Unfolding and aggregation during the thermal denaturationof streptokinase

Ana I. Azuaga1, Christopher M. Dobson2, Pedro L. Mateo1 and Francisco Conejero-Lara1

1Departamento de Quımica Fısica e Instituto de Biotecnologıa, Facultad de Ciencias, Universidad de Granada, Granada, Spain;2Oxford Centre for Molecular Sciences and New Chemistry Laboratory, University of Oxford, UK

The thermal denaturation of streptokinase from Strepto-coccus equisimilis (SK) together with that of a set of frag-ments encompassing each of its three domains has beeninvestigated using differential scanning calorimetry (DSC).Analysis of the effects of pH, sample concentration andheating rates on the DSC thermograms has allowed us tofind conditions where thermal unfolding occurs unequivo-cally under equilibrium.Under these conditions, pH 7.0 anda sample concentration of less than � 1.5 mgÆmL)1, orpH 8.0, the heat capacity curves of intact SK can be quan-titatively described by three independent two-state transi-tions, each of which compares well with the two-statetransition observed for the corresponding isolated SKdomain. The results indicate that each structural domain ofSK behaves as a single cooperative unfolding unit underequilibrium conditions. At pH 7.0 and high sample con-centration, or at pH 6.0 at any concentration investigated,the thermal unfolding of domain Awas accompanied by the

time-dependent formation of aggregates of SK. Thisproduces a severe deformation of the DSC curves, whichbecome concentration dependent and kinetically controlled,and thus precludes their proper analysis by standarddeconvolution methods. A simple model involving time-dependent, high-order aggregation may account for theobserved effects. Limited-proteolysis experiments suggestthat in the aggregates the N-terminal segment 1–63 and thewhole of SKdomain Care at least partially structured,whiledomain B is highly unstructured. Unfolding of domain A,under conditions where the N-terminal segment 1–63 has ahigh propensity for b sheet structure and a partially formedhydrophobic core, gives rise to rapid aggregation. It is likelythat this region is able to act as a nucleus for the aggregationof the full-length protein.

Keywords: protein unfolding; protein aggregation; differen-tial scanning calorimetry; streptokinase; domains.

Streptokinase (SK) is a bacterial exoprotein from Strepto-coccus equisimilis consisting of a single chain of 414 aminoacid residues [1]. SK and human plasminogen form anequimolar high-affinity complex that directly catalyzes theproteolytic conversion of plasminogen to plasmin [2]. Thedomain organization of SK has been delineated previouslyby a combination of limited proteolysis studies andbiophysical methods [3,4] and confirmed later in the crystalstructure of the complex between SK and the catalyticdomain of plasmin, also known as microplasmin [5]. SKconsists of three well-defined domains (A, B and C)consecutive in the sequence, and an unstructured tail atthe C-terminus [3,5]. The three domains are folded similarlyand the crystal structure shows few contacts between them

[5], consistent with the high flexibility of the isolated proteinin solution [6]. SK domains play diverse and complementaryroles in SK–plasminogen complex formation, in thegeneration of the proteolytic active site in the plasminogenmoiety and in substrate plasminogen docking and process-ing by the activator complex [3,7–12].A variety of techniques, including DSC, CD and NMR,

have been used previously to investigate the thermalunfolding and stability of intact SK and a number offragments prepared either by limited proteolysis or recom-binant methods [4,13–20]. The unfolding profiles of intactSK have been interpreted in the literature as consisting ofone, two, three or even four independent transitions,depending on the experimental conditions and on thetechnique used. These results have led to significantdiscrepancies between different studies in the number ofunfolding units present in the SK structure. Furthermore,under some experimental conditions the correspondencebetween the number of structural domains (three) and thenumber of unfolding transitions observed (up to four)remains unclear.The aim of this work was to obtain new evidence that

could serve to shed light on the interpretation of thethermal transitions of SK and their correspondence with itsstructural domains. We have investigated the thermaldenaturation of SK and a set of fragments correspondingto isolated domains using DSC at several pH, scan rateand sample concentration values. The thermal denatura-tion profiles are reinterpreted in the light of new evidenceobtained in the present work together with the results of

Correspondence to F. Conejero-Lara, Departamento de Quımica

Fısica e Instituto de Biotecnologıa, Facultad de Ciencias,

Universidad de Granada, Granada, 18071 Spain.

Fax: + 34 958272879, Tel.: +34 958242371,

E-mail: conejero@ugr.es

Abbreviations: SK, Streptococcus equisimilis streptokinase; SKA,

recombinant SK fragment of sequence 1–146 plus an N-terminal

methionine; SKA1, SK fragment of sequence 1–63; SKB, SK fragment

of sequence 147–287; SKC, SK fragment of sequence 288–380; SKBC,

SK fragment of sequence 147–380; DSC, differential scanning calori-

metry; ESI-MS, electrospray ionization mass spectrometry; ANS,

8-anilino-1-naphthalenesulfonic acid.

(Received 21 January 2002, revised 14 June 2002,

accepted 11 July 2002)

Eur. J. Biochem. 269, 4121–4133 (2002) � FEBS 2002 doi:10.1046/j.1432-1033.2002.03107.x

previous studies. We demonstrate that under certainexperimental conditions, where thermodynamic equilib-rium is unequivocally established within the wholetemperature range of the DSC experiments, the unfoldingprofiles of SK are quantitatively described by threeindependent two-state transitions. In contrast, under otherconditions of pH and moderate-to-high sample concentra-tions, time-dependent, transient protein aggregation occursduring the thermal denaturation of intact SK and of theisolated A domain. The presence of these aggregationprocesses has a profound effect on the DSC curves andprecludes their analysis by standard equilibrium deconvo-lution methods. The results presented here on the thermaldenaturation of SK and its domains help to clarifyinconsistencies existing in previous reports concerning thenumber of cooperative folding units in this multidomainprotein. We have also carried out a preliminary character-ization of the thermally induced aggregation of SK using avariety of techniques. The results provide us with some ofthe properties of these high molecular mass aggregates andhelp to delimit the regions of the SK sequence responsiblefor aggregation.

M A T E R I A L S A N D M E T H O D S

Protein sample preparation

Purified streptokinase from culture filtrates of S. equisim-ilis was supplied by SmithKline Beecham Pharmaceuticals(Gronau, Germany). The protein purity (assessed by SDS/PAGE) was greater than 95%. SK fragments corres-ponding to the sequences 1–63 (SKA1), 147–287 (SKB),288–380 (SKC) and 147–380 (SKBC) were obtained byproteolytic cleavage of the intact protein and purified tohomogeneity as described elsewhere [3]. The recombinantMet-SK1-146 (SKA) fragment corresponding to SKdomain A was cloned, overexpressed in Escherichia colicells and purified as described previously [19]. All sampleswere stored frozen at )20 �C. All chemicals used were ofanalytical grade.Prior to the experiments, protein samples were extensively

dialysed against the appropriate buffer at 4 �C. Sampleconcentrations were determined by absorbance at 280 nm,using the following extinction coefficients (e0.1%), whichwere determined here as described by Gill & von Hippel[21]: intact SK, 0.72; SKA1, 0.84; SKA, 0.60; SKB,0.62; SKC, 0.72; SKBC, 0.73. Freshly purified proteinsamples were confirmed as monomeric by gel-filtrationchromatography.

Differential scanning calorimetry

Calorimetric experiments were made using a DASM4instrument [22]. DSC scans were conducted between 0 and110 �C. Instrumental baselines, obtained by filling bothcalorimeter cells with the corresponding buffer, weresystematically subtracted from the sample experimentalthermograms. The reversibility of protein denaturationwas assessed by comparing the thermograms obtained intwo consecutive scans with the same sample. The occur-rence of time-dependent denaturation processes accom-panying the thermal unfolding was investigated byrepeating the DSC experiment using different heating

rates within the range 0.25–2.0 �CÆmin)1 [23,24]. DSCtraces were corrected for the effect of the calorimeterresponse as reported elsewhere [25]. The temperaturedependence of the molar partial heat capacity, Cp, of theproteins was calculated from the DSC data as describedelsewhere [26], using a partial specific volume of0.73 mLÆg)1, which is the average value observed forglobular proteins. For thermal unfolding occurring atequilibrium, the Cp curves of single-domain fragmentswere fitted using the two-state model as describedelsewhere [27]. In these analyses, the Cp functions of thenative states are assumed to be linear, whereas those ofthe unfolded states are described by quadratic functions;the latter were determined from the sequence of each SKfragment according to Makhatadze & Privalov [28]. Forthe multidomain proteins, the equilibrium Cp curves werefitted to the sum of a number of two-state transitions. Inthese fittings the heat capacity change, DCp, for theunfolding of each domain was fixed by using the valuesobtained from the analysis of the Cp curves of thecorresponding single-domain, isolated fragment.

Gel-filtration chromatography

Aggregation of intact SK induced by heating at pH 7.0was checked by gel-filtration chromatography, using a1 · 30 cm Superose-12 column (Pharmacia, Uppsala,Sweden) attached to a Gilson HPLC instrument equippedwith an automatic sample injector. The column wasequilibrated at room temperature in 50 mM sodiumphosphate, pH 7.0, and calibrated with gel-filtrationstandards from Biorad and Sigma. SK samples of 20 lLin 20 mM phosphate, pH 7.0, were incubated in Eppen-dorf tubes in a thermostatic bath at different temperaturesfor 10 min and immediately cooled on ice. The sampleswere then injected into the column and eluted at a flowrate of 0.8 mLÆmin)1. Elution profiles were recorded bymonitoring absorbance at 220 and 280 nm. Peak areasand elution times were determined by using the manu-facturer’s software.

Limited proteolysis

The structural properties of heat-induced SK aggregateswere probed by limited proteolysis. A 10 mgÆmL)1 sampleof intact SK in 20 mM phosphate, pH 7.0, was heated to65 �C for 10 min to induce aggregation (see Results) andthen cooled on ice. The sample was immediately submittedto proteolysis with a-chymotrypsin (10 lgÆmL)1) at 23 �C.Aliquots were removed at different times, 20 mM phenyl-methanesulfonyl fluoride added to stop the proteolysis, andthen analysed by SDS/PAGE. The time-course of proteol-ysis of an identical unheated SK sample was also followedas a reference. An aliquot obtained after 10 min ofproteolysis of the heated SK sample was analysed by RP-HPLC using a C18 Dynamax-300 column as describedelsewhere [3]. The samples corresponding to the majorpeaks in the HPLC chromatograms were separated andanalysed by SDS/PAGE and electrospray ionization massspectrometry (ESI-MS). ESI-MS spectra were acquired on aBioA triple quadrupole atmospheric pressure mass spectro-meter from VG Biotech, equipped with an electrosprayinterface.

4122 A. I. Azuaga et al. (Eur. J. Biochem. 269) � FEBS 2002

CD spectroscopy of the 1–63 SK fragment

Far-UV CD spectra of the isolated 1–63 fragment (SKA1)were acquired on a JASCO J-720 spectropolarimeter at20 �C. Measurements were made between 190 and 250 nmat different pH values between 3.0 and 8.0, in 10 mMglycine, acetate or phosphate buffers. Sample concentra-tions were 0.1 mgÆmL)1. Data were recorded using a scanrate of 50 nmÆmin)1 and a response time of 1 s. Cuvettepath lengths were 0.1 cm. An average of 10 scans wasobtained. A baseline was subtracted from the spectra of thesamples and finally the result was smoothed. The meanresidue ellipticity, [Q], was calculated in units of degÆcmÆdmol)1. Near-UV spectra of the 1–63 SK fragment werealso recorded at pH 4.5 between 250 and 320 nm, using asample concentration of 1.0 mgÆmL)1 and a cuvette with apath length of 0.5 cm.

Fluorescence enhancement of ANS by the 1–63SK fragment

Fluorescence spectra of 8-anilino-1-naphthalenesulfonicacid (ANS) both in the presence and absence of the SKA1fragment were measured at 20 �C in a PerkinElmer LS-50spectrofluorimeter. The excitation wavelength was 380 nmand spectra were recorded between 400 and 600 nm. Theconcentrations of ANS and the SK fragment in the cuvettewere 10 lM. Fluorescence spectra were corrected using thespectra obtained for solutions in the absence of dye orprotein.

R E S U L T S

Thermal unfolding of SK under equilibrium

The thermal denaturation of intact SK and a set of SKfragments including either one or two SK domains wasfollowed by DSC at pH 7.0 in 20 mM sodium phosphatebuffer. Experiments at pH 6.0 and pH 8.0 were alsocarried out for intact SK and some of the fragments.The effects of sample concentration were also investi-gated.The concentration of all the samples was initially kept

to � 1 mgÆmL)1. Figure 1 shows the Cp curves corres-ponding to intact SK, each of the isolated SK domains(SKA, SKB and SKC) and a fragment consisting of SKdomains B and C (SKBC) at pH 7.0. The data for SKAat pH 7.0 and 0.88 mgÆmL)1 have been taken fromAzuaga et al. [19]. Fragments SKA, SKB and SKC showsingle unfolding transitions with high reversibility and noevidence of protein aggregation, even after heating tohigh-temperature. This indicates that the thermal unfold-ing of all the isolated SK domains at pH 7.0 occursessentially at equilibrium. Fragment SKBC unfolds intwo reversible transitions, corresponding to the consecu-tive unfolding of domain B and C, as described elsewhere[4]. At this low protein concentration (0.94 mgÆmL)1)intact SK also unfolds reversibly in two well-separatedpeaks.When the sample concentration of SK or SKA is raised

above 1.5 mgÆmL)1 at pH 7.0 the DSC profiles are clearlymodified, due to the presence of protein aggregationprocesses (see below). Similar effects of concentration or

aggregation processes were not observed in the rest of thefragments (results not shown). The DSC curves of SK atpH 6.0 are also affected by extensive aggregation under allconcentrations (1.0–10.3 mgÆmL)1) investigated. On theother hand, at pH 8.0 the thermal unfolding of both intactSK and SKA is fully reversible at all concentrations used inthis study (1.0–10 mgÆmL)1 for SK and 0.9–5.5 mgÆmL)1

for SKA).The DSC curves of each protein moiety for which

thermodynamic equilibrium conditions are unequivocallyverified (those measured at pH 8.0 or pH 7.0 and lowsample concentrations) have been fitted assuming that eachprotein domain unfolds independently in a two-statetransition. In the fits of the DSC curves of multidomainmoieties [intact SK (three domains) and SKBC (twodomains)], the heat capacity increment, DCp, for theindependent unfolding of each domain has been fixed byusing the values obtained from the fits corresponding tosingle-domain fragments. All the fits are good, as can beseen for pH 7.0 in Fig. 1. Figure 2 shows the deconvolutionof the heat capacity curves for intact SK at pH 7.0 andpH 8.0 into three independent two-state transitions, whichcan easily be identified as corresponding to each SKdomain. The parameters obtained from these fits are listed

Fig. 1. Partial molar heat capacity curves, Cp, of intact SK and frag-

ments SKBC, SKA, SKB and SKC obtained by DSC at pH 7.0, 20 mM

sodium phosphate. Experiments were performed at a heating rate of

2 �CÆmin)1 Sample concentrations employed were about 1 mgÆmL)1

(see text). Open circles represent the experimental data. Lines corres-

pond to the best fittings using equilibrium models of single or multiple

two-state transitions (see text for details).

� FEBS 2002 Unfolding and aggregation of streptokinase (Eur. J. Biochem. 269) 4123

in Table 1. These results indicate that during the equilibriumunfolding of SK, each domain behaves as a single cooper-ative unit, regardless of whether it is isolated or linked toother domains.

The effect of protein concentration and scan rateon the DSC curves

At pH 7.0 and sample concentrations higher than� 1.5 mgÆmL)1, the DSC curves of intact SK show a clearconcentration effect (Fig. 3B); the second peak observed atlow concentration, which corresponds to the sum of theequilibrium unfolding transitions of domainsA andC, splitsinto two well-separated peaks as the sample concentrationincreases. This effect is even more pronounced at pH 6.0(Fig. 3C), where protein precipitates are also visuallyevident after heating in the DSC cell. At pH 7.0, on theother hand, the samples remain transparent during theheating although soluble, high molecular mass aggregatesare formed (see below). At pH 8.0 the DSC curves areindependent of protein concentration (Fig. 3A) showingthat aggregation does not occur at concentrations below10 mgÆmL)1.Fig. 4 shows the results of a set of DSC experiments

carried out with SK to assess the reversibility of each of thetransitions under different conditions. At pH 7.0 and lowsample concentration (1.04 mgÆmL)1; Fig. 4A) or at pH 8.0even at relatively high protein concentration (3.3 mgÆmL)1;Fig. 4C), the peaks observed are highly reversible. AtpH 7.0 and sample concentration of 3.4 mgÆmL)1 (Fig. 4B),only the peak corresponding to the unfolding of domain Bis highly reproducible in a consecutive scan. Moreover,heating the sample to higher temperatures results in a majorloss of area for the transitions in a further scan. At pH 6.0the irreversibility is even more pronounced (Fig. 4D). Theseresults indicate that irreversible denaturation processesconcomitant with the thermal unfolding of SK occur atpH 7.0 and high sample concentrations and at pH 6.0 at allconcentrations.The effect of the temperature scan rate on theDSC curves

of SK at pH 7.0 and 3.4 mgÆmL)1 has also been investigatedto check whether the irreversible processes result in a kineticcontrol of the DSC traces [23,24] (Fig. 5). The unfoldingtransition corresponding to domain B is not affected by thescan rate, indicating that it occurs under equilibriumconditions. On the other hand, there is a significant effect

Fig. 2. Partial molar heat capacity curves, Cp, of intact SK at pH 7.0

and 8.0 showing the result of the fitting of the curves using an equilibrium

model with three two-state transitions. Symbols stand for the experi-

mental Cp data. Continuous lines correspond to the best fittings.

Dashed lines represent the predicted Cp curve of each of the two-state

transitions in which the global curves can be deconvoluted.

Table 1. Thermodynamic parameters for the independent thermal unfolding of the three SK domains observed by DSC. DCp values marked with (f)

were fixed in the fitting and correspond to the values obtained for the isolated domains. The uncertainties of the parameters correspond to the

standard errors obtained in the fittings. Reproducibility of Tm and DH values in different experiments was better than 0.5 �C and 10 kJ mol)1,respectively.

Domain A Domain B Domain C

Tm(�C)

DH (Tm)

(kJÆmol)1)DCp (Tm)

(kJÆK)1Æmol)1)Tm(�C)

DH (Tm)

(kJÆmol)1)DCp (Tm)

(kJÆK)1Æmol)1)Tm(�C)

DH (Tm)

(kJÆmol)1)DCp (Tm)

(kJÆK)1Æmol)1)

pH 7.0

SKA 52.3 ± 0.1 281 ± 1 6.1 ± 0.3 – – – – – –

SKB – – – 45.14 ± 0.03 377 ± 1 6.0 ± 0.3 – – –

SKC – – – – – – 71.2 ± 0.2 201 ± 1 0.8 ± 0.2

SKBC – – – 43.52 ± 0.05 320 ± 2 (f) 69.6 ± 0.2 224 ± 3 (f)

Intact SK 61.4 ± 0.2 319 ± 10 (f) 46.2 ± 0.1 363 ± 3 (f) 69.7 ± 0.7 199 ± 5 (f)

pH 8.0

SKA 47.1 ± 0.1 233 ± 1 5.5 ± 0.1 – – – – – –

SKC – – – – – – 67.8 ± 0.1 193 ± 1 0.6 ± 0.1

Intact SK 57.0 ± 0.2 270 ± 5 (f) 45.9 ± 0.1 363 ± 3 (f) 69.9 ± 0.5 194 ± 2 (f)

4124 A. I. Azuaga et al. (Eur. J. Biochem. 269) � FEBS 2002

of scan rate for the rest of the DSC curves. A decrease in thescan rate shifts the second peak towards lower temperaturestogether with a reduction in its area. The scan rate alsoaffects the high-temperature transition.These results indicate that time-dependent aggregation

processes are involved in the thermal denaturation ofSK at pH 7.0 and sample concentrations higher than� 1.5 mgÆmL)1, and at pH 6.0 at all concentrations studied.This results in considerable modification of the shape of theDSC curves, which become kinetically controlled andtherefore impossible to analyse on thermodynamic groundsalone. The most pronounced effects are observed attemperatures at which domain A unfolds suggesting aparticularly significant role for this domain in the overallaggregation of SK.

Thermal denaturation of isolated SK domain A

Amarked concentration effect on the DSC curves was alsofound for SKAat pH 7.0 (Fig. 6). At sample concentrationsequal to or higher than 2 mgÆmL)1, the DSC traces showtwo well-resolved peaks. The increase of sample concentra-tion shifts the first peak towards lower temperatures. Thispeak also becomes narrower and has a smaller area than thesingle two-state unfolding transition observed at a concen-tration of 0.88 mgÆmL)1. The partial development ofdenaturation heat suggests the formation of partiallyunfolded forms. The first peak is essentially irreversible ina second consecutive scan (result not shown). A secondtransition at a higher temperature (about 75 �C) appearsapproximately to complete the total heat of unfolding.Similarly to intact SK, the solution remains clear after

heating. Although the DSC curves are much simpler thanfor intact SK, the concentration effects are very similarunder the same conditions. Therefore an aggregationprocess similar to that found for intact SK appears tooccur with the isolated A domain. This result indicates thatthe aggregation tendency observed for intact SK resides atleast in part within domain A.The high-temperature transition occurring at high sample

concentrations is partially reversible for both SKA andintact SK. In a previous paper, we analysed the thermalunfolding of SK by one-dimensional NMR under the sameconditions studied here and at high sample concentrations[4]. We found that at temperatures near to 65 �C the NMRsignals became very broad and further heating at 85 �Cproduced a sharpening of the NMR signals, the spectrumbecoming similar to that expected for an unfolded poly-peptide chain. This line broadening of the NMR signals cannow be attributed to the aggregation processes that we haveseen here. These observations suggest that the high-temperature transition at around 75–80 �C observed forSK, and in all probability for SKA, corresponds to theunfolding and dissociation of protein aggregates, leadingfinally to the fully unfolded state.

A simple model for transient, kinetically controlledaggregation

A simple model can explain the effect of concentration ontheDSC curves of SKA. The thermal unfolding of fragmentSKA at low concentrations is very well described by a two-state transition, without the presence of intermediates with asignificant population. Therefore, the monomeric states in

Fig. 3. The effect of sample concentration on the DSC curves of intact SK at pH 8.0 (A), 7.0 (B) and 6.0 (C). Sample concentrations in mg per mL are

indicated along each curve. Curves have been displaced in the vertical axis for clarity. The length of the vertical segment in each panel represents

30 kJÆK)1Æmol)1 on the vertical axis.

� FEBS 2002 Unfolding and aggregation of streptokinase (Eur. J. Biochem. 269) 4125

equilibrium at low concentration are the native, N, and theunfolded, U.

N ����! UIt can be assumed that the unfolded state, U, forms

n-order aggregates, An.

nU ����!k1k2An

The aggregation process is considered to be reversiblebecause state An can dissociate and unfold at hightemperatures. Nevertheless, association and dissociationcan be slow at certain temperatures and therefore kineticallycontrolled. Constants k1 and k2 are the association anddissociation rate constants, respectively, related by theequilibrium constant of the aggregation process, KA. Thus,aggregation will be detected only for large k1 values andhigh concentrations of the state that undergoes aggregation.The equations of this simple model have been included in

the Appendix. The heat capacity curves, Cp, can bepredicted from these equations using the following set ofparameters: a linear heat capacity function for the nativestate, Cp(N); the enthalpies and the heat capacities of theunfolded state, DHU and DCpU, and of the aggregate, DHAandDCpA, all them relative to the native state, expressed permol of monomer at a given reference temperature, T0; thetemperature at which the Gibbs energy of unfolding is zero,Tm; the activation enthalpy for the aggregation process,DH 6¼1 ; the values of k1 and KA at T0; and finally theaggregation order, n. It should be pointed out thataccording to these equations, the DSC curves will dependon both total protein concentration and scan rate, as seen inour experimental data.Using this model we carried out the simultaneous fitting

of the DSC profiles obtained for SKA at pH 7.0 anddifferent sample concentrations, using only the DSC datacorresponding to the first of the two transitions present in

Fig. 5. The effect of the scan rate of the DSC calorimeter on the heat

capacity curves of intact SK at pH 7.0 and 3.36 mgÆmL)1. Scan ratesare: continuous line, 2.04 �CÆmin)1; dashed line, 1.03 �CÆmin)1; dottedline, 0.51 �CÆmin)1; dashed-dotted line, 0.25 �CÆmin)1.

Fig. 4. Tests of reversibility of the DSC transitions of intact SK by consecutive heating of the same sample in the calorimeter. The sample concen-

trations and pH values are indicated in the panels. First heatings of the sample are represented in continuous line, second heating in dashed lines,

third heating in dotted lines and fourth heating in dashed-dotted lines.

4126 A. I. Azuaga et al. (Eur. J. Biochem. 269) � FEBS 2002

the curves at high sample concentrations. To reduce thenumber of fitting parameters, Tm, DHU and DCpU werefixed in the fits using the values in Table 1 for SKA atpH 7.0. In addition, for the sake of simplicity the relativeheat capacity function of the aggregate, DCp,A, and theactivation enthalpy for the aggregation process, DH 6¼1 , werefixed to zero. The last assumption implies a temperature-independent k1, which is a reasonable approximationconsidering the narrow temperature interval in whichassociation is taking place. With these approximations,the number of adjustable parameters is reduced to five,which is a reasonable number taking into account that asingle two-state transition also requires five parameters to becorrectly described. The aggregation order, n, has beenmodified in different fitting attempts starting from n ¼ 2 ton ¼ 10. Higher values of n were not used due to numericalproblems in the computer fitting procedure. The fit for n¼ 8is represented in Fig. 6 together with the experimentalcurves. Despite the large number of simplifications, themodel is consistent with the effect of concentration on theshape and Tm of the first transition occurring at � 50 �C.Good descriptions of theDSC curves are obtained when then values are higher than 6. The parameters obtained fromthese fits are listed in Table 2.

Fitting the DSC curves including the second transition athigh sample concentrations gives poor results. The modelpredicts the second transition to be much sharper (morecooperative) than it proved experimentally. This discrep-ancy may be due to the fact that our model assumes a singletwo-state process for the association–dissociation reaction,whereas this process is very likely to be much morecomplicated, probably including many heterogeneous as-sociation/dissociation steps. Nevertheless, the general fea-tures of the experimental DSC curves are satisfactorilyrepresented by the model in spite of its simplicity and thenumber of approximations considered in the analysis.

Detection of temperature-induced SK aggregationby gel-filtration chromatography

With the aim of identifying the nature of the irreversibleprocesses occurring during the thermal denaturation of SK,several aliquots of protein in 20 mM phosphate, pH 7.0, atdifferent concentrations of between 0.05 and 18.5 mgÆmL)1

were incubated at 90 �C for 10 min and immediately cooledon ice. This procedure was based on the supposition that theassociation–dissociation equilibrium becomes effectivelyfrozen at low temperatures. To estimate the percentage ofaggregated protein the samples were subsequently analysedby gel-filtration chromatography at room temperature(Fig. 7A). At concentrations lower than 2.0 mgÆmL)1, theelution profiles consist of a single peak corresponding to thenative protein. At higher sample concentrations, however,an additional peak appears at the exclusion volume of thecolumn, which for the Superose 12 column corresponds toaggregates of at least 40 molecules of SK. No peaks ofintermediate mass were detected. The percentage of proteinin the aggregated form increasedwith sample concentration,reaching nearly 100% at the highest concentration investi-gated.Another set of SK samples of 9.9 mgÆmL)1 in 20 mM

phosphate, pH 7.0, were incubated for 10 min at differenttemperatures and immediately cooled on ice. For SKsamples incubated at temperatures below 45 �C, no aggre-gation was detected. At higher temperatures, the percentageof protein in the aggregated form increased (Fig. 7B),reaching a maximum at between 55 and 70 �C, where up to90% of the protein was aggregated. At higher incubationtemperatures the percentage of protein in the aggregatedecreased and was only about 42% at 100 �C. This isconsistent with the proposal that the aggregates unfold and

Fig. 6. The effect of sample concentration on the DSC curves of SK

domain A (SKA) at pH 7.0. Sample concentrations in mg per mL are

indicated along each DSC curve. Symbols represent the experimental

Cp data. Lines correspond to the simultaneous fitting of the three Cp

curves using the model described in the text. The parameters of the

fitting are: n ¼ 8; Cp(N) ¼ 33.1 + 0.11ÆT (kJÆK)1Æmol)1);DHA(50 �C) ¼ 177 kJÆmol)1; lnKA(50 �C) ¼ 76.2; lnk1 ¼ 60.7.

Table 2. Parameters resulting from the simultaneous fitting of the DSC

curves of SKA at pH 7.0 and different sample concentrations, using the

equations of the model described in the text. All parameters correspond

to T ¼ 50 �C. The uncertainties of the parameters correspond to thestandard errors obtained in the fittings.

n

DHAna

(kJÆmol)1) ln KA

DGA–DGUa

(kJÆmol)1) ln k1b

6 182 ± 6 54.4 ± 0.5 )24.4 42.8 ± 0.2

7 179 ± 5 65.3 ± 0.7 )25.1 51.7 ± 0.2

8 177 ± 5 76.2 ± 0.5 )25.6 60.7 ± 0.2

9 174 ± 6 87.1 ± 0.9 )26.0 69.6 ± 0.2

10 172 ± 6 110 ± 5 )29.6 78.5 ± 0.3

a Expressed per mol of monomer. b k1 units are mol–(n)1)Æmin)1.

� FEBS 2002 Unfolding and aggregation of streptokinase (Eur. J. Biochem. 269) 4127

dissociate at high temperatures. Nevertheless, during thisprocedure, in which samples are cooled from 100 to 0 �C,

some additional aggregation of SK cannot be avoidedunless the cooling is extremely fast.These results allow us to identify the irreversible process

induced by high temperatures with the formation of highmolecular mass aggregates of SK. The maximum degree ofaggregation occurs at high sample concentrations and attemperatures where domain A unfolds, and decreases athigher temperatures.

Limited proteolysis of the SK aggregates

Limited proteolysis with a-chymotrypsin was used tocharacterize the heat-induced aggregates of intact SK.Figure 8 shows SDS/PAGE gels monitoring the course ofproteolysis of a 10 mgÆmL)1 SK sample in 20 mM phos-phate, pH 7.0, which was heated to 65 �C for 10 min andthen cooled on ice. The proteolytic behaviour of anunheated identical sample is also shown for comparison.During the course of chymotryptic proteolysis of native

SK, several fragments accumulated as reported elsewhere[3]. The pattern of proteolysis of the aggregated SK samplewas, however, dramatically different. Despite forming highmolecular mass aggregates, its sensitivity to proteolysis wasmuch higher than that of native monomeric SK. Further-more, the SK chain was cleaved much more heterogene-ously. This indicates that the accessibility of the chain toproteolytic attack and therefore its structural disorder ishigher than in the native protein. In contrast to native SK,the 16 kDa fragment, corresponding to domain B, is notresistant to proteolysis, meaning that this domain isunstructured in the SK aggregates.The twomost highly populated fragments were generated

very quickly, within 2 min of proteolysis, corresponding tomolecular masses of approximately 7 and 12 kDa, andremained in the proteolytic mixture for up to 60 min. Thissuggests that both fragments might be involved in stablestructures in the protein aggregates. ESI-MS analysis ofthese fragments revealed a mass of 6765.6 ± 0.2 Da for the7 kDa fragment, whereas the 12 kDa fragment is in fact amixture of two fragments with masses of 12 265.2 ±0.2 Da and 12 428.3 ± 0.3 Da. These experimental

SK [1-414]

SKB [147-287]

SKC [288-380]

[1-63][1-63]

[275(6)-380]

0 2 5 10 20 60 0 2 5 10 20 60Time (min)

A B

SK [1-414]

MolecularMass (Da)

4500030000

25000

17000

12000

6000

[64-380]SKBC [147-380]

Fig. 8. SDS/PAGE gels monitoring the time course of proteolysis of native SK (A) and aggregated SK (B). Labels adjacent to the gels indicate the

sequence of some fragments. The molecular mass scale has been obtained using SDS/PAGE protein standards.

Fig. 7. Gel filtration analysis of the percentage of heat induced SK

aggregation at pH 7.0. (A) Aliquots of SK were preincubated at 90 �Cfor 10 min at different sample concentrations prior analysis. (B)

Aliquots of SK of 9.9 mgÆmL)1 were preincubated for 10 min at dif-ferent temperatures prior analysis.

4128 A. I. Azuaga et al. (Eur. J. Biochem. 269) � FEBS 2002

masses, together with the sequence specificity of chymo-trypsin, allowed us to identify the sequences of the threefragments as: 1–63 (defined previously as fragment SKA1),276–380 and 275–380, respectively.A sample of aggregated SK was subjected to proteolysis

for 10 min as described above, filtered and then analysed bygel-filtration chromatography. Aliquots were collected andanalysed by SDS/PAGE. It was observed that the fragments1–63 and 275(6))380 migrated together in the chromato-grams (results not shown), indicating that these twofragments interact in the proteolysed mixture.

Structural characterization of SK fragment 1–63

Isolated SK fragment 1–63 (SKA1) was structurally char-acterized in solution using a variety of techniques. Far-UVCD spectra of SKA1 were obtained at a series of pH valuesbetween 2.0 and 8.0 (Fig. 9A). The shape of the CD spectrawas strongly dependent on pH, changing from a typicalb sheet spectrum at pH 4.0 and 5.0 to the characteristic

random-coil spectrum at both pH 2.0 and pH 8.0. Thenear-UV spectrum of SKA1 at pH 4.5, 10 mM acetatebuffer and a sample concentration of 1.0 mgÆmL)1, how-ever, shows very little ellipticity in the 320–250 nm wave-length range (results not shown), suggesting that thefragment has only a small amount of fixed tertiary structureeven when it contains a large amount of secondarystructure.In the light of this latter observation, we investigated the

interaction between the SKA1 fragment and the hydropho-bic dye ANS. ANS has a strong tendency to interact withhydrophobic clusters exposed to the solvent, resulting in astrong enhancement in the fluorescence of the dye and ablue shift of the wavelength of the maximum, kmax, of thefluorescence spectrum. ANS has been frequently used tomonitor conformational changes and to characterize parti-ally folded states in proteins [29–31]. Figure 9B shows thefluorescence spectrum of a 10 lM solution of ANS in thepresence and absence of 10 lM of SKA1, at pH 7.0 and 4.4.The presence of the fragment produces a large increase inthe intensity of ANS fluorescence, which is higher at pH 4.4than at pH 7.0. kmax also changes in the presence of SKA1,with shifts of up to)50 nm compared to free ANS (see insetin Fig. 9B). The maximum shift occurred at around pH 5,the pH at which the fragment has the greatest amount ofsheet b structure.These results indicate that under mildly acid conditions

SKA1 has a significant amount of b sheet structure, apartially exposed hydrophobic core but little or no fixedtertiary structure. These features are characteristic of thecompact denatured states often known as molten globules.

D I S C U S S I O N

In this paper we have described how the thermal denatur-ation of SK is highly affected by pH and sample concen-tration. The most significant effect is the occurrence of high-order aggregation processes accompanying the unfolding ofthe protein, which are enhanced by lowering the pH orincreasing the sample concentration. The presence ofaggregation has a significant effect on the shape of theDSC curves, which become both concentration dependentand kinetically controlled. The primary consequence ofthese effects is the unsuitabilility of using standard,thermodynamics-based deconvolution methods to analysethe curves.At pH 7.0 and a sample concentration of less than

� 1.5 mgÆmL)1, the thermal unfolding of SK occursunequivocally under equilibrium conditions. This conclu-sion is also valid for pH 8.0 and sample concentrationsbetween 1.0 and 10 mgÆmL)1. The DSC curves obtained forSK under these conditions are accurately described by thesum of three two-state transitions, indicating that SKcontains three independent cooperative folding units. Thisfinding agrees with our previous studies [3,4,19] andwith thenumber of structural domains observed in the crystalstructure of SK complexed with microplasmin [5].Previous reports on studies into the thermal unfolding of

SK made by several authors using different techniquesreveal significant discrepancies in their account of thenumber of unfolding units involved [4,13–20]. One of thereasons for this disagreement might arise from the fact thatin some of these studies the number of independent

Fig. 9. Structural properties of SK fragment 1–63. (A) Far-UV CD

spectra of SK fragment 1–63 at different pH values. Symbols are:

pH 2.0 (j); pH 3.0 (h); pH 4.0 (d); pH 5.0 (s); pH 6.0 (m); pH 7.0

(n); pH 8.0 (r). (B) Fluorescence spectra of mixtures of 10 lM ANSand 10 lM SK fragment 1–63, at pH 4.4 (dashed line) and 7.0 (dottedline). Spectra in continuous lines represent the corresponding spectra

of 10 lM ANS in the absence of the SK fragment. (B inset) Depend-ence with the pH of the wavelength of the maximum of the fluores-

cence spectra for the ANS + SK 1–63 mixtures relative to the

spectrum of free ANS.

� FEBS 2002 Unfolding and aggregation of streptokinase (Eur. J. Biochem. 269) 4129

unfolding transitions of intact SK has been inferred bystandard deconvolution methods of the complex DSCcurves, without recourse to any additional informationexternal to these curves. Direct deconvolution can some-times suffer from uncertainties in the chemical baselinecorrections to the thermograms, which may bias theresulting number of unfolding transitions. We have shownhere that the use of changes in the heat capacity of unfoldingdetermined independently for the isolated domains allowsthis difficulty to be circumvented without having to resort tochemical baseline corrections. Using this procedure, thecomplex DSC profiles of SK can be perfectly explained interms of three independent transitions. Additionally, undersome of the experimental conditions used in previousstudies, aggregation processes, such as those shown here,may severely deform the DSC curves, which if unnoticedcould lead to misleading results when deconvolutionprocedures are applied.The unfolding temperatures, Tm, of the SK domains

decrease in the order: C>A > B. This order is contrary tothat of the values of the specific enthalpy of unfolding whencompared at the same temperature (B > A > C). Thevalues of the specific DCp for the unfolding of domains Aand B are similar (about 0.4 JÆK)1Æg)1), consistent with theirhigh structural homology, and fall within the range ofvalues observed for small globular proteins [32]. In contrast,the specific DCp for the unfolding of domain C is very low(about 0.06 JÆK)1Æg)1). This value, together with the lowunfolding enthalpy of the domain, is consistent with itslower degree of structure [5].Under the same equilibrium conditions, the values of Tm

and DHm for either of the isolated domains B and C agreewell with the values obtained when these domains form partof larger protein moieties. These values also agree well withthose already published derived from studies of theirthermal unfolding followed by CD and NMR [4]. Thus,the stabilities of domains B and C are not significantlyaffected by their detachment from the remainder of theprotein. Domain A, on the other hand, is destabilized by9–10 �C when excised from the rest of the chain. Visualinspection of the crystal structure of SK [5] indicates thatthere are significant contacts between domains A and B. It issurprising that removal of these interactions does not affectthe stability of domain B. Nevertheless, interdomain con-tacts may be conditioned by the complex formation withmicroplasmin in the crystal structure. An alternativeexplanation could be that some interactions internal todomain A are affected by chain excision. Domain C is, inconstrast, relatively isolated from the rest of the SKstructure and the linker with domain B appears to be veryflexible.An increase in pH from 7.0 to 8.0 does not affect the

stability of either domain B or C; only domain A shows aclear reduction of its Tm when the pH is raised from 7.0 to8.0. This dependence of the stability upon pH suggests thatunfolding is coupled to the change in ionization of theHis140 sidechain, which in the crystal structure forms a cleardouble salt bridge with the Asp32 and Asp106 sidechainswithin domain A [5], although we cannot exclude theparticipation of other ionisable groups.On the other hand, the results described here demonstrate

that under certain experimental conditions, i.e. pH 7.0 andsample concentrations higher than a few mg per mL, or

pH 6.0 at all the concentrations investigated, the thermalunfolding of SK domain A, either isolated or when part ofthe intact protein, is accompanied by formation of highmolecular mass aggregates. Further heating, however,produces dissociation and unfolding of these aggregates,which result in a cooperative transition in the DSC curves.A very simple model reproduces well the effects that the

kinetically controlled aggregation process exert over theunfolding transition of SK domain A. The enthalpy ofthe aggregate per mol of monomer unit (177 kJÆmol)1) liesbetween the enthalpies of the native state (the referencestate) and the unfolded state (267 kJÆmol)1), indicating thatthe aggregate contains a significant degree of structure. Thisconclusion is consistent with the development of anadditional cooperative transition accompanying the disso-ciation of the aggregates, and suggests that at least some ofthe structure within the aggregates could be specific.The aggregation process at pH 7.0 is slow enough at theintermediate temperatures where it occurs to lead to thekinetic control of the DSC curves. We have described asimilar slow association process for a thermolysin fragmentin a previous paper [33].Different values for the aggregation order, n, in our

model give good fits for the first transition of the DSCcurves of SKA. This finding could be interpreted inprinciple as an indication of the insensitivity of this modelto the value of n, although it could also suggest that theaggregation process is more heterogeneous and complexthan represented by this simple model. In spite of this, thethermodynamic parameters of the aggregated state, whenexpressed per mol of monomer, are essentially independentof the aggregation order (See Table 2). It is interesting tonote that at 50 �C, close to the unfolding temperature ofdomain A, aggregation is highly favoured, as the Gibbsenergy change of the aggregation process is about )26 kJper mol of monomer.The gel filtration study indicates that the SK aggregates at

room temperature consist of at least 40 molecules. Asmentioned in the Results, we could not test high values of nin our fittings of theDSC curves due to numerical problems.We should bear in mind, however, that the value of n in themodel actually represents an apparent average of themolecularity of the rate-limiting step of aggregation, which,depending on the specific aggregation mechanism, could bemarkedly different from the size of the final aggregates thatare formed after cooling.The most significant resistance of the SK aggregates to

limited proteolysis is located in two separate sequenceregions: segment 1–63, within domain A, and segment275(6))380, which corresponds principally to domain C(residues 292–380). As the isolated A domain also under-goes an aggregation process similar to that of intact SK, it isvery likely that the region that principally stabilizes theaggregated state resides within the segment 1–63.We cannotexclude, however, the participation of domain C in theseinteractions because domain C and fragment 1–63 migratetogether in the gel-filtration chromatography of a proteo-lysed sample of aggregated SK. Nevertheless, the presenceof domain C is not necessary for aggregation, while region1–63 of domain A is both necessary and sufficient.It is interesting that the two 12 kDa fragments that

accumulate during proteolysis of the aggregate encompassthe whole of domain C (starting at Leu292) plus an

4130 A. I. Azuaga et al. (Eur. J. Biochem. 269) � FEBS 2002

additional segment [274(5))292], which includes the linkerbetween domains B and C and extends into the domain Bstructure in native SK. Indeed, proteolytic cleavages occurat positions Tyr274–Tyr275 and Tyr275–Val276, instead ofat position Phe287–Asp288, in the flexible linker betweendomains B and C, as it is the case in native SK. Tyr274 andTyr275 are located in the middle of a b sheet withindomain B in the crystal structure of SK [5]. It appears thenthat domain B is not correctly folded in the aggregatesbecause part of its sequence is involved in interactionsassociated with the formation of the aggregate.At mildly acidic pH, the fragment SKA1 in isolation can

adopt a conformation rich in b sheet, containing a hydro-phobic patch that can interact with ANS. This conforma-tion is destabilized by an increase in pH, and SKA1 is fullyunfolded at pH 8.0. This structural change correlates withthe pH dependence of the aggregation of both SK andSKA, the extent of which is reduced by an increase in pHfrom 6.0 to 8.0. This result suggests that unfolding ofdomain A under experimental conditions stabilizing theb sheet rich conformation of segment 1–63 and its hydro-phobic cluster produces SK species prone to high orderaggregation. This is highly consistent with the generalconclusion that protein aggregation is promoted by anincreased propensity of b sheet structures that dominatestable intermolecular interactions [34].

C O N C L U S I O N S

When the thermal denaturation of SK occurs unequivocallyunder equilibrium conditions (in this work pH 7.0 and lowprotein concentration, or pH 8.0) an analysis of the DSCprofiles in conjunction with those of the isolated domainsshows that this protein consists of three independentunfolding units, each of which corresponds to one of itsthree structural domains.At pH 6.0 or at pH 7.0 with high sample concentrations,

the thermal unfolding of SK domain A, either in isolationor when forming part of intact SK, is accompanied by theformation of high molecular mass aggregates. This associ-ation is reversible at pH 7.0 but at low and intermediatetemperatures the equilibrium slows down, severely deform-ing the unfolding profiles and rendering them kineticallycontrolled, which precludes their simple interpretation byequilibrium models.The SK aggregates are largely unstructured but remain

resistant to proteolysis in specific sequences of the proteinchain [1–63, 274(75))380]. Of these, the N-terminal region1–63 is both necessary and sufficient for the formation ofaggregates and appears to be the nucleus of aggregation. Aswell as shedding light on the specific events associated withthe aggregation of SK, this study provides insight into thenature of protein aggregation more generally. In particularit supports real evidence for the specificity of the aggregationprocess, and for the role of nucleation events in thismechanism.

A C K N O W L E D G E M E N T S

We thank Dr Richard A.G. Smith of AdProTech Limited for

supplying the streptokinase. We also thank Dr J. Trout for revising

the English text. This work has been financed by the European Union

Network ERB4061-PL-950200 and by grants PB96-1446 and

BIO2000-1459 of the Spanish Ministry of Science and Technology.

The Oxford Centre for Molecular Sciences is funded by BBSRC,

EPSRC and MRC. The research of CMD is also supported by the

Wellcome Trust.

R E F E R E N C E S

1. Malke, H., Roe, B. & Ferretti, J.J. (1985) Nucleotide sequence of

the streptokinase gene from Streptococcus equisimilisH46A. Gene

34, 357–362.

2. Christensen, L.R. (1945) Streptococcal fibrinolysis: a proteolytic

reaction due to a serum enzyme activated by streptococcal

fibrinolysin. J. Gen. Physiol. 28, 363–383.

3. Parrado, J., Conejero-Lara, F., Smith, R.A.G., Marshall, J.M.,

Ponting, C.P. & Dobson, C.M. (1996) The domain organization

of streptokinase: nuclear magnetic resonance, circular dichroism,

and functional characterization of proteolytic fragments. Protein

Sci. 5, 693–704.

4. Conejero-Lara, F., Parrado, J., Azuaga, A.I., Smith, R.A.G.,

Ponting, C.P. & Dobson, C.M. (1996) Thermal stability of the

three domains of streptokinase studied by circular dichroism and

nuclear magnetic resonance. Protein Sci. 5, 2583–2591.

5. Wang, X., Lin, X., Loy, J.A., Tang, J. & Zhang, X.C. (1998)

Crystal structure of the catalytic domain of human plasmin

complexed with streptokinase. Science 281, 1662–1665.

6. Damaschun, G., Damaschun, H., Gast, K., Gerlach, D., Missel-

witz, R., Welfle, H. & Zirwer, D. (1992) Streptokinase is a flexible

multi-domain protein. Eur. Biophys. J. 20, 355–361.

7. Young, K.C., Shi, G.Y.,Wu, D.H., Chang, L.C., Chang, B.I., Ou,

C.P. & Wu, H.L. (1998) Plasminogen activation by streptokinase

via a unique mechanism. J. Biol. Chem. 273, 3110–3116.

8. Conejero-Lara, F., Parrado, J., Azuaga, A.I., Dobson, C.M. &

Ponting, C.P. (1998) Analysis of the interactions between strep-

tokinase domains and human plasminogen. Protein Sci. 7, 2190–

2199.

9. Nihalani, D., Kumar, R., Rajagopal, K. & Sahni, G. (1998) Role

of the amino-terminal region of streptokinase in the generation of

a fully functional plasminogen activator complex probed with

synthetic peptides. Protein Sci. 7, 637–648.

10. Chaudhary, A., Vasudha, S., Rajagopal, K., Komath, S.S., Garg,

N., Yadav, M., Mande, S.C. & Sahni, G. (1999) Function of the

central domain of streptokinase in substrate plasminogen docking

and processing revealed by site-directed mutagenesis. Protein Sci.

8, 2791–2805.

11. Reed, G.L., Houng, A.K., Liu, L., Parhami-Seren, B., Matsueda,

L.H., Wang, S. & Hedstrom, L. (1999) A catalytic switch and the

conversion of streptokinase to a fibrin-targeted plasminogen

activator. Proc. Natl Acad. Sci. USA 96, 8879–8883.

12. Lin, L.F., Houng, A. & Reed, G.L. (2000) Epsilon amino caproic

acid inhibits streptokinase–plasminogen activator complex for-

mation and substrate binding through kringle-dependent mecha-

nisms. Biochemistry 39, 4740–4745.

13. Radek, J.T. & Castellino, F.J. (1989) Conformational properties

of streptokinase. J. Biol. Chem. 264, 9915–9922.

14. Welfle, H.,Misselwitz, R., Fabian, H., Damerau,W., Hoelzer,W.,

Gerlach, D., Halnin, N.N. & Venyaminov, S.Y. (1992) Con-

formational properties of streptokinase – secondary structure and

localization of aromatic amino acids. Int. J. Biol. Macromol. 14,

9–18.

15. Welfle, K., Pfeil, W., Misselwitz, R., Welfle, H. & Gerlach, D.

(1992) Conformational properties of streptokinase – differential

scanning calorimetric investigations. Int. J. Biol. Macromol. 14,

19–22.

16. Misselwitz, R., Kraft, R., Kotska, H., Fabian, K., Welfle, K.,

Pfeil, W., Welfle, H. & Gerlach, D. (1992) Limited proteolysis of

streptokinase and properties of some fragments. Int. J. Biol.

Macromol. 14, 107–116.

� FEBS 2002 Unfolding and aggregation of streptokinase (Eur. J. Biochem. 269) 4131

17. Teuten, A.J., Broadhurst, R.W., Smith, R.A.G. & Dobson, C.M.

(1993) Characterization of structural and folding properties of

streptokinase by n.m.r. spectroscopy. Biochem. J. 290, 313–319.

18. Medved, L.V., Solovjov, D.A. & Ingham, K.C. (1996) Domain

structure, stability and interactions in streptokinase. Eur. J.

Biochem. 239, 333–339.

19. Azuaga, A.I., Woodruff, N., Conejero-Lara, F., Cox, V.F., Smith,

R.A.G. & Dobson, C.M. (1999) Expression and characterization

of the intact N-terminal domain of streptokinase. Protein Sci. 8,

443–446.

20. Beldarrain, A., Lopez-Lacomba, J.L., Kutyshenko, V.P., Serrano,

R. & Cortijo, M. (2001) Multidomain structure of a recombinant

streptokinase. A differential scanning calorimetry study. J. Protein

Chem. 20, 9–17.

21. Gill, S.C. & von Hippel, P.H. (1989) Calculation of protein

extinction coefficients from amino acid sequence data. Anal.

Biochem. 182, 319–326.

22. Privalov, P.L. & Plotnikov, S.A. (1989) Three generations of

scanning microcalorimeters for liquids. Thermochim. Acta 139,

257–277.

23. Sanchez-Ruiz, J.M., Lopez-Lacomba, J.L., Cortijo, M. & Mateo,

P.L. (1988) Differential scanning calorimetry of the irreversi-

ble thermal denaturation of thermolysin. Biochemistry 27, 1648–

1652.

24. Conejero-Lara, F., Sanchez-Ruiz, J.M.,Mateo, P.L., Burgos, F.J.,

Vendrell, J. & Aviles, F.X. (1991) Differential scanning calori-

metric study of carboxypeptidase B, procarboxypeptidase B and

its globular activation domain. Eur. J. Biochem. 200, 663–670.

25. Lopez-Mayorga, O. & Freire, E. (1987) Dynamic analysis of dif-

ferential scanning calorimetry data. Biophys. Chem. 27, 87–96.

26. Privalov, P.L. & Khechinashvili, N.N. (1974) A thermodynamic

approach to the problem of stabilization of globular protein

structure: a calorimetric study. J. Mol. Biol. 86, 665–684.

27. Viguera, A.R., Martınez, J.C., Filimonov, V.V., Mateo, P.L. &

Serrano, L. (1994) Thermodynamic and kinetic analysis of the

SH3 domain of spectrin shows a two-state folding transition.

Biochemistry 33, 2142–2150.

28. Makhatadze, G.I. & Privalov, P.L. (1990) Heat capacity of pro-

teins. I. Partial molar heat capacity of individual amino acid

residues in aqueous solution: hydration effect. J. Mol. Biol. 213,

375–384.

29. Semisotnov, G.V., Rodionova, N.A., Razgulyaev, O.I., Uversky,

V.N., Gripas¢, A.F. & Gilmanshin, R.I. (1991) Study of the

�molten globule� intermediate state in protein folding by a hydro-phobic fluorescent probe. Biopolymers 31, 119–128.

30. Guijarro, J.I., Sunde, M., Jones, J.A., Campbell, I.D. & Dobson,

C.M. (1998) Amyloid fibril formation by an SH3 domain. Proc.

Natl Acad. Sci. USA 95, 4224–4228.

31. Van Nuland, N.A., Meijberg, W., Warner, J., Forge, V., Scheek,

R.M., Robillard, G.T. & Dobson, C.M. (1998) Slow cooperative

folding of a small globular protein HPr.Biochemistry 37, 622–637.

32. Privalov, P.L. (1979) Stability of proteins. Adv. Prot. Chem. 33,

167–241.

33. Conejero-Lara, F. & Mateo, P.L. (1996) Presence of a slow

dimerization equilibrium on the thermal unfolding of the

205–316 thermolysin fragment at neutral pH. Biochemistry 35,

3477–3486.

34. Chiti, F., Taddei, N., Baroni, F., Capanni, C., Stefani, M.,

Ramponi, G. & Dobson, C.M. (2002) Kinetic partitioning of

protein folding and aggregation. Nat. Struct. Biol. 9, 137–143.

A P P E N D I X

We present here the mathematical development of themodel introduced in Results.The equilibrium between N and U can be treated

separately by defining the following partition function:

Qm ¼ 1 þ KU ¼ 1 þ e�DGURT ð1Þ

whereKU is the equilibrium constant of unfolding and DGUis the change in the standardGibbs energy of unfolding. Thefractions of N and U, relative to the total concentration ofmonomeric species, are given by:

yU ¼e�

DGURT

QmyN ¼

1

Qmð2Þ

The fraction of protein monomers in each state relative tothe total concentration of protein monomers, C0, is definedas:

xN ¼½N�C0

xU ¼½U�C0

xA ¼n½An�C0

ð3Þ

Taking into account that the total concentration of proteinis C0 ¼ [N] + [U] + n[An], we can relate the fractions, yiand xi by

xi ¼ yi � ð1� xAÞ ð4Þ

where i stands for either N or U.We can define the average enthalpy of the monomeric

species, relative to N by:

hDHim ¼X

i

yi � DHi ¼ yU � DHU ð5Þ

where DHU is the enthalpy change of unfolding.The average enthalpy of the whole system is:

hDHi ¼ xADHA þ xUDHU ¼ xADHA þ ð1� xAÞyUDHU

¼ hDHim þ xAðDHA � hDHimÞ ð6Þ

where DHA is the enthalpy of the aggregate relative toN, expressed per mol of monomer. Differentiating thisexpression with respect to temperature provides theheat capacity relative to N, also known as the excessheat capacity. The addition of the heat capacity of thenative state, Cp(N), gives the total heat capacity of theprotein:

Cp ¼ CpðNÞ þ Cp;m þ xAðDCp;A � Cp;mÞ

þ ðDHA � hDHimÞdxAdT

ð7Þ

HereCp(N) + Cp,m is the heat capacity curve that would beobserved in the absence of aggregation, i.e. at pH 7.0 andlow protein concentrations. Cp,m can be easily obtained bydifferentiating Eqn (5) with respect to temperature.

Cp;m ¼ yUDCp;U þ1

RT 2yUDHUðDHU � hDHimÞ ð8Þ

For the purpose of these equations DCp,U, DCp,A, DHUand DHA are taken to be in general functions of tempera-ture.The rate of formation of the aggregate can be expressed

as:

dxAdt¼ nCn�1

0 k1ynUð1� xAÞn � k2xA ð9Þ

Taking into account the constant scan rate, v ¼ dT/dt, in aDSC experiment:

4132 A. I. Azuaga et al. (Eur. J. Biochem. 269) � FEBS 2002

dxAdT¼ 1

v½nCn�1

0 k1ynUð1� xAÞn � k2xA� ð10Þ

The association and dissociation rate constants, k1 and k2,can be related by the equilibrium constant of the aggrega-tion process:

KA ¼k1k2¼ ½An�½U�n ð11Þ

The van’t Hoff equation gives the temperature dependenceof KA:

d ln KAdT

¼ nðDHA � DHUÞRT2

ð12Þ

and the aggregation rate constant, k1, changes withtemperature, as given by the equation:

ln k1 ¼ ln k1ðT0Þ �n � DH 6¼1

R

1

T� 1

T0

� �þ ln T

T0

� �

ð13Þ

DH 6¼1 represents the activation enthalpy of the associationprocess, expressed per mol of monomer, and T0 is areference temperature, chosen as 50 �C here for the sake ofconvenience.Eqn (10) can be integrated numerically to calculate xA

and dxA/dT as functions of temperature. The Cp curvespredicted by the model can be calculated by substitutingthese two functions into Eqn (7).

� FEBS 2002 Unfolding and aggregation of streptokinase (Eur. J. Biochem. 269) 4133