Effect of PEGylation on the solution conformation of antibody … · Keywords: PEGylation; immunoglobulin; antibody fragment; conformation INTRODUCTION Biomedical exploitation of

Effect of PEGylation on the Solution Conformationof Antibody Fragments

YANLING LU,1 STEPHEN E. HARDING,1 ALISON TURNER,2 BRYAN SMITH,2 DILJEET S. ATHWAL,2

J. GÜNTER GROSSMANN,3 KENNETH G. DAVIS,1 ARTHUR J. ROWE1

1National Centre for Macromolecular Hydrodynamics, University of Nottingham, Sutton Bonington LE12 5RD, England, UK

2UCB-Celltech, 216 Bath Road, Slough, Berkshire SL1 1NH, UK

3CCLRC Daresbury Laboratory, Synchrotron Radiation Department, Warrington, Cheshire WA4 4AD, England, UK

Received 16 March 2007; revised 18 May 2007; accepted 13 June 2007

Published online in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/jps.21170

Corresponde115-951-6148; FE-mail: steve.ha

Journal of Pharm

� 2007 Wiley-Liss

2062 JOURN

ABSTRACT: Covalent attachment of poly(ethylene glycol) (PEG) to therapeutic anti-body fragments has been found effective in prolonging the half-life of the proteinmolecule in vivo. In this study analytical ultracentrifugation (AUC) in combinationwith small angle X-ray scattering (SAXS) has been applied to a number of antibodyfragments and to their respective PEGylated conjugates. Despite the large increase inmolecular weight due to the attachment of a 20–40 kDa PEG moiety, the PEGylatedconjugates have smaller sedimentation coefficients, s, than their parent antibodyfragments, due to a significant increase in frictional ratio f/fo (from �1.3 to 2.3–2.8):the solution hydrodynamic properties of the conjugates are clearly dominated by thePEG moiety ( f/fo �3.0). This observation is reinforced by SAXS data at high values of r(separation of scattering centres within a particle) that appear dominated by the PEGpart of the complex. By contrast, SAXS data at low values of r suggest that there are nosignificant conformational changes of the protein moiety itself after PEGylationThe location of the PEGylation site within the conjugate was identified, and found tobe consistent with expectation from the conjugation chemistry. � 2007 Wiley-Liss, Inc. andthe American Pharmacists Association J Pharm Sci 97:2062–2079, 2008

Keywords: PEGylation; immunoglobul
in; antibody fragment; conformation
INTRODUCTION

Biomedical exploitation of the discriminationand affinity of antibody binding properties hasbeen advanced by biotechnological developments.There are now 18 commercially available mono-clonal antibody products and more than 100 inclinical development. In the near future, engineer-ed antibodies are predicted to account for >30%

nce to: Stephen E. Harding (Telephone: þ44-0-ax: þ44-0-115-951-6142;[email protected])

aceutical Sciences, Vol. 97, 2062–2079 (2008)

, Inc. and the American Pharmacists Association

AL OF PHARMACEUTICAL SCIENCES, VOL. 97, NO. 6, JUN

of all revenues in the biotechnology market.1

High-level expression systems have been develop-ed for mammalian cell expression of antibodies;however, the large quantities needed in thera-peutic doses make them prohibitively costly.Antibody fragments (Fab’, Fv and scFv)2–4 andengineered variants (diabodies, triabodies, mini-bodies and single-domain antibodies)1,5–7 that canbe readily produced in large-scale in microbialsystems are now emerging as credible alternativesfor in vitro immunoassays and in vivo tumour-targeting therapy.8

The improved pharmacokinetics associated withantibody fragments in tumour penetration areoften modulated by the tendency of such fragments

E 2008

SOLUTION CONFORMATION OF ANTIBODY FRAGMENTS 2063

to have very short circulation times in vivo.6

However, enhancement of the pharmaceuticalproperties of antibody fragments by ‘‘PEGylation’’,or attachment of poly(ethylene glycol) (PEG) to thefragment, appears to confer a prolonged circulatinghalf-life on the antibody fragments.9–13 ProteinPEGylation modulates many properties of bio-medical significance, including reduced toxicity,reduced immunogenicity and antigenicity; slowingrates of clearance and proteolysis, and enhancingsolubility and stability (for reviews, see Refs. 14,15).

The molecular basis of the beneficial effects ofPEGylation of therapeutic proteins can be attri-buted to the unique physicochemical properties ofPEG itself (for reviews, see Refs. 16,17). One of themost distinct features of PEG is its propensity tooccupy a large volume in an aqueous environmentthrough time-averaged water association and pre-sents a volume 5–10 times larger than a solubleprotein of comparable molecular weight,11,14,16,18,19–21

and this can confer on a protein conjugated to itadvantageous features.

Numerous functionalised PEG molecules arenow available commercially.14,18,22 An appro-priate choice of the size and structure of thePEG moieties and the conjugation chemistry hasbeen found crucial to balance the desired pro-longed half-life in vivo, while maintaining anacceptable level of relevant biological activity oractivities (including not only antigen binding andantibody effector function, but also the ability tolocalise to certain tissues such as tumours).10

Hence, the molecular weight of the PEG moiety,its structure, the number and location of the

Figure 1. Schematic representation of an infragments. (a) IgG; (b) (Fab’)2; (c) Fab: no hingare linked through one disulfide bond; (d) Facysteine.

DOI 10.1002/jps J

PEG moieties attached to the protein, as wellas the chemical method of attachment, alldetermine the physicochemical and pharma-cological properties of the resulting proteinconjugate.10,23

Currently there are two different kinds ofPEG structure—linear or branched—which canbe attached to proteins. A branched structure witha single attachment site could be more favourablethan the linear structure since a high molecularweight PEG moiety can apparently be obtainedwithout increasing the number of attachmentsites.24,25 Moreover, branched chain PEGylatedproteins have been found to be more stable againstenzyme proteolysis than linear moieties.23,25

Mono-site attachment of a limited number oflonger chain PEG molecules rather than multi-site attachment of a greater number of smallerPEG molecules has been found most beneficial forretaining a higher level of biological activity.9,10

Generation of PEGylated protein requires site-specific conjugation strategies to minimise theamounts of unmated isomers produced.19,23,26 Inthe case of Fab’ (Fig. 1) it is possible to engineer acysteine residue in the ‘residual’ hinge region ofthe molecule (or more than one if required): tofacilitate specific PEGylation through reactionbetween the thiol group of the cysteine and themaleimide group present in a PEG-maleimideconstruct to form a stable thioether linkage.9,10,24

This reaction can generate site-specific PEGyla-tion of the Fab’ (Fig. 2) resulting in yielding ahomogeneous PEGylated product preserving thefull binding activities of the antibody fragment.9

tact antibody molecule and some antibodye region, the heavy chain and light chainb’: has a short hinge region with a free

OURNAL OF PHARMACEUTICAL SCIENCES, VOL. 97, NO. 6, JUNE 2008

Figure 2. Schematic illustration of site-specific Fab’ PEGylation. PEG-maleimidereacts with the thiol group in the hinge region of a Fab’ fragment forming a stablethioether linkage.

2064 LU ET AL.

Whilst the pharmacokinetics and pharma-codynamics of PEGylated proteins have beenextensively studied, their solution properties arenot as well characterised. Using thiol/maleimideconjugation chemistry a group of PEGylatedantibody products has been generated to allowinvestigation of their solution properties and toassess if PEGylation has influenced proteinconformation.

The samples were characterised using thecomplementary probes of analytical ultracentri-fugation (AUC) and small angle X-ray scattering(SAXS). AUC can tell us about the overallhydrodynamic properties of the complexes andwhether, from measurement of the sedimentat-ion coefficient and the translational frictionalratio whether it is the protein or PEG componentswhich dictate the properties.27,28 This techniqueoffers significant advantages compared to chro-matographic based methods for the study ofcomplex systems like these in that it coversa very wide range of molecular size without anyrequirement for known standards for calibration.There are no losses arising from macromolecule/gel matrix interactions caused by macromoleculesirreversibly binding to chromatographic media, orfrom the failure of the conjugate molecules to be‘‘electrophoresed’’ into polyacrylamide gels.

Another powerful tool we employ in thisstudy is small angle SAXS, which probes thedistribution of SAXS density in the scatteringparticle (see, e.g., 29,30), allowing deductions to bemade about the conformation of the scatteringparticles, and in particular if the effect of com-plexation changes the conformation of the proteincomponent in anyway. By combining ultracen-trifugation with SAXS, it is possible to make anassessment of the effect of PEGylation on themolecular integrity of a selection of antibodyfragments.

JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 97, NO. 6, JUNE 2008

MATERIALS AND METHODS

Production and Purification of Antibody Fragment

Three antibody fragments, human g1 Fab’; humang1 F(ab’)2 and murine g1 Fab’ and associatedPEGylated products were selected for study.

Human g1 Fab’ was derived from E. coli: cellslurry was extracted in 0.1 M Tris/10 mM EDTA,pH 7.4 for 16 h at 308C with constant agitation.The cells were removed by centrifugation at10,000 rpm, for 30 min, followed by 0.22 mmfiltration and adjustment to pH 7.0 using 2MTris-HCl, pH 8.5. The Fab’ was purified byapplication of the clarified periplasmic extractto a GammaBind G (GE-Healthcare, Chalfont St.Giles, UK) column previously equilibrated withphosphate buffered saline (PBS). The boundproduct was eluted with 0.1 M glycine HCl, pH2.7, then neutralised with 2 M Tris-HCl, pH 8.5.The Fab’ was quantified by absorbance at 280 nm,concentrated four-fold via ultrafiltration (UF,10 kDa molecular weight cut-off membrane) andbuffer exchanged into 20 mM Tris-HCl, pH 8.0.The diafiltered product was applied to a pre-equilibrated (20 mM Tris-HCl, pH 8.0) anionexchange column, Poros 50HQ (Applied Biosys-tems, Warrington, UK). The Fab’ was collected inthe unbound fraction, concentrated and bufferexchanged into 100 mM sodium phosphate buffer,pH 6.0 containing 2 mM EDTA, to a stock con-centration of 19.5 mg/mL.

Human g1 F(ab’)2 was obtained by pepsin digestof IgG and purified by gel filtration (S-200HRAmersham Pharmacia Biotech, Little Chalfont,UK). The product was concentrated and bufferexchanged into 100 mM sodium phosphate buffer,pH 6.0 containing 2 mM EDTA, to a stockconcentration of 21.0 mg/mL.

Murine g1 Fab’ was mammalian derived using aNS0 cell line, grown in serum-free conditions in a

DOI 10.1002/jps


100 L fermenter. The fermenter supernatantwas concentrated 10 times and then applied to apreequilibrated (20 mM sodium phosphate/150 mM sodium chloride, pH 7.1) GammaBindPlus Sepharose (GE-Healthcare) column. TheFab’ was eluted with 0.1 M glycine-HCl, pH 2.8,concentrated and buffer exchanged into 100 mMsodium phosphate buffer, pH 6.0 containing 2 mMEDTA, to a stock concentration of 19.6 mg/mL.

The concentrations of the protein solutions weredetermined by measuring absorbance at 280 nmwith extinction coefficients calculated from theamino acid compositions.

PEGylation of the Fab’ fragments was achievedby site specific attachment of PEG-maleimidederivatives purchased from Nektar Therapeutics(Mountain View, CA) and NOF Corporation(Tokyo, Japan). to selectively reduced thiols.The PEGylated Fab’ products used in this studyare illustrated in Figure 3. hFab’-(L)PEG25k(Fig. 3a) was prepared by the reduction of humang1 Fab’, in 100 mM sodium phosphate buffer/2mM EDTA, pH 6.0, using 2-mercaptoethylamine(2-MEA) at a final concentration of 5 mM for30 min at 378C. The reductant was removed viadiafiltration (10 kDa molecular weight cut-off)into 100 mM sodium phosphate buffer/2 mMEDTA, pH 6.0. The resultant thiol was titrated

Figure 3. Schematic illustration of (a) (hF20k); (c) (mFab’-(L)PEG2� 20k); (d) (DFM-the general structure for monomethoxy PEG

DOI 10.1002/jps J

using 4,40-dithiodipyridine (DTDP) by measure-ment of the thiopyridine released at 324 nm; aratio of approximately one thiol per Fab’ fragmentwas obtained. A mono-maleimide PEG derivativewith a 25 kDa linear PEG chain was added to thereduced Fab’ using a molar ratio of 1.0:2.4 (Fab’/PEG) and incubated for 18 h at ambient tem-perature. The Fab’–PEG mixture was thenadjusted to pH 4.5 by the addition of glacial aceticacid and applied to a preequilibrated (50 mMsodium acetate, pH 4.5) cation exchange column,SP Sepharose HP (Amersham Pharmacia Bio-tech). The product was eluted using a lineargradient of 0–250 mM sodium chloride over 20column volumes (CV) then concentrated andbuffer exchanged using a stirred cell (Amicon;10 kDa molecular weight cut-off membrane) into50 mM sodium chloride, 125 mM sodium chloride,pH 5.5 to a final concentration of 4.8 mg/mL, asmeasured by the absorbance at 280 nm. The finalproduct was characterised by sodium dodecylsulphate–polyacrylamide electrophoresis (SDS–PAGE) and gel filtration–high performance liquidchromatography (GF–HPLC).

hFab’-(Br)PEG2� 20k (Fig. 3b) was preparedas described for hFab’-(L)PEG25k, except that abranched mono-maleimide PEG derivative (con-sisting of two linear 20 kDa chains) of overall size

ab’-(L)PEG25k); (b) (hFab’-(Br)PEG2�(Br)PEG2� 20k). CH3–(CH2CH2O)n– is(mPEG).


2066 LU ET AL.

40 kDa was used as the PEGylation reagent. Thefinal product was presented in 50 mM sodiumacetate, 125 mM sodium chloride, pH 5.5 to aconcentration of 18 mg/mL and characterised bySDS–PAGE and GF–HPLC.

mFab’-(L)PEG2� 20k (Fig. 3c) was prepared byreduction of murine g1 Fab’ in 100 mM sodiumphosphate buffer, pH 6.0 containing 2 mM EDTAusing tris (hydroxypropyl)phosphine (THP,0.5 Min water, Cytec, Bradford, UK) at a finalconcentration of 2.5 mM for 1 h at ambienttemperature. The reductant was removed viadiafiltration (10 kDa molecular weight cut-off)into 100 mM sodium phosphate buffer, pH 5.7containing 2 mM EDTA. Titration of the gener-ated thiols with DTDP resulted in approximatelytwo thiols per Fab’ molecule due to the reductionof the interchain disulfide bond between the heavyand light chain of the Fab’. The reduced Fab’ wasincubated with a mono-maleimide PEG derivativewith a 20 kDa linear PEG, at a molar ratio of1.00:3.25 (Fab’/PEG) at ambient temperature for18 h. This resulted in a product where two 20 kDaPEG chains were attached to the Fab’ via the thiolat the C terminal end of the heavy chain andthe light chain. Hence, there was no interchaindisulphide bridge between the heavy and lightchains. The Fab’–PEG mixture was then adjustedto pH 4.5 by the addition of glacial acetic acid andapplied to a preequilibrated (50 mM sodiumacetate, pH 4.5) SP Sepharose HP column. Theproduct was eluted using a linear salt gradient, of0–125 mM sodium chloride over 20 CV, concen-trated and buffer exchanged into 50 mM sodiumacetate, 125 mM sodium chloride, pH 5.5 to a finalstock concentration of 31.7 mg/mL. The finalproduct was characterised by SDS–PAGE andGF–HPLC.

DFM-(Br)PEG2� 20k (Fig. 3d): The reductionwas performed as described for hFab’-(L)PEG25 k.PEGylation was achieved by incubation of abis-maleimide derivative of a branched PEG

Table 1. Molecular Weights M and PartiaAntibody Fragments

Sample Attached PEG (kDa

Murine g1 Fab’ 0Human g1 Fab’ 0Human g1 F(ab’)2 0mFab-(L)PEG2� 20k 40hFab’-(L)PEG25 k 25hFab’-(Br)PEG2� 20k 40DFM-(Br)PEG2� 20k 40


(composed of two 20 kDa linear chains) with thereduced Fab’ using a molar ratio of 2.2:1.0 (Fab’/PEG) for 18 h at ambient temperature. ThediFab’–PEG mixture was purified as describedfor hFab’-(L)PEG25 k, but using a linear saltgradient of 0–250 mM sodium chloride over 30 CV.The final product was concentrated and bufferexchanged into 50 mM sodium acetate, 125 mMsodium chloride, pH 5.5 to a final concentration of13.5 mg/mL and was characterised by SDS–PAGEand GF–HPLC.

Concentrations of the PEGylated protein solu-tions were measured from measurement of ultra-violet (UV) absorbance at a wavelength of 280 nm.The extinction coefficient of the PEG is negligibleas pure PEG revealed a minimal signal contribu-tion,11 therefore a weight-averaged extinctioncoefficient was used for each PEGylated moleculeby taking into account the contribution of themass of the PEG to the conjugate. For instance, inthe case of mFab-(L)PEG2� 20k, at 280 nm theextinction coefficient of the Fab’ alone is 1614mL g�1 cm�1, and taking into account the masscontribution of the 40 kDa PEG, the weight-averaged extinction coefficient for mFab-(L)PEG2� 20k is 873 mL g�1 cm�1.

The commercial manufacturers claim for themolecular weight of the branched PEG (2� 20 k)of 40 kDa was checked by both size exclusionchromatography coupled to multi-angle laser lightscattering (courtesy of Dr. G. Morris, NCMH)yielding a value for the weight average molecularweight Mw of (40.5� 1.2) kDa.

Table 1 lists the PEGylated conjugates usedin this study. The partial specific volume n ofeach antibody molecule was calculated usingthe routine SEDNTERP31 from the amino acidcomposition. Following Lepori and Mollica32 andNichol et al.,33 a value of 0.83 mL/g was usedas the partial specific volume for the all thePEG molecules evaluated. The weight-averagedpartial special volume for each of the PEGylated

l Specific Volumes n of PEGylated

) Molecular Weight M (kDa) n (mL/g)

47 0.72648 0.72696 0.72687 0.77473 0.76288 0.777

136 0.759

DOI 10.1002/jps


conjugates (see Table 1) was then calculated usingthe following relation:34

nc ¼XNi¼1

fini ¼ fpnp þ fnpnnp (1)

where fp and fnp are the weight fractions of proteinand nonprotein components respectively, and npand nnp their partial specific volumes.

SDS–PAGE and HPLC Analysis of PEGylatedAntibody Fragments

The Fab’ fragments and their associated PEGy-lated products were analysed for purity by SDS–PAGE and GF–HPLC prior to the hydrodynamicstudies.

SDS–PAGE was performed using Novex Tris-Glycine (4–20%) gels (Invitrogen EC60255BOX,Paisley, UK) under both nonreducing and redu-cing conditions. For nonreducing conditions 10 mLof a 0.4 mg/mL solution of the sample was diluted1:1 with Tris-Glycine SDS Sample buffer x2(Invitrogen LC2676) and boiled for 2 min; forreducing conditions 2.2 mL of NuPage SampleReducing Agent� 10 (Invitrogen NP0004) wasincluded in the nonreducing mixture and boiledfor 3 min. In both cases 10 mg was loaded onto thegel and ran according to the manufacturer’sinstructions (a constant voltage of 125 V for 90min). The bands on the gel were visualised byCoomassie Blue staining.

GF–HPLC was performed using an HPLCsystem (Agilent 1100, Stockport, UK) with ZorbaxGF450 (Agilent 884973-902) and Zorbax GF250(Agilent 884973-901) columns connected in series.The elution buffer was 0.2M sodium phosphate,pH 7.0 containing 10% ethanol. The sampleswere analysed at 1.0 mL min�1 isocratically with a30 min run time, at a wavelength of 280 nm.

Sedimentation Velocity and Equilibrium in theAnalytical Ultracentrifuge

All Fab’ fragments and their associated PEGy-lated products were diluted to between 0.5 and1.6 mg/mL at six different concentrations prior tosedimentation velocity experiments in an OptimaXLA (Beckman Instrument, Palo Alto, CA), at208C using an An60Ti four-place rotor running atvarious rotor speed (45000 rpm for velocity runsand 12000–14000 rpm for equilibrium runs). Thesedimentation process was monitored by UV

DOI 10.1002/jps J

absorbance with a scan interval of 4 min, withthe buffer as the reference solvent. Absorptionoptics was chosen for studying PEGylated anti-bodies as PEG had minimal signal contribution at280 nm11: any unconjugated PEG would not bedetected.

We also investigated the sedimentation coeffi-cient behaviour of the linear ‘‘20 kDa’’ (molecularweight estimated by the manufacturer) PEG re-agent (see Fig. 3b) and the branched ‘‘40 kDa’’ PEGreagent (see Fig. 3c) using the interference opticalsystem on an Optima XLI (Beckman Instrument).The 20 and 40 kDa PEG samples (in powder form)were dissolved separately in 50 mM sodium acetate,125 mM sodium chloride buffer (pH5.5) andconcentrations were determined using an AtagoDD-5 differential refractometer (Jencons Scien-tific, Leighton Buzzard, UK) (dn/dc¼ 0.134 mL/g).Samples were then diluted to between 0.3 and2.3 mg/mL prior to the velocity run at 50000 rpm.

Sedimentation velocity data were analysed bythe so-called least squares g�(s) or ‘‘ls-g�(s)’’procedure as implemented in SEDFIT.35,36 g�(s)profiles confirmed the monodispersity and theaggregate-free nature of the solutions (see Dis-cussions later), which are critical for the sub-sequent interpretation of the SAXS data. Theexperimental sedimentation coefficient valuess20,b (the designation s20,b to indicate that thevalue reported applies to conditions of 20.08C in asolvent b, which in this case is the formulationbuffer) obtained from SEDFIT have been cor-rected to s20,w (standard solvent conditions—thedensity and viscosity of pure water at 20.08C)using SEDNTERP.31 The s20,w values obtained atdifferent concentrations were extrapolated to zeroconcentration to yield s020;w thereby eliminatingsolution nonideality effects.37

Sedimentation equilibrium data were evaluatedby the model-independent routine MSTARA38—no model (ideal, single solute, etc.) has to beassumed a priori, as required by other software.Determination of the baseline absorption byover-speeding the centrifuge was necessary tocorrect for UV-absorbance from nonmacromole-cular material. At each loading concentration, theMSTARA program provides a plot of M�(r) versusj(r) for determination of molecular weight, whereM�(r) is an operational point average molecularweight and j(r) is the radial displacement squaredparameter defined by:

jðrÞ ¼ ðr2 � r2aÞ

r2b � r2a(2)


2068 LU ET AL.

where ra, rb are the radial positions of the cellmeniscus and cell base, respectively. Althoughthe M�(r) values themselves have no physicalmeaning, the M�(r) extrapolated to the cellbase (j(r)¼ 1) yields the apparent weight-averagemolecular weight over all the macromolecularcomponents in the ultracentrifuge cell, Mw.

39 Mwat different concentrations were extrapolated tozero concentration to eliminate any contributionfrom thermodynamic nonideality.37

Small Angle X-Ray Scattering (SAXS)

The SAXS data were obtained at Station 2.1 at theSynchrotron Radiation Source (SRS, Daresbury,UK), employing sample-to-detector distances of1 m (to cover a scattering range of 0.04

Figure 4. SDS–PAGE analysis (NOVEXTM, Tris/Gly-cine 4–20%) of human g1 Fab’ and associated PEGylatedconjugates under nonreducing and reducing conditions.Lane 1, molecular weight marker (Mark12TM,NOVEXTM); (Lane 2) human g1 Fab’; (Lane 3) hFab’-(Br)PEG2� 20k; (Lane 4) DFM-(Br)PEG2� 20k. Leftpanel, nonreducing conditions; (right panel) reducingcondition.

Figure 5. SDS–PAGE (Tris-Glycine 4–20%) analysisof murine g1 Fab’ and its associated PEG conjugatesunder nonreducing and reducing conditions. Lane 1,molecular weight markers (Mark 12); (Lane 2) humang1 Fab’; (Lane 3) murine g1 Fab’; (Lane 4) mFab’-(L)PEG2� 20k


in Lane 2 was seen to contain some F(ab’)2evidenced by a band at approximately 100 kDa:there was also evidence for some free heavyand light chains at approximately 25 kDa. Theassociated PEGylated components in Lanes 3 and4 are shown to be >95% pure. The contamina-tion of the human g1 Fab’ has no consequencesin terms of interpretation of sedimentationcoefficients as the other components resolve awaybut renders opaque to interpretation the SAXSrecords. By contrast, both the murine g1 Fab’ andmFab’-(L)PEG2� 20k were >99% pure (Fig. 5).The mobility of the mFab’-(L)PEG2� 20k (Lane 4)was equivalent in both nonreducing and reducingconditions confirming that there was no disul-phide bridge between the heavy and light chainsand that a 20 kDa linear PEG chain was attachedto the C terminal end of both the heavy and lightchain.

The results of GF–HPLC analysis for bothhuman g1 Fab’ and murine g1 Fab’ withassociated PEGylated products can be seen inFigures 6 and 7. GF–HPLC analysis of human g1

DOI 10.1002/jps J

Fab’ (Fig. 6), confirmed the presence of F(ab’)2(28%). The PEGylated species, hFab’-(Br)PEG2�20k, DFM-(Br)PEG2� 20k and hFab’-(L)PEG25 k(not shown) were shown to be>98% pure. For themurine g1 Fab’ and mFab’-(L)PEG2� 20k (Fig. 7),both species were shown to be >99% pure.

Hydrodynamic Behaviour of PEGs

Figure 8 shows an example of the g�(s) distribu-tion for the branched 40 kDa PEG plotted withrespect to s (Fig. 8a) and log es

2 (Fig. 8b), where thelatter takes into account we are dealing with aquasi-continuous log-normal distribution of mole-cular weights for the PEG.

The conformation of the PEG chain in aqueoussolution is still a matter of debate. X-raystructural analysis has shown that crystallinePEG chains can adopt two extreme structuralconformations: a zigzag, random coil structure forshorter chains, or a winding, helical structure forlonger chains.49 It has been argued from earlierstudies that viscosity50–53 and diffusion coef-ficient data50 support a random coil conformation,


Figure 6. GF–HPLC analysis of (a) human g1 Fab’; (b) hFab’-(Br)PEG2� 20k;(c) DFM-(Br)PEG2� 20k. Zorbax GF450/GF250 HPLC columns in series with isocraticelution in 0.2 M sodium phosphate, pH 7.0/10% ethanol at 1.0 mL min�1, detected at awavelength of 280 nm.

2070 LU ET AL.

whereas it has been claimed from calorimetricdata54 that it adopts a helical conformation. Basedon volumetric studies, Lepori and Mollica32 favoura compromise model for PEG in solution is ahydrated flexible coiled polymer with some helicalsegments.

In many cases with high polymeric solutes thereis a unique relation between the sedimentationcoefficient s and the molecular weight M. Forexample the extremes are as follows (see, e.g.Ref. 55):

Compact sphere : s � M0:667 (3)

Rigid rod : s � M0:15 (4)


For a flexible coil polymer, s and M follows therelation that55

s � M0:4�0:5 (5)

or to an approximation

s2 � M (6)

In all three cases loge(s) is the appropriateabscissa. Therefore if the PEG moiety investi-gated in this study is approximately a randomcoil conformation, and contains a continuous log-normal distribution of polymer chain lengths,56

we would expect the g�(s) distribution plot on alog(s2) scale (or 2 log(s)) which in turn mirrors the

DOI 10.1002/jps

Figure 7. GF–HPLC analysis of (a) murine g1 Fab’ and (b) mFab’-(L)PEG2� 20k.Zorbax GF450/GF250 HPLC columns in series; isocratic elution in 0.2 M sodiumphosphate, pH 7.0/10% ethanol at 1.0 mL min�1; detected at a wavelength of 280 nm.


normalised molecular weight distribution to be aunimodal Gaussian distribution. This unimodalGaussian distribution feature (see, e.g. Fig. 8b) isclearly evident within the range of concentrationsstudied (from 0.6 to 2.3 mg/mL). The slightdeviation could due to the diffusive effects.

The g�(s) distribution yields a modal sedimenta-tion coefficient. For both PEG reagents, con-siderable nonideality effects were observed asthe sedimentation coefficient showed a markeddecrease with increasing concentrations (data notshown). The concentration-dependence coefficientks estimated from the extrapolation of 1/s versusconc. plot to zero concentration yields a value of(81� 6) mL/g using the relation

1

s¼ 1

s0þ ks

s0c (7)

This high ks value found for the 40 kDa PEG ismuch larger than for globular proteins whichconfirms the flexible expanded conformation of the

DOI 10.1002/jps J

molecule. Extrapolations of the 1/s versus conc.plot to zero concentration yields an s020;w value of(0.64� 0.01)S for the linear 20 kDa PEG reagentand (0.82� 0.01)S for the branched 40 kDa PEGreagent. Interestingly, it is possible to estimatethe molecular weight M based on the s020;w and ks(after correction for radial dilution) measure-ments:37 for the 40 kDa PEG reagent, a value ofM �39000 g/mol is returned.

The sedimentation coefficient is inversely pro-portional to the frictional ratio following therelation:

s020;w ¼Mwð1 � nr0Þ

6pNAh0

4pNA3nMw

� �1=3 1f=f0

(8)

where Mw is the weight averaged molecularweight (g/mol), n is the partial specific volume(mL/g), NA is Avogadro’s number (6.02205� 1023mol�1), r0 (g/mL) and h0 (Poise) are the densityand viscosity of water at 20.08C respectively.


Figure 9. The ls-g�(s) distribution for a 1.2 mg/mLhFab’-(Br)PEG2� 20k solution. (a): g�(s) versus s dis-tribution. (b): log-normal distribution. In (b) the � line isthe experimental distribution and the red solid lineshows a single Gaussian curve.

Figure 8. ls-g�(s) distribution of a 0.57 mg/mL 40 kDaPEG solution. (a) plotted versus s; (b) plotted versuslog(s2), Green: experimental g�(s) distribution; red:simulated data from SEDFIT35,36 and black solid linesshows a single Gaussian curve. Simulated data generat-ed from SEDFIT is based on a single-component system(using Mw¼ 40 kDa and s¼ 0.8S), showing the broad-ening of the peak is due to diffusion alone.

2072 LU ET AL.

Following this relation, high frictional ratios(ratio of the actual friction experienced by aparticle moving through a fluid to the frictionwhich it would experience if it were a compactsphere of the same mass)57 were found for the20 kDa PEG ( f/fo �2.5) and the 40 kDa PEG( f/fo �3.0) using the algorithm UNIVERSAL_PARAMS.58

Hydrodynamic Behaviour of PEGylated Conjugates

Interestingly, sedimentation velocity analyses ofthe PEGylated antibody solutions reveal theirhomogeneous nature. Moreover, the g�(s) dis-


tribution of these PEGylated antibodies mirrorsthe effect seen in the analysis of PEG that thedistribution is log-normal with respect to s or s2

(an example see Fig. 9). This suggests that thehydrodynamic behaviour of PEGylated antibodyis predominantly dictated by the attached PEGmoiety. Again, similar to the 40 kDa PEG, themodal sedimentation coefficients obtained fromg�(s) distribution at different concentrations wereextrapolated to zero concentration to eliminatethe effect of nonideality, and the s020;w values arereported in Table 2.

The weight-average molecular weights ofthe PEGylated conjugates from sedimentationequilibrium are also shown in Table 3. Despite

DOI 10.1002/jps

Table 2. Experimental Sedimentation and X-Ray Scattering Data for the Antibody Fragments and their PEGylatedConjugates

Sample s020;w (S) f/fo Rg (Å) Dmax (Å)

Murine g1 Fab’ 3.67� 0.02 1.30 26.4� 1.0 76� 2Human g1 Fab’ 3.95� 0.01 1.23 — —Human g1 F(ab’)2 5.49� 0.01 1.40 41.0� 1.0 117� 3mFab-(L)PEG2� 20k 2.12� 0.02 2.75 29.5� 1.0 94� 2hFab’-(L)PEG25 k 2.40� 0.01 2.29 31.2� 1.0 96� 2hFab’-(Br)PEG2� 20k 2.16� 0.02 2.68 — —DFM-(Br)PEG2� 20k 3.41� 0.03 2.47 49.6� 1.0 145� 3

s020;w: sedimentation coefficient; Rg: radius of gyration, Dmax: maximum dimension of the scattering particle.


the polydisperse nature of the PEG reactants, themeasured molecular weight and the theoreticalvalues for the PEGylated molecule are in reason-able agreement.

Solution Conformation of Antibody Fragmentsand PEGylated Conjugates

Sedimentation Coefficients

The sedimentation coefficient ðs020;wÞ is a mani-festation of the overall size and shape of amacromolecule in solution and the data fromour experiments are summarised in Table 2. Thevalue for human g1 Fab’ (3.95� 0.01)S is similarto previous determinations.59 With its short hingeremoved, murine g1 Fab’ sediments at the samerate (3.67� 0.02)S as murine classic g2a Fab’(3.68� 0.01)S.60 Comparison of these last twovalues indicate that the existence of the hinge maynot alter the overall conformation of the antibodyfragment and the difference in the hydrodynamicbehaviour between human g1 Fab’ and murine g1Fab’ is ascribed to the different antibody speciesused. It is worth pointing out that the murine g1Fab’ has no upper or middle hinge region since astop codon was inserted in the gene after the CYS

Table 3. Molecular Weights of the PEGylatedConjugates

SampleTheoretical

M (Da) Mw (Da)

mFab-(L)PEG2� 20k 87200 78000� 3000hFab’-(Br)PEG2� 20k 88100 82000� 4000hFab’-(L)PEG25 k 73100 73000� 2000DFM-(Br)PEG2� 20k 136100 125000� 4000

Mw: experimental (weight-average) molecular weight fromsedimentation equilibrium.

DOI 10.1002/jps J

residue in CH1. Interestingly, human g1 F(ab’)2has an identical sedimentation coefficient (withinexperimental error) to the DFM (covalently linkeddiFab’-maleimide) of (5.53� 0.04)S (Harding, S.E.and Rhind, S.K. unpublished work).

Despite a large increase in molecular weightdue to the conjugation of either linear 25 kDaor branched 40 kDa PEG moieties with theproteins, both hFab’-(L)PEG25k and hFab’-(Br)PEG2� 20k sediment at a slower rate com-pared to the parent antibody human g1 Fab’ (seeTable 2). Similar sedimentation behaviour hasbeen observed for the mFab’-(L)PEG2� 20k andDFM-(Br)PEG2� 20k. These data suggest a sig-nificant effect of PEG on the conformation of allthe antibody fragments studied, with the increasein frictional ratio f/fo (from �1.3 to 2.3–2.8):the solution hydrodynamic properties of theconjugates are clearly dominated by the PEGmoiety ( f/fo �3.0).

Unfortunately quantifying further the effect onconformation is rendered difficult because ofthe effects of addition of the PEG on the waterassociation or time averaged ‘‘hydration’’ of theantibodies—which also increases the frictionalratio. The large hydrodynamic size or volume ofPEGylated antibodies in solution has alreadybeen reported.10

Small Angle X-Ray Scattering

Although the complexities of hydration andflexibility render it difficult to comment on thedetailed conformation of the protein-PEG con-jugates it is nonetheless possible to providelimited but nonetheless useful information onthe conformation of the protein moiety andthe location of the PEGylation site within thecomplex, using SAXS to extend the sedimentationvelocity data.


Figure 10. Superimposition of the experimental p(r)functions (with error bars) of murine g1 Fab’ (- - -) andmFab-(L)PEG2� 20k (—). Both molecules have thesame peak M at �30Å. The maximum dimensions Dmaxare 76 and 95 Å respectively.

Figure 11. Superimposition of the experimental p(r)functions (with error bars) of human g1 F(ab’)2 (- - -) andDFM-(Br)PEG2� 20k (—). For human g1 F(ab’)2, peaksM1 and M2 occur at �32 Å and �75 Å respectively, withDmax �120 Å. For DFM-(Br)PEG2� 20k peaks M1 andM2 occur at rM1 �35 Å and rM2�90 Å respectively, withDmax �145 Å.

2074 LU ET AL.

Small angle SAXS measures the relativescattering intensity between the scattering par-ticle and the solvent.29 The Guinier approxima-tion42 of the scattering profile yields the radiusof gyration (Rg) of the particle in solution andindicates its elongation or compactness. Fouriertransformation of the scattering profile gives thedistance distribution function p(r) as a function ofscattering distance r. This is characterised by oneor maxima (M) and the maximum dimension ofthe scattering particle (Dmax) both of which arecharacteristic features of the particle in solution:the algorithm in GNOM41 allows to calculate p(r)as well as a further estimate for Rg (encouraginglyRg values from this procedure which wereconsistent with those from the Guinier analysis).

Using these procedures a radius of gyration of26.4 Å was found for the murine g1 Fab’, similar tothe previous determination.61 Comparing thePEGylated conjugates with their parent anti-bodies (Table 2), scattering data have shown thatthe effect of PEGylation is to contribute to a more‘‘extended’’ conformation in solution as indicatedby the higher Rg and Dmax values. By attachingtwo linear 20 kDa PEG to the C-termini of theheavy and light chains of the murine g1 Fab’, theresulting conjugate mFab’-(L)PEG2� 20k has aDmax of 94 Å, suggesting a �20 Å increasecompared to the murine g1 Fab’. The radius ofgyration data are consistent with the Dmaximplying that the PEGylated molecule is moreelongated. More pronounced differences werefound between the human g1 F(ab’)2 and DFM-(Br)PEG2� 20k, with a �9 Å increase in Rg and a�30 Å increase in Dmax.

The distance distribution function p(r) of ascattering particle reflects the shape and massdistribution of the molecule.29,62 Figures 10 and11 show the superimpositions of the p(r) functionof the antibody fragments and their PEGylatedconjugates. The p(r) function represents thedistribution of all intra-particle scattering vectorsand therefore the maximum M in p(r) correspondsto the most frequently occurring interatomic dis-tance within the molecular structure.29,63 Murineg1 Fab’ and mFab’-(L)PEG2� 20k both have thismaximum M at a value of r of �30 Å (Fig. 10): thisvalue refers to the most commonly occurringdistance between the scattering centres in a singleFab’. Thus we can conclude that no significantconformational change to the antibody fragmentoccurs upon PEGylation, that is, the structure ofthe Fab’ fragment itself remains essentiallyunaltered. Parts of the PEGylated molecule extend


over larger distances (e.g. between the PEG moietyand the protein) and therefore contribute to thedistribution of longer distances in the p(r) functionwhich result in an increase in Rg and Dmaxcompared to the isolated Fab’ fragment.

Similarly, there is little tendency for the Fab’domain within DFM-(Br)PEG2� 20k to undergoconsiderable conformation changes when com-pared to the human g1 F(ab’)2. Analogous posi-tions of the first maxima M1 are found at (32–35)Å (Fig. 11) that can be assigned to the mostcommonly occurring distance within a single Fab’.The second maximum peak M2 describes the mostcommon distance within the whole molecule andtherefore can be ascribed to the distance between

DOI 10.1002/jps


the two Fab’s. As a bis-maleimide linker wasused for the synthesis of DFM-(Br)PEG2� 20k,an increase of 15 Å in peak M2 of DFM-(Br)PEG2� 20k (90 Å) compared to the humang1 F(ab’)2 (75 Å) illustrates the conformationalvariation between these two molecules due to theadduct of the bis-maleimide derivative of thebranched PEG.

Figure 13. Superimposition of experimental p(r)function (with error bars) of the murine g1 Fab’ withp(r) functions of two Fab’ crystallographic models cal-culated from HYDROPRO46: 1UCB (— ��— ��) and 1BBJ( ). For the murine g1 Fab’, the maximum M occursat �30 Å and the maximum dimensions Dmax occurs at�76 Å respectively.

DISCUSSION

Ab Initio Modelling

Ab initio shape reconstruction using the programDAMMIN43 was carried out to visualise theconformations of the antibodies and their PEGy-lated conjugates in solution For each molecule,10 low resolution models obtained from differentDAMMIN runs were analysed and the stableconstructions (NSD< 0.8) were averaged thatyielded a model representing the most probablefeatures of the solution (see Materials andMethods Section).

To examine the validity of the shape reconstruc-tion, the ab initio model of the murine g1 Fab’ wascompared with the crystallographic models of twoFab’ molecules (accession code 1UCB and 1BBJrespectively). Superimpositions of the ab initiomodel of murine g1 Fab’ with the crystallographicmodels using SUPCOMB47 and MASSHA48 are invery good agreement with all three Fab’ modelsand have a similar volume (an example seeFig. 12). The scattering properties of the crystal-

Figure 12. Superimposition of the Fab’ crystal struc-ture (PDB accession code 1UCB, red cartoon) with theab initio model of murine g1 Fab’ (grey spheres). Thecysteine in the light chain at position 214 is highlightedby green spheres.

DOI 10.1002/jps J

lographic models of two Fab’ molecules (accessioncode 1UCB and 1BBJ respectively) were calculat-ed using CRYSOL45 and HYDROPRO46 (data notshown). Superimposition of the p(r) functions (seeFig. 13) for the three different Fab’ models yieldsimilar positions for the maxima M and Dmax andreflect the excellent harmony revealed betweenab initio models and crystal structures (Fig. 12).

Correspondingly, ab initio models of mFab-(L)PEG2� 20k and hFab’-(L)PEG25k were calcu-lated and superimposed with the model ofthe murine g1 Fab’ (see Figs. 14 and 15).Distinct structural features are revealed in the

Figure 14. Superimposition of ab initio models ofmFab-(L)PEG2� 20k (transparent cyan spheres) withmurine g1 Fab’ (transparent grey spheres). For orienta-tion the red ribbon model represents the Fab’ crystalstructure as shown in Figure 9.


Figure 15. Superimposition of ab initio models ofhFab’-(L)PEG25 k (transparent pink spheres) withmurine g1 Fab’ (transparent grey spheres). For orienta-tion the red ribbon model represents the Fab’ crystalstructure in analogy to Figures 9 and 11. Figures 9,11and 12 were produced with the routine PYMOL.65

2076 LU ET AL.

superimpositions of the Fab’ with PEGylationconjugates and clearly demonstrates that theFab’ is essentially unchanged by PEGylation. Theextension in the PEGylated model is attributed tothe conjugation of PEG. However, it is apparentthat the increase in volume of the PEGylatedmodel is smaller than expected considering thesize of the PEG moiety. With the significantincrease in frictional ratio as observed by sedi-mentation velocity, it might be anticipatedthat the PEGylated molecule being much more‘‘extended’’ or occupying a larger excluded volume.An explanation can be found in the fact that thePEG molecules are highly dynamic and flexible insolution and primarily the immediate interactionregions with the antibody will contribute to astrong scattering contrast between solvent andmacromolecule. Besides, the shape restorationprocedure with DAMMIN works more reliablyfor compact and rigid molecular configurationsand it may be less sensitive to loosely boundand dynamically disordered structural arrange-ments.64 Moreover, as the electron density withina PEGylated molecule is not homogeneous and ischaracterised by high solvation, the scatteringcontrast between the PEG moiety and the solventis smaller compared to the one between proteinand solvent.

This deviation from the expected particlevolume suggests that the ab initio restorationdoes not represent fully the effective volume of


the whole PEGylated molecule. Considering theabove reasons it could be inferred that only themore ordered and rigid parts of the PEG moietieshave been manifested in the shape calculations:contributions on the other hand from the dyna-mically disordered PEG segments had a contrastfor X-rays that were only marginally higher thanthat for the buffer solution. Further studiesinvolving contrast variation small angle neutronscattering (SANS) may help to resolve the issue onthe conformation and contribution of the PEGmoiety. Although SAXS fails to reveal detailsabout the conformation of the whole PEG compo-nent, nonetheless it is still possible to commenton the way PEG is bound to the protein. Forinstance, models of mFab’-(L)PEG2� 20k andhFab’-(L)PEG25k have shown that the attachedPEG moiety extends from one end of the Fab’. Asuperimposition of the crystallographic modelswith the ab initio model of hFab’-(L)PEG25k isachievable so that the PEG moiety is attached tothe Fab’ fragment through the free cysteine thatis exposed at the C-termini of the Fab’ molecule(see Fig. 9). This is consistent with the conjugationchemistry applied in site-specific PEGylation(Fig. 2).

Concluding Remarks

AUC and measurement of the translational fric-tional ratio has shown that for all PEGylated-antibody fragment complexes studied, the hydro-dynamic properties are dominated by the PEGmoiety. SAXS at low values of r (distance betweenscattering centres) as manifested by the rM and Rgvalues, indicate that the conformation of Fab’ itselfis essentially unchanged by incorporation into thecomplex. The globular Fab’ therefore appears to‘‘sit’’ inside the flexible PEG: although attachedat a specific point, because of its much greaterflexibility the time-averaged flexible conformationof the PEG may effectively cover part or all ofthe Fab’.

ACKNOWLEDGMENTS

We thank the United Kingdom Engineering andPhysical Science Research Council for supportingthis work, and the Council for the Central Labora-tory of the Research Councils for access to the SRS,Daresbury. We thank Alistair Henry at UCB-Cell-tech (Slough, UK) for useful discussions, Helen

DOI 10.1002/jps


Brand and Gayle Phillips, also at UCB-Celltech(Slough, UK) for help in the purification and PEGy-lation of the Fab’ fragments and Dr. Gordon Morris(NCMH) for the SEC-MALLs check on our mole-cular weights.

REFERENCES

1. Holliger P, Hudson PJ. 2005. Engineered antibodyfragments and the rise of single domains. Nat Bio-technol 23:1126–1136.

2. King DJ, Adair JR, Angal S, Low DC, Proudfoot KA,Lloyd JC, Bodmer M, Yarranton GT. 1992. Ex-pression, purifucation and characterisation of amouse:human chimeric antibody and chimericFab’ fragment. Biochem J 281:317–323.

3. Sun CZ, Wirsching P, Janda KD. 2003. EnablingScFv as multi-drug carriers: A dendritic approach.Bioorg Med Chem 11:1761–1768.

4. Skerra A, Pluckthun A. 1988. Assembly of a func-tional immunoglobulin Fv fragment in Escherichiacoli. Science 240:1038–1041.

5. Carter P, Kelley RF, Rodrigues ML, Snedecor B,Covarrubias M, Velligan MD, Wong WLT, RowlandAM, Kotts CE, Carver ME, Yang M, Bourell JH,Shepard HM, Henner D. 1992. High level Escher-ichia coli expression and production of a bivalenthumanized antibody fragment. Bio/Technology10:163–167.

6. King DJ, Turner A, Farnsworth AP, Adair JR,Owens RJ, Pedley RB, Baldock D, Proudfoot KA,Lawson AD, Beeley NR. 1994. Improved tumortargeting with chemically cross-linked recom-binant antibody fragments. Cancer Res 54:6176–6185.

7. Casey JL, King DJ, Chaplin LC, Haines AM, PedleyRB, Mountain A, Yarranton GT, Begent RH. 1996.Preparation, characterisation and tumour target-ing of cross-linked divalent and trivalent anti-tumour Fab’ fragments. Br J Cancer 74:1397–1405.

8. Chen J, Jaracz S, Zhao X, Chen S, Ojima I. 2005.Antibody-cytotoxic agent conjugates for cancertherapy. Expert Opin Drug Deliv 2:873–890.

9. Chapman AP, Antoniw P, Spitali M, West S,Stephens S, King DJ. 1999. Therapeutic antibodyfragments with prolonged in vivo half-lifes. NatBiotechnol 17:780–783.

10. Chapman AP. 2002. PEGylated antibodies andantibody fragments for improved therapy: A review.Adv Drug Deliver Rev 54:531–545.

11. Koumenis IL, Shahrokh Z, Leong S, Hsei V, DeforgeL, Zapata G. 2000. Modulating pharmacokinetics ofan anti-interleukin-8 F(ab’)2 by amine-specificPEGylation with preserved bioactivity. Int J Pharm198:83–95.

DOI 10.1002/jps J

12. Tsutsumi Y, Onda M, Nagata S, Lee B, KreitmanRJ, Pastan I. 2000. Site-specific chemical modi-fication with polyethylene glycol of recombinantimmunotoxin anti-Tac(Fv)-PE38 (LMB-2) improvesantitumor activity and reduces animal toxicity andimmunogenicity. PNAS 97:8548–8553.

13. Delgado C, Pedley RB, Herraez A, Boden R, BodenJA, Keep PA, Chester KA, Fisher D, Begent RHJ,Francis GE. 1996. Enhanced tumour specificity ofan anti-carcinoembrionic antigen Fab’ fragmentby poly(ethylene glycol) (PEG) modification. Br JCancer 73:175–182.

14. Bailon P, Berthold W. 1998. Polyethylene glycol—conjugated pharmaceutical proteins. PharmaceutSci Technol Today 1:352–356.

15. Harris JM, Chess RB. 2003. Effect of PEGylation onpharmaceuticals. Nat Rev Drug Discov 2:214–221.

16. Harris JM. 1992. Introduction to biotechnical andbiomedical applications of poly(ethylent glycol). In:Harris JM, editor. Poly(ethylene glycol) chemistry:Biotechnical and biomedical applications. PlenumPress. New York: pp 1–29.

17. Zalipsky S, Harris JM. 1997. Introduction tochemistry and biological applications of poly(ethy-lene glycol). In: Harris JM, Zalipsky S, editors.Poly(ethylene glycol) chemistry and biologicalapplications. Washington, DC: American ChemicalSociety. pp 1–5.

18. Roberts MJ, Bentley MD, Harris JM. 2002. Chem-istry for peptide and protein PEGylation. Adv DrugDeliver Rev 54:459–476.

19. Wang Y, Youngster S, Grace M, Bausch J, BordensR, Wyss DF. 2002. Structural and biological char-acterization of pegylated recombinant interferonalpha-2b and its therapeutic implications. AdvDrug Deliver Rev 54:547–570.

20. Fee CJ, Van Alstine JM. 2006. PEG-proteins: Reac-tion engineering and separation issues. Chem EngSci 61:924–939.

21. Fee CJ, Van Alstine JM. 2004. Prediction of theviscosity radius and the size exclusion chromato-graphy behavior of PEGylated proteins. Bioconju-gate Chem 15:1304–1313.

22. Kodera Y, Matsushima A, Hiroto M, Nishimura H,Ishii A, Ueno T, Inada Y. 1998. PEGylaion of pro-tiens and bioactive substances for medical andtechnical applications. Prog Polym Sci 23:1233–1271.

23. Rajender Reddy K, Modi MW, Pedder S. 2002. Useof peginterferon alfa-2a (40 KD) (Pegasys1) for thetreatment of hepatitis C. Adv Drug Deliver Rev54:571–586.

24. Charles SA, Harris JM, Pedder S, Kumar S. 2000.Improving hepatitis C therapy: Attaching a poly-ethylene glycol ‘‘tail’’ to interferon for better clinicalproperties. Mod Drug Discov 5:59–67.

25. Monfardini C, Schiavon O, Caliceti P, MorpurgoM, Harris JM, Veronese FM. 1995. A branced


2078 LU ET AL.

monomethoxypoly(ethelene glycol) for protein mod-ification. Bioconjugate Chem 6:62–69.

26. Veronese FM, Harris JM. 2002. Introduction andoverview of peptide and protein pegylation. AdvDrug Deliver Rev 54:453–456.

27. In: Scott D, Harding SE, Rowe AJ, editors. 2005.Analytical ultracentrifugation: Techniques andmethods. Cambridge: Royal Society of Chemistry.

28. In: Harding SE, Rowe AJ, Horton JC, editors.1992. Analytical ultracentrifugation in biochemis-try and polymer science. Cambridge: Royal Societyof Chemistry. 629.

29. Glatter O, Kratky O. 1982. Small angle X-ray scat-tering. London: Academic Press.

30. Koch MHJ, Vachette P, Svergun DI. 2003.Small-angle scattering: A view on the properties,structures and structural changes of biologicalmacromolecules in solution. Q Rev Biophys 36:147–227.

31. Laue TM, Shah BD, Ridgeway TM, Pelletier SL.1992. Computer-aided interpretation of analyticalsedimentation data for proteins. In: Harding SE,Rowe AJ, Horton JC, editors. Analytical ultracen-trifugation in biochemistry and polymer science.Royal Society of Chemistry. Cambridge: pp 90–125.

32. Lepori L, Mollica V. 1978. Volumetric properties ofdilute aqueous solutions of poly(ethylene glycols).J Polym Sci 16:1123–1134.

33. Nichol LW, Ogston AG, Wills PR. 1981. Effect ofinert polymers on protein self-association. FEBSLett 126:18–20.

34. Durchschlag H, Zipper P. 2005. Calculation ofvolume, surface, and hydration properties of biopo-lymers. In: Scott D, Harding SE, Rowe AJ, editors.Analytical ultracentrifugation: Techniques andmethods. Royal Society of Chemistry. Cambridge:pp 389–431.

35. Schuck P. 2000. Size distribution analysis of macro-molecules by sedimentation velocity ultracentrifu-gation and Lamm equation modeling. Biophys J78:1606–1619.

36. Dam J, Schuck P. 2004. Calculating sedimentationcoefficient distributions by direct modeling of sedi-mentation velocity concentration profiles. MethodsEnzymol 384:185–212.

37. Rowe AJ. 1992. The concentration dependence ofsedimentation. In: Harding SE, Rowe AJ, HortonJC, editors. Analytical ultracentrifugation in bio-chemistry and polymer science. Royal Society ofChemistry. Cambridge: pp 394–406.

38. Cölfen H, Harding SE. 1997. MSTARA andMSTARI: Interactive PC algorithms for simple,model independent evaluation of sedimentationequilibrium data. Eur Biophys J 25:333–346.

39. Creeth JM, Harding SE. 1982. Some observationson a new type of point average molecular weight.J Biochem Biophys Methods 7:25–34.


40. Boulin CJ, Kempf R, Gabriel A, Koch MHJ. 1988.Data acquisition systems for linear and area x-raydetectors using delay-line readout. Nucl Instr MethPhys Res 269:312–320.

41. Semenyuk AV, Svergun DI. 1991. GNOM-a pro-gram package for small-angle scattering data pro-cessing. J Appl Crystal 24:537–540.

42. Guinier A, Fournet G. 1995. Small angle scatteringof X-rays. Wiley: New York.

43. Svergun DI. 1999. Restoring low resolution struc-ture of biological macromolecules from solutionscattering using simulated annealing. Biophys J76:2879–2886.

44. Volkov VV, Svergun DI. 2003. Uniqueness ofab initio shape determination in small-angle scat-tering. J Appl Cryst 36:860–864.

45. Svergun DI, Barberato C, Koch MHJ. 1995. CRY-SOL-a program to evaluate x-ray solution scatter-ing of biological macromolecules from atomiccoordinates. J Appl Cryst 28:768–773.

46. Garcia de la Torre J, Huertas ML, Carrasco B. 2000.Calculation of hydrodynamic properties of globularproteins from their atomic-level structure.Biophys J 78:719–730.

47. Kozin MB, Svergun DI. 2000. Automated matchingof high- and low-resolution structural models.J Appl Cryst 34:33–41.

48. Konarev PV, Petoukhov MV, Svergun DI. 2001.MASSHA-a graphics system for rigid-bodymodelling of macromolecular complexes againstsolution scattering data. J Appl Cryst 34:527–532.

49. Heymann B, Grubmuller H. 1999. Elastic proper-ties of poly(ethylene-glycol) studied by moleculardynamics stretching simulations. Chem Phys Lett307:425–432.

50. Chew B, Couper A. 1976. Diffusion, Viscosity andsedimentation of poly(ethylene oxide) in water.J Chem Soc, Faraday Trans 72:382.

51. Branca C, Magazu S, Maisano G, Migliardo P,Migliardo F, Romeo G. 2002. Hydration parametersof aqueous solutions of poly(ethylene glycol)s byviscosity data. Physica Scripta 66:175– 179.

52. Bailey FE, Callard RW. 1959. Some properties ofpoly(ethylene oxide) in aqueous solution. J ApplPolym Sci 1:56–62.

53. Branca C, Magazu S, Maisano G, Migliardo F,Migliardo P, Romeo G. 2003. Study of conforma-tional properties of poly(ehtylene oxide) by SANSand PCS techniques. Physica Scripta 67:551–551.

54. Maron SH, Filisko FE. 1972. Heats of solution anddilution for poly(ethylene oxide) in several solvents.J Macromol Sci Phys 6:79–90.

55. Smidrsød O, Andresen IL. 1979. Biopolymerkjemi.Trondheim, Norway: Tapir Press.

56. Fujita H. 1962. Mathematical theory of sedimenta-tion analysis. Academic Press. New York: 315.

DOI 10.1002/jps


57. van Holde KE. 1985. Chapter 5: Sedimentation.Physical Biochemistry. Prentice-Hall, Inc. NewJersey: pp 110–133.

58. Harding SE, Cölfen H, Aziz Z. 2005. The ELLIPSsuite of whole-body protein conformation algo-rithms for Microsoft Windows. In: Scott D, HardingSE, Rowe AJ, editors. Analytical ultracentrifuga-tion techniques and methods. Royal Society ofChemistry. Cambridge: pp 468–483.

59. Carrasco B, Garciade la Torre J, Byron O, King D,Walters C, Jones S, Harding SE. 1999. Novel size-independent modeling of the dilute solution confor-mation of the immunoglobulin IgG Fab’ domainusing SOLPRO and ELLIPS. Biophys J 77:2902–2910.

60. Lu Y. 2007. PhD Thesis: Solution conformation ofengineered antibodies. School of Biosciences, ed.,Nottingham: University of Nottingham

61. Morgan PJ, Byron OD, Harding SE. 1992. Thesolution conformation of novel antibody fragmentsstudied using analytical ultracentrifugation.

DOI 10.1002/jps J

Discovery 1–4. (Beckman Instruments, Palo Alto,USA) DS-834.

62. Svergun DI. 1992. Determination of the re-gularisation parameter in indirect transformusing perceptual criteria. J Appl Crystal 25:495–503.

63. Furtado PB, Whitty PW, Robertson A, Eaton JT,Almogren A, Kerr MA, Woof JM, Perkins SJ. 2004.Solution structure determination of monomerichuman IgA2 by X-ray and neutron scattering, ana-lytical ultracentrifugation and constrained model-ling: A comparison with monomeric human Ig A1.J Mol Biol 338:921–941.

64. Zipper P, Durchschlag H, Krebs A. 2005. Model-ling of Biopolymers. In: Scott D, Harding SE, RoweAJ, editors. Analytical ultracentrifugation: Techni-ques and methods. Royal Society of Chemistry.Cambridge: pp 320–371.

65. DeLano WL. 2002. The PyMOL Molecular GraphicsSystem. DeLano Scientific. San Carlos. CA, USA,http://wwwpymolorg.


Documents

Effect of PEGylation on the solution conformation of antibody … · Keywords: PEGylation; immunoglobulin; antibody fragment; conformation INTRODUCTION Biomedical exploitation of