Proc. Natl. Acad. Sci. USAVol. 90, pp. 2025-2029, March 1993Biophysics

Protein dynamics in minimyoglobin: Is the central core ofmyoglobin the conformational domain?

(heme proteins/flash photolysis/kinetics)

ERNESTO E. Di IoRIo*, WEIMING YU*, CLAUDIO CALONDER*, KASPAR H. WINTERHALTER*,GIAMPIERO DE SANCTISt, GIANCARLO FALCIONIt, FRANCA ASCOLIt, BRUNO GIARDINA0§,AND MAURIZIO BRUNORIT*Laboratorium fOr Biochemie I, Eidgenossische Technische Hochschule (ETH), UniversitAitsstrasse 16, 8092 Zurich, Switzerland; tDipartimento di BiologiaMolecolare, Celiulare e Animale, Universita di Camerino, Via F. Camerini, 62032 Camerino, Italy; tDipartimento di Medicina Sperimentale e ScienzeBiochimiche, Universita "Tor Vergata," Via 0. Raimondo, 00173 Rome, Italy; and lDipartimento di Scienze Biochimiche "Alessandro Rossi-Fanelli"Universita di Roma "La Sapienza," Piazzale A. Moro 5, 00185 Rome, Italy

Communicated by Hans Frauenfelder, November 18, 1992 (received for review June 6, 1992)

ABSTRACT The kinetics of CO binding to the horsemyoglobin fragment Mb-(32-139), the so-called "mini-Mb,"were investigated by laser flash photolysis in 0.1 M phosphatebuffer and in buffer with 75% (vol/vol) glycerol. The reactiondisplays complex time courses that can be approximated sat-isfactorily only with a sum of five exponentials. The features ofthe kinetic components and a comparison of the deoxy-minus-carbonyl difference spectra ofmini-Mb and horseMb obtainedunder equilibrium conditions, with the kinetic difference spec-tra resulting from the global analysis of the traces recordedbetween 400 and 450 nm, show that CO binding to mini-Mb isaccompanied by large structural changes. In view of the factthat mini-Mb is an approximation ofthe Mb-(31-105) fragmentencoded by the central exon of the Mb gene, this finding isparticularly relevant. On the basis of our data and previousreports [De Sanctis, G., Falcioni, G., Giardina, B., Ascoli, F.& Brunori, M. (1988) J. Mol. Biol. 200, 725-733; De Sanctis,G., Falcioni, G., Grelloni, F., Desideri, A., Polizo, F., Giar-dina, B., Ascoli, F. & Brunori, M. (1992) J. Mol. Biol. 222,637-643], we propose that the protein fragment encoded by thecentral exon of the Mb gene is the domain responsible forligand-linked conformational transitions, while the two termi-nal fragments dampen the amplitude of the structural changesthat accompany ligand binding, thus rendering the proteinstable and kinetically more efficient in its physiological func-tion.

Dynamic fluctuations are of vital importance for proteinsbecause they guarantee the exchange of ligand or substratemolecules, or both, between the solvent and the active site (1)as well as allosteric transitions or large-scale motions such asthose occurring in pore-forming proteins (2). Interesting openquestions concern the correlation between structural anddynamic features and the role played by the various domainsof a globular protein in determining its dynamic behavior.These questions are closely related to another one-i.e.,what is the role of dynamic fluctuations in determining thefunctional properties of a given protein. An original approachto investigate these fundamental aspects of protein structureand dynamics is to see if the domain encoded by the centralexon of the myoglobin (Mb) gene, which consists of threeexons and two long introns (3), behaves dynamically like theparent protein or if it displays peculiar properties. An answerto this question is also necessary to better understand whythere are three exons in the genes encoding Mb and thehemoglobin (Hb) chains (4-7).

Mini-Mb is a proteolytic fragment of horse-heart Mb com-prising residues 32-139 [Mb-(32-139); refs. 8 and 9] andrepresents an acceptable approximation of the domain en-coded by the central exon of the Mb gene.¶ Mini-Mb resem-bles the native protein functionally (8, 9), even though thestability of its oxygenated derivative is drastically reduced(10). Given the similarity in the three-dimensional folding ofMb and Hb subunits (11) and in the exon-intron structure ofthe genes encoding for these proteins, 11 mini-Mb serves as amodel ofgeneral validity. The central exons ofthe Mb and Hbchain genes code for the domain that provides the hydro-phobic crevice where heme is bound, and the role proposedfor the C- and N-terminal fragments is to optimize the fitbetween crevice and prosthetic group (6, 7, 12).To gain insight on the dynamic behavior of the central Mb

core, we have investigated by flash photolysis the CO bindingkinetics to mini-Mb under various experimental conditions.

MATERIALS AND METHODSHorse heart Mb and bovine hemin were purchased fromSigma. Purity was checked respectively by isoelectric focus-ing (13) and by the pyridine-hemochromogen method (ref. 14,pp. 10 and 11). Clostripain (EC 3.4.22.8; Clostridium pepti-dase B) was obtained from Sigma and had an activity of 145units/mg. All other chemicals were of analytical grade andpurchased either from Fluka or from Merck.

Absorption spectra were recorded on a Cary 219 (Varian)or on a HP-8452A (Hewlett-Packard) spectrometer.Apo-Mb was prepared as described by Rossi Fanelli et al.

(15); the only modification was subjecting the resulting globinto a single 4-h dialysis against a solution (100 mg/liter) ofsodium bicarbonate.Mini-Mb was prepared as described (8, 9) except for the

reconstitution with heme, which was done in the presence ofcyanide to increase the stability of the product (9).For flash-photolysis measurements, horse Mb or freshly

reconstituted mini-Mb were diluted with either phosphatebuffer alone or with a phosphate buffer/glycerol mixture thathad been preequilibrated with CO and contained traceamounts of sodium dithionite. Final concentrations were ca.65 ,uM protein, 0.1 M phosphate (pH 7 at 20°C),.and, whenpresent, 75% (vol/vol) glycerol. Right after preparation thesamples were transferred anaerobically into a gold-platedcopper cell with a path length of 1.3 mm, which was mounted

§Present address: Istituto di Chimica FacoltA di Medicina, UniversitaCattolica, L.go F.Vito 1, 00100 Rome, Italy.'The central exon ofthe Mb gene codes for the fragment Mb-(31-105)(3), whereas those of the a and , human Hb subunit genes,respectively, code for fragments Hb-(32-99) and Hb-(31-104) (7).

2025

The publication costs of this article were defrayed in part by page chargepayment. This article must therefore be hereby marked "advertisement"in accordance with 18 U.S.C. §1734 solely to indicate this fact.

Proc. Natl. Acad. Sci. USA 90 (1993)

in a CF1204 cryostat equipped with an ITC-4 temperaturecontroller (Oxford). CO binding after photolysis was mea-

sured in the 400 to 450-nm region by using the instrumenta-tion and procedure described elsewhere (16, 17).Mini-Mb and horse-Mb concentrations were determined

spectrophotometrically in the reduced CO form on the basisof extinction coefficients at the Soret maximum of 150A'mM-1cm-1 (9) and 207 A-mM-1cm-1 (ref. 14, p. 19),respectively, with 0.1 M phosphate buffer (pH 7) as solvent.The analysis of the experimental data was performed by a

global least-squares-fitting procedure as described (16, 17).For the kinetic measurements on mini-Mb, a multi-exponential function of the type

AA(T, t) = AAi exp{-ko,.exp[-Hj/(R.T)].t} [1]

was used in which R is the gas constant, T and t are

respectively absolute temperature and time, AA(T, t) is theoverall absorbance change, and, for each kinetic component

0

i, AAi is the absorbance change at time zero, ko0 is thepreexponential, and Hi is the activation enthalpy. The num-

ber of kinetic components to be used for the fittings wasdetermined on the basis of an F test (18).Room temperature kinetics of CO binding to horse Mb

were fitted to a sum of one distributed and one exponentialprocess:

AA(T, t) = AA1 J dH-g,(H)-expf-k0I exp[-H/(R T)] t}

+ AA2.exp{-k%.exp[-H2/(R.T)].t} [2]

by using the g(H) parameterization described elsewhere (16).For all fittings, the error analysis on the individual param-

eters was done as described elsewhere (16).

RESULTSCO recombination to mini-Mb after nanosecond-flash-

a photolysis can be fitted satisfactorily only by using a sum of*1 < five exponentials (Eq. 1 with i ranging from 1 to 5). This2 applies both to samples in buffer and in glycerol/buffer

mixtures, even though the traces in the latter solvent indicate

a more complex situation, as shown in Fig. 1. Here repre-*4 / 80 noo sentative time courses for CO rebinding to mini-Mb in 0.1 M3 phosphate buffer and in that buffer with 75% glycerol are

depicted; the fitted curves are reconstructed from the pa-

c, u = ggooo Brameters listed in Table 1. An attempt to fit the mini-Mb data

with a sum of one distributed and one (or more) exponential

_7 -6 -5 -4 -3 -2 -1 ° 1 processes was not successful. On the contrary, horse Mb7-6 -5 -'4 -3 -2 -1 1traces could be well approximated with Eq. 2.

The effect ofCO and glycerol concentration on the reaction1LA b rates of the five kinetic components is illustrated in Table 1.

In phosphate buffer, a reduction of the free CO concentration

-4 -3 -2Time, log s

FIG. 1. Time courses for the binding of CO to mini-Mb afterphotolysis monitored at 408 nm (o), 422 nM (A), and 442 nm (O) ina and b and 410 nm (o), 422 nm (A), 426 nm (O) and 438 nm (o) inc. Symbols refer to the experimental data, while the continuouscurves were computed from the reaction rates listed in Table 1.Measurements were performed in 0.1 M phosphate buffer (pH 7) (aand b) and in 75% glycerol/25% phosphate buffer (c). Traces weretaken at 273 K (a and b) and at 290 K (c). Prior to the addition of theconcentrated protein stock, the solvent was saturated at 293 K with1 atmosphere (atm; 1 atm = 1.01 x 105 Pa) (a and c) or 0.5 atm (b)of CO. Trace amounts of dithionite were also present.

Table 1. Effect of CO concentration and of the presence of 75%glycerol in the solvent on the apparent reaction rates

Reaction rates, s-1 at 273 K

Kinetic 1 atm COcomponent 0.5 atm CO No glycerol 75% glycerol

I 1.7 ['9 104 2.5 [401 4 4.9 40[2

0.10 0.14

II 2.1 6].103 4.3 ].103 21 [1 91030.19 0.37

III 3.8 .102 6.6 5.810 2.4 72.1J20.40 0.32

IV 4[49] 99 [118 18 r20]L41J [ ~83] 116

0.25 0.12

V 2.9[] 3 [29] 5.6[ ].1020.06 0.03

Rates are for the five kinetic components needed to account for thetime courses of the reaction between mini-Mb and CO. The 65%confidence limits are given in square brackets, while the fraction ofthe five processes, as obtained from their relative amplitudes at 422nm, are in italic numbers. The relative amplitude of the five com-ponents observed in 75% glycerol cannot be computed because inthis case the time courses are always a mixture of bleaching andabsorbing signals (Figs. lc and 4). Furthermore, they would not becomparable with those in the absence ofglycerol because ofthe largespectral changes observed when the cosolvent is present (Figs. 2 and4). More details on the experimental conditions are given in Materialand Methods.

2026 Biophysics: Di lorio et al.

Proc. Natl. Acad. Sci. USA 90 (1993) 2027

420 430Wavelength, nm

FIG. 2. Difference spectra between deoxy and carbonyl deriva-tives of mini-Mb measured under equilibrium conditions in phos-phate buffer (continuous line without symbols) or from the total toabsorption change in flash-photolysis kinetic measurements in phos-phate buffer (continuous line with circles) and in 75% glycerol/25%phosphate buffer (continuous line with squares), both equilibrated at273 K with 1 atm of CO. The difference spectrum of horse Mb(dashed line) is also reported and refers to equilibrium conditions inbuffer as well as to kinetic measurements in buffer with 75% glycerol.Normalization of the spectra has been done on the basis of theirintegrated areas.

from ca. 1 mM to ca. 0.5 mM produces a roughly 50%decrease in the rate coefficients of the three intermediateprocesses (II, III, and IV) but has no significant effect on theslowest one. The errors on the estimates of the reaction ratesfor process I are too large to be able to exclude the possibilitythat this kinetic component is affected by CO concentration.The increase in velocity ofprocess I is remarkable when 75%glycerol is added to the solvent.From the global least-squares analysis of the traces re-

corded for mini-Mb and horse Mb between 400 nm and 450nm at regular intervals of2 nm, the difference spectra relativeto the overall CO combination process and to the variouskinetic components have been obtained. Most of them areshown in Figs. 2, 3, and 4, along with those measured underequilibrium conditions and those computed from histidine-heme-CO flash-photolysis measurements from the literature(19). In the case of horse Mb the kinetic and equilibriumdifference spectra are the same in both the presence and

Wavelength, nm

absence ofglycerol (Fig. 2). For mini-Mb the spectra stronglydepend on solvent composition. In phosphate buffer, amarked red shift of the positive band in the kinetic differencespectrum is observed relative to the equilibrium situation(Fig. 2). Furthermore, the spectrum of mini-Mb is quitesimilar to that ofhorse Mb, even though its bands are broader(Fig. 2). A completely different picture emerges from thedifference spectra of mini-Mb when 75% glycerol is present(Fig. 2). The positive region of the spectrum appears inter-mediate between the static (equilibrium) and kinetic differ-ence spectra measured in phosphate buffer, with a broadpositive peak centered around 440 nm and a shoulder at ca.432 nm. In the negative region, a broad peak centered around410 nm is observed. Only a barely visible shoulder is presentat ca. 420 nm, where the major mini-Mb-CO band is seen inphosphate buffer.

In the absence of glycerol, the difference spectra for thefirst three kinetic components do not show any particularlyremarkable feature (Fig. 3), all being very similar in terms ofpeak positions (ca. 420 nm and 440 nm) and isosbestic points(ca. 430 nm). Only component IV displays a moderatesplitting of the positive band with a structure that is affectedby the CO concentration (Fig. 3). Process V in phosphatebuffer has an isosbestic point at ca. 410 nm and a negativepeak at ca. 420 nm. The amplitude of all five processes isaffected by CO concentration (Table 1 and Fig. 3).

In buffer containing 75% glycerol, processes III and IVhave difference spectra similar to those of the correspondingcomponents detected in phosphate buffer alone, even thoughthe splitting of the positive band of process IV is morepronounced in the presence of glycerol (Figs. 3a and 4a). Onthe contrary, the spectroscopic properties of the second (II)and fifth (V) kinetic components are markedly influenced bythe addition of glycerol, whereas not much can be said forprocess I in view of the poor signal-to-noise ratio. In moredetails, process II in glycerol/phosphate buffer displays anegative band centered at ca. 410 nm, a positive one at ca. 426nm, and an isosbestic point at ca. 417 nm, while process V haspeaks at ca. 410 nm and 424 nm and an isosbestic point at ca.416 nm (Fig. 4b). The features of processes II and V in bufferwith 75% glycerol closely resemble those reported in theliterature (19) for CO binding to the heme-water complex(Fig. 4B, dashed line) and for the replacement of water byhistidine in the water-heme-CO complex (Fig. 4b, continu-ous line) except for a systematic shift of ca. 2 nm most likelydue to a mismatch between monochromators.

Kinetic traces for the reaction ofmini-Mb with CO in bufferwith 75% glycerol recorded between 40K and 180 K could besatisfactorily approximated only with a sum of three en-thalpy-distributed kinetic species (data not shown). The

00000***jR°369@92%*R FIG. 3. Difference spectra of the in-

iS! dividual kinetic components obtained- 7xKXXXXX) from global least-squares fittings of the

++4O kinetic traces recorded at different wave-+600 lengths in 0.1 M phosphate equilibrated6 o at 273 K with 1 atm (a) or with 0.5 atm (b)x ofCO. The spectra are normalized on the0 basis of the total to absorbance changes.

Spectral data shown for processes II (+)and III (*) are direct measurements,whereas the amplitudes of the spectrareferring to processes I, IV, and V have

430 440 450 been amplified as follows: I x S (o), IV x4 (o), and V x 10 (x).

Biophysics: Di lorio et al.

Proc. Natl. Acad. Sci. USA 90 (1993)

410 420 430 440 450 400Wavelength, nm

FIG. 4. Difference spectra of the ki-netic components obtained by deconvo-lution of the progress curves for the re-action of mini-Mb with CO in phosphatebuffer with 75% glycerol at 290 K. (a)Spectra ofprocesses I (o, amplitude x 5),HI (*), and IV (o, amplitude x 4). (b)Spectra of processes II (+) and V (x;amplitude x 10) along with those com-puted from the absolute spectra of thewater-heme complex minus the water-heme-CO complex (dashed line) and ofthe water-heme-CO complex minus thehistidine-heme-CO complex (continu-ous line) given in figure 7 of ref. 19.

enthalpy distribution of the major kinetic component closelyresembles that reported in the literature for free heme (20).

DISCUSSIONA striking feature ofthe data depicted in Fig. 1 is that as manyas five kinetic components, all exponential in time, areneeded to describe the absorbance changes following nano-second-flash photolysis of carbonyl mini-Mb. To clarify theorigin of such a complex situation, different measurementswere performed that varied either the free ligand concentra-tion or the solvent composition. In the absence of glycerol,the three intermediate kinetic processes (II-IV) follow bimo-lecular kinetics, as clearly indicated by the roughly 50%odecrease in their rate coefficients when the free CO concen-tration was reduced to about half (Table 1). Therefore, theycan be ascribed to ligand-binding steps. It is just as clear thatthe slowest process (V) is monomolecular, since its velocityis independent of free-ligand concentration (Table 1). Theresults are not conclusive for the fastest kinetic component (I)because the large error in estimates of the rate coefficients atthe two CO concentrations (Table 1) does not allow one todecide if process I is monomolecular or bimolecular. Aplausible explanation, consistent with these results, is thatphotolysis of mini-Mb-CO produces transient conforma-tional species that rebind CO with distinct rates (processes II,III, and IV) and then slowly reconvert to the equilibriumligated product (process V). The longer the protein stays inthe ligand-free form after photolysis, the larger will be thefraction of molecules that undergo the conformational tran-sition and, therefore, the larger will be the amplitude ofprocess V; this interpretation is supported by the inverserelationship between the relative amplitude of process V andCO concentration (Table 1 and Fig. 3). The available exper-imental data do not provide a conclusive explanation aboutthe nature ofthe three bimolecular processes; however, somepossibilities can be considered. One is related to the exposureof hydrophobic patches of the protein to the solvent as aconsequence of the removal of the two terminal segments.This situation could give rise to molecular aggregates ofdifferent size, and these in turn may account for the observedmultiplicity of bimolecular processes. However, this explana-tion is inconsistent with the dependence on CO concentrationof the amplitude of the three bimolecular processes (Table 1and Fig. 3), unless we postulate the very unlikely situation inwhich the equilibrium between different aggregation states isaltered when the CO concentration is lowered from ca. 1 mMto ca. 0.5 mM. A second interpretation is that the threebimolecular processes correspond to different conformational

states either of the CO-bound structure or of the photolyzedstate. This hypothesis may be tested by IR spectroscopy;however, because of its low solubility, mini-Mb cannot beobtained at the high concentrations needed for this type ofinvestigation. The already mentioned dependence of the rel-ative amplitude of the three bimolecular processes on COconcentration favors the second hypothesis (i.e., heterogene-ity of the photolyzed state) over the first one, especially whenone considers the occurrence of large structural transitionsupon ligand dissociation and recombination in mini-Mb.That changes in the ligation state are associated in mini-Mb

with large conformational transitions is indicated by severallines of evidence. The a-helical content in the equilibriumdeoxy derivative of mini-Mb is ca. 40%o and increases to ca.60%o upon CO binding (9), contrary to a-helical content in Mbor isolated hemoglobin chains. EPR investigations on mini-Mbreconstituted with Co-heme show that the signal in the giregion of the deoxy spectrum is much broader for the proteinfreshly reconstituted under anaerobic conditions than for thatobtained from the oxy derivative by treatment with dithionite(10). Fig. 2 shows that the discrepancy between the spectralproperties ofhorseMb (dashed line) and mini-Mb in phosphatebuffer is much smaller when the difference spectra are recon-structed from the to absorbance changes induced by flashphotolysis (continuous line with circles) and is larger whenobtained by subtraction ofthe equilibrium spectra ofthe deoxyand carbonyl derivatives of the protein (continuous line with-out symbols). This finding shows that after nanosecond-flashphotolysis, the protein does not assume the equilibrium deoxyconfiguration and that the photo-induced unliganded speciesare more similar to the equilibrium ligand-bound than to theligand-free form. Furthermore, the data of Fig. 2 support theCD observations that mini-Mb is structurally closer to itsparent protein horse-Mb in the carbonyl form than in its deoxystate. The presence of a monomolecular process (V) with avery low rate constant and an amplitude that plots inverselywith CO concentration (Table 1 and Figs. 1 and 3) is alsoindicative of large ligand-linked structural transitions. Fur-thermore, the bands in the kinetic difference spectrum ofmini-Mb are considerably broader than those in the equilib-rium and horse-Mb spectra; this indicates a larger contributionofinhomogeneous broadening to the kinetic spectral bands, inkeeping with a more distributed structure of the protein (16).The red shift of the positive band in the kinetic differencespectrum ofmini-Mb, compared with that ofhorseMb (Fig. 2),can be attributed to the fact that the deoxy heme in thetransient mini-Mb species is in a ligand-bound-like structuralarrangement but also could be related to an alteration of thedielectric constant within the heme pocket.

a

).1 - oO0

0

* *0

0*0 [l *

*4j j 3-*0 0%

o **0

0 **00

0 0

0 00

0

-O

400

2028 Biophysics: Di lorio et al.

Proc. Natl. Acad. Sci. USA 90 (1993) 2029

The kinetic measurements performed in 0.1 M phosphatebuffer containing 75% glycerol, though considerably morecomplex than those in buffer alone, also point to large,ligand-linked, structural changes in mini-Mb. Three of thefive processes needed for the deconvolution of the high-temperature CO-binding kinetics display peculiar features.We start with component II, which accounts for a largefraction of the total absorbance changes induced by photol-ysis (Fig. 4b). Its difference spectrum is very close to thatreported in Fig. 4b as a dashed line, which refers to thebinding of CO to the heme-water complex detected as atransient when histidine-heme-CO is subject to photolysis(19). Furthermore, the difference spectrum of the slowestkinetic component, obtained from the deconvolution of theCO rebinding kinetics to mini-Mb in the presence of 75%glycerol (process V in Fig. 4b), closely resembles that ob-served when water is replaced by histidine as the axial ligandof a heme-CO complex (shown in Fig. 4b as a continuousline; ref. 19). Finally, the rate coefficient listed in Table 1 forprocess I in the presence of 75% glycerol is practicallyidentical to that reported in the literature for the binding ofwater to the tetracoordinated heme (19). Thus, it appears thatwhen the conversion of mini-Mb from the ferric to thecarbonyl derivative is carried out in the presence of 75%glycerol, the protein remains in a more unfolded state wherephotolysis induces the rupture ofboth axial heme bonds-theone with CO and that with the proximal histidine. Furthersupport to this interpretation is given by flash-photolysismeasurements at cryogenic temperatures (data not shown),since the major kinetic component obtained from the CObinding traces measured between 40 K and 180 K is charac-terized by an enthalpy distribution similar to that previouslyreported for free heme (20).None of the kinetic processes used for the deconvolution of

the CO binding kinetics to mini-Mb in phosphate bufferindicates rupture upon photolysis of the bond between theheme and the proximal histidine. Only processes III and IVappear to be of the same nature in the presence and absenceof glycerol, as can be judged from their spectral properties. Inparticular, process IV displays difference spectra alwayscharacterized by a splitting of the positive band, with variousrelative intensities of the two peaks, when the CO concentra-tion or the solvent composition is changed (Figs. 3 and 4a), asif it were a mixture between two interconverting speciescharacterized by different spectra but with the same bindingrate.The scenario that we propose from the above discussion

can be summarized as follows. In aqueous phosphate buffer,processes II, III, and IV represent CO binding to structuralintermediates between the equilibrium CO-bound and thedeoxy conformations of mini-Mb; process V is a monomo-lecular rearrangement of the protein, which brings it back tothe CO-bound equilibrium configuration; and process I cor-responds to geminate CO recombination. In the presence of75% glycerol, the conversion of the heme from the met to thecarbonyl form is accompanied by only partial refolding of theprotein compared with that in buffer alone (9), possiblybecause of the high viscosity of the solvent. Under theseconditions, photolysis of CO from the heme-iron of onlypartially refolded molecules is associated with rupture of theproximal bond with His-F8 accompanied by binding of wateras the proximal ligand (process I), CO combination to theheme-water complex (process II), and replacement of theproximal water by His-F8 (process V); in contrast processesIII and IV refer to binding of CO to structural intermediates,analogous to those observed in the absence of glycerol.

CONCLUDING REMARKSAn important aspect of this investigation on mini-Mb isrelated to its similarity with the protein fragment encoded by

the central exon of the Mb gene. The results reported hereand those of previous investigations (9, 10) show that thecentral core of the Mb molecule has the features of a"conformational domain" responsible for the transmission tothe whole protein of changes in the ligation state at theheme-much in keeping with the "protein-quake" modelproposed by Ansari et al. (21). However, the amplitude ofthestructural fluctuations accompanying ligand binding and dis-sociation to this central core is too large and renders thesystem kinetically ineffilcient and structurally unstable. Ad-dition of the two terminal polypeptides (coded by exons oneand three) reduces the structural degrees of freedom of thecentral domain, thus producing a damping effect on theamplitude of the conformational changes and largely improv-ing the efficiency of the protein as a biochemical machinery.This interpretation implies that a critical mass is needed toachieve optimal protein function. Furthermore, in view oftheknown similarity between Mb and the Hb chains, in terms ofboth their overall protein fold and structure of their genes,this correlation between structure and dynamics of the pro-tein domains may be crucial to the onset of allosteric controlmediated by quaternary changes.

The authors thank Anton Lehman for his skillful technical help.This work has been supported by Swiss National Science FoundationGrant 31.9423.88 and by the Eidgenossische Technische HochschuleSpecial Credit 03586/41-1080.5, by the Italian "Ministero per l'Uni-versita e per la Ricerca Scientifica e Tecnica" (40%6 "Live protein"),and by a short term fellowship from the "SocietA Italiana di Bio-chimica" to G.D.S.

1. Case, D. A. & Karplus, M. (1979) J. Mol. Biol. 132, 343-368.2. Parker, M. W., Tucker, A. D., Tsernoglou, D. & Pattus, F. (1990)

Trends Biochem. Sci. 15, 126-129.3. Lewin, R. (1983) Science 219, 1312-1312.4. Blanchetot, A., Wilson, V., Wood, D. & Jeffreys, A. J. (1983)

Nature (London) 301, 732-734.5. G6, M. (1981) Nature (London) 291, 90-92.6. Eaton, W. A. (1980) Nature (London) 284, 183-185.7. Craik, C. S., Buchman, S. R. & Beychok, S. (1980) Proc. Natl.

Acad. Sci. USA 77, 1384-1388.8. De Sanctis, G., Falcioni, G., Giardina, B., Ascoli, F. & Brunori, M.

(1986) J. Mol. Biol. 188, 73-76.9. De Sanctis, G., Falcioni, G., Giardina, B., Ascoli, F. & Brunori, M.

(1988) J. Mol. Biol. 200, 725-733.10. De Sanctis, G., Falcioni, G., Grelloni, F., Desideri, A., Polizio, F.,

Giardina, B., Ascoli, F. & Brunori, M. (1992) J. Mol. Biol. 222,637-643.

11. Dickerson, R. E. & Geis, I. (1983) Hemoglobin: Structure, Func-tion, Evolution, and Pathology (Benjamin/Cummings, Menlo Park,CA).

12. Craik, C. S., Buchman, S. R. & Beychok, S. (1981) Nature (Lon-don) 291, 87-90.

13. Pharmacia Fine Chemicals (1982) Isoelectric Focusing: Principlesand Methods (Ljungfortagen AG, Orebro, Sweden).

14. Antonini, E. & Brunori, M. (1971) in Hemoglobin and Myoglobin inTheir Reactions with Ligands, eds. Neuberger, A. & Tatum, E. L.(North-Holland, Amsterdam).

15. Rossi Fanelli, A., Antonini, E. & Caputo, A. (1958) Biochim.Biophys. Acta 30, 608-615.

16. Di lorio, E. E., Hiltpold, U. R., Filipovic, D., Winterhalter, K. H.,Gratton, E., Vitrano, E., Cupane, E., Leone, M. & Cordone, L.(1991) Biophys. J. 59, 742-754.

17. Richter, C. & Di Iorio, E. E. (1991) in Frontiers in Biotransforma-tion: Membrane Organization and Phospholipid Interaction ofCytochrome P450, eds. Ruckpaul, K. & Rein, H. (Akademie,Berlin), Vol. 5, pp. 72-93.

18. Bevington, P. R. (1969) in Data Reduction and Error Analysis forthe Physical Sciences (McGraw-Hill, New York), pp. 195-202.

19. Huang, Y., Marden, M. C., Lambry, J. C., Fontaine-Aupart, M. P.,Pansu, R., Martin, J. L. & Poyart, C. (1991) J. Am. Chem. Soc. 113,9141-9144.

20. Alberding, N., Austin, R. H., Shirley, S. C., Eisenstein, L.,Frauenfelder, H., Gunsalus, I. C. & Nordlund, T. M. (1976) J.Chem. Phys. 65, 4701-4711.

21. Ansari, A., Berendzen, J., Bowne, S. F., Frauenfelder, H., Iben,I. E. T., Sauke, T. E., Shyamsunder, E. & Young, R. D. (1985)Proc. Natl. Acad. Sci. USA 82, 5000-5004.

Biophysics: Di lorio et al.