Raman dispersion spectroscopy probes heme distortions indeoxyHb-trout IV involved in its T-state Bohr effectReinhard Schweitzer-Stenner, Michael Bosenbeck, and Wolfgang DreybrodtInstitute of Experimental Physics, University of Bremen, 2800 Bremen 33, Germany

ABSTRACT The depolarization ratios of heme protein Raman lines arising from vibrations of the heme group exhibit significant depen-dence on the excitation wavelength. From the analysis of this depolarization ratio dispersion, one obtains information about symmetry-lowering distortions 5Qr of the heme group that can be classified in terms of the symmetry races r = Alg, Blg, B2g, and A2g in D4hsymmetry. The heme-protein interaction can be changed by the protonation of distinct amino acid side chains (i.e., for instance the Bohrgroups in hemoglobin derivates), which gives rise to specific static heme distortions for each protonation state. From the Ramandispersion data, it is possible to obtain parameters by fifting to a theoretical expression of the Raman tensor, which provide informationon these static distortions and also about the pK values of the involved titrable side chains. We have applied this method to the v4 (1,355cm-') and vlo (1,620 cm-') lines of deoxygenated hemoglobin of the fourth component of trout and have measured their depolarizationratio dispersion as a function of pH between 6 and 9. From the pH dependence of the thus derived parameters, we obtain pK valuesidentical to those of the Bohr groups, which were earlier derived from the corresponding 02-binding isotherms. These are pKa1 = pKa2 =8.5 for the a and pK,,1 = 7.5, pK2 = 7.4 for the ,3 chains. We also obtain the specific distortion parameters for each protonation state. Asshown in earlier studies, the v4 mode mainly probes distortions from interactions between the proximal histidine and atoms of the hemecore (i.e., the nitrogens and the C. atoms of the pyrroles). Group theoretical argumentation allows us to relate specific changes of theimidazole geometry as determined by its tilt and azimuthal angle and the iron-out-of-plane displacement to distinct variations of thenormal distortions bQr derived from the Raman dispersion data. Thus, we found that the pH dependence of the heme distortions 6QA1g(totally symmetric) and 6QB1g (asymmetric) is caused by variations of the azimuthal rather than the tilt angle of the Fe-His (F8) bond. Incontrast to this, the v10 line mainly monitors changes resulting from the interaction between peripheral substituents of the porphyrinmacrocycle (vinyl). From the pH dependence of the parameters, it is possible to separately identify distortions 6Qr affecting the hemesin the a and , chains, respectively. From this, we find that in the a subunit structural changes induced on protonation of the correspond-ing Bohr groups are mainly transferred via the Fe- N, bond and give rise to changes in the azimuthal angle. In the : subunit, however, inaddition, structural changes of the heme pocket arise, which most probably result from protonation of the imidazole of the COOH-ter-minal His (HC3 13). This rearranges the net of H bonds between His HC3 f3, Ser (F9 (3), and Glu (F7 (B).

INTRODUCTIONThe understanding of the molecular basis of the func-tionally relevant allosteric mechanism (i.e., cooperati-vity, Bohr and Root effect) in hemoglobin (Hb)' mole-cules is one ofthe major goals ofthe research work dedi-cated to these proteins. Numerous experimental andtheoretical studies have provided substantial insight intothe basic mechanism governing cooperativity of ligandbinding to various Hb derivates ( 1-13 ). Our knowledgeabout the processes that transduce structural changes in-duced by proton binding to amino acid residues in-volved in the Bohr ( 14, 15) or Root effect ( 16), however,is still rather limited (17-19). Proton binding to Bohrgroups, for example, may influence the apparent oxygenaffinity of Hb by two different processes. It eitherchanges the equilibrium between the low affinity tertiary

'Abbreviations used in this article: CTT, chironomous thummithummi; deoxyHbA, deoxygenated human hemoglobin; deoxyHb-trout IV, deoxygenated hemoglobin of the fourth component of trout;DPD, depolarization ratio dispersion; DPR, depolarization ratio; Hb,hemoglobin; HbA, human hemoglobin; HLC model, Heme-ligandcoupling model; oxyHbA, oxygenated human hemoglobin; oxyHb-trout IV, oxygenated hemoglobin of the fourth component of trout;REP, resonance excitation profile; RTC, Raman titration curve; Tris,tris(hydroxymethyl)-aminomethane.Address correspondence to Dr. Reinhard Schweitzer-Stenner,Institute of Experimental Physics, University of Bremen, 2800Bremen 33, Germany.

t and the high affinity tertiary r state or directly affectsthe ligand affinity of the heme within a given t or r state( 1, 2, 10). Whereas in the former case the heme groupsin the deoxygenated (deoxy)Hb-t and oxygenated(oxy)Hb-r subunits remain unaffected by pH, somestructural changes of prosthetic groups can be expectedto occur in the latter case.

Such distortions of the active sites can be detected byresonance Raman dispersion spectroscopy (RDS) (20).It involves the measurement ofpolarized resonance exci-tation profiles (REPs) ofRaman lines arising from skele-tal modes of the porphyrin in the region between the Band Q absorption bands. From these data the corre-sponding depolarization ratio dispersion (DPD) can bederived. The latter reflects asymmetric distortions oftheporphyrin macrocycle caused by the peripheral substitu-ents attached to its ,B carbons and various noncovalentheme-protein interactions (3, 21-24). An analysis ofthese data in terms of an earlier derived theory on Ra-man scattering in the weak coupling regime (20) yieldsvibronic coupling parameters that depend on symmetryclassified normal distortions of the heme. Proton bind-ing to amino acid side chains may cause changes in theprotein conformation that are transferred to the hememoiety, thus giving rise to distortions ofthe heme groupspecific to each titration state. In this case the total scat-tered Raman intensity has to be considered as an incoher-

0006-3495/93/04/1194/16 $2.00 Biophys. J. © Biophysical Society

Biophys. J. Biophysical SocietyVolume 64 Apnl 1993 1 194-1209

1194 0006-3495/93/04/1194/16 $2.00

ent superposition of the scattering at proteins existing indifferent titration states, the concentration of which isdetermined by pH, and the pK values of the above sidechains. As a consequence the DPD becomes pH depen-dent, and the pKvalue ofthe distorting protonation pro-cesses can be derived by analyzing the pH dependence ofthe corresponding effective vibronic coupling parame-ters (20).As previously shown, RDS can be applied successfully

to elucidate various aspects of allosteric interactions be-tween protein binding sites and the porphyrin macrocy-cle in Hb molecules (20). To this end, we measured theDPDs of the oxidation marker line V4 and the spinmarker line v10 of their Raman spectra. For human he-moglobin (HbA), we found that protonation of Bohrgroups (i.e., for instance the COOH-terminal His HC3d) induces significant symmetry-lowering distortions ofthe heme groups in oxyHbA (19), whereas deoxyHbAremains unaffected (24). On the other hand, it is knownthat all the corresponding Adair constants depend on pH(25), indicating that the Bohr effect is operative in the Tand R state of the protein. Thus, our results suggest thatthe mechanisms governing the Bohr effect are differentin the quaternary states T and R. The T-state Bohr effectchanges mainly the equilibrium between the t and r

states of the subunits (i.e., proton binding stabilizes t),whereas the R-state Bohr effect directly affects the con-

formation of the ligated r state ( 18, 19). RDS experi-ments on oxyHbA derivatives (26-28) suggest that inthe R-state Bohr effect, allosteric coupling mechanismsare involved in the ,B rather than in the a subunits. This iscorroborated by earlier time-resolved Raman experi-ments on photodissociated HbA hybrids a(Fe) f(Co)2and a(Co)2 f(Fe) * that revealed different time depen-dences of the apparent frequency of the structural sensi-tive Fe-N, (His F8)-stretching mode at different pHon relaxation of the protein structure after photolysis( 17, 29). Although the relaxation into the deoxygenatedT state was found to be pH independent for the a sub-units, it exhibits a considerable pH dependence for the fi

subunits. Moreover, RDS experiments on isolatedoxy-HbA chains revealed that the DPDs ofthe oxidationand spin marker lines V4 and v10 of the a chain vary as afunction ofpH, whereas the corresponding DPDs ofthea chain are pH independent (30). In the physiologicalrelevant region between pH 6 and 9, distorting interac-tions between proton binding sites and heme groupsseem to be confined to those amino acid residues that areinvolved in the allosteric linkage between proton andligand binding (i.e., Bohr and Root groups) ( 1, 2, 5, 10,16, 31, 32). Ifthe latter is absent as is the case, for exam-ple, in myoglobin (33) and Hb-BME (bis(N-maleimido-dimethyl )ether; cf. reference 34), the DPDs ofthe corre-

sponding V4 and v10 lines are found to be pH independentbetween 6 and 8 (26, 27, 35, 36).

Recently, we also investigated the influence of protonbinding to the structure of oxyHb of the fourth compo-

nent oftrout (oxyHb-trout IV). This particularHb mole-cule is of considerable interest because its binding iso-therms for 02 and CO indicate a complete breakdown ofcooperativity at pH below 6.5 (Root effect) ( 16, 32, 37,38). The analysis of these binding curves for various pHbetween 6.1 and 7.7 (16, 37) by virtue of an extendedHerzfeld-Stanley model (2) reveals that the fully oxy-genated molecule undergoes a conformational R -* Ttransition on lowering the pH below 7.0 (39). This canbe understood in terms ofproton binding to yet unidenti-fied amino acid residues (Root groups), which lower thefree energy of cooperativity by stabilizing the quarter-nary T state. In addition, proton binding to Bohr groupslowers the apparent ligand affinity at pH above 7.0.The deoxygenated molecule ofHb-trout IV is in the T

state, and all its subunits exhibit their tertiary structuresin the t state, independent ofpH. In contrast to what hasbeen found for HbA, the Bohr effect in Hb-trout IV de-pends on the nature ofthe ligand ( 16, 32, 37). This indi-cates that in contrast to what has been observed for HbA,the Bohr effect of Hb-trout IV operates by direct cou-pling between proton and ligand binding sites (2, 39).Consequently, one expects that this type of allosteric in-teraction is involved in both, the T- and R-state Bohreffect of Hb-trout IV.To investigate the corresponding allosteric coupling

mechanism, RDS was first applied to oxyHb-trout IV atpH between 6.7 and 8.5. From the pH dependence ofthevibronic coupling parameters determining the Ramanintensity of the v4 mode, we found that in particular theprotonation of the Root groups cause a considerablystrong asymmetric Blg-type perturbation of the pyrrolenitrogen N(III) of the porphyrin macrocycle (Fig. 1) inthe ,B subunits. The underlying mechanism involves ther -* t transition of the ligated A subunits that forces theimidazole of the corresponding proximal His F8 into a

more tilted position with respect to the heme normal,thus increasing repulsive interactions between C. of theimidazole and N(III) ofthe heme. The a subunits do notexhibit these conformational changes because the qua-ternary transition is not capable to switch their tertiarystructures from r -* t in the fully oxygenated state (40).

As mentioned above, deoxyHb-trout IV is in the qua-

ternary T state with its subunits in the t state indepen-dent of pH. Each of the subunits carries two Bohrgroups. For the deoxygenated a subunits, the pK valuesderived from the binding isotherms are pKa1 = pKa2 =

8.5 and correspondingly for the : subunits pKlI = 7.4and pK2 = 7.5 (39). Since no changes of quaternary-tertiary or tertiary-tertiary interaction occur on varia-tion ofpH (39), each subunit can be regarded as isolatedfrom the others with respect to the conformationalchanges caused by the protonation of one of the aboveBohr groups. For this reason, deoxyHb-trout IV is a suit-able system to study the heme-protein coupling mecha-nism involved in possible direct interactions betweenBohr groups and heme within the same subunit.

ScwizrSenre l eeDstrin nDoyH ru V19

Schweitzer-Stenner et a]. Heme Distortions in Deoxy Hb-trout IV 1195

Recently, Bosenbeck et al. (41) have measured theband shape of the Raman line resulting from theFe NE (His F8) stretching mode in deoxyHb-trout IVas a function of pH. They found that this band changesits line shape with its peak position shifting from v = 217cm-' at pH = 6.1 to 223 cm-' at pH = 8.0. Such shiftsseen earlier by many authors (42) have been attributedto changes from the T structure to the more relaxed Rstructure. Thus, a contradiction arises, because it hasbeen shown from the thermodynamic analysis ofthe 02-

and CO-binding curves that the molecule remains in itsquaternary T and tertiary t structure irrespective ofpH(39). This has been resolved by analysis ofthe line shapeof the VFe-His Raman line, which could be decomposedinto five sublines at 202, 211, 217, 223, and 228 cm-.The frequencies of these sublines remain fixed on varia-tion of pH. Their intensities, however, show significantchanges. The latter could be explained by assuming thateach subline is a superposition of sub-sublines, withidentical frequencies but different Raman scattering in-tensities, due to different protonation states of the sub-units. By using a titration model to calculate the molarfractions of the four different protonation states withinthe a and d subunits, all the intensities ofthe sublines intheir dependence on pH could be fitted with the pK val-ues introduced above. It turned out that the lines at 202and 21 1 cm-l showed variations of intensities governedby pK values of 8.4/8.5 and could therefore be assignedto the a subunits, whereas pK values of 7.4/7.5 resultedfrom the intensity variations of the sublines at 223 and227 cm-', relating these lines to the A subunits. Theseresults were interpreted by use of a model proposed byBangcharoenpaurpong et al. (43) and recently modifiedby Friedman et al. (44), which describes the couplingbetween the T* bond of Fe2+- NE and the lr* orbitals ofthe pyrrole nitrogen atoms in terms of the polar angle 0of the Fe- NE bond with respect to the normal of theheme-plane and the azimuthal angle sp between theN(I) Fe N(III) direction in the heme-plane andthe projection of the Fe-NE direction onto this plane.In this picture the sublines are due to conformationswith different values of the polar angle 0, and the sub-sublines are due to conformations at a given 0 but withdifferent values of the azimuthal angle (p (Fig. 1 ) due toconformational changes on titration ofthe Bohr groups.

The frequency of the subline is mainly determined bythe polar angle, whereas a decrease in the azimuthal an-

gle enhances the intensities of the corresponding lines.Although the VFe-His band ofHbA also has been found tobe composed of different sublines, their intensities re-

main pH independent between 6 and 9 at low Cl--con-centrations, in accordance with the results obtainedfrom earlier RDS studies (24).

Thus, a considerable heterogenity is present in theheme-protein linkage in Hb-trout IV, which is in-fluenced by proton binding to Bohr groups. This clearlysuggests that in contrast to HbA the latter is also directly

coupled to the binding sites in the deoxygenated t state.Now the question arises as to how this affects the struc-ture ofthe prosthetic heme groups. To address this point,we measured the depolarization ratio dispersion ofthe V4and v10 mode of deoxyHb-trout IV as a function of pH.They can be shown to monitor structural changes of theheme group due to protonation of the above Bohrgroups. Our results detail the role of the Fe-His F8linkage in this direct type of allosteric interaction.

THEORETICAL BACKGROUNDDerivation of the polarizibility tensorTo describe the experimentally obtained REPs andDPDs, it is necessary to extend the theory of Peticolas etal. (20), which is based on Loudon's formalism (45)into fifth order. This accounts for the vibrational side-bands of the B and Q bands by considering the creationand subsequent annihilation of vibrations giving rise toabsorption in these bands. Symmetry perturbationsfrom D4h are introduced into this formalism by expand-ing the vibronic coupling operator in the Hamiltonianwith respect to equivalent normal distortions 3Qr. Thisleads to the expression

aH__

C2HOH,>eR OHrR + E Qri, (la)OQr, aQrR 6Q= C QrROIQrj 6=

aH aH a2H .5Ii (bQpeQrd 6Q=O rQ Q =o

RQR and 3Qr relate to the normal coordinate ofa Ramanvibration and the second vibrational quantum, respec-tively. 3Qf represents normal distortions of symmetrytype rj, which can be written as

jQrj = z bQlri (2)

where 6Qf represents distortions proportional to the am-plitudes of various different normal coordinates of thesymmetry type Fi = A,g, Big, A2g, and B2g. Introducingthis into the fifth-order expansion of the Raman tensoryields the perturbed tensor as

#pa z #pMg es T)es]e,s=Q,B r

+ 1A, MGE es~Ie,s,t,u=Q,B r

X( cst T.(tug)

+ (2 crs;t )(zf CrtR )

x ( crtT)F2 + ( crst )

(3)

where ,ugp,, g are dipole transition matrix elements con-necting the electronic ground state and the excited elec-

1196 Biophysical Journal Volume 64 April 1993

x r,A,.fr 2: cr,Rtr F32: cst tur r

1196 Biophysical Journal Volume 64 April 1993

tronic states Ie >, s >. (p, a = x, y, z label the polarizationstate of the transition).The frequency functions F, F1, F2, and F3 are defined

by

F = {(e + OR VL + iY )Ckj- VL + ijS)l1F, = {(Pe + UR - VL + ijY)(vs- VL + iY )

X (Pt + ,- VL + iYt)(vu - VL + j,y)}1

F2 = {(Pe + OR VL + iye)(vs + UR + O - VL + ijS)X (Pt + Q- VL + iyt)(Vu - VL + ijY)},

F3 = {(We + OR- VL + i_ye)(Vs + UR + A VL + iys)X (Pt + UR - VL + iYt)(vu - VL + iYu)}1* (4)

Antiresonant terms have been omitted for simplicity; ve,vS, vP, vu are the wavenumbers of the electronic transi-tions from the groundstate Ig> into le>, Is>, It>, andu > excited electronic states, which are indicating Gou-

terman's unmixed Q and B states of the porphyrin sys-tem (46), respectively. )ye, ys, y t, and y label the corre-sponding halfwidths, VL is the frequency of the incidentlaser light, QR( Q,,) is the frequency of the Raman mode(the second vibration). The constants cr?R, cer, are re-lated to the following matrix elements of the vibroniccoupling operators described by Eqs. la and lb:

( I =aOH a2H .|Qrj sQOI,Ce,s e 0gQrR IQO+ RQR

( =5eOH a2 bQrj s/Q0i. (5)

Q = 11 QrR 10> and QO' label the vibrational matrixelements. rR( FA) are the representations of the Ramanmode (second vibration) andrj that for the normal dis-tortion 6Qrj. r labels the product representations rR x

rj and rF x rj for each rF.It should be noted that the above matrix elements are

related to the distortion ofthe equilibrium configurationof the excited electronic states e> and s> with respectto the electronic ground state along the normal distor-tions bQri. Consequently, variations in cr may reflectconfigurational changes in both the ground state and theexcited electronic states involved (47). The vibroniccoupling parameters can be classified as follows:(a) c"g and c"g reflect intrastate Frank-Condon (FC)coupling in the Q and B state,

(b)CQB(F=Q Al, B15, B2g, A2g) describes interstateHerzberg-Teller coupling between Q and B, and(C) CQQ and CBB (r = B15, B2g) are related to intrastateJahn-Teller coupling within Q and B.Using the above expression for f3, and Placzek's (48)

formalism for calculating REP and DPD of distinct Ra-man lines from their tensor components, we are able tofit the theory to the experimental data of the polarizedREPs with one common set of fit parameters crR andcer, (e, s = Q, B; r= Al., B,g, A2g, and B2g). Theseparameters c~er,1Ic' are in first order linearly related to

the perturbations 6Q'j. Thus, they contain informationabout possible asymmetric heme-protein interactions.

It should be mentioned that in case of strict D4h sym-

metry, the depolarization ratio is p = 0. 125 for A1g lines,p = 0.75 for B1g and B2g lines, and p = oo for A2g linesindependent of the excitation wavelengths (49). Sym-metry-lowering distortions 3Qr introduce dispersion ofthe depolarization ratio, which by this mechanism is anideal tool to detect interactions between the heme groupand the protein inducing such distortions.So far the theory considers only one conformational

type of Hb molecules to be present. In reality, due tovarious subunit heterogenity protonation processes andconformational transitions, many conformational statesof the molecules with different polarizibility tensors aresimultaneously present in the solution. The total scatter-ing intensity is therefore an incoherent superposition ofthe intensities of these differing species of molecules.Therefore, as has been shown elsewhere (20), each ten-sor element of the effective Raman tensor can be ex-pressed by:

1/2

II3eIffl = X )

I(6)

where X1 denotes the mole fraction of the Ith conforma-tion, the corresponding Raman tensor is denoted by(fl,)1. Some straightforward but lengthy calculationslead to the following expression for the effective distor-tion parameters, which are the result of a fitting proce-dure of the theoretical expression to the observed REPsand DPD (20).

I 1/2(p 1/2

I e,sP ) = 2: XI(Cel,s 1l

(7)

(Crr,) are the vibronic coupling parameters of the Ra-man fundamental related to the Ith conformation.

MATERIALS AND METHODS

Preparation of Hb-trout IVThe blood for the preparation was obtained from commercially pur-chased trout salmo iredius. It was washed several times with 0.9%aqueous solution of NaCI containing ethylenediaminetetraacetate toavoid coagulations and afterward hemolized with distilled water. Theobtained Hb solution was then dialyzed against 0.1 M tris(hydroxy-methyl)-aminomethane (Tris)-HCI buffer at pH 9.1. After equilibra-tion it was applied to a DEAE-Sephadex A50-column from SigmaChemical Co. (St. Louis, MO) (dimension 40 x 2 cm). The pH gra-dient elution was carried out using two containers, one with 0.1 MTris-HCI buffer at pH 9.1 and the other with 0.1 M KH2PO4 at pH 4.5.The different pH values were adjusted dialyzing against 0.1 M Tris-HCIbuffer (pH > 7.0) and 0.1 M bis-Tris-HCI buffer (pH < 7.0). Thesamples were deoxygenated by adding a few grains ofNa2S204 and werethermostated at 4 ± 0.5°C. The concentration ofeach sample was 1.0 x10-' M and monitored by measuring the optical absorbance with anHP-diode array spectrometer.

Experimental arrangementThe exciting radiation was obtained using an argon ion laser fromSpectra Physics (Darmstadt, Germany). The laser beam was polarized

Schweitzer-Stenner et al. Heme Distortions in Deoxy Hb-trout IV 1197Schweitzer-Stenner et al. Heme Distortions in Deoxy Hb-trout IV 1197

perpendicularly to the scattering plane and focused by a cylindrical lensonto a sample. The Raman radiation was measured in backscatteringgeometry. A polarization analyzer between sample and spectrometerenabled us to measure the intensity ofthe two components perpendicu-lar and parallel to the scattering plane. A polarization scrambler wasused to avoid different transmissions of the spectrometer for differentpolarization. The Raman radiation was analyzed with a Czerny-Turner double monochromator (Spex, Munich, Germany) collectedby a photocounting system (Ortec, Munich, Germany) and digitizedby a microcomputer, where it was stored for further analysis. To calcu-late the correct height of the Raman lines, a program was used to sub-tract fluorescence background and to decompose complex spectra intodistinct Lorentzian lines of defined widths and heights. The intensitiesof the Raman lines were compared with the intensity of the SO42- lineat 980 cm-' in a separate sample containing (NH4)2SO4 and 1 mMdeoxyHb. By this method, the S042- line was used as internal standardfor correction of the Raman intensities with respect to laser power andtransmission of the spectrometer. Since at the given concentration ofdeoxyHb corrections for absorption are not necessary (35), we did notcorrect the intensities of the SO42- line and of the Hb lines for differentabsorption due to their differing wavelengths ofthe Raman light. Depo-larization ratios ofthe Raman lines were measured in the absorption ofthe Q band by using an excimer laser (EMG MSC; Lambda Physics,Gottingen, Germany) and pumped dye laser (model FL 2001; LambdaPhysics) with a conventional Raman spectrometer ( 1877 Triple Mate;Spex) and a diode array camera (0-SMA; Spectroscopy Instruments)for detection of the signal. Details are given by Bobinger et al. (50).

Fitting procedureA program called MINUITL from the CERN library (ComputerCenter, University of Bremen) was used in a least-squares fit to theexperimental DPDs and REPs. In a first approximation, all fits werecarried out in the framework of Gouterman's four-orbital model,which imposes the following restrictions on the vibronic coupling pa-rameters (49):

cBig = CBlg; cB2g = -Msg (8)

In most cases a satisfactory fit could not be obtained under these restric-tions. Consequently, they have been relaxed, and cr and CrB (r = Big,B2g) were used as free parameters. The quality of the fits was judged bymeans of the error value calculated byf, which is a function of the fitparameters

f= f2 xi2(fv 1/2 (9)

where x? is the x2 value of the ith data point with respect to the fit andis its statistical error. N denotes the number of data points. The best

fit is selected by the minimum of fin dependence on the fitting parame-ters. To estimate the statistical errors of the vibronic coupling parame-ters, we calculated the normalized error valuef/fmin as a function ofthedistinct parameters cr in the vicinity of the respective minimum of thex2 function. It turned out that inappropriate fits are obtained forf/lfilarger than 1.02. Thus, we considered bcr = cr (fmin)Cr( 1.02 fmin ) as the statistical error of the vibronic coupling parame-ter cr.

Since Bjg7 and B2.-type contributions to the polarizability tensor can-not be discriminated by polarization experiments on randomly ori-ented molecules, we omitted all CBlg parameters in the fit to the data.Thus, the vibronic coupling parameters cBlg may also reflect some

minor contributions from B24-type coupling to the scattering tensor.B,l-type coupling, however, can be expected to be predominant forboth lines investigated, because the vio mode exhibits Bj, symmetry inD4h, and the Np atoms mainly involved in the heme-protein interac-tion affecting the v4 mode cannot contribute to BU-type coupling. Toexpress the coupling parameters cr in absolute units (cm- ), we consid-ered explicitly their contribution to the fifth-order term ofthe polariza-bility tensor as described in detail by Schweitzer-Stenner et al. (51 ) and

Bobinger et al. (52). The remaining multimode contributions to thefifth-order term are represented by an effective helping vibration (51 ).Our experimental data sets cover the region between 22,000 and

18,900 cm-'. The vibronic couplingparameters determine these DPDsand REPs to a different extent. For the V4 mode, CQB and Cr providelarger contributions to its Raman cross-section than crQ. In contrast,CQQ and CQB mainly determine the cross-section of v10, for whichCBB isof minor importance. As a consequence, statistical errors of the cBB(v1o) and cr (v4) values are largerthan those ofthe remaining parame-ters. The accuracy to which the cr can be determined is demonstratedby earlier Raman dispersion studies on oxyHb-BME (26, 28). Sincethe Bohr effect is absent in this modified Hb, the DPDs and REPs ofv4and v10 exhibit only small variations in dependence of pH, which arewithin the experimental errors. The corresponding cr also show onlyslight changes with pH (26).

RESULTS

Depolarization dispersion of the v4modeWe have measured the polarized REPs of the V4 line ofthe porphyrin macrocycle at 1,355 cm-'. The line resultsfrom an Alg normal vibration determined by a halfpyrrole stretching vibration consisting of out-of-phasemotions of N- Ca and Ca- C,0 bonds (53 ). Its disper-sion ofthe DPD mainly probes distortions ofthe pyrrolenitrogens and the Ca carbons caused by their interactionwith the Fe2" ligands (20, 54). This is due to the largeamplitudes ofthe involved N- Ca stretching vibrations.Since by group theory the symmetry r of the vibroniccoupling matrix elements cr is related to the representa-tion rR ofthe Raman mode and the representation rj ofthe distortion 6Q"i by r = rR X rj, CAlg elements reflectdistortions of Alg symmetry. Consequently, the matrixelements CB'g and CA2g result from symmetry-loweringdistortions of Big and A2g type, respectively.As shown in a recent article (54), the above Blg and

A2g distortions can be related to the geometry of theheme-Im F8 complex (shown in Fig. 1) by means ofthefollowing coupling model. If the imidazole eclipses theline N(I)- Fe2+-N(III) (azimuthal angle (p = 0), Bjg-type distortions are induced by repulsive interactions be-tween N(I), N(III) ofthe porphyrin, and CE, C, ofIm F8( 3, 21 ), respectively. They increase with increasing tiltangle 0 and decreasing Fe displacement (. If the azi-muthal angle of the proximal imidazole increases, A2g-and B2g-type distortions are induced by perturbations ofthe a carbons, whereas the Bjg distortions decrease, ow-ing to the reduced perturbations ofthe pyrrole nitrogens.At a fixed azimuthal angle 4), the magnitude of A2g andB2g distortions depends on the tilt angle 0 and the Fedisplacement (. This model is refered to as heme-ligandcoupling (HLC) model.The fits to the DPDs and REPs were performed using

the vibronic coupling matrix elements crs as free parame-ters. To express them in absolute units, we consideredexplicitly their contribution to the fifth-order term ofthepolarizability tensor and the vibronic side band Q, oftheoptical absorption spectrum as described in detail by

1 no _.J 1119B Biophysical Journal Volume 64 April 1993

Im (His F8)

/(III)

FIGURE I Schematic representation ofthe heme-N,( His F8) complex.N(I), N(II), N(III), and N(IV) are the pyrrole nitrogens, e is the tiltangle ofthe Fe2+- Nf(His F8) bond with respect to the heme normal,and 4) is the azimuthal angle formed by the line N(I)- Fe- N(III)and the projection of the imidazole of the heme.

Schweitzer-Stenner et al. (51). Electronic distortionshave not been considered explicitly because their deter-mination requires REPs and DPD data covering the Q0-band region (54). To derive the correct energy positionsand the corresponding transition dipole moments oftheQ and B band (cf. Eq. 3), the latter were deconvolutedinto two Lorentzians according to the Q0 and Qv bandsand the Bo and B, bands, respectively. By integratingover the Lorentzians ofthe Q0 and Bo bands, the respec-tive dipole transition moments were derived. The param-eters thus obtained are EQ. = 17,035 cm-', EBO = 23,042cm-', 1ugQ = 1.23 Debye, and ygB = 7.26 Debye (55).These values have been used as fixed parameters in ourfit to the data.The use of Lorentzian profiles for the analysis of the

optical bands requires some further comments. It is wellknown from several studies on optical spectra of deoxy-genated heme proteins that at least their B bands aresubject to inhomogeneous broadening (56-58). Hence,an exact fit to their band shape required either a decon-volution into two or more Gaussian subbands or the ap-

plication of an asymmetric distribution function (56).In our case, however, we only need a good estimation ofthe ratio ,ugQ/,gB to obtain a consistent set of cl (54).For this purpose, Lorentzian fits to the absorption bandsprovide sufficient information.Three data sets consisting of polarized REPs and

DPDs measured at pH 6.4, 7.2, and 8.1 are displayed inFig. 2. The solid lines were obtained by using the theoreti-cal expressions for the DPD and REPs derived from thepolarizability tensor in Eq. 3 in the fits to the data. Ow-ing to the low energy ofQ0 state, the data points observed

by an argon ion laser cover only the preresonant regionof the Qv and Bo band, in contrast to the situation inoxyHb-trout IV, where Qv-resonance enhancement canbe probed using an excitation wavelength of 528 nm(40). One may therefore doubt whether fits to the aboveDPDs and REPs yield the correct vibronic coupling pa-rameters. To give more evidence for the reliability ofourfits, we measured some DPR values between the reso-nance positions Q0 and Qv at the above pH values byusing an excimer-pumped dye laser. As shown in Fig. 2,these data points (indicated by arrows) nicely agree withthe DPD curves calculated from the fits to the corre-sponding data observed with the argon ion laser.

Altogether we fitted DPDs and REPs measured at 16different pH between 6.0 and 9.0. All data sets could befitted using only c' with r = Alg and Big.The cllg are considerably large, especially in the alka-

line region. This is reflected by the DPDs that exhibitrather large DPR values at their maxima positioned be-tween the Qol/Qlo- andQ 1 resonance positions (DPR =

0.4-0.5) compared with corresponding values found fordeoxyHbA and deoxyMb (DPR = 0.2-0.3) (54). A2g-type contributions are weak and cannot be resolved un-ambiguously from the experimental data. This clearlysuggests that the v4 mode is subject to asymmetric Bigperturbations. The pH dependence of the coupling pa-rameters (Raman titration curves [RTC]) is shown inFig. 3. The full lines herein result from a titration model,which will be discussed below.

Depolarization dispersion of the ivlo lineNormal coordinate analysis on Ni(II)-octaethylpor-phyrin has shown that the structurally sensitive v10 mode(Big in D4h) results from Cm- Ca stretching vibrations,Cm H bending, and C, vibrations along the C, PSbonds. The latter provide only small contributions to thepotential energy distribution of the mode but this doesnot imply that its contribution to vibronic coupling issmall. The pyrrole nitrogens do not show any significantdisplacements (53). As a consequence one expects thatthe DPD of the v 0 line mainly probes peripherical cou-pling between the macrocycle (i.e., the pyrrole carbons)and the corresponding PS. This can be modulated bynoncovalent interactions between the PS and aminoacid residues in the heme cavity (20). As shown in ourrecent article (30), the PS can be expected to cause sym-metry-lowering distortions of B2g and A2g symmetry,provided that identical PS are sterically equivalent. Thisgives rise to vibronic coupling parameters cr with r =

A2g (for B2g distortions) and B2g (for A2g distortions).Asymmetric Big distortions are only imposed ifidenticalbut sterically inequivalent PS cause different perturba-tions ofthe corresponding pyrroles. The symmetry ofthethus induced cr is Aig. Fig. 4 depicts DPDs and REPs ofthe vio mode measured at pH 6.4, 7.2, and 8.1. The fulllines result from fits to the data obtained with an argon

Schweitzer-Stenner et al. Heme Distortions in Deoxy Hb-trout IVSchweitzer-Stenner et al. Heme Distortions in Deoxy Hb-trout IV 1199

0r-CL

w

-i

18.1 20.9 18.1 20.9 18.1 20.9

ExcitationWavenumber [10Ocm ]

FIGURE 2 DPDs and polarized REPs ofthe P4 line (1,355 cm-') ofdeoxyHb-trout IV measured at pH 6.4, 7.2, and 8.1. The solid lines result from afit to the data. The arrows label the data points measured by means ofan excimer pumped dye laser. The resonance positions QOl, Q1O, and Q1, aremarked in the DPD diagrams.

ion laser. They give again a good prediction to the DPRvalues in the region between the Qoo and Q1I resonance

positions as shown by the corresponding DPR valuesindicated by arrows.

Only coupling parameters of Aig and Big symmetrywere necessary for the reproduction of the experimental

data, suggesting that the vlo mode is affected by Blg-rather than by B2g-type distortions. The c'lg are in thesame order ofmagnitude as the corresponding c s, indi-cating that the asymmetric Blg distortions are consider-ably strong. Fig. 5 displays the RTCs ofthe c' , which areagain derived from DPD and REPs measured at 16 dif-

1200 Biophysical Journal Volume 64 April1 200 Biophysical Journal Volume 64 April 1993

J uv..nl

15000

_40000

r

7 50.0e

on .,,,i, i, .....--6 70 60o 9.X

pit-value

1 5n on

ion on

emm

v

C)

300.00

200.00

"00.00m

.C)

0.0

300 00

200.00

o0cr

m

6.on 4- ... ,,p

r, no. .;6..oe e oo..

9s.oo

pit- value

m

6 p0 a 00

p}l -valtie

0 ;to 600 9.00

p1I-value

p,-value3...66.... ..i. ,, .............,

pli- value

FIGURE 3 RTCs ofthe effective vibronic coupling parameters cr derived from the fits to the DPDs and REPs ofthe a.0 mode ofdeoxyHb-trout IV.The solid lines result from a fit to a titration model described in the text.

ferent pH. The solid lines in the diagrams result fromfits, which will be discussed below.

Analysis of Raman titration curves

To rationalize the pH dependence ofthe effective distor-tion parameters, we consider the titration states of the aand ,B subunits, which can be regarded asjointly indepen-dent. From Eq. 7 the pH dependence of the couplingparameters is related to the molar fractions X.(i, j ) ofthetitration states in a and subunits with PK100 = PK2a =

8.5 and pK,, = 7.4, pK2 = 7.5, respectively. (i, j ) denotethe occupation of the titratable sites with protons (i, j =0, 1) and u denotes the subunit a or 1. Therefore, one

expects variations of the c' in the region of pH 7.5

and also pH - 8.5. Inspection of the RTCs of the V4

mode in Fig. 3, however, shows that the parameters CA,cBlg, Bl show strong variations with a maximum or an

inflection point at pH 7.5, whereas the parameters cAJgand cAIg show no variations in this region. Instead, a

slight decrease of their values is observed for pH > 8.3.All RTCs of the vio mode exhibit the strongest varia-

tions in the region at 7.5. From this we try the followingreasoning. For cAIg, /B19 and clg, we assume that theirRTCs mainly reflect a dependence of the correspondingvibronic coupling parameters cr (i, j), on the titrationstate (i, j ), in the subunits. For reasons of simplicity,we further assume that for the two a chains, the corre-

sponding c' (i, j )a = Ce,Sa are identical, i.e., independent

Schweitzer Stenner at al. Heme Distortions in Deoxy Hb trout IV 1201

r,

c

7rnn nn

,0

150.00 -

100 00 -

Eu

50 00333r

r-

~O

Heme Distortions in Deoxy Hb-trout IVSchweitzer-Stenner et al. 1201

-4 1

1 i-11I

CKQ~

-

LL

181 20.9 i8.1 2 ,9 10.1 20.9

Excitation Wavenumber [103cm' I

FIGURE 4 DPDs and polarized REPs ofthe vlo line (1,607 cm') ofdeoxyHb-trout IV measured at pH 6.4,7.2, and 8.1. The solid lines result froma fit to the data. The arrows label the data points measured by means ofan excimer dye laser. The resonance positions Q01, Q10, and Q,, are markedin the DPD diagrams.

of the titration state (i, j )a even though we cannot ex- A corresponding expression is obtained by interchangingclude that they are slightly different. In this case, by sim- a and A3, in the case that the a subunits exhibit c' (i, j )a,ple algebra from Eq. (7) one finds which are dependent on their titration states, whereas

the parameters of the a chains are constant. This hase =Xs(i, je)[(Cs(i j)2 + (Cra)2] been assumed to be approximately valid for cQlJ and cB

i j ofthe v4 mode. Based on the above approximations, Eq.= z X~(i, j)c(i, j)2. (10) 7a has been used to fit the pH dependence ofthe effective

i,j

1202 Biophysical Journal Volume 64 April 1993

200 00

10000 1 -

o" t)..00.00

I L 1 W1pHl-value pi--value

300~0.024j 0

-~~~~~~~~~~~~~~~~~~~~~~~~~~~00

50.00~~~~~~~~~~~~~~~~~~~~00

.111~~~~~~~~~~~~~~~~~~~~~~~~~~~~~.

0.00 006. 0 91J7. 7.00a0.00.v

p11-value p11--value

300__00 -0

200.00~ ~ ~ ~ ~ 00

30000~~~~~~~~~100

WI10000- m 60.00 1O-W6.00~~~T~~O 500 9~~ 6007.0 80 90

pH-value p11-valvauei

FIUR RCso te feciv ibonccopln praeer o ' ervd ro hefistoth PD adREs fth300.00fdexybtru

IV.Th sli lne rsut foma ittoa itatonmodl esrie20000 txt

Schweitzer-Stenner et at. Heme Distortions in Deoxy Hb-trout IV 1203

I SO.OO

coupling parameters of the v4 (Fig. 3) and the vl0 line(Fig. 5 ). This was done by using the c' (i, j )u as fit param-eters. These parameters are listed in Table 1 and 2 for theV4 and P10 line, respectively. It should be noted that due tothe almost identical pK values in the a and chains,respectively, Xu(0, 1) = X.( 1, 0). Therefore, only thesum of cl (0, 1) and c' (1, 0) can be determined. Asdemonstrated by the solid lines in Fig. 3 and 5, this pro-cedure yields satisfactory fits to the RTCs. Some devia-tions from the data points for c'lg (V4) and cAIg (v10)seem to indicate that in contrast to our assumptions, thecorresponding CA,g are not independent of the titrationstate [i, j]. It should be noted, however, that the corre-

sponding data are subject to larger statistical errors forreasons outlined under Material and Methods.

DISCUSSION

Heme-protein coupling monitored bythe DPD of the v4 modeIt is known for several Hb derivatives that the v4 mode isresponsive to conformational changes of the Fe- HisF8 linkage. Its frequency exhibits an inverse correlationwith that ofthe vFe-His stretching mode ( 59 ). This obser-vation can be rationalized by assuming that a larger outofplane displacement 6 ofthe iron increases the popula-tion of the anti-bonding a, (dZ2-UL)MO of the Fe-Imcomplex and reduces the eg(d)-eg( 7r*h) overlap.Whereas the first effect reduces the VFe_Im frequency, thelatter one increases the V4 frequency (6). As also has beenshown for several heme proteins, the DPD of the V4mode is responsive to interactions between Cf and C5 ofthe proximal imidazole and the atoms ofthe heme core,namely the pyrrole nitrogens N( I), N( III), and the adja-cent C,|, (20, 54). Owing to the HCL model briefly de-scribed in the result section (54), this type ofheme-pro-tein coupling depends on the parameters 6 (Fe2 displace-ment), 0 (tilt angle), and 4) (azimuthal angle) of the

TABLE 1 Vibronic coupling parameters cr.(i, j), of the V4 mode

(0, 0) (0, 1) + (1, 0) (1, 1) Subunit

C Blg 230 560 120 ,Bc Blg 45 240 35 /CAlg 80 120 10 /1223 70 160 40 A1228 60 92 25 /

CQBAg 140 360 180 aBBAlg 50 180 90 a

I202 6 20 65 aI211 15 100 69 aI217 30 60 80 a

Vibronic coupling parameters cH(i, j)u of the P4 mode derived fromfitting the titration model described in the text to the RTCs in Fig. 3.The parameters are expressed in units of cm-'. The intensities I (inarbitrary units) of the sub-sublines of the five sublines from the VFe-Hisvibration [41 ] are given for comparison.

TABLE 2 Vibronic coupling parameters cr (i, j), of the v1* mode

(0,0 ) (0, 1) + (1, 0) (1, 1) Subunit

cAlg 130 160 20 ,BC BAlg 20 140 20 /CBlg 170 500 300 ,B

QBAlg 80 100 170 dCQBg 70 80 220 d

Vibronic coupling parameters cr(i, j), of the vl0 mode derived fromfitting the titration model described in the text to the RTCs in Fig. 5.The parameters are expressed in units of cm-'.

Fe- His F8 complex (cf. Fig. 1). Consequently, oneexpects some correlation between the pH dependence ofthe c' ( V4) and that of vibronic coupling parameters de-termining the intensity of lih VFe-His stretching band.

In a recent study from (. laboratory (41 ), we haveshown that the profile ofth His band can be decom-posed into five different su i s. Although their intensi-ties are found to be a fun; nii of pH, their frequenciesremain unaltered. This w. ; i'01(.rpreted by use ofthe cou-pling model suggested by V I ciw man et al. (44), which isbased on earlier consideriumons by Bangcharoenpaur-pong et al. (43 ). It relates flAq t iency changes to the varia-tions of the tilt angle 0, whereas the azimuthal angle 4bwas assumed to determine the intensity. The protona-tion of Bohr groups was assumed to change the azi-muthal angle sp and thus the intensities of the sub-sub-line. The different sublines were related to different tiltangles 0 of the Fe- N,( His F8) bond.

This interpretation is consistent with the RDS data onthe V4 mode owing to the following considerations. Inview of the HLC model, a decrease in 4) results in largerperturbations of the pyrrole nitrogens N(I) and N(III)that enhance asymmetric Blg distortions of the v4 mode.This should be reflected by the vibronic coupling parame-ters CB'g, which must be considered as effective ones withrespect to the sublines. From Table 1 we read thatCBg( 1. 1) <c Bg(O.0) <c1Bg(O. I ) + CBlg( 1. 1). As shownabove, this reflects different titration states of the A sub-unit. According to Bosenbeck et al. (41 ), the intensitiesIci,j of the VFe-His sublines at 223 cm-' (Q,,j) and 227cm-l mainly depend on the protonation of the Bohrgroups in the d subunit. From Table 1, one reads thatI (1, 1 ) < I(0, 0) < I(0, 1 ) + I( 1, 0) for both lines. Thisshows that the pH dependence of the subline intensitiesparallel that ofthe CBlg ofthe V4 mode, in full accordancewith what one would expect if both parameters are gov-erned by the azimuthal angle (D. The heme-imidazoleinteraction also changes the symmetric coupling parame-ters c"9. While CA'g exhibit a pronounced variation onprotonations of the Bohr groups in the 13 subunit, cAIgand cBB show only a slight pH dependence above 8.0,which mainly reflects different titration states of the asubunit. Table 1 shows that CAlg( 1, 1) < Cllg(0, 0) <coo( 1, 0) + C0(0, 1), indicating that this parameteralso depends on 4).

120 Bipya Jora Voum 64 Api 1993.1 204 Biophysical Journal Volume 64 April 1993

For cA)g and cAIg, one finds a different relation,namely CA18(0 O) < CAlg81, I ) < CAs 8(°, 1I) + CAlg( 1, O).We compare this with the intensities of the vFe-His SUhlines at 202, 21 1, and 217 cm-l which are mainly in-fluenced by protonations in the a subunit (41). For eachofthese sublines, one finds I(0, 0) < I(1, 1), which againcorrelates nicely with the corresponding coupling param-eters CA19. Thus, the comparison of the pH dependenceofCV(V4) and the intensities of the sub-sublines provideevidence that both are governed by the same structuralparameter, which may be identified with the azimuthalangle 4) of the Fe N( His F8) bond.Even though the coupling model of Friedman et al.

(44) provides a reasonable and consistent description ofour Raman data, it should not be overlooked that it doesnot describe the physical relationship between the geo-metric parameters of the Fe-His complex and the vi-bronic coupling determining the intensities ofthe VFe -His

sublines. This recently has been attempted in anothercoupling model proposed by Stavrov (6 1 ) and Bersukerand Stavrov (62). It considers that the A2U orbitals oftheporphyrin macrocycle and the dZ2-UL orbitals of theFe Nf (His F8) bond are mixed by pseudo-Jahn-Teller coupling induced by out of plane modes of A2Usymmetry and static distortions caused by the axial ori-ented proximal imidazole. In a pentacoordinated Fe-porphyrin, this interaction stabilizes the high spin config-uration ofiron and causes its out ofplane displacement 6(63). The overlap between the above A2U and Alg or-bitals can be shown to increase with increasing 6. Excita-tion into the B state changes the population of the A2Uorbitals, thus reducing the Fe displacement 6. As a conse-

quence, the adiabatic potential minimum moves withrespect to the ground state along the coordinate 6, givingrise to the Raman intensity of the VFe-His mode. Thismodel also can be used to correlate our Raman data onv4 and VFe-His . Since it predicts that the frequency of theVFeHis mode decreases with increasing 6, it relates thefive sublines derived from the VFe-His band to differentFe displacements. This also may give rise to differentangles 0 (63), in accordance with Friedman's model(44). The pH-induced variations in the intensities ofthesublines should be attributed to changes in the mini-mum position of the excited B state with respect to 6 oranother internal coordinate determining the normal co-ordinate of the VFe-His mode (i.e., 0 and p.)The decrease in 6 on excitation into the B state would

reduce the distances N(I) C. and N(III) CE, thusgiving rise to an enhancement ofBlg-type distortions (cf.the HLC model). In this case the minimum ofthe B stateis shifted along a normal distortion 6QBQg, and as a con-

sequence Big-type vibronic coupling is admixed to theRaman tensor of the v4 mode (64). Hence, the couplingmodel proposed by Stavrov also predicts that an increasein Cs ( V4) should correlate with an increase in the inten-sities of corresponding VFe-His sublines in full accor-dance with our results.

Now we discuss some further aspects of the vibroniccoupling parameters derived from the DPDs of the v4

mode. First, the absence ofA2g distortions requires somefurther comments. Due to the HLC model, variations in4) should be reflected by changes of antisymmetric cou-pling (c2Bg). Significant contributions to cA2!, however,are only provided if the azimuthal angle is large(>100) and the Fe displacement 6 is small (<0.2 A) toenable interactions between imidazole and Ca (54).Thus, the absence ofA2g coupling implies that these con-ditions are not met. This parallels findings on deoxyHb,deoxyMb (54), and deoxyHb-chironomous thummithummi (CTT)III (65), the v4 mode of which showsonly small contributions from A2g coupling. The Blgcontributions to the IV4 mode are exceptionally strong,compared with corresponding parameter values foundfor other heme proteins (deoxyHbA, oxyHbA, and itsisolated subunits, deoxyMb). A similar finding has beenmade on oxyMb (54), but in this molecule Big distor-tions are induced by interactions between dioxygen andN(III). In the case of deoxyHb-trout IV, the large CBilgsuggest comparatively small values of 4) that can be ex-pected to give rise to particular strong Big perturbationsof the pyrrole nitrogens in the tilted gemometry of theFe- His bond.

This interpretation parallels findings Friedman et al.(44) made on the VFe-His band of a Hb from a deep seafish, which exhibits a very low t-state affinity. Theyfound that the corresponding VFe-His mode exhibits anunusually strong intensity, which the authors related to asmall azimuthal angles of the Fe N,(His F8) bond.Moreover, they suspect that the distortions imposed onthe heme in such a contrained geometry gives rise to thelow t-state affinity. As derived from the binding iso-therms of Hb-trout IV and HbA, the equilibrium con-stant KT of Hb-trout IV is an order of magnitude lowerthan that ofHbA (20,40). The vibronic coupling param-eters found for the v4 mode ofdeoxyHb-trout IV indicatethat the low KT values ofHb-trout IV also may be relatedto a small azimuthal angle ofthe Fe- NE(His F8) bond.As shown above, allosteric coupling between Bohr

groups and heme groups is much more effective in the ,Bthan in the a subunits. This parallels findings on HbA(17, 26, 29, 30, 40). In the A subunits, protonation af-fects the vibronic coupling parameters to a different ex-tent. Although cQ%g significantly depends on the protona-tion state (i, j), c"g and c"g remain unaffected in thelimit of accuracy. The very same finding recently hasbeen made on deoxyHb-CTT III (66). This is contradic-tory to Gouterman's four-orbital model (46), which pre-dicts

QQ= ag - sin (2v)a,g,CAlg = COS (2v)aig,CAlg==dg + sin (2v)a,g. (11)

v = 0.17 is the unmixing parameter, whereas dig and aig

Schweitzer Sienner et si. Heme Distortions in Deoxy Hb-irout IV 1205Schweitzer-Stenner et a]. Heme Distortions in Deoxy Hb-trout IV 1 205

describe the FC and HT coupling in the framework ofthe 50:50 states. Eq. 8 implies that at least variations incQj4 and c"g should be correlated, in contrast to ourobservation. Moreover, one would expect that cAl -

cBAlg = (cos 2v - sin 2v)a,g, which is also not consistentwith our data.Two possible explanations may be given to explain

this discrepancy. First, one may consider that two-elec-tron matrix elements may contribute to cQj4 and cBB(51 ). In Gouterman's two-electron picture, they wouldread as (51):

a,g = (A glA2,Egc>,P, = x,y. (12)

This matrix element is positive in cAQ and negative incBB . Therefore, if dig and alg exhibit a similar depen-dence of the protonation state (i, j), their variations as afunction ofpH add up for CA'g but may nearly eliminatefor cBAIg. Two-electron matrix elements for vibronic cou-pling have been shown to exist ifthe electronic distribu-tions follow the nuclei in a model for a vibrating mole-cule. Consideration of such "flocking basis sets" are nec-essary for the convergence of perturbation calculationsbased on the conventional Herzberg-Teller approach(66). Second, one may assume that out of plane modesof the porphyrin macrocycle are capable to vibronicallycouple the Q states (symmetry EJ) with the dX states(symmetry Eg) ofFe2 ( 7r-d interaction) (67). This typeof interaction would mix dcI orbitals into the Q states,thus rendering them sensitive to structural changes inthe Fe -His F8 complex. At present, we believe thatboth ir-d interactions and two-electron matrix elementscontribute to vibronic coupling of the V4 mode. Furthertheoretical and experimental studies are apparently nec-essary to get more insight into these mechanisms.

Heme-protein coupling monitored bythe DPD of the v10 modeAs mentioned in the result section, the DPD of the vlomode monitors interactions that mainly result from the(asymmetric) influence ofthe PS on the porphyrin mac-rocycle. These can be modified by noncovalent interac-tions between the PS and the heme environment.

All RTCs can be described solely in terms of protona-tions of the Bohr groups in the : subunits. This suggeststhat the influence on the corresponding vl0 by protonbinding to Bohr groups in the a subunits is weak. Similarfindings have been obtained for oxyHbA and its isolatedsubunits (19, 30). The influence of heme-imidazolecoupling on the v10 mode is expected to be weak for rea-sons given in Results. As a consequence, there is no strictcorrelation between the cl (vlo) and the intensities oftheVFe-HiS sublines. Moreover, in contrast to what has beenobtained for V4, no unique correlation between thec' (vlo) exists. The parameter Cl which depends on

Aig distortions, and parameter cAg (reflecting Big distor-

tions) exhibit c ( 1.1 ) > cr(l1.0) + cl (0. ) > cl (0.0),whereas the remaining parameters (i.e.,cA'g and c Alg)correlate with the cl of the v4 mode.The PS may influence the porphyrin macrocycle di-

rectly or in an indirect way. The direct coupling includeselectron withdrawing from the Alu orbitals of C. atomsby the carbon hydrogen chains of the PS (68). The dif-ferent PS can be expected to differ in terms oftheir elec-tron withdrawing capacity, thus giving rise to asymmet-ric distortions of the pyrroles. As shown by Schweitzer-Stenner et al. (30), one expects B2g and A2g distortions ifidentical PS cause the same distortions. Our data clearlyshow, however, that Alg coupling is admixed to the Ra-man tensor of the vlo mode that results from Blg distor-tions. Thus, some of the identical PS interact differentlywith the C. Ideal candidates for this are the two vinylsattached to the pyrroles II and III (V(II) and V( III)) forthe following reasons. First, there is some electronic mix-ing between the exited states 11* states of the vinyls andthe porphyrins (69). Second, it is known that V( II) andV(III) have different orientations with respect to hemein deoxy- and oxyHbA (3, 70, 71), which can be ex-pected to cause different distortions ofthe correspondingC,. Third, V(III) exhibits a van der Waals bond to ValFG4 , that is located close to the amino acids that maybe involved in the subunit Bohr effect, namely Ser F9 aand Glu FG1 a (72), which are H-bonded to the car-boxyl group and the imidazole of the COOH-terminalHis HC3 (. Furthermore, it is known that Val FG5 (B isH-bonded to the penultimate Tyr HC2 d (12, 14, 70),which is most probably also the case in Hb-trout IV. ThisH-bond serves as a direct linkage between the COOH-terminal His HC3 and the flexible FG-helix and is ofutmost importance for the R-state Bohr effect in oxy-HbA ( 19). One may therefore expect that the protona-tion ofBohr groups in the , subunit rearranges the struc-ture close to His F8 and Val FG5, thus giving rise tochanges in the orientation ofthe proximal imidazole andV(III). The latter is then reflected by the RTCs ofthe vi0mode.An indirect influence of the PS on the structure of the

macrocycle may be caused by noncovalent PS-PS inter-actions that have been shown to cause strong nonplanardistortions of metalloporphyrins in solution (73, 74),thus lowering their symmetry from D4h to D2d or S4 (52).Since the PS are not equivalent, these distortions wouldbe asymmetric and the symmetry of the macrocyclewould be lowered further to C2 or Cs. The nonplanaritydepends on the length ofthe PS and on their orientationwith respect to each other. Changes of the latter maytherefore cause changes in nonplanarity and thus varia-tions ofthe DPD ofthe Vlo mode. Support for this modelcomes from the dispersion experiments on the isolatedsubunits ofoxyHbA (30), which showed that asymmet-ric perturbations affect the v10 mode in (3SH_oxyHb ratherthan in aSH_oxyHbA. Crystallographic data on the intactoxyHbA had previously revealed that the macrocycle in

1 1(UaQ-Blphsia JornlVoue1 ZLC Biophysical Journal Volume 64 April 1993

the a subunits is planar, whereas it exhibits significanttilting of its pyrroles in the : subunit (70).

SUMMARYRaman dispersion experiments on deoxyHb-trout IVhave shown that proton binding to Bohr groups causesconsiderable asymmetric distortions ofthe heme groups.This process of heme-protein coupling is much morepronounced in the : than in the a subunits. It involvesheme-protein interactions between the proximal His F8and the pyrrole nitrogens and noncovalent binding be-tween the FG helix in the A subunit and the PS of theprosthetic heme group. It could be shown further thatthe pH dependence ofthe vibronic-coupling parametersof the v4 mode correlate nicely with the pH-dependentvariations in the intensities ofthe sublines into which thePFe-His band profile was recently decomposed (41 ). Thisfinding can be explained by two different coupling mod-els designed to explain the resonance enhancement ofthe PFe-His mode in the B-band region.

R. Schweitzer-Stenner thanks Dr. Solomon Stavrov for helpful discus-sions on vibronic coupling processes in Fe-porphyrins.

We gratefully acknowledge the support by a grant from the DeutscheForschungsgemeinschaft.

Receivedfor publication and infinalform 6 August 1992.

REFERENCES1. Szabo, A., and M. Karplus. 1975. Analysis of cooperativity in he-

moglobin. Valency hybrids, oxidation and methemoglobin re-placement reactions. Biochemistry. 14:931-1940.

2. Herzfeld, J., and E. Stanley. 1974. A general approach to coopera-tivity and its application to oxygen equilibrium of hemoglobinand its effectors. J. Mol. Biol. 82:231-265.

3. Gellin, B., and M. Karplus. 1977. Mechanism oftertiary structuralchange in hemoglobin. Proc. Natl. Acad. Sci. USA. 74:801-805.

4. Baldwin, J. L., and C. Chotia. 1979. Haemoglobin, the structuralchanges related to ligand binding and its allostreric mechanism.J. Mol. Biol. 129:175-200.

5. Johnson, M. L., and G. K. Ackers. 1982. Thermodynamic analysisof human hemoglobins in terms of the Perutz mechanism: ex-tensions of the Szabo-Karplus model to include subunit assem-bly. Biochemistry. 21:201-211.

6. Friedman, J. M. 1985. Structure, dynamics and reactivity in hemo-globin. Science (Wash. DC). 228:1274-1280.

7. Ackers, G. K., and F. R. Smith. 1986. Resolving pathways offunc-tional coupling within protein assemblies by site specific struc-tural perturbation. Biophys. J. 49:155-165.

8. Murray, L. P., J. Hofrichter, E. R. Henry, and W. A. Eaton. Timeresolved optical spectroscopy and structural dynamics followingphotodissociation of carbonmonoxy hemoglobin. Biophys.Chem. 29:63-76.

9. Rousseau, D. L., and J. M. Friedman. 1988. Transient and cryo-genic studies of photodissociated hemoglobin and myoglobin.In Biological Application ofRaman Spectroscopy. T. G. Spiro,editor. John Wiley & Sons Inc., New York/Chichester. 133-216.

10. Lee, A. W., M. Karplus, C. Poyart, and E. Bursaux. 1988. Analysisof Proton Release in Oxrygen Binding by H[emoglobin: lImplica-

tion for the Cooperative Mechanism. Biochemistry. 27:1285-1301.

1 1. Eaton, W. A., E. R. Henry, and J. Hofrichter. 1991. Application oflinear free energy relations to protein conformational changes:the quaternary structural change of hemoglobin. Proc. Natl.Acad. Sci. USA. 88:4472-4475.

12. Perutz, M. F. 1989. Mechanism ofcooperativity and allosteric reg-ulation in proteins. Q. Rev. Biophys. 22:139-236.

13. Zhang, M., F. A. Ferrone, and A. J. Martino. 1990. Allosterickinetics and equilibria differ for carbon monoxide and oxygenbinding to hemoglobin. Biophys. J. 58:333-340.

14. Perutz, M. F. 1970. The Bohr effect and combination with organicphosphates. Nature (Lond.). 228:734-739.

15. Kilmartin, J. V., J. H. Fogg, and M. F. Perutz. 1980. Role ofC-ter-minal histidine in the alkaline Bohr effect of human hemoglo-bin. Biochemistry. 19:3189-3193.

16. Wyman, J., J. Gill, H. T. Gaud, A. Colosimo, B. Giardina, H. A.Kuiper, and M. Brunori. 1978. Thermodynamics ofligand bind-ing and allosteric transitions in hemoglobins. Reaction of Hbtrout IV with CO. J. Mol. Biol. 124:161-175.

17. Scott, T. W., J. M. Friedman, M. Ikeda-Saito, and T. Yonetari.1983. Subunit heterogeneity in the structure and dynamics ofhemoglobin. A transient Raman study. FEBS (Fed. Eur. Bio-chem. Soc.) Letts. 158:68-71.

18. Kwiatkowski, L., and R. W. Noble. 1982. The contribution ofhistidine (HC3) (146 f,) to the R-state Bohr effect of humanhemoglobin. J. Biol. Chem. 257:8891-8895.

19. Schweitzer-Stenner, R., D. Wedekind, and W. Dreybrodt. 1986.Correspondence of the pK-values of oxyHb-titration states de-tected by resonance Raman scattering to kinetic data of liganddissociation and association. Biophys. J. 49:1077-1088.

20. Schweitzer-Stenner, R. 1989. Allosteric linkage-induced distor-tions of the prosthetic group in haem proteins as derived by thetheoretical interpretation of the depolarization ratio in reso-nance Raman scattering. Q. Rev. Biophys. 22:381-479.

21. Warshel, A. 1977. Energy-structure correlations in metallopor-phyrins and the control ofoxygen binding by hemoglobin. Proc.Natl. Acad. Sci. USA. 74:1789-1793.

22. Ten Eyck, L. F. 1979. Hemoglobin and myoglobin. In The Por-phyrins. Vol. III. D. Dolphin, editor. Academic Press, NewYork. 445-472

23. Collins, D. W., P. Champion, and D. B. Fitchen. 1976. ResonantRaman scattering from heme proteins. Polarization dispersionand a-band splitting. Chem. Phys. Letts. 40:416-420.

24. Schweitzer-Stenner, R., W. Dreybrodt, and S. el Naggar. 1984.Investigation of pH-induced symmetry distortions of the pros-thetic group in deoxyhemoglobin by resonance Raman scatter-ing. Biophys. Struct. Mech. 10:241-256.

25. Shih, T. B., R. T. Jones, J. Bonaventura, and C. Bonaventura.1984. Involvement of His HC3(146),B in the Bohr effect of hu-man hemoglobin. Studies of native and N-ethyl-maleimido-treated hemoglobin A and hemoglobin cowtown (f3146His -Leu). J. Biol. Chem. 259:967-974.

26. Wedekind, D., R. Schweitzer-Stenner, and W. Dreybrodt. 1985.Heme-apoprotein interaction in the modified oxyhemoglobin-bis(N-maleimidomethyl)ether and in oxyhemoglobin at highCi-concentration detected by resonance Raman scattering. Bio-chim. Biophys. Acta. 830:224-232.

27. Wedekind, D., U. Brunzel, R. Schweitzer-Stenner, and W. Drey-brodt. 1986. Correlation of pH-dependent resonance Ramanand optical absorption data reflecting haem-aproprotein inter-action in oxyhemoglobin. J. MoL Struct. 143:457-460.

28. Schweitzer-Stenner, R., D. Wedekind, and W. Dreybrodt. 1989.The influence of structural variations in the F- and FG-helix ofthe ,8-subunit modtified oxy:Hb-NEFS o1n the heme structure de-

Schweitzer-Stenner et al. Heme Distortions in Deoxy Hb-trout IV 1207

tected by resonance Raman spectroscopy. Eur. Biophys. J.17:87-100.

29. Chavez, M. D., S. H. Courtney, M. R. Chance, D. Kiula, J. Nocek,B. M. Hofman, and J. M. Friedman. 1990. Structural and func-tional significance ofinhomogenous line broadening ofband IIIin hemoglobin and Fe-Mn hybrid hemoglobin. Biochemistry.29:4844-4852.

30. Schweitzer-Stenner, R., U. Dannemann, and W. Dreybrodt. 1992.Investigation ofHeme Distortions and Heme-Protein Couplingin the Isolated Subunits of Oxygenated Human Hemoglobin byResonance Raman Dispersion Spectroscopy. Biochemistry.31:694-702.

31. Wyman, J. 1966. Allosteric linkage. J. Am. Chem. Soc. 89:2202-2218.

32. Ascoli, T. G., B. Falcioni, B. Giardina, and M. Brunori. 1986.Thermodynamic characterization of the allosteric transition introut hemoglobin. E. Biophys. J. 13:245-249.

33. Antonini, E., and M. Brunori. 1970. Hemoglobin and Myoglobinin their Reaction with Ligands. Elsevier, Amsterdam. 175-183.

34. Moffat, J. K. 1971. Structural and functional properties of chemi-cally modified horse hemoglobin. J. Mol. Biol. 58:79-88.

35. El Naggar, S., W. Dreybrodt, and R. Schweitzer-Stenner. 1985.Haem-protein interactions detected by resonance Raman scat-tering in Mb- and Hb-derivates lacking the saltbridge His 146b-Asp94b. Eur. Biophys. J. 12:43-49.

36. El Naggar, S., R. Schweitzer-Stenner, W. Dreybrodt, and A.Mayer. 1984. Determination of the Raman Tensor ofthe HaemGroup in Myoglobin by Resonance Raman Scattering in Solu-tion and Single Crystals. Biophys. Struct. Mech. 10:257-273.

37. Brunori, M. 1975. Molecular adaption to biophysical require-ments. Curr. Top. Cell. Regul. 9:1-39.

38. Brunori, M., M. Coletta, B. Giardina, and J. Wyman. 1978. Amacromolecular transducer as illustrated by trout hemoglobinIV. Proc. Natl. Acad. Sci. USA. 75:4310-4312.

39. Schweitzer-Stenner, R., and W. Dreybrodt. 1989. An extendedMonod-Wyman-Changeaux model expressed in terms of theHerzfeld-Stanley formalism applied to oxygen and carbonmonoxide binding curves of hemoglobin trout IV. Biophys. J.55:691-701.

40. Schweitzer-Stenner, R., D. Wedekind, and W. Dreybrodt. 1989.Detection ofheme perturbations caused by the quaternary R -0

T transition in oxyhemoglobin trout IV by resonance Ramanscattering. Biophys. J. 55:703-712.

41. Bosenbeck, M., R. Schweitzer-Stenner, and W. Dreybrodt. 1992.pH-induced conformational changes of the Fe2" - N (His F8)linkage in deoxyhemoglobin trout IV detected by the Ramanactive Fe2 - Nf(His F8) stretching mode. Biophys. J. 61:31-41.

42. Kitagawa, T. 1988. Heme protein structure and the iron histidinestretching mode. In Biological Application of Raman Spectros-copy. T. G. Spiro, editor. John Wiley & Sons Inc. New York/Chichester. 97-132.

43. Bangcharoenpaurpong, O., K. T. Schomaker, and P. Champion.1984. A resonance Raman investigation of myoglobin. J. Am.Chem. Soc. 106:5688-5698.

44. Friedman, J. M., B. F. Campbell, and R. W. Noble. 1990. A possi-ble new control mechanism suggests by resonance Raman spec-tra from a deep ocean fish hemoglobin. Biophys. Chem. 37:43-59.

45. Loudon, R. 1979. Quantum Theory of Light. Clarendon Press,New York.

46. Gouterman, M. 1959. Study of the effects of substitution on theabsorption spectra of porphyrin. J. Chem. Phys. 30:1139-1161.

47. Shelnutt, J. A. 1981. The Raman excitation spectra and absorption

spectrum of a metalloporphyrin in an environment of low sym-metry. J. Chem. Phys. 72:3948-3958.

48. Placzek, G. 1934. Rayleighstreuung und Ramaneffekt. In Hand-buch der Radiologie, Bd. 6. E. Marx, editor. Akademische Ver-lagsanstalten, Leipzig.

49. Spiro, T. G., and T. C. Strekas. 1974. Resonance Raman spectra ofheme proteins: effects of oxidation and spin state. J. Am. Chem.Soc. 96:338-345.

50. Bobinger, U., R. Schweitzer-Stenner, and W. Dreybrodt. 1989.Highly resolved depolarization ratio dispersion and excitationprofiles of Raman fundamentals of protoporphyrin IX in aheme protein matrix. J. Raman Spectrosc. 20:191-201.

51. Schweitzer-Stenner, R., U. Bobinger, and W. Dreybrodt. 1991.Multimide Analysis ofDepolarization Ratio Dispersion and Ex-citation Profiles of Seven Raman Fundamentals from the HaemGroup in Ferrocytochrome c. J. Raman Spectrosc. 22:65-78.

52. Bobinger, U., R. Schweitzer-Stenner, and W. Dreybrodt. 1991.Investigation of asymmetric perturbations of Ni(II)-octaethyl-porphyrin in CH2Cl2 by Raman dispersion spectroscopy. J.Phys. Chem. 95:7625-7634.

53. Li, X.-Y., R. S. Czersnuszewicz, J. R. Kincaid, P. Stein, and T. G.Spiro. 1990. Consistent Porphyrin Force Field. 2. Nickel Oc-taethylporphyrin Skeletal and Substituent Mode Assignmentsfrom '5N, Meso-d4, and Methylene-d,6 Raman and Infrared Iso-tope Shifts. J. Phys. Chem. 94:47-61.

54. Schweitzer-Stenner, R., and W. Dreybrodt. 1992. Investigation ofHaem Protein Coupling and Structural Heterogenity in Myoglo-bin and Haemoglobin by Resonance Raman Spectroscopy. J.Raman Spectrosc. 23:539-550.

55. Bosenbeck, M. 1990. Resonanz-Raman-Streuung an Hamoglobinder Forelle salmo irideus. Dr. rer. nat. Dissertation. Universityof Bremen, Germany. 33-37.

56. Srajer, V., K. T. Schomaker, and P. M. Champion. 1986. Spectralbroadening in biomolecules. Phys. Rev. Lett. 57:1267-1270.

57. Cordone, L., A. Cupane, M. Leone, and E. Vitrano. 1986. Opticalabsorption spectra ofdeoxy- and oxyhemoglobin in the tempera-ture range 300-20 K. Relation with protein dynamics. Biophys.Chem. 24:259-275.

58. Srajer, V., L. Reinish, and P. M. Champion. 1988. Protein fluctua-tions, distributed coupling, and the binding of ligands to hemeproteins. J. Am. Chem. Soc. 110:6656-6666.

59. Ondrias, M. R., D. L. Rousseau, J. A. Shelnutt, and S. R. Simon.1982. Quaternary transformation induced changes at the hemein deoxyhemoglobins. Biochemistry. 21:3420-3437.

60. Ahmed, A. M., B. F. Campbell, D. Caruso, M. R. Chance, M. D.Chavez, S. H. Courtney, J. M. Friedman, I. E. T. Iben, M. R.Ondrias, and M. Yang. 1991. Evidence for proximal control ofligand specificity in hemeproteins: absorption and Raman stud-ies ofcyrogenically trapped photoproducts ofligand bound myo-globin. Chem. Phys. 158:329-351.

61. Stavrov, S. A. 1992. The contribution of the out-of-plane irondisplacement to the strong coupling of the Fe- His(F8 )-modein the resonance region of the Soret band. Proc. Symp. RamanSpectrosc. Biol. Syst. Bremen. 19.

62. Bersuker, I. B., and S. S. Stavrov. 1988. Structure and Properties ofMetalloporphyrins and Hemproteins: The Vibronic Approach.Coord. Chem. Rev. 88:1-68.

63. Powers, L., B. Chance, M. Chance, B. Campbell, J. M. Friedman,S. Khalid, C. Kumar, A. Naqui, K. S. Reddy, and Y. Zhou.1987. Kinetic, Structural and Spectroscopic Identification ofGeminate States of Myoglobin: A Ligand Site on the ReactionPath. Biochemistry. 26:4785-4796.

64. Tsuboi, M., and A. Hirakawa. 1976. Adiabatic potential and Ra-man scattering: a theoretical background for a proposed rule. J.Raman Spectrosc. 5:76-86.

1208 Biophysical Journal Volume 64 April 1993

65. Jentzen, W., W. Dreybrodt, R. Schweitzer-Stenner, and K. Ger-sonde. 1990. Heme distortions in monomeric insect Hb-CTTIII: resonance Raman scattering and its relation to biologicalfunction. Proc. XII Int. Conf Raman Spectrosc. Columbia Uni-versity Press, New York. 680-681.

66. Curtis, W. C., and 0. E. Weigang. 1975. Theory of vibronic inter-actions. The importance of floating basis sets. J. Chem. Phys.63:2135-2143.

67. Mineyev, A. P., Y. A. Sharanov, N. A. Sharanova, and Y. P. Ly-sov. 1983. p-d-Interaction and the Temperature DependentMagnetic Circular Dichroism Spectra of Low Spin Fe(III)-Heme Compounds. Theor. Chim. Acta. 65:421-438.

68. Shelnutt, J. A., and V. Ortiz. 1985. Substituent Effects on the Elec-tronic Structure of Metalloporphyrins: A Quantitative Analysisin Terms of Four Orbital Model Parameters. J. Phys. Chem.89:4733-4739.

69. DeVito, V. L., and S. A. Asher. 1989. UV-resonance enhancementof vinyl stretch in ferric protoporphyrin IX: conjugation or pres-

ervation of the vinyl 7r -- 7r* transitions. J. Am. Chem. Soc.26:9143-9156.

70. Shaanan, B. 1983. Structure of human oxyhemoglobin at 2.1 Aresolution. J. Mol. Biol. 171:31-59.

71. Fermi, G., M. F. Perutz, B. Shaanan, and R. Fourme. 1984. Hu-man deoxyhaemoglobin at 1.74 A resolution. J. Mol. Bio.175:159-174.

72. Perutz, M. F., and M. Brunori. 1982. Sterochemistry of coopera-tive effects in fish and amphibian hemoglobin. Nature (Lond.).299:241-426.

73. Shelnutt, J. A., C. J. Medforth, M. D. Barber, K. M. Barkigia, andK. M. Smith. 1991. Relationship between structural parametersand Raman frequencies for some planar and nonplanar Nickel(II) Porphyrins. J. Am. Chem. Soc. 113:4077-4087.

74. Shelnutt, J. A., J. D. Hobbs, S. A. Majumder, L. D. Sparks, C. J.Medforth, M. 0. Senge, K. M. Smith, M. Miura, and H. M. E.Quirke. 1992. Resonance Raman Spectroscopy of Non-PlanarNickel Porphyrins. J. Raman Spectrosc. 23:523-529.

Schweitzer-Stenner et al. Heme Distortions in Deoxy Hb-trout IV 1209