Pressure-Induced Dissociation of Antigen-Antibody Complexes

Srikanth Sundaram,† Charles M. Roth,‡ and Martin L. Yarmush*,†,‡

Department of Chemical and Biochemical Engineering, Rutgers University, Piscataway, New Jersey 08854, andCenter for Engineering in Medicine/Surgical Services, Massachusetts General Hospital, Harvard Medical School,and Shriners Burns Hospital, Boston, Massachusetts 02114

Pressures on the order of 1000-4000 bar have been reported to reversibly dissociatea number of oligomeric protein complexes without gross changes in protein structure.Here, we report that hydrostatic pressure can also dissociate some antigen-antibodycomplexes in solution. The association of fluorescent-labeled antigens with monoclonalantibodies was monitored via increases in the fluorescence anisotropy upon binding.Previously, we had found that pressures of 2000 atm were able to dissociate bovineserum albumin (BSA) from immunoadsorbents formed from certain antibodies butnot others. In this study, we have found that the sensitivity to pressure in solutionis present for the interaction of BSA with MAb 9.1 and absent for the interaction ofBSA with MAb 6.1; this behavior is consistent with the immunoadsorbent study. Theinteraction of hen egg white lysozyme with two monoclonal antibodies was alsomeasured. Interestingly, the complex with the greater electrostatic character (HyHEL-5) did not exhibit pressure sensitivity, as would be expected due to electrostrictioneffects, whereas the more hydrophobic complex (HyHEL-10) exhibited a strong pressuresensitivity. In each of the systems displaying pressure sensitivity, the free energy ofassociation was found to increase linearly with pressure, indicating a constant changein volume between the free and bound states. Overall, these results indicate thatsome antigen-antibody complexes exhibit significant sensitivity to pressure, whereasothers do not; the mechanisms that discriminate between these cases remainunresolved. Understanding and manipulation of this phenomenon may prove usefulin a variety of processes involving the recovery from antigens of antibodies.

Introduction

The first reported observation of pressure-inducedchanges in proteins was the coagulation of egg white(albumin) under pressures of 5000-7000 atm (1). Yearslater, Kauzmann (2) predicted that protein denaturationwould be accompanied by large reaction volumes. Sincethen, pressure has been used as a tool to study bothunfolding and association processes, and several reviewsare available (3-6). The response of biological processesto pressure provides information regarding the relativevolumes of different states, e.g., associated vs dissociated.Increase in the fundamental understanding of the mech-anisms involved in the effect of pressure on proteinstructure and function has potential implications in manyareas. These include bioaffinity separations and biosen-sors (7), protein crystallization (8-10), viral (11, 12),tumor and T-cell vaccination (13, 14), enzymology (15-17), and food processing (18, 19).

We have previously studied the pressure elution char-acteristics of monoclonal antibody immunoadsorbents toa model protein antigen, BSA. These studies includedinitial observations on the pressures required to eluteBSA from a panel of anti-BSA monoclonal antibody

immunoadsorbents and the stability of the immobilizedmonoclonal antibodies at the high pressures used for thedissociation (20). For one of the MAb immunoadsorbents(MAb 9.1), over 75% of the reversibly bound antigen wasrecovered following a single incubation at 2000 atm, andover 90% recoveries were obtained by repeated pres-surizations. In addition, repeated pressurizations to 2000atm exerted no detrimental effect on immunoadsorbentbinding properties, whereas immunoadsorbent bindingcapacity was significantly (by about 70%) reduced upontreatment with a common eluent (glycine/HCl at pH 2.5).These results clearly demonstrated that pressure canprovide an effective yet mild elution scheme for therecovery of molecules bound to immunospecific supports.Many questions remain, however, regarding the mech-anisms driving pressure-induced antibody-antigen dis-sociation.

Fluorescence polarization spectroscopy has been usedto measure the dissociation constant of a variety of high-affinity macromolecular association reactions, at bothatmospheric and elevated pressures. Among the reac-tions that have been studied by this technique are thepressure-induced dissociation of enolase (21), tryptophansynthase (22), and the lac repressor (23) and histonesubunit interactions (24). In the present study, weutilized high-pressure fluorescence polarization spectros-copy to characterize the effect of pressure on MAb-Agsystems involving two well-characterized protein anti-gens. Specifically, we have (1) developed, characterized,and validated reagents and methodology for studying

* To whom correspondence should be addressed: Center forEngineering in Medicine, Massachusetts General Hospital, Big-elow 1401, Boston, MA 02114. Telephone: 617-726-3474. Fax: 617-374-5665.

† Rutgers University.‡ Massachusetts General Hospital.

773Biotechnol. Prog. 1998, 14, 773−781

S8756-7938(98)00066-6 CCC: $15.00 © 1998 American Chemical Society and American Institute of Chemical EngineersPublished on Web 08/26/1998

antigen-antibody interactions in solution under highhydrostatic pressure, (2) studied the effect of pressureon the model antigens used in this study, namely BSAand HEWL, and (3) screened a selection of monoclonalantibodies to BSA and HEWL for pressure sensitivity.

Materials and Methods

Proteins. BSA with its free sulfhydryl group boundby a cysteine molecule (Cys-BSA) was obtained fromSigma Laboratories (St. Louis, MO). HPLC analysisusing a TSK 2000 SWXL column did not reveal asignificant amount of oligomer; thus, Cys-BSA was usedfor labeling without further purification. HEWL wasobtained from Boehringer-Mannheim and was purifiedusing SEC with a Superose 12 column to remove oligo-mers and some low-molecular-weight contaminants. BSAand HEWL concentrations were determined by UVabsorbance at 280 nm assuming extinction coefficientsof 0.66 and 2.64 cm2/mg, respectively.

Monoclonal Antibodies. Anti-BSA monoclonals usedin this study (6.1 and 9.1) were selected from a panel ofMAbs which had been previously characterized in thislaboratory according to their physicochemical and bindingproperties (25, 26) as well as their pressure behaviorupon immobilization. The cell lines were donated byNEN Research (Boston, MA) and grown in mouse ascites,as previously described (25). The anti-HEWL mono-clonals (HyHEL-5, HyHEL-10) used in this study werea gift from Dr. Sandra Smith-Gill (National Institutesof Health, Bethesda, MD) and were also grown in mouseascites. All anti-BSA and anti-HEWL MAbs were murineIgG1, κ. Anti-BSA MAbs were isolated by affinity chro-matography as previously described (25, 27).

Anti-HEWL antibodies were purified using a combina-tion of ion-exchange chromatography and protein Achromatography. First, the ascites fluid was extensivelydialyzed against PBSA and filter-sterilized through a0.22-µm filter. The immunoglobulin fraction was pref-erentially precipitated using ammonium sulfate precipi-tation; an equal volume of saturated ammonium sulfatewas added dropwise with constant stirring to ascitesfluid, and the reaction was allowed to proceed overnightat 4 °C. The mixture was then centrifuged to removethe precipitated immunoglobulin. The resuspended an-tibody fraction was dialyzed extensively against 20 mMTris, pH 8.0, to remove any traces of ammonium sulfate.The antibody fraction was then applied to a Mono-Q (orQ-Sepharose fast flow) column (Pharmacia, Piscataway,NJ) and eluted with a slow salt gradient. Most of theantigen-specific antibody eluted between 150 and 200mM NaCl as measured by ELISA. Fractions that pos-sessed anti-HEWL activity were pooled together andanalyzed by isotyping (Amersham, Piscataway, NJ),which revealed the presence of some contaminatingIgG2a and IgG2b. The contaminating immunoglobulinswere then removed by passing the sample over a 1-mLProtein A cartridge, which has no affinity for murineIgG1 under low salt conditions (Pharmacia). FollowingProtein A treatment, no contaminating IgG2 could bedetected in the flow-through by isotyping, while theeluent from the Protein A column (eluted with 0.1 Mcitrate buffer, pH 3.0) contained mostly IgG2. Thepurified anti-HEWL MAbs were monomeric, as indicatedby HPLC analysis using a TSK 3000 SWXL column.

Labeling Reagents. All fluorophores used in thisstudys5-((((2-iodoacetyl)amino)ethyl)amino)naphthalene-1-sulfonic acid (IAEDANS), iodoacetoamido fluorescein(IAF), and pyrene maleimide (PM)swere obtained from

Molecular Probes (Eugene, OR). 2-Iminothiolane (Traut’sreagent), 5,5′-dithiobis(2-nitrobenzoic acid) (DTNB, Ell-mans reagent), and 2-mercaptoethanol were obtainedfrom Pierce Chemicals (Rockford, IL). For fluorescencemeasurements, ultrapure Hepes was obtained fromSigma Chemicals (St. Louis, MO), and triple-crystallizedTris was obtained from Fisher Scientific (Springfield, NJ).Ellman’s reaction was used to estimate the free thiols inproteins.

BSA Labeling. While BSA is supposed to have onefree sulfhydryl (-SH) per molecule due to the free Cysresidue at position 34, commercially available prepara-tions (even high-grade “monomer” preparations) meas-ured using Ellman’s assay gave -SH values ranging from0.3 to 0.5 per molecule. Therefore, cysteinal BSA withno free thiols was used as the starting reagent. Treat-ment of Cys-BSA with â-mercaptoethanol for 1 h at 0 °Cand pH 6.3 was found to regenerate the blocked Cys34thiol. BSA stocks treated under these conditions exhib-ited labeling ratios of 0.85-1.0 thiol group per molecule(as determined with the Ellman’s assay), whereas theuntreated Cys-BSA exhibited no free thiols. The pH ofthe mercaptoethanol was a significant factor in theregeneration of mercaptalbumin, whereas the durationof treatment beyond 1 h did not greatly change the extentof reaction. The number of free thiol groups regeneratedat pH 6.3 was observed to decrease with time of storageat 4 °C; consequently, labeling reactions were carried outimmediately after regeneration of mercaptalbumin.

Preliminary efforts to label BSA under native bufferresulted in much greater than 1 (∼34) labels per BSAmolecule and yet incomplete reaction at Cys34 or otherlocations, as indicated by residual free thiol ratios of 0.4-0.6 (28). Therefore, the labeling reaction was conductedunder denaturing conditions (6 M GuHCl in 20 mM Tris,5 mM EDTA, pH 8.0) with a 20-fold molar excess ofIAEDANS, at which the labeling reaction was found toproceed efficiently to completion. BSA is known torecover its native conformation after such a treatmentand retain greater than 90% of its immunoreactivity uponrefolding. The reaction was conducted for 1 h at roomtemperature, following which excess dye and GuHCl wereremoved in a PD-10 desalting column and the labeledprotein was recovered in native buffer (20 mM Tris, 5mM EDTA, pH 8.0). After labeling, the number ofresidual free thiol groups was close to zero (less than0.05), and the IAEDANS/BSA ratio was close to one(greater than 0.90). Size exclusion HPLC analysisindicated the presence of a small amount of dimer (lessthan 2%). The dimers were removed by SEC with aSuperose 12 column.

HEWL Labeling. HEWL does not have any free -SHgroups; therefore, a thiol group was introduced by react-ing lysozyme with 2-iminothiolane (Traut’s-reagent) (29).The Traut’s labeled HEWL served as the starting mate-rial for labeling with IAF. To minimize the chance ofcompromising the immunoreactivity of the labeled pro-tein, the Traut’s reaction was targeted to the R-aminoterminus of the protein by conducting the reaction at pH6. Using 10 mM of Traut’s reagent at 0 °C for 24 h, about0.7-0.8 -SH group was attached per HEWL molecule.

Following Traut’s reaction, HEWL was reacted with a20-fold molar excess of IAF. Contrary to the problemsencountered with BSA, HEWL labeling with IAF pro-ceeded to completion even under native conditions. Thefinal dye-to-protein ratio (usually about 0.7) correspondedclosely to the starting -SH value of Traut’s-labeledHEWL; residual -SH values after incorporation of thedye were close to zero. No significant amount of dimers

774 Biotechnol. Prog., 1998, Vol. 14, No. 5

could be detected by HPLC; thus, labeled HEWL wasused without further purification.

Immunoactivity. Immunoactivity of labeled proteinswas determined via an ELISA assay. ELISA plates werecoated overnight at 4 °C with 100 µL of a protein solution(1 µg/mL for lysozyme and 10 µg/mL for BSA, each inTBS). The wells were then washed thoroughly with TBS-Tween and blocked with either 1% BSA (for lysozymeELISA) or 1% thyroglobulin (for BSA ELISA) in TBS.After thorough washing, combinations of the appropriateantibody and different amounts of the labeled proteinbeing tested as well as controls (unlabeled protein) werepremixed and preincubated at room temperature for 1 hand then added to the plates and incubated for anadditional 2 h. Enzyme-labeled anti-mouse secondaryantibodies were then added, and detection of the productformation upon substrate addition was determined spec-trophotometrically using a plate reader. Specifically, agoat anti-mouse alkaline phosphatase secondary antibody(Sigma) diluted 1:500 in TBS-Tween was used for thelysozyme assays. The substrate was preformulatedpNPP (Sigma) dissolved in 5 mL of 10% DEA, pH 9.8.Absorbance was determined at 405 nm about 30 minafter addition of substrate. Because the aforementionedgoat anti-mouse antibody comes with BSA as a preserva-tive, an anti-mouse alkaline phosphatase antibody thatdoes not contain BSA was obtained from BoehringerBiochemicals and used for the BSA assays. Otherwise,the BSA ELISA was performed in like manner to thatfor lysozyme.

High-Pressure Fluorescence Polarization Spec-troscopy. All fluorescence measurements, with theexception of lifetimes, were performed on a custom-designed photon-counting spectrofluorometer, model PC1(ISS, Champaign, IL). Fluorescence lifetime measure-ments were carried out on a Koala-PC (ISS) at theLaboratory of Fluorescence Dynamics at the Universityof Illinois, UrbanasChampaign, IL. The model systemsused were all of high affinity, with equilibrium dissocia-tion constants on the order of nanomolar to picomolar.This necessitated that the pressure experiments beconducted at very low concentrations (usually 1-50 nM)to accurately measure the pressure-induced phenomena.

A fluorescence cell (ISS) specifically designed for high-pressure experiments and rated to 4 kbar was used forthese experiments. The cell has three optical ports (oneexcitation and two emission) and one hydrostatic port topermit its pressurization. The three optical windows are10 mm in diameter and placed in the optical plane toallow the use of the cell for both L- and T-formatfluorescence measurements. High pressure was gener-ated by fluid compression, with high-grade, low-fluores-cence 200-proof ethanol as the pressurizing medium andcontrolled using an automated pressure control andgeneration system (APP Systems, Ithaca, NY).

High-Pressure Cell Window Birefringence Cor-rection. It has been previously reported that quartzwindows used in high-pressure devices show a pressure-dependent scrambling of polarization, presumably as aresult of strain-induced anisotropies in the windowmaterial (the so-called photoelastic effect) (30). Severalexperiments were performed to determine the magnitudeof pressure-induced depolarization and the best methodto correct measured fluorescence polarization (or anisot-ropy) data for this effect. The fluorophore for whichcorrection factors were to be determined (IAEDANS orIAF) was reacted with cysteine and dissolved in concen-trated glycerol (about 90%). A set of correction factors

was determined, and these were used to correct allpressure data for the systems containing the fluorophore.

The determination of correction factors using confor-mationally pressure-invariant samples was accomplishedby the method of Paladini and Weber (30). Briefly, thetrue polarization (Pt) is related to the measured polariza-tion (Pe) for L-format measurements as (30)

where

and R is the window correction factor (assuming that boththe excitation and emission windows are identical), I|| isthe intensity of emitted light parallel to the incidentpolarized light, and I⊥ is the emitted intensity perpen-dicular to the incident polarized light. The two keyassumptions are that there is negligible birefringence atatmospheric pressure and that the birefringence isinsensitive to wavelength in the relevant range; the latterhas been tested and found to be accurate (30). Theanisotropy (r) is given by

Measured polarization values for glycerol-based sampleswere corrected for viscosity-induced changes before es-timating window correction factors by using the Perrinequation (30).

Analysis of Binding Parameters from AnisotropyData. The fraction bound λ can be determined from thecorrected anisotropies in the poised, free, and boundstates from

where rf, rλ and rb are the anisotropies at the free,partially bound, and completely bound states, respec-tively, and Y is the ratio of the quantum yields of thebound to free labeled antigen, which was measured andfound to be unity for the systems studied here. As themeasured anisotropy is a function of pressure due tobirefringence and, to lesser extent, viscosity, two separatemethods were used to calculate the fraction bound. First,the anisotropies of the free and bound states weredetermined at each pressure, and eq 4 was applieddirectly. Second, the anisotropies at each pressure andcomposition were corrected as described above, in whichcase the anisotropy of the free and completely boundstates needed only to be measured at atmospheric pres-sure.

The dissociation constant Kd corresponding to eachbound fraction λ was calculated from the equation

where MT and XT are the concentrations of antibody sitesand antigen molecules, respectively. The BSA antibodiesused here have been extensively characterized and foundto exhibit noncooperative binding such that there are ca.2 binding sites for each antibody molecule (25). For each

Pe ) Pt(1 - R)(1 - 2R)

(1 - RPt)(1)

Pe )I| - I⊥

I| + I⊥(2)

r )2Pt

3 - Pt(3)

λ )rλ - rf

(rb - rλ)Y + (rλ - rf)(4)

Kd )(MT - λXT)(1 - λ)

λ(5)

Biotechnol. Prog., 1998, Vol. 14, No. 5 775

protein, we assume that MT is equal to twice the antibodyconcentration.

The increase in dissociation constant with pressure isused to define the reaction volume or dissociation volume,∆Vd, via the thermodynamic relationship

For a macroscopic change in pressure, the reactionvolume can, in principle, be a function of pressure; thefirst-order change in reaction volume is identified witha change in isothermal compressibility (21, 31). In thisstudy, free energy versus pressure data were fitted usinglinear regression and, where the reaction volume wasstatistically different from zero, also to a quadraticregression. In each case, analysis of variance of theregressed parameters was used to define an F statistic.A correlation was deemed significant if it passed theF-test at the p ) 0.05 level and highly significant if itpassed at the p ) 0.01 level.

ResultsChange in Fluorescence Anisotropy upon Anti-

body Binding. To be useful for determination ofantibody-antigen association/dissociation behavior, thelabeled antigens must exhibit a distinguishable changein fluorescence anisotropy upon antibody binding. Thefluorophores IAEDANS, IAF, and pyrene were eachconjugated to BSA, and IAEDANS and IAF were eachconjugated to lysozyme. The changes in anisotropy wereconsidered acceptable for all the antigen-label conju-gates except BSA-IAF (28). On the bases of fluorescencesensitivity, change in anisotropy, and ease of labeling,conjugates of BSA-IAEDANS and HEWL-IAF werechosen for further study.

Immunoreactivity of Labeled Antibodies. Com-petitive ELISAs were performed to evaluate the effect oflabeling on the antigenicity of lysozyme and BSA. Theantibodies MAb 5.1 and MAb 9.1 are known to bind todomains I-C and III-C of BSA, respectively (25). As canbe seen in Figure 1, the presence of the IAEDANS labeldid not affect the binding of BSA to antibody MAb 5.1and only slightly affected the binding to MAb 9.1.Likewise, only minor differences in binding occurredbetween native and IAF-labeled lysozyme with theantibodies HyHEL-5 and HyHEL-10 (Figures 2a and b,respectively). These results suggest that the ability ofthe antibodies to bind their antigens is not affectedsignificantly by the presence of the fluorophores.

Effect of Pressure on Fluorescence Properties.At elevated pressures, birefringence in the windows ofthe pressure cell results in a scrambling of the fluores-cence polarization. This effect was accounted for by themethod of Paladini and Weber (30) using Cys-IAEDANSin glycerol as a pressure-invariant standard. Thedecrease in anisotropy of Cys-IAEDANS with pressure,corrected for the pressure-dependent change in viscosityusing the Perrin equation and the viscosity data ofMcDuffie and Kelly (32), was used to determine thevalues of the window correction factors as the solutionof eq 1 for the window correction (or scrambling) factorR at each pressure. Using the window correction factorsdetermined for our apparatus using Cys-IAEDANS, theeffect of birefringence for other molecules was deter-mined. The absolute and corrected anisotropies of BSA-IAEDANS and HEWL-IAF are shown in Figure 3. Foreach molecule, the window correction factors accounted

for the pressure dependence of the observed anisotropies.Furthermore, the fluorescence maxima and intensitiesof BSA-IAEDANS and HEWL-IAF were independentof pressure. This behavior indicates that gross confor-mational changes did not alter the environment of thefluorophore as a function of pressure. Upon addition ofexcess antibody (MAb 9.1 for BSA-IAEDANS and Hy-HEL-5 or HyHEL-10 for HEWL-IAF), no changes influorescence intensity or emission maximum were ob-served, indicating that the binding also did not affect theenvironment around the fluorescent label. This is con-sistent with the known epitopes of the antigens beingdistal from the putative binding sites of the fluorescentlabels.

Pressure Behavior with Anti-BSA Antibodies.Having validated the labeled antigens, we sought toinvestigate the behavior of the interaction in solutionbetween BSA and some of its MAbs under pressure todetermine whether our previous results on immuno-adsorbents were due to the immobilization procedure.The degree of association of antigen and antibody is foundby comparing the anisotropies under conditions of partialdissociation (“poised” sample) to those where the antigenis fully bound and where the antigen is fully dissociated(i.e., in the absence of antibody). These results are shownfor the interaction of BSA-IAEDANS with MAb 9.1 inFigure 4. For the concentrations used (50 nM BSA-IAEDANS and 37.5 nM MAb 9.1), the BSA was mostlybound at low pressures and exhibited an increasingdegree of dissociation as pressure increased over therange of 1-2000 bar. The fraction of BSA bound vspressure was calculated from the anisotropy data usingeq 4. Under these conditions, approximately 80% of theantigen was bound at atmospheric pressure, whereasonly about 5% was bound at 2000 bar.

∆Vd ) (∂∆Gd

∂p )T

) -RT(∂ ln Kd

∂p )T

(6)

Figure 1. Competitive ELISA of labeled and unlabeled BSA.The indicated concentrations of BSA-Cys or BSA-IAEDANSwere added to BSA-coated wells containing a constant amountof (A) MAb 5.1 or (B) MAb 9.1. After subsequent washing,incubation with secondary antibody, and development, theabsorbance of each sample was measured using a plate reader.Values represent the mean ( SD for triplicate measurements,normalized to the value with no competing antigen added.

776 Biotechnol. Prog., 1998, Vol. 14, No. 5

A quantitative measure of the change in dissociationbehavior can be made by expressing the association datain terms of thermodynamics quantities, such as thedissociation constant, Kd, or the association free energy(∆Gassoc ) RT ln Kd). For the interaction of BSA-IAEDANS with MAb 9.1, the dissociation constant de-termined at atmospheric pressure (14 nM) is comparableto the previously reported value of 6.3 nM (25). Theassociation free energy increased by about 2.7 kcal/mol,corresponding to an increase in dissociation constant of2 orders of magnitude, with pressure increasing from 1to 2000 bar (Figure 5). From the slope of the best-fit line(r2 ) 0.954) to the data in Figure 5, the volume decreaseupon dissociation is computed to be ∆Vd ) -57 ( 5 mL/mol BSA.

Pressure-induced dissociation of antigen-antibodycomplexes is not, however, a universal phenomenon. Thefree energy of association for binding of BSA-IAEDANSto MAb 6.1, an antibody that recognizes an epitope onthe I-C domain of BSA, was found to be insensitive topressure over the same range of 1-2000 bar (Figure 5),even at concentrations where the complexes were wellpoised (i.e., λ ≈ 0.5). The dissociation constant deter-mined at atmospheric pressure (115 nM) is somewhatgreater than the known value of approximately 25-30nM (25). This small discrepancy could be attributableto the presence of the fluorophore at Cys34, which is nearthe putative epitope of MAb 6.1 on BSA. Increasingpressure also did not significantly affect the binding ofBSA-IAEDANS with the MAbs 3.1 or 5.1 (data notshown).

Pressure Behavior with Anti-HEWL Antibodies.The dissociation constant for the binding of HEWL toHyHEL-5 was determined by the same methods as for

BSA to its antibodies. The free energy of association forthis antibody-antigen pair increased only very slightlywith pressure (Figure 6). The volume change on dis-sociation of -4 ( 1 mL/mol HEWL-IAF is statisticallysignificant (p < 0.01) but, in light of the magnitudes ofreaction volumes determined in this and other studies,is inconsequential. The measured dissociation constantof 520 ( 150 pM at atmospheric pressure agrees withthat previously reported for this system (50).

Figure 2. Competitive ELISA of labeled and unlabeled HEWL.The indicated concentrations of HEWL or HEWL-IAF wereadded to HEWL-coated wells containing a constant amount of(A) MAb HyHEL-5 or (B) MAb HyHEL-10. After subsequentwashing, incubation with secondary antibody, and development,the absorbance of each sample was measured using a platereader. Values represent the mean ( SD for triplicate measure-ments, normalized to the value with no competing antigenadded.

Figure 3. Fluorescence properties of labeled antigens underpressure. (A) Measured ([) and corrected (9) anisotropy valuesof 1 µM BSA-IAEDANS as a function of pressure at 25 °C.Excitation was at 340 nm with 8-nm band-pass through a UG11band-pass filter, and fluorescence emission was monitored at485 nm through a 485DF22 interference filter. Correction factorswere determined using eqs 1-4 and measurements on apressure-invariant sample (Cys-IAEDANS in concentratedglycerol). (B) Measured ([) and corrected (9) anisotropy valuesof 1 nM HEWL-IAF as a function of pressure at 25 °C. Excitationwas at 485 nm through a 485DF22 interference filter, andfluorescence emission was monitored at 530 nm through a530DF30 interference filter. Correction factors were determinedas for BSA.

Figure 4. Anisotropy measurements for various proportionsof BSA-IAEDANS and MAb 9.1. Measured anisotropies as afunction of pressure at 25 °C for concentrations of 1000 nMBSA-IAEDANS with no antibody (“free”; 9), 1000 nM BSA-IAEDANS with 10 µM MAb 9.1 (“bound”; b), and 50 nM BSA-IAEDANS with 37.5 nM MAb 9.1 (“poised”; 2). Values representthe mean ( SD for duplicate measurements.

Biotechnol. Prog., 1998, Vol. 14, No. 5 777

The binding of another anti-HEWL antibody, HyHEL-10, exhibited considerable pressure dependence. Atprotein concentrations of 1.33 nM HEWL-IAF and 0.75nM HyHEL-10, a dramatic change in dissociation con-stant was observed, from 8.7 pM at atmospheric pressureto 1420 pM at 2000 bar (Figure 6). This corresponds toa volume decrease upon dissociation of ∆Vd ) -58 ( 4mL/mol HEWL. Quadratic regression of the 1.33 nMHEWL-IAF data resulted in a statistically significant(although not highly significant, i.e., 0.025 < p <0.05)change in reaction volume with pressure (d∆Vd/dp) of0.012 mL mol-1 kbar-1. Experiments at protein concen-trations of 13.3 nM HEWL-IAF and 7.5 nM HyHEL-10resulted in slightly different absolute values of dissocia-tion constant but a nearly identical dissociation volumeof -58 ( 3 mL/mol HEWL. For this data set, quadraticregression did not result in a statistically significantchange in reaction volume with pressure.

DiscussionThe effect of hydrostatic pressure on a number of

protein oligomers has been extensively studied and

reviewed. However, the same cannot be said for Ag-MAb systems, especially those involving protein antigens.High-pressure studies on antigen-antibody systems havebeen generally limited to qualitative observations utiliz-ing immobilized antibodies. For example, high pressurewas observed to increase the solubility of immunecomplexes and to reduce the binding of polyclonal anti-bodies to their antigen (33, 34). In recent years, therehave been reports concerning MAb-hapten systems,particularly fluorescein-binding MAb 4-4-20 (35-37). Wehave previously characterized the pressure behavior ofa panel of MAb immunoadsorbents to BSA and estab-lished that high pressure can be an effective and mildmeans of eluting bound antigen. We now report on theeffect of pressure on MAb-Ag systems involving proteinantigens in solution. A panel of four MAbs directed totwo different antigens (BSA and HEWL) was studied.

To address the issue of pressure-induced conforma-tional changes in the antigens, the effect of pressure onfluorescence properties of labeled antigens was studied.The fluorescence intensity, lifetime, and anisotropy alldepend on the local environment of the fluorophore. Thatno detectable changes in these properties for eitherlabeled antigen occurred in solution under pressureindicates strongly that no significant conformationalchanges occurred. There have been a number of reportsregarding pressure-induced conformational changes ofHEWL using a variety of techniques such as fluorescencespectroscopy (38), quasi-elastic light scattering (39),Raman spectroscopy (40), and X-ray crystallography athigh pressure (41). These studies clearly establishedthat, at pressures below 2000-3000 bar, only very minorchanges occur in the protein conformation. In addition,all such changes were completely reversible. The resultsobtained in this study using fluorescence spectroscopyof HEWL-IAF are consistent with that conclusion.

The extent and nature of pressure-induced conforma-tional changes in BSA is less clear from the existingliterature. Paladini et al. (42) found that the electro-phoretic mobility of BSA decreased by about 22% whensubjected to hydrostatic pressures of 1-2 kbar. Nystromand Roots (43, 44) used quasi-elastic light scattering tostudy the effect of pressure-induced denaturation ofbovine serum albumin in buffers of pH 4.7 and 7.4. Ineach case, the molecules appeared to contract slightly atpressures from 1 to 800 bar. At pressures above 800 bar,Rh increased roughly linearly with increasing pressure;over the pressure range 1-4000 bar, Rh increased byabout 30%. When compared with lysozyme, it appearsthat BSA undergoes a larger conformational change thanlysozyme in similar solvents. Recently, the effect ofpressure on the fluorescence intensity of commerciallyavailable BSA-FITC was reported to decrease to about60% of the atmospheric value after a 1-h treatment at2000 bar (45). Nonetheless, the immunoreactivity of thepressure-treated BSA-FITC with MAb 9.1 was notsignificantly affected. No significant conformationaleffects were observed for singly labeled BSA (BSA-IAEDANS) used in this study for monitoring pressure-induced dissociation of Ag-MAb complexes. BSA labeledwith other fluorophores of different fluorescence lifetimes(pyrene maleimide, IAF) at the same position (Cys34)likewise did not exhibit variations in fluorescence proper-ties with pressure.

Of the six Mab-Ag pairs tested in this study, only twoshowed significant pressure sensitivity. This finding isconsistent with previous reports on the use of pressureto elute protein antigens from immunoadsorbents (20,46). From a selection of five immobilized monoclonal

Figure 5. Free energy of association of BSA-IAEDANS withMAbs 9.1 and 6.1 vs pressure. The free energies of associationfor the interaction of 50 nM BSA-IAEDANS with 37.5 nM MAb9.1 (b) and 250 nM BSA-IAEDANS with 100 nM MAb 6.1 (4)as a function of pressure were determined from the anisotropydata (e.g., Figure 4) and eqs 4 and 5. Values represent the mean( SD for duplicate measurements. The linear regression of ∆Gvs pressure for the BSA-IAEDANS/9.1 system is included andindicates a volume decrease upon dissociation of magnitude 57( 5 mL/mol BSA. No significant correlation between free energyand pressure exists for BSA-IAEDANS/6.1.

Figure 6. Free energy of association of HEWL-IAF withHyHEL-5 and HyHEL-10 vs pressure. The free energies ofassociation for the interaction of 1 nM HEWL-IAF with 1 nMHyHEL-5 (2), 1.33 nM HEWL-IAF with 0.75 nM HyHEL-10(b), and 13.3 nM HEWL-IAF with 7.5 nM HyHEL-10 (O) as afunction of pressure are shown. Values represent the mean (SD for duplicate measurements. Linear regressions of ∆G vspressure for each system are included, indicating a volumedecrease upon dissociation of magnitude 4 ( 1 (HyHEL-5), 58( 4 (0.75 nM HyHEL-10), and 58 ( 3 mL/mol HEWL (7.5 nMHyHEL-10).

778 Biotechnol. Prog., 1998, Vol. 14, No. 5

antibodies to BSA, MAb 9.1 showed the greatest sensitiv-ity to pressure; other antibodies, including MAb 6.1,exhibited little to no pressure sensitivity (46). Theresults of the present study serve to verify those resultsand, through their consistency, indicate that the pressureeffect was not related to immobilization of the antibodies.The interaction of lysozyme with different antibodieslikewise produced different dependencies. These resultsdemonstrate that pressure-induced dissociation is not aparticular property of an antigen. Furthermore, it is notparticular to an antibody; we have found that theinteraction between human serum albumin and MAb 9.1(not shown) is not sensitive to pressures up to 2000 bar,despite the marked sensitivity of the BSA-MAb 9.1interaction. Therefore, we conclude that pressure sen-sitivity is a function of the physicochemical nature of theantigen-antibody interface.

The extent of pressure-induced dissociation observedfor the interaction of HEWL and HyHEL-10 was notdependent on protein concentration. While there appearsto be a consistent difference of about a factor of 2 inequilibrium constant between the two measured concen-trations, this corresponds to a rather small difference infree energies of about 0.4 kcal/mol; furthermore, thereaction volumes computed from the two data sets areidentical (-58 mL/mol HEWL). This expected resultserves as added confirmation of the accuracy of ourmeasurements and of the notion that pressure affectsonly the intermolecular association of the antigen-antibody interface. The reaction volumes computed herefor the HEWL-HyHEL-10 and BSA-MAb9.1 systemswere similar in magnitude (∼-60 mL/mol) and withinthe range observed for protein dimeric systems (e.g., arange of ∼-55 mL/mol for enolase (21) to ∼-170 mL/mol for tryptophan synthase (22)).

For the pressure-dependent systems studied here, thefree energy was essentially linear with pressure. Thereaction volume was regressed as the slope of thisrelationship (eq 6) and appears to be a meaningfulparameter to describe the pressure sensitivity of theantigen-antibody association. While eq 6 dictates thatlocally the free energy will be a linear function ofpressure, for some molecules changes in reaction volumewith pressure do occur. First-order corrections arecharacterized by a change in isothermal compressibilitywhich can be significant for certain association reactions,such as dimerization of carboxylic acids (31, 47). For theprotein dimer enolase (21), a change in reaction volumewith pressure of -30 mL mol-1 kbar-1 was reported,albeit that the regression was performed from fractionbound vs pressure rather than free energy vs pressure.In only one of the systems that we studied was there asomewhat significant compressibility change, as deter-mined by quadratic regression of the free energy vspressure data. The change in reaction volume deter-mined here, 0.012 mL mol-1 kbar-1, is 3 orders ofmagnitude less than the value obtained by Paladini andWeber (21) for enolase and is even about 2 orders ofmagnitude less than the values obtained for carboxylicacids (31, 47); thus, we conclude that the change incompressibility is negligible for the systems under studyhere. Furthermore, we conclude that the reaction volumeis a meaningful parameter which is characteristic of aparticular antigen-antibody system. The pressure de-pendence of association of varying concentrations ofantigen and antibody is described accurately by a singlereaction volume. The dependence of the reaction volumeon environmental variables (temperature, ionic strength,pH) remains an issue for further study.

The only unexpected finding in this study comes fromthe fact that the interaction of HEWL with HyHEL-10exhibited significant pressure sensitivity, whereas thatof HEWL with HyHEL-5 was almost completely insensi-tive to pressure. The solvation of charged groups givesrise to a large negative change in volume and is thusconsidered to be a major contributor to reaction volumes(see, e.g., 48), and hence to pressure sensitivity. TheHEWL-HyHEL-5 interaction is known to be electro-statically driven (49, 50); there are three salt links at theinterface resulting from the interaction among fourcharged residues buried at the interface (Arg 68 and Arg45 of HEWL and Glu 35 and Glu 50 of the MAb heavychain). Yet, HyHEL-5 showed little or no pressuresensitivity. On the other hand, the binding site ofHyHEL-10 is dominated by primarily hydrophobic resi-dues; for example, 6 out of the 13 residues in HyHEL-10that contact the antigen are tyrosines (51, 52). Nonethe-less, HyHEL-10 exhibited strong pressure sensitivity(∆Vd ≈ -58 mL/mol).

The finding that some antigen-antibody complexes canbe dissociated under hydrostatic pressure has potentialapplication to a number of bioprocessing and biotechnol-ogy areas. Clearly, however, much more needs to belearned regarding the fundamental mechanisms of thepressure effect, as well as the influence of environmentalvariables (e.g., pH, ionic strength, temperature, chemicalpotential of other species in solution) before this phe-nomenon can be applied in a practical setting. Thekinetics of pressure-induced dissociation are also likelycritical to its applicability. The results presented hereraise questions regarding the mechanism of pressure-induced dissociation that can only be answered throughmore extensive modeling and experiments.

Notation

BSA bovine serum albuminDEA diethanolamineHEWL hen egg white lysozymeMAb monoclonal antibodyAg antigenIAEDANS 5-((((2-iodoacetyl)amino)ethyl)amino)naph-

thalene-1-sulfonic acidIAF iodoacetamido fluoresceinPBSA phosphate-buffered saline with 0.02% sodium

azidePCS photon correlation spectroscopypNPP p-nitrophenyl phosphateRh hydrodynamic radiusSEC size exclusion chromatography-SH sulfhydrylTBS Tris-buffered saline

Acknowledgment

This work was partially supported by a grant from theNational Science Foundation (BES-9696066).

References and Notes(1) Bridgman, P. W. The coagulation of albumen by pressure.

J. Biol. Chem. 1914, 19, 511-512.(2) Kauzmann, W. Some factors in the interpretation of protein

denaturation. Adv. Protein Chem. 1959, 14, 1-67.(3) Weber, G.; Drickamer, H. G., The effect of high pressure

upon proteins and other biomolecules. Q. Rev. Biophys. 1983,16, 89-112.

Biotechnol. Prog., 1998, Vol. 14, No. 5 779

(4) Heremans, K. Pressure effects on the secondary and tertiarystructure of biopolymers. In Current Perspectives in High-Pressure Biology; Jannasch, H. W., Marquis, R. E., Zimmer-man, A. M., Eds.; Academic Press: New York, 1987; pp 225-234.

(5) Silva, J. L.; Weber, G. Pressure stability of proteins. Annu.Rev. Phys. Chem. 1993, 44, 89-113.

(6) Gross, M.; Jaenicke, R. Protein under pressure: The influ-ence of high hydrostatic pressure on structure, function, andassembly of proteins and protein complexes. Eur. J. Biochem.1994, 221, 617-630.

(7) Cheung, C. Y.; Green, D. J.; Litt, G. J.; Laugharn, J. A. High-pressure-mediated dissociation of immune complexes dem-onstrated in model systems. Clin. Chem. 1998, 44, 299-303.

(8) Visuri, K.; Kaipainen, E.; Kivimaki, J.; Niemi, H.; Palosaari,S. A new method for protein crystallization using highpressure. Bio/Technology 1990, 8, 547-549.

(9) Gross, M.; Jaenicke, R. Growth inhibition of lysozymecrystals at high hydrostatic pressure. FEBS Lett. 1991, 284,87-90.

(10) Schall, C.; Wiencek, J. M.; Yarmush, M. L.; Arnold, E.Lysozyme crystal growth reduced at high pressure. J. CrystalGrowth 1994, 135, 548-554.

(11) Nakagami, T.; Shigehisa, T.; Ohmori; T., Taji, S.; Hase,A.; Kimura, T.; Yamanishi, K. Inactivation of herpes virusesby high hydrostatic pressure. J. Virol. Methods 1992, 38,255-262.

(12) Silva, J. L.; Luan, P.; Glaser, M.; Voss, E. W.; Weber, G.Effects of hydrostatic pressure on a membrane envelopedvirus: High immunogenicity of the pressure inactivated virus.J. Virol. 1992, 66, 2111-2117.

(13) Lider, O.; Karin, N.; Shinitzky, M.; Cohen, I. R. Therapeuticvaccination against adjuvant arthritis using autoimmune Tcells pretreated with hydrostatic pressure. Proc. Natl. Acad.Sci. U.S.A. 1987, 84, 4577-4580.

(14) Gesmundo, N.; Calonghi, N.; Sartor, G.; Farrugia, G.;Miotti, S.; Canevari, S.; Colnaghi, M. I.; Masotti, L. Effect ofhydrostatic pressure and cholesterol depletion on the expres-sion of a tumor-associated antigen. Eur. J. Biochem. 1993,217, 337-343.

(15) Kunugi, S. Effect of pressure on activity and specificity ofsome hydrolytic enzymes. Colloq. INSERM 1992, 224, 129-137.

(16) Nakamura, K. Biocatalysis in supercritical fluids. TrendsBiotechnol. 1990, 8, 288-292.

(17) Kim, J.; Dordick, J. S. Pressure affects enzyme functionin organic media. Biotechnol. Bioeng. 1993, 42, 772-776.

(18) Hayashi, R. Utilization of pressure in addition to temper-ature in food science and technology. Colloq. INSERM 1992,224, 185-194.

(19) Mozhaev, V. V.; Heremans, K.; Frank, J.; Masson, P.;Balny, C. Exploiting the effects of high hydrostatic pressurein biotechnological applications. Trends Biotechnol. 1994, 12,493-501.

(20) Olson, W. C.; Leung, S. K.; Yarmush, M. L. Recovery ofantigens from immunoadsorbents using high pressure. Bio/Technology 1989, 7, 369-373.

(21) Paladini, A. A.; Weber, G. Pressure-induced reversibledissociation of enolase. Biochemistry 1981, 20, 2587-2593.

(22) Silva, J. L.; Miles, E. W.; Weber, G. Pressure dissociationand conformational drift of the â dimer of tryptophan syn-thetase. Biochemistry 1986, 25, 5780-5786.

(23) Royer, C. A.; Weber, G.; Daly, T. J.; Matthews, K. S. Thedissociation of lactose repressor tetramer using hydrostaticpressure. Biochemistry 1986, 25, 8308-8315.

(24) Scarlata, S. F.; Roop, T.; Royer, C. A. The energetics ofhistone subunit interactions as studied by high pressure.Biochemistry 1989, 28, 6631-6637.

(25) Morel, G.; Yarmush, D. M.; Benjamin, D.; Colton, C. K.;Yarmush, M. L. Monoclonal antibodies to bovine serumalbumin: affinity and specificity determinations. Mol. Im-munol. 1988, 25, 7-15.

(26) Olson, W. C.; Spitznagel, T. M.; Yarmush, M. L. Dissocia-tion kinetics of antigen-antibody interactions: Studies on apanel of anti-albumin monoclonal antibodies. Mol. Immunol.1989, 26, 129-136.

(27) Antonsen, K. P.; Colton, C. K.; Yarmush, M. L. Elutionconditions and degradation mechanisms in long-term immu-noadsorbent use. Biotechnol. Prog. 1991, 7, 159-172.

(28) Sundaram, S. Effect of Pressure on Antigen-AntibodyComplexes. Ph.D. dissertation, Rutgers, The State Universityof New Jersey, 1997.

(29) Jue, R.; Lambert, J. M.; Pierce, L. R.; Traut, R. R. Additionof sulfydryl groups to Escherichia coli ribosomes by proteinmodification with 2-iminothiolane. Biochemistry 1978, 17,5399-5406.

(30) Paladini, A. A.; Weber, G. Absolute measurements offluorescence polarization. Rev. Sci. Instrum. 1981, 52, 419-427.

(31) Payne, V. A.; Matubayasi, N.; Murphy, L. R.; Levy, R. M.Monte Carlo study of the effect of pressure on hydrophobicassociation. J. Phys. Chem. B 1997, 101, 2054-2060.

(32) McDuffie, G. E., Jr.; Kelly, M. V. Effect of pressure on theviscosity and dielectric relaxation time in glycerol. J. Chem.Phys. 1964, 41, 2666-2670.

(33) Campbell, D. H.; Johnson, F. H. Pressure and specificprecipitation. J. Am. Chem. Soc. 1946, 68, 725.

(34) Tsuji, F. I.; Davis, D. L.; Donald, D. H. The effect ofhydrostatic pressure on the rate of inactivation of cypridinaluciferase by specific antibody. J. Immunol. 1966, 96, 614-621.

(35) Herron, J. N.; Kranz, D. M.; Jameson, D. M.; Voss, E. W.Thermodynamic properties of ligand binding by monoclonalanti-fluorescyl antibodies. Biochemistry 1986, 25, 4602-4609.

(36) Coelho-Sampaio, T.; Voss, E. W. Pressure-induced dissocia-tion of fluorescein from the anti-fluroescein single-chainantibody 4-4-20. Biochemistry 1993, 32, 10929-10935.

(37) Carrero, J.; Voss, E. W. Temperature and pH dependenceof fluorescein binding within the monoclonal antibody 9-40active site as monitored by hydrostatic pressure. J. Biol.Chem. 1996, 271, 5332-5337.

(38) Chryssomallis, G. S.; Torgerson, P. M.; Drickamer, H. G.;Weber, G. Effect of hydrostatic pressure on lysozyme andchymotripsinogen detected by fluorescence polarization. Bio-chemistry 1981, 20, 3955-3959.

(39) Nystrom, B.; Roots, J. Quasielastic light scattering studiesof the pressure-induced denaturation of lysozyme. Makromol.Chem. 1984, 185, 1441-1447.

(40) Heremans, K.; Wong, P. T. T. Pressure effects on theRaman spectra of proteins: Pressure-induced changes in theconformation of lysozyme in aqueous solutions. Chem. Phys.Lett. 1985, 118, 101-104.

(41) Kundrot, C. E.; Richards, F. M. Changes in the high-resolution structure of crystalline hen egg-white lysozymeproduced by a hydrostatic pressure of 1000 atm. In CurrentPerspectives in High-Pressure Biology; Jannasch, H. W.,Marquis, R. E., Zimmerman, A. M., Eds.; Academic Press:New York, 1987; pp 245-255.

(42) Paladini, A. A.; Silva, J. L.; Weber, G. Slab gel electro-phoresis at high hydrostatic pressure: A new approach forthe study of oligomeric proteins. Anal. Biochem. 1987, 161,358-364.

(43) Nystrom, B.; Roots, J. Study of pressure-induced denatur-ation of bovine serum albumin by photon-correlation spec-troscopy. Chem. Phys. Lett. 1982, 91, 236-240.

(44) Nystrom, B.; Roots, J. Dynamic light scattering studies ofprotein solutions under high pressure. J. Chem. Phys. 1983,78, 2833-2837.

(45) Howlett, J. R.; Ismail, A. A.; Armstrong, D. W.; Wong, P.T. T. Pressure-induced conformational changes in an antigenand an antibody and the implications for their use forhyperbaric immunoadsorption. Biochim. Biophys. Acta 1992,1159, 227-236.

(46) Yarmush, M. L.; Antonsen, K. P.; Sundaram, S.; Yarmush,D. M. Immunoadsorption: Stategies for antigen elution andproduction of reusable adsorbents. Biotechnol. Prog. 1992, 8,168-178.

(47) Suzuki, K.; Taniguchi, Y.; Watanabe, T. The effect ofpressure on the dimerization of carboxylic acids in aqueoussolution. J. Phys. Chem. 1973, 77, 1918-1922.

780 Biotechnol. Prog., 1998, Vol. 14, No. 5

(48) Hamann, S. D. The role of electrostriction in high-pressurechemistry. Rev. Phys. Chem. Jpn. 1980, 50, 147-168.

(49) Smith-Gill, S. J.; Wilson, A. C.; Potter, M.; Prager, E. M.;Feldmann, R. J.; Mainhart, C. R. Mapping the antigenicepitope for a monoclonal antibody against lysozyme. J.Immunol. 1982, 128, 314-322.

(50) Sheriff, S.; Silverton, E. W.; Padlan, E. A.; Cohen, G. H.;Smith-Gill, S. J.; Finzel, B. C.; Davies, D. R. Three-dimensional structure of an antigen-antibody complex. Proc.Natl. Acad. Sci. U.S.A. 1987, 84, 8075-8079.

(51) Smith-Gill, S. J.; Mainhart, C.; Lavoie, T. B.; Feldmann,R. J.; Drohan, W.; Brooks, B. R. A three-dimensional model

of an anti-lysozyme antibody. J. Mol. Biol. 1987, 194, 713-724.

(52) Padlan, E. A.; Silverton, E. W.; Sheriff, S.; Cohen, G. H.;Smith-Gill, S. J.; Davies, D. R. Structure of an antibody-antigen complex: Crystal structure of the HyHEL-10 Fab-lysozyme complex. Proc. Natl. Acad. Sci. U.S.A. 1989, 86,5938-5942.

Accepted July 20, 1998.

BP980066M

Biotechnol. Prog., 1998, Vol. 14, No. 5 781