) cellulose nanobre adsorbents, detailingir linear regression t.

COO cellulose

5 mmol 10 mmol 20 mmol

11.8 18.9 47.50.065 0.034 0.049

80 S.R. Dods et al. / J. Chromatogr. A 1376 (2015) 7483

Fig. 5. Transb lose n(b) Eight-layer orben

mass transand rangedfor the highing study a21 mg lysozadsorbents typically infor COO andsurface area large capplateaued fbed chroma

Fig. 6. Equilib5 MPa and mo5.3 was used fed pressures of electrospun diethylaminoethyl (DEAE) and carboxylate (COO) cellu DEAE and COO adsorbents at varying compressions. (c) and (d) DEAE and COO adsfer of the target protein with a nanobre adsorbent from 4 s to 0.3 s. The highest DBCs were recordedest functionalisations found in the equilibrium bind-nd were 12 mg BSA/mL for 2 200 mmol DEAE andyme/mL for 20 mmol COO. At these DBCs, nanobrecompare poorly against packed-bed media, which are

the 3085 mg/mL range for DEAE and 40100 mg/mL carboxymethyl resins, despite suggesting reasonable

a for binding in the equilibrium study [32]. There wasacity drop between the Qmax values and DBCs. DBCsor increasing owrates, where in traditional packed-tography we would expect continual loss through ow

distributionhas been padsorbent tproperties explanationlter holdedistributiondyes. Such tstructure ofor bindingtions. Anotof the ion-

rium binding adsorption isotherms of 8-layer electrospun diethylaminoethyl (DEAE) andied under different reactant concentrations. (a) BSA in 10 mM Tris buffer at pH 8 was uor COO. Error bars indicate SD of the average Q and C values taken from three replicatesanobre adsorbents of varying bed layers and compressions. (a) andts of varying bed layers compressed at 5 MPa. effects on the diffusion mass transfer. This differencereviously reported for a non-pressed DEAE nanobreested under similar conditions [19]. Flow distributionof the custom lter holder used were considered as an

for the difference between Qmax values and DBCs. Ther was developed using spacer frits for complete ow

across an adsorbent and tested visibly using colouredests would not be able to reveal the internal microscalef nanobre matrix and some areas may be unreachable

under ow conditions but available under static condi-her contributing factor would be the chemical natureexchange groups. The hydrophilic nature of the COO

d carboxylate (COO) cellulose nanobre adsorbents compressed atsed for DEAE testing and (b) lysozyme in 20 mM acetate buffer at pH.

S.R. Dods et al. / J. Chromatogr. A 1376 (2015) 7483 81

Fig. 7. Dynamic binding capacities at 10% breakthrough of 8-layer electrospun (a) diethylaminoethyl (DEAE) and (b) carboxylate (COO) cellulose nanobre adsorbentscompressed at 5 MPa and modied under different reactant concentrations using identical binding conditions as before with NaCl elution. Error bars indicate SD.

group and the swelling effect attributed to causing higher transbedpressure may allow for greater capacity under static conditions ifa gel-like layer was formed around nanobre strands. However,this reasoning can only be applied to the COO adsorbents andnot DEAE. Further investigation into the structure of ion-exchangenanobre adsorbents is required. Convective mass transfer mediaand particularly the non-dead-end structure of nanobre adsor-bents, have the ability to operate at considerably higher owrateswhich benets the overall productivity to separate proteins [19,33].Therefore a high dynamic capacity was less important than

attaining a repeatable capacity at higher owrates where in largescale devices, capacity can be circumnavigated with higher volumesof adsorbent.

3.7. DBCs of varying bed layers and compressions

DBCs at 10% breakthrough for 8-layer adsorbents compressedat loads of 1, 5 and 10 MPa applied after electrospinning and adsor-bents of varying bed layers (4, 8 and 12) compressed at 5 MPawere investigated for functionalisations 2 200 mmol DEAE and

Fig. 8. Dynamlayers and com(d) DEAE and Cic binding capacities at 10% breakthrough of electrospun diethylaminoethyl (DEAE) anpressions using identical binding conditions as before with NaCl elution. (a) and (b) EigOO adsorbents of varying bed layers compressed at 5 MPa. Error bars indicate SD.d carboxylate (COO) cellulose nanobre adsorbents for varying bedht-layer DEAE and COO adsorbents at varying compressions. (c) and

82 S.R. Dods et al. / J. Chromatogr. A 1376 (2015) 7483

20 mmol COO (Fig. 8). The DEAE and COO chemistry protocolswere different in capacities with adsorbents compressed at 1 MParecording the highest DBCs at the lowest owrate tested of 900 CV/hrecording 20 mg BSA/mL and 27 mg lysozyme/mL, respectively. Adecreasing present andThe ow disfor this initgenerally colow transbecreate littleas a cause oDBC was fodecreasing for DEAE anthe adsorbetein bindingall owratestrength pradequate tr

DEAE an5 MPa showbents wherrespectivelybents showsimilar DBCdecrease andue to the ha second cocompressiothe bed vol

4. Conclus

Compresnanobre atowards theical properscale devicehence batchthese bioseand COO mphologies, strengths wincrease wiwith no sigbed pressuloads of DEto the hydrthe highestas a repeateand 20 mmthe lowest lthe lowest binding. At increasing However, Dat 5 and 102000 CV/h.

This studfunctionalisical to the rand operatibre materiadevelopme

balance of capacity and material strength and tailor the material tothe application.

Acknowledgements

thanarlysistaical S

nces

trickla. Aze

reconds BiBalasutegie485eng, Es to p. Etzejugate. Liu, Jnt for uang

ions iniu, X.ue enhang, ctrosp480onwasent s13) 42. Tijintillatio462ucche

enhahnol. Frey, 08) 37ardic

hum. 46 (2. Herm, 2013

Menk, et aimensh cap037a, M

U) be (200hang

of elmbranHardih pro911hang,

memmbr. SHiroteneraluloseaito, Ase ox190aito, Mizationng TEM219a, C.

le mat Chen,luloseaito, Ars by

(200capacity from 900 CV/h to around 2000 CV/h was then DBC appeared to stabilise for up to 12,000 CV/h.tribution through the lter holder could be responsibleial capacity drop for increasing owrate but DBCs weremparable for increasing owrates over 2000 CV/h. Thed pressures of 1 MPa DEAE and COO suggested they

resistance to ow and discourage channelling effectsf the DBC drop. A detrimental effect of compression onund for increasing loads 5 MPa and 10 MPa with DBCsfrom 12 to 9 mg BSA/mL and 20 to 17 mg lysozyme/mLd COO, respectively. The increasing loads would packnts more and this reduced available surface area for pro-

would lead to lower DBCs. The DBCs were stable acrosss tested and combined with the improved mechanicaloperties, and then a loss in DBC could be considered anade-off for large scale application.d COO adsorbents of varying bed layers compression ated high DBCs at 10% breakthrough for the 4-layer adsor-e the bed volume was very low at 0.07 mL and 0.09 mL,. Increasing the number of bed layers for DEAE adsor-

ed reduced DBCs. For COO 12-layer adsorbents recordeds to 8-layer up to 3000 CV/h but then DBCs began tod may suggest that channelling effects were presentigh transbed pressures. The bed layer results indicatednsideration in fabricating nanobre adsorbents wheren at 5 MPa can create a robust material but increasingume with layers further reduced DBCs.

ions

sion and heat treatment steps during the fabrication ofdsorbents allow their physical properties to be tunedir application as a chromatography medium. Mechan-ties are critical for handling and packing into larges and impact operable owrates, column capacity and

operation time. Functionalisations also directly affectparation properties. The differences between DEAEodications are clear as shown by changes in mor-

transbed pressures and capacities. Absorbent tensileere similar for DEAE and COO and were found to

th greater levels of compression after electrospinningnicant difference between functionalisations. Trans-res show seemingly little effect between compressingAE and yet large changes for COO, which is attributedophilic COO groups. When studying protein separation

attainable capacities by functionalisation were foundd treatment of 200 mmol/g adsorbent DAECH for DEAEol/g NaClO for COO adsorbents. Nanobres prepared atevel of compression (1 MPa) yielded the highest DBCs atowrate, which indicates the available surface area for5 and 10 MPa compressions capacity was decreased andbed layers compressed at 5 MPa also decreased DBCs.BCs recorded remained stable for increasing owrate

MPa compressions while 1 MPa was only stable above

y shows that the interactions between fabrication andation in the synthesis of nanobre adsorbents are crit-equired physical properties of the material for packingng a bioseparation medium. This requires that nano-ls properties are measured and understood alongsidents in surface chemistry, in order to strike the correct

Wein the eand asBiolog

Refere

[1] I. S[2] A.M

freeTre

[3] B. stra477

[4] X. Ztion

[5] M.Rcon

[6] H.Fme

[7] Z. Hcat

[8] H. Ltiss

[9] L. Zele463

[10] R. Kpre(20

[11] L.Ddis435

[12] A. ZforTec

[13] M. (20

[14] O. HandSci

[15] G.TMA

[16] T.J.Liu3-dhig372

[17] Z. M(PS272

[18] H. Ztionme

[19] O. hig111

[20] L. ZtiveMe

[21] M. regCel

[22] T. Slulo183

[23] T. Sualusi199

[24] H. Msca

[25] L.F.cel

[26] T. Sbe289k the Research Complex at Harwell for use of facilities stages of the study and Dr. Robert McKean for guidancence. The study was funded by the Biotechnology andciences Research Council.

nd, World Preview 2018 Embracing the Patent Cliff, London, 2012.vedo, P.J. Rosa, I.F. Ferreira, M.R. Aires-Barros, Chromatography-very of biopharmaceuticals through aqueous two-phase processing,otechnol. 27 (2009) 240247.ndaram, S. Harrison, D.G. Bracewell, Advances in product releases and impact on bioprocess design, Trends Biotechnol. 27 (2009).. Ruckenstein, Membrane chromatography: preparation and applica-rotein separation, Biotechnol. Progr. 15 (1999) 10031019.l, T. Bund, Monoliths for the purication of whey protein-dextrans, J. Chromatogr. A 1218 (2011) 24452450.. Ma, C. Winter, R. Bayer, Recovery and purication process develop-monoclonal antibody production, MAbs 2 (2010) 480499., A review on polymer nanobers by electrospinning and their appli-

nanocomposites, Compos. Sci. Technol. 63 (2003) 22232253. Ding, G. Zhou, P. Li, X. Wei, Y. Fan, Electrospinning of nanobers forgineering applications, J. Nanomater. 2013 (2013) 111.A. Aboagye, A. Kelkar, C. Lai, H. Fong, A review: carbon nanobers fromun polyacrylonitrile and their applications, J. Mater. Sci. 49 (2013).rh, N. Karak, M. Misra, Electrospun cellulose acetate nanobers: thetatus and gamut of biotechnological applications, Biotechnol. Adv. 311437.

g, J.S. Choi, S. Lee, S.H. Kim, H.K. Shon, Recent progress of membranen using electrospun nanobrous membrane, J. Membr. Sci. 453 (2014).lli, M.L. Focarete, C. Gualandi, S. Ramakrishna, Electrospun nanobersncing structural performance of composite materials, Polym. Adv.22 (2011) 339349.Electrospinning cellulose and cellulose derivatives, Polym. Rev. 488391.k, B. Stevens, D.G. Bracewell, Nanobre fabrication in a temperatureidity controlled environment for improved bre consistency, J. Mater.011) 38903898.anson, Bioconjugate Techniques, third ed., Academic Press, Waltham,.haus, H. Varadaraju, L. Zhang, S. Schneiderman, S. Bjustrom, L.l., Electrospun nanober membranes surface functionalized withional nanolayers as an innovative adsorption medium with ultra-acity and throughput, Chem. Commun. (Cambridge, UK) 46 (2010)22.. Kotaki, S. Ramakrishna, Surface modied nonwoven polysulphoner mesh by electrospinning: a novel afnity membrane, J. Membr. Sci.

6) 179187., H. Nie, D. Yu, C. Wu, Y. Zhang, C.J.B. White, et al., Surface modica-ectrospun polyacrylonitrile nanober towards developing an afnitye for bromelain adsorption, Desalination 256 (2010) 141147.

ck, S. Dods, B. Stevens, D.G. Bracewell, Nanober adsorbents forductivity downstream processing, Biotechnol. Bioeng. 110 (2013)28.T. Menkhaus, H. Fong, Fabrication and bioseparation studies of adsorp-branes/felts made from electrospun cellulose acetate nanobers, J.ci. 319 (2008) 176184.a, N. Tamura, T. Saito, A. Isogai, Surface carboxylation of porousted cellulose beads by 4-acetamide-TEMPO/NaClO/NaClO2 system,

16 (2009) 841851.. Isogai, Ion-exchange behavior of carboxylate groups in brous cel-

idized by the TEMPO-mediated system, Carbohydr. Polym. 61 (2005).. Hirota, N. Tamura, S. Kimura, H. Fukuzumi, L. Heux, et al., Individ-

of nano-sized plant cellulose brils by direct surface carboxylationPO catalyst under neutral conditions, Biomacromolecules 10 (2009)

96. Burger, B.S. Hsiao, B. Chu, Ultra-ne cellulose nanobers: new nano-erials for water purication, J. Mater. Chem. 21 (2011) 75077510.

G.T. Tsao, Chemical procedures for enzyme immobilization on porous beads, Biotechnol. Bioeng. 19 (1977) 14631473.. Isogai, Introduction of aldehyde groups on surfaces of native cellulose

TEMPO-mediated oxidation, Colloids Surf. A Physicochem. Eng. Asp.6) 219225.

S.R. Dods et al. / J. Chromatogr. A 1376 (2015) 7483 83

[27] A. Isogai, T. Saito, H. Fukuzumi, TEMPO-oxidized cellulose nanobers,Nanoscale 3 (2011) 7185.

[28] D. Ishimura, Y. Morimoto, H. Saito, Inuences of chemical modications on themechanical strength of cellulose beads, Cellulose (1998) 135151.

[29] A. Staby, J.H. Jacobsen, R.G. Hansen, U.K. Bruus, I.H. Jensen, Comparison of chro-matographic ion-exchange resins V. Strong and weak cation-exchange resins,J. Chromatogr. A 1118 (2006) 168179.

[30] H.T. Chiu, J.M. Lin, T.H. Cheng, S.Y. Chou, Fabrication of electrospun poly-acrylonitrile ion-exchange membranes for application in lysozyme adsorption,Express Polym. Lett. 5 (2011) 308317.

[31] B.V. Bhut, J. Weaver, A.R. Carter, S.R. Wickramasinghe, S.M. Husson, The roleof polymer nanolayer architecture on the separation performance of anion-exchange membrane adsorbers: I. protein separations, Biotechnol. Bioeng. 108(2011) 26452653.

[32] W.W. Ward, G. Swiatek, Protein purication, Curr. Anal. Chem. 5 (2009)85105.

[33] S. Xie, F. Svec, J.M. Frchet, Rigid porous polyacrylamide-based monolithiccolumns containing butyl methacrylate as a separation medium for the rapidhydrophobic interaction chromatography of proteins, J. Chromatogr. A 775(1997) 6572.

Fabricating electrospun cellulose nanofibre adsorbents for ion-exchange chromatography1 Introduction2 Materials and methods2.1 Fabricating cellulose nanofibre adsorbents2.2 DEAE modification2.3 TEMPO-mediated oxidation2.4 Morphological, chemical and tensile strength analyses2.5 Equilibrium adsorption capacities2.6 DBCs and transbed pressures

3 Results and discussion3.1 Surface morphology3.2 FTIR-ATR3.3 Tensile strength3.4 Transbed pressures of varying bed layers and compressions3.5 Equilibrium absorption capacities of varying chemical modifications3.6 DBCs of varying chemical modifications3.7 DBCs of varying bed layers and compressions

4 ConclusionsAcknowledgementsReferences