Dissociation Equilibrium of HumanRecombinant Interferon 'Y

Raina Boteva, Theodora Ziateva, Victoria Dorovska-Taran,Antonie J. W. G. Visser, Roumen Tsanev, and Benedetto SalvatoInstitute of Molecular Biology and Institute of Organic Chemistry,

Bulgarian Academy of Sciences, Sofia 1113, Bulgaria,CNR Centre of Metalloproteins, Department of Biology,

University of Padua, 35121 Padua, Italy,and Department of Biochemistry, Agricultural University, Dreijenlaan 3,

NL-6703 HA Wageningen, The Netherlands

___ • __ • T ~~ __ ~-_-.~~-.-~-_::_- , _

Biochenrlstry®Reprinted from

Volume 35, Number 47, Pages 14825-14830

Biochemistry 1996, 35, 14825-14830 14825

Dissociation Equilibrium of Human Recombinant Interferon ytRaina Boteva, *.t Theodora Zlateva.! Victoria Dorovska-Taranj+ Antonie J. W. G. Visser," Roumen Tsanev.t and

Benedetto Salvato!

Institute of Molecular Biology and Institute of Organic Chemistry, Bulgarian Academy of Sciences, Sofia 1113,Bulgaria, CNR Centre of Metalloproteins, Department of Biology, University of Padua, 35121 Padua, Italy, and Department of

Biochemistry, Agricultural University, Dreijenlaan 3, NL-6703 HA Wageningen, The Netherlands

Received November 21, 1995; Revised Manuscript Received June 27, 199(j®

ABSTRAer: The biologically active form of interferon y is a dimer composed of two noncovalently boundidentical polypeptide chains of 17 kDa each. In this study, it was found that dissociation of the dimerinto monomers significantly reduced the fluorescence quantum yield and the efficiency of the intermolecularTyr to Trp radiationless energy transfer. The same process caused significant changes in the fluorescencedecay and in the fluorescence anisotropy decay. The kinetic and thermodynamic parameters of the dimer-monomer equilibrium were determined by fluorescence measurements at different temperatures and by atheoretical mathematical model. Dissociation of the dimers into monomers was an endothermic processand was favored by concentrations of the protein lower than 1¡.tM and by increasing the temperature. Itwas accompanied by formation of aggregates, a slow and partially reversible process leading to inactivationof the interferon. It is suggested that certain monomeric conformers are competent for aggregation.

Interferon y (lFN-r)l is a cytokine produced in minutequantities by some activated T lymphocytes. Due to itsvarious activities-antiviral, antiproliferative, immunoregu-latory, and modulation of gene expression-this protein isused for the treatment of different diseases. Unfortunately,storage of unfrozen highly purified IFN-y preparations evenat low temperature and in the presence of stabilizing agentsis accompanied by a gradual decrease of their antiviralactivity (pestka et al., 1987). Possible reasons for thisspontaneons inactivation could be either slow irreversiblechanges in the active conformation of the molecule and/orprocesses of aggregation.

The X-ray crystal structures of recombinant human andrabbit IFN-y have shown that they are a. helical proteinscontaining six helices and no ß sheet structure. Their activeform is a dimer composed of two identical polypeptidechains, tightly associated by numerous íntemelical interac-tions (Balick et al., 1992; Samudzi et al., 1991). Sedimenta-tion studies on human IFN-y equilibria have registereddimers and oligomers but no monomers (Yphantis &Arakawa, 1987). Refolding of the protein denatured by urea,acid, or guanidine hydrochloride was found to yield twodifferent forms: dime ric (active) and aggregated with a 4-8-fold-reduced activity (Arakawa et al., 1985; Arakawa & Hsu,1987). Aggregation appears to be a complex process,

t This work was supported by a grant from the Bulgarian NationalCommittee for Science, partially supported by a grant from PaduaUniversity to T.z., and supported by a fellowship of the WageningenAgricultural University to V.D.-T.

• Corresponding author. Telephone: (359-2) 713-2641. Fax: (359-2) 72 35 07.* Institute of Molecular Biology, Bulgarian Academy of Sciences.

f University of Padua."Agricultural University..L Institute of Organic Chemistry, Bulgarian Academy of Sciences.e Abstract published in Advance ACS Abstracts, October I, 1996.1Abbreviations: hrIFN-y, human recombinant interferon Yi HPLC,

high-pressure liquid chromatography: MEM, maximal entropy method:PAGE, polyacrylamide gel electrophoresis; N-Ac-Trp-~. N-acetyl-L-tryptophanamide; VSV. vesicular stomatitis virus.

probably involving various conformational intermediates anddifferent phases. It is initiated during a lag phase whichcould last from a few minutes up to many days dependingon the temperature (Mulkerrin & Wetzel, 1989). The dimersand aggregates have similar secondary but different tertiarystructures in the local microenvironment of the aromaticamino acid residues as shown by UV circular dichroic spectra(lIsu & Arakawa, 1985).There is no doubt that the stability of IFN-y dimers should

play a central role in maintaining the biological activity ofthe protein. However, extensive physicochemical studies onthe dissociation of the dimer are lacking to date. In thepresent paper, we have studied and characterized the dimer--monomer equilibrium of recombinant human IFN-y by bothstatic and dynamic fluorescence measurements. We havefollowed the effects of temperature, protein concentration,and time on the intrinsic protein fluorescence and haveregistered conformational changes which parallel the forma-tion of different forms: dimers, monomers, and aggregates.

MATERIALS AND METHODS

Human recombinant inteiferon y (hrIFN-y) was a productof Pharmagen Ltd., Sofia It does not contain the leadersequence Cys- Tyr-Cys or the N-terminal methionine and hasa purity of 99.5%, as judged by HPLC and gel electrophore-sis. Its specific activity is about 5 x 107 lU/mg asdetermined by using the 50% protection value of interferondilution against the cytopathic effect of VSV on WISH cells(Rubinstein et al., 1981).Protein concentration was determined by the procedure

of Lowry (1951) and by using a molar extinction coefficientat 280 um of 1.1 x 10"M-l cm"! calculated on the basis ofthe aromatic amino acid content (Beaven & Holiday, 1952).

Fluorescence was measured with a Perkin-Elmer modelMPF-43 spectrofluorometer equipped with a thermostaticallycontrolled cuvette holder. Excitation was at either 270 or295 um. The relative Trp quantum yield was calculated bycomparing the integrated emission spectrum of the protein

SOOO6~2960(95)02759-0 CCC: $12.00 © 1996 American Chemical Society

14826 Biochemistry, Vol. 35, No. 47, 1996

300 340 380 420Wavelength ( nm )

FIGURE 1: Fluorescence spectra of hrIFN-y in 10 mM Tris-HCl atpH 7.7 excited at 270 run (a and e) and at 295 run (b and t). Spectrae and f are obtained after incubation of the protein solution at 50oc for 90 min. The pairs a and b and e and f are normalized toabsorbancy at 380 run. The difference spectra c (between a and b)and d (between e and t) express the contribution of Tyr residues.Emission spectra a+c characterize predominantly the dimeric formsand d-f the monomeric forms.

excited at 295 nm with that of the standard N-Ac-Trp-NH2normalized for the absorbancy at 295 nm. Protein solutionswith a concentration higher than 5 ,uM were used for thisexperiment. Measurements were performed immediatelyafter dilution of stock protein solutions (15-30 ,uM) at 20"C, A value of 0.13 was used for the quantum yield of thestandard (Lehrer, 1971). The efficiency of the Tyr to Trpenergy transfer was estimated as described by Eisinger(1969).

Time-resolved fluorescence and fluorescence anisotropydecays were measured by using the time-correlated single-photon counting technique (Dorovska- Taran et al., 1994; Papet al., 1993). The measuring system was able to resolveaccurately fluorescence lifetimes of 30 ps. The excitationand the emission wavelengths were 300 and 348. nm,respectively. The experiments were carried out at twotemperatures: 20 and 50 "C, The samples incubated for 90min at 50 "C were measured at this temperature and at 20"C after cooling. The lifetime and correlation time distribu-tions were determined from the total fluorescence decay andthe anisotropy decay by the MEM program (MaximalEntropy Data Consultants Ltd., Cambridge, U.K.). Detailsare given by Livesey and Brochon (1987).Native polyacrylamide gel electrophoresis in a homoge-

neous medium was performed with the Pharmacia FastSystem at 15 "C and with a 40 min running time. Fast GelMedia (Homogeneous 20) with stacking gel zones T 7.5%,C 3% and separating gel zones T 20.0%, C 2% were used.The buffer system in the gel was 0.112 M sodium acetateand 0.112 M Tris-Ha at pH 6.5. Fast gel native buffer was0.88 M r-alanine and 0.25 Tris-Ha at pH 8.8. Proteinsamples with a concentration of about 1,uM were analyzed.The protein bands were visualised by silver staining.

RESULTS

Fluorescence Properties of Human Recombinant Inter-feron y. Fluorescence emission spectra of predominantlydimeric hrIFN-y are shown in Figure 1 (curves a-c). Whenexcited either at 270 nm (where both Tyr and Trp residuesabsorb) or at 295 nm (where predominantly Trp absorbs),the spectra are dominated by Trp emission (Figure 1, curves

Boteva et al.

1.0

0.9

e- 0.6

oEt........¡ 0.7

260 290 300270 280

À (nm)

FIGURE 2: Wavelength dependence of the relative quantum yield4>(Ä)/4>'li of l!-lM hrIFN-y (10 mM Tris-HCI at pH 7.7) measuredafter incJbation of the protein for 90 min at 4 "C (.), at 30 "C(O), and at 47 "C (.A.). The solid lines are theoretical curves obtainedaccording to Eisinger (1969) for different values of the energytransfer efficiency e. All theoretical and experimental data arenormalized to absorbancy at 295 nm.

a and b, respectively). The position of the spectral maximumis at 343 nm, and a relative Trp emission quantum yield of0.042 was calculated. By differential computation of theemission spectra excited at 270 and 295 nm, a weakcontribution of Tyr residues (less than 10% of the overallprotein fluorescence) could be detected (Figure I, curve c).

The efficiency of the radiationless energy transfer fromTyr to Trp was calculated from the corresponding fluores-cence excitation spectra. The best fit of the experimentaldata to the theoretical curves calculated according to Eisinger(1969) was obtained for a 62% efficiency of the transfer(Figure 2). Thus. about 62% of the excitation energyabsorbed by the Tyr residues is transferred via dipole-dipoleinteractions to the Trp chromophore.

Dependence of the Trp Emission on the Concentration ofhrIFN-y. Dilution of the stock interferon solutions led to asignificant reduction of the Trp emission, which was notproportional to the decrease of the protein concentration. Thiseffect became pronounced at concentrations lower than 1¡,lM. The fluorescence measured at different dilutions wasnormalized to 1 p.M IFN-y and plotted against the proteinconcentration. As seen in Figure 3, the experimental pointsbest fit a theoretical curve of a dimer-monomer equilibriumderived by our theoretical model (see Appendix) assumingan equilibrium constant of 3.8 x 10-7 M and a 50% lowerquantum yield of the monomer as compared to that of thedimer.

The results from PAGE experiments showed that even at4 "C a band corresponding to the monomer appears in thegel together with that of the dimer (Figure 5~): !hisconfirms the existence of a dimer-monomer equilibnum.

Kinetic and Thermodynamic Parameters of hrIFN-y Dis-sociation. Incubation of 1 ¡,lM hrIFN-y at three differenttemperatures (13, 30, and 47 "C) led to a decrease of the

. --._-_._------.. ---'(-.--

..: .

. Dissociation Equilibrium of Interferon 'Y

~ 100 r----------------,:i.!!. 90Ec 60Il)'<tCO)- 70cuIIIUc 60fiVII!! 50O:::Iii:

1.0 2.0

( !FN J 11MFIGURE 3: Fluorescence maximal intensity at 345 nm, measuredafter 24 h of incubation of !FN-r solutions of different concentra-tions at 13 "C, Excitation was at 295 nm, and the data werenormalized to a 1 I'M protein concentration. The solid line is atheoreticalcurve of a monomer-dimer dissociationcalculated withan equilibrium constant of 3.8 x 10-7M-l.

3.0 4.0

100

GI

:a 80~o.,"O

8c 608!!o::lii: •40

20 40 60 80 100 120

lime (min)FIGURE 4: Fluorescence intensity decrease at 345 run of a 1 I'M!FN-y solution incubated at 13 "C (O), at 30 "C (A), and at 47 oc(.). The solid lines are theoretical curves determined by solvingthe equations in the Appendix by means of a computer program.They fit well the equilibrium constants in Table L

fluorescence intensity. Its value reached a final plateaudepending on the temperature (Figure 4). The solid linesdescribe the dimer-monomer equilibrium of IFN-l' accord-ing to the theoretical kinetic model given in the Appendix.They were obtained by selecting the dissociation andassociation rate constants kl and "-l to best fit the experi-mental data. Their values and that of the equilibriumconstants Kc for the corresponding temperatures are givenin Table 1. The equilibrium constant at 47 "C is about 2-and 6.6-fold higher than those at 30 and 13 "C, respectively.This was due to the fact that the higher temperature mainlyincreased dissociation rate constants (fable 1).

The thermodynamic parameters of the dissociation werecalculated from the equilibrium constants in Table 1. Thelatter were derived from the fluorescence decrease takingplace during a period of 20-25 min (Figure 4) which isshorter than the time required for a detectable aggregationto start. The free energy change (åG) was 8.2 kcal/mol,the enthalpy change (Mf) 10.1 kcal/mol, and the entropychange (åS) on the order of 6 cu. The activation energy(Erø) had a value of 12.0 kcal/mol.

-"'-~'-~, ,.,_.-" "'-'1------:"_·,'--'-. r, 41,.~ '.. '

Biochemistry, Vol. 35, No. 47, 1996 14827

....s...._b

.... m

FIOURE 5: Native PAGelectrophoresis of I I'M hrlFN-r analyzedimmediately after preparation (A) or after 60 min of incubation at47 "C (B): s, start; b, boundary between stacking and separatinggels; d, dimers; and ro, monomers. Aggregatesare seen on the startin part B.

Table 1: SpecificActivityand Kinetic Parametersof hrIFN-yDissociation at Different Temperatures"T (K) Ke (M) kl (S-l) k-l (M-l çl) activity(lU/mg)286 3.8 x 10-7 3.8 X 10--4 1.0 x l()3 1.7 x J01303 1.4 x 10-6 1.3 X 10-3 9.5 x 1()2320 2.5 x 10-6 3.5 X 10-3 1.4 X 103 0.4 X 107

" 14 is the equilibrium constant,and kl and k-l are the dissociationand association rate constants, respectively, calculated from thefluorescencedata (Figure 4) and from the theoreticalkinetic model(see Appendix)by a computerprogram. Standard errors for kineticand activitymeasurementswere about 5%.

Fluorescence Decay and Fluorescence Anisotropy Decay.The polarized fluorescence decays were measured at 20 and50 "C and then back at 20 "C, Figure 6 contains a summaryof fluorescence and anisotropy decay curves and their MEMreconstructions into lifetime distribution [a.('r) versus 'f] andcorrelation time distribution [ß(l/J) versus Ø]. All relevantparameter values are presented in Table 2. Although theprotein monomer contains a single Trp, the fluorescencelifetime distribution consisted of two major components. Inthe two samples (untreated and incubated at 50 "C), bothmeasured at 20 "C, long-lived components with barycentersat 3.1 and 3.3 ns (after heating) were predominant At 50"C, this major lifetime decreased to 2.0 ns and shorterlifetimes of 0.57 and of 0.11 ns appeared (Figure 6 and Table2).

The fluorescence anisotropy decay at 20 "C was com-pletely dominated by a rotational correlation time with thebarycenter at 25 ns (fable 2). It arose from the tumbling ofhrIFN-l' dimers and possibly tetramers as suggested by thebroadness of the distribution. Incubation of the protein at50 "C for 90 min caused a dramatic change in the anisotropydecay pattern (Figure 6). The process was characterized bya rapid decay superimposed on a constant anisotropy whichwas not decaying on the 30 ns experimental time scale. Theshort correlation time was centred at 3.2 ns. The long decaycould not be resolved and was arbitrarily assigned to 100 ns

• 00':"

14828 Biochemistry, Vol. 35, No. 47, 1996

,-...<I)

§ 104

o~ 103CI,)U

ß102U

<I)

~o101::s

li::10°

O

0.4

0.3>-Po.g00.2<I)

~0.1

0.0

O

5 10 15 20Time (ns)

25

5 10Time (ns)

15

Boteva et al.

1.0

0.8

,-... 0.6~ö 0.4

0.2

30 0.1 11: (ns)

10

1.0

0.8-273K- 323K.•... 273 K (back)

,-... 0.6 .~co..

0.4

0.2

20 10010ej> (ns)

FIGURE 6: (Top) Experimental and fitted fluorescence decays (left) of hrIFN-r under different conditions: curve l, at 293 K; curve 2, at323 K; curve 3, at 293 K after heating at 50 oc for 90 min. The MEM reconstructions are shown on the right Barycenters (r) and relativeamplitudes (a) are given in Table 2. (Bottom) Experimental anisotropy decay curves constructed from the individual polarized decay,curves for hrIFN-r at 293 and 323 K. The third anisotropy decay curve did not differ from the one at 323 K and is not presented. The MEMreconstructions are shown on the right Barycenters (¢) and relative amplitudes (fJ)are preseted in Table 2.

Table 2: Fluorescence Lifetimes and Rotational Correlation Timesof Tryptophan in hrIFN-y"

T 'rI 'r2 1'3 ,pl t/J2(K) al (ns) 0.2 (ns) 0.3 (ns) ßl (ns) ßz (ns) iH293 0.10 0.11 0.90 3.1 0.006 1.6 0.246 2S 0.99323 0.15 0.11 0.69 2.0 0.16 0.57 0.096 3.2 0.147 100 1.15293c 0.15 0.15 0.83 3.3 0.02 0.15 0.071 5.4 0.190 100 1.02

" Obtained from the main contributions of the MEM of fluorescencedecay [a(r) versus -r] and fluorescence anisotropy decay analysis rJj(¢)versus ¢]. b The reduced X2 indicates the quality of the flt. cAfterheating for 90 min at 50 "C.

(Table 2). The short correlation time was probably arisingfrom monomers and confirmed the conclusion that the highertemperature favored the dissociation of the dimers. Afterthe sample was cooled to 20 "C, a correlation time of 5.4 nswas obtained for the monomer. This value is close to thetheoretical one for a hydrated, spherical protein of 17 kDawhich is the molecular mass of the IFN-y monomer. Theincrease from 3.2 to 5.4 ns is in keeping with the dependenceof the rotational correlation time on 'I/T (t¡=viscosity, T =absolute temperature).

A nonresolved very long correlation time of hrIFN-" waspredominant both in the heated and in the subsequentlycooled samples (Figure 6). The appearance of this compo-nent strongly suggests the formation of aggregates uponheating which persist after cooling to 20 "C, These datawere confirmed by the results from the PAGE performedafter incubation of a I¡,tM interferon solution at 47 "C for60 min. Two bands corresponding to protein monomers andaggregates were observed on the gel (Figure 5B).

Biological Activity of hr/FN-". hrIFN-y (1¡,tM) incubatedfor 90 min at 13 oC had a specific antiviral activity of 1.7 x

••• - ~.. ¡. •, ,".' 'I·". "

107 lU/mg. This activity dropped to 4 x 1()6 lU/mg afterincubation of the protein at 47 "C for 90 min; i.e. the heatingled to a 4-fold activity reduction (fable 1).

DISCUSSION

The biologically active form of IFN-y is a homodimer.Each subunit contains one TIp at position 36 and four Tyrresidues. The unique TIp is highly conserved in all inter-ferons" sequenced until now. According to the crystal-lographic data, it is located at the end of helix B in a burriedhydrophobic region formed by Phe 29, Leu 30, TIe 32, andLeu 34 of the same helix (Ealick et al., 1991).

As shown in this study, the fluorescence spectrum ofhrIFN-" is dominated by the emission of the TIp residuewith a maximum at 343 nm, This emission position is typicalof a chromophore partially exposed to the external medium.A significant part of the TIp fluorescence was found to bedue to energy transfer from Tyr residues. Evaluation of theefficiency of this process shows that 62% of the excitationenergy absorbed by Tyr residues is transferred via radia-tionless dipole-dipole interactions to the TIp chromophore(Figure 2). By using the crystallographic coordinates of theC-a. atoms (Ealick et al., 1991), distances ranging from 18.5to 30.1 Å were calculated between the single TIp and thefour Tyr residues which theoretically could be involved inthe energy transfer within the same domain of the dimer.For the intersubunit energy transfer, the most probablecandidate seems to be Tyr 98 (helix E) of the other subunitwhich is the Tyr residue nearest to TIp 36 (18.5 Å calculateddistance). For the intrasubunit interactions, Tyr 53 (helixC) of the same subunit is the chromophore nearest to Trp36 (calculated distance is 19.2 Å). However, it should be

Dissociation Equilibrium of Interferon y

pointed out that an efficient energy transfer requires not onlya short distance but also a proper mutual orientation of theindole and phenol rings.

Sedimentation studies on !FN-y equilibrium have led tothe conclusion that human recombinant IFN-y is a stabledimer and does not dissociate into monomers at neutral pH(Yphantis & Arakawa, 1987). The crystallographic data haveshown that the dimer is stabilized by multiple intersubunitinteractions. This has suggested that the separation of thetwo subunits would be difficult without a significant disrup-tion of their tertiary structure (Ealiek et al., 1991).

However, our data show that the dimer exists in a dynamicequilibrium with the monomer. The dimers and the mono-mers possess different fluorescence properties which alloweddetailed studies on the kinetics of the dissociation process.Accumulation of monomers was associated with a significantdecrease (about 50%) of the Trp emission (Figure 1, comparecurves b and t). However, the position of the emissionmaximum remained almost constant (343-344 urn), indicat-ing that no dramatic conformational changes in the microen-vironment of the single Trp chromophore took place afterdissociation of the dimer. This fact should be of criticalimportance for the reversibility of the association-dissocia-tion process.

Dissociation of the dimer into monomers abolishes theenergy transfer between the Trp chromophore and Tyrresidues belonging to different subunits of the dimer .. Whilethe efficiency of the energy transfer in the dimer was 62%,the efficiency of the same process in the monomer wasreduced to 20-25%. This residual efficiency was due tosome intrasubunit energy transfer (Figure 2). As expected,this was accompanied by an enhancement of the fractionalcontribution of the Tyr residues to the overall emission (fromabout 10% in the dimer to about 30% in the monomer, Figure1).The fluorescence decay of the single Trp in hr!FN-y is

heterogeneous. The heterogeneity increases when the proteinis heated and subsequently cooled. If it is assumed thatdimers and tetramers have a comparable fluorescence lifetimepattern, then this finding is in agreement with the conceptthat the other forms, monomers and aggregates, havesignificantly different lifetimes. These results show that the50% decrease in the Trp emission upon dissociation of thedimer is due not only to a reduction of the Tyr to Trp energytransfer but also to changes in the fluorescence lifetimedistribution of excited monomeric species with somewhatshorter lifetimes.

The value calculated for the free energy changes of thedimer dissociation is positive (8.2 kcal/mol). This meansthat the dimer is the energetically favored species. Thedissociation reaction is endothermic (enthalpy changes of10.1 kca1Imol) and is mainly driven by the entropy changesof the system (an increase of about 6 eu). The relativelyhigh value of the activation energy of the dissociation process(EIICt = 12.0 kcal/mol) is in agreement with the fact that it isfavored by the dilution of the protein solution and by anincrease of the temperature. The association rate constantswere practically not affected by the temperature within thetemperature range studied.

Assuming a value of 0.735 cm3/g for the partial specificvolume and of 0.2 cm3/g for the degree of hydration at 20oC (Vos et al., 1987), a correlation time of 13 ns could betheoretically predicted for a rigid sphere with a 35 kDa

Biochemistry, Vol. 35, No. 47, 1996 14H2Y

molecular mass, corresponding to that of an !FN-y dimer.From MEM analysis of the anisotropy decay at 20 oC, abroad distribution was recovered with the barycenter at 25ns (Table 2). This is about 2 times higher than the predictedcorrelation time for the dimer. Even if hrIFN-y were not asphere but a flattened ellipsoid with dimensions of about 60x 40 x 30 Å as suggested by the crystallographic model(Ealick et al., 1991) and if a somewhat higher degree ofhydration were assumed, the 25 ns correlation time couldnot be reconciled with only dimeric species. Given thebroadness of the correlation time distribution, it is moreprobable that a mixture of dimers and tetramers prevails,similarly as found previously for interferon <l2 (Vincent etal., 1992). Dissociation of dimers into monomers and thepresence of aggregates is seen when hrIFN-y is heated at50 "C for 90 min. This process seems to be irreversible; nocorrelation time characteristic of dimeric species was found,and no band of the dimer in PAGE was observed aftercooling (Figure 5B). In fact, aggregation caused a significantreduction (about 75%) of the antiviral activity of !FN-y.

In agreement with the data of others (Mulkerrin & Wetzel,1989), our observations also show that aggregation is arelatively slow process requiring a population of certainmonomeric forms competent for aggregation. Thus, duringthe first 20 min, it does not interfere with the dimer-monomer equilibrium. This explains the very good fittingof the experimental data to a simple monomer-dimerequilibrium model.

Incubation of the protein for 90 min at 50 "C does notseem to thermally denature the sample for the followingreasons: (1) a significant antiviral activity is still preservedand (2) the fluorescence parameters of the heated proteinare characteristic of a native state. It should be noted thatthe emission intensity of the thermally denatured protein isreduced by about 30% and the position of its emissionmaximum is blue-shifted by 11 urn (spectra not shown).Studies on IFN-y aggregation by other authors (Mulkerrin&Wetzel, 1989) have also shown that thermal denaturationof this protein occurs above 50 "C,

Our data suggest a central role of the dynamic conforma-tional distribution of the monomers for the aggregationprocess, and we propose the following model for the IFN-yequilibrium:

D <=> Mt<:::> •.. <=> MI <:::> ..• <=>Mn

~tA

where D is the dimer, Mi represents monomers in differentconformational states (conformers), and A represents ag-gregates. Probably, some monomeric conformers which aremore populated at higher temperatures are competent foraggregation. The aggregation of the monomers could beexplained by the effect of heating on the <l helical structureswhich before full denaturation may expose hydrophobiesurfaces, leading to irregular aggregation.

Human IFN-y possesses a very high specific activity, andits biological effect in vivo shows a concentration optimum(Kleinerman et al., 1986). Thus, the physiological meaningof the monomer-dimer equilibrium (with an inactivemonomer) is probably the role of preventing unwanted effectsof too high local concentrations before the interferonmolecules reach the target cells. In the latter case, the

14830 Biochemistry, Vol. 35, No. 47, 1996

formation of a dimer+receptor complex can allow theselection of dimers even if the interferon concentration isvery low, due to the high-affinity constant of this complex.Aggregates probably do not play an important role for thesystem in vivo being unfavored at physiological temperatureand concentrations.

ACKNOWLEDGMENTThe authors are very grateful to Dr. R. Mironova for

providing highly purified hrIFN-y. We thank D. Givkova,A. Vanthoch, and E. Dainese for technical assistance andProf. M. Beltramini and Dr. F. Ricchelli for their interest.

APPENDIXThe dynamic equilibrium of the dissociation of the dimer

A2 to monomers A

according to the law of mass action is described by thefollowing equations:

klA2 = k_IA2

s,= k¡lk_l

(1)

(2)

(3)

whereAT is the totaI protein concentration as monomers andK¿ is the equilibrium constant.

The changes of the dimer and of the monomer concentra-tions could be followed by the differential equations express-ing the corresponding reaction rates:

dA/dt = -k}A2 + k_IA2

dA/dt = klA2 - k_lA2

(4)

(5)Substituting Az and A in eqs 4 and 5 from eq 3, we

obtain:

dA/dt =4k_IA/ - (kl +4k_IAT)A2 + k_IAT2 (6)

dA/dt = -k_IA2 - k'¡A + k'IAT (7)

where Ttl = kll2.Taking into account that (Tt1)2« 4k-1Tt tATand that K¿ is

on the order of 10-8, as indicated by sedimentation experi-ments (Yphantis & Arakawa, 1987), the integration of eq 7

Boteva et al.

from O to A gives the following expression for the kineticsof monomer formation:

(8)

where q' = -k-¡kIAT.Equation 8 allows us to calculate the concentration A of

the monomer as a function of time. The parameters K¿ andq' have been computed by fitting the theoretical function(eq 8) to experimental data (Figure 4). A is related to theregistered fluorescence F as follows:

(9)

or, if we introduce the total concentration AT, eq 9 becomes:

(10)

where ¢Z and Øl are the quantum yields of the dimer and ofthe monomer, respectively.

REFERENCES

Anderson, P., Yip, Y. K., & Vileek, J. (1982) J. Biol. Chern. 257,11301-11304.

Beaven, G. H., & Holiday, E. R. (1952) Adv. Protein Chern. 7,319-325.

Dorovska-Taran, V., van Hoek, A, Link, T. A., Visser, A J. W.G., & Hagen, W. R (1994) FEBS Lett. 348, 305-310.

Ealick, S. E., Cooc, W. J., Vijay-Kumar, S., Carson, M., Naga-bhushan, T. L., Trotta, P. P., & Bugg, C. E. (1991) Science 252,698-702.

Eisinger, G. (1969) Biochemistry 8, 3902-3907.Hsll, Y.-R., & Arakawa, T. (1985) Biochemistry 24, 7959-7%3.Kleinerman, E. S., Kurrock, R., Wyatt, D., Quwesada, R., Gutter-

man, U., & Fidler, L J. (1986) Cancer Res. 46, 5401-5405.Lehrer, S. S. (1971) Biochemistry 10, 3254-3263.Livesey, A. K., & Brochon,J. C. (1987) Biophys. J. 52,693-706.Lowry, O. H., Rosebrough, N. L., Farr, A I., & Randall, B. J.

(1951) J. Bioi. Chern. 193, 265-275.MuIkerrin, M. G., & Wetzel, R. (1989) Biochemistry 28,6556-

6561.Pap, E. H. W., Bastiaens, P. I. H., Borst, J. W., van den Berg, P.

A W., van Hoek, A, Snoek, G. T., Wirtz. K. w. A., & Visser,A. J. W. G. (1993) Biochemistry 32, 13310-13317.

Pestka, S., Langer, J. A., Zoon, K. C., & Sammel, C. E. (1987)Annu. Rev. Biochern. 56, 727-777.

Rubinstein, S., Familletti., P. C., & Pestka, S. (1981) J. Virol. 37,755-759.

Vincent,M., Li De La Sierra, I. M., Berberan-Santos, M. N., Diaz,A, Díaz, M., Padron, G., & Gallay, J. (1992) Eur. J. Biochem.210, 953-961.

Vos, K., van Hoek, A., & Visser, A. J. W. G. (1987) Eur. J.Biochem. 165,55-63.

Ypbantis, D. A, & Arakawa, T. (1987) Biochmústry 26, 5422-5427.

B19527597