Biochem. J. (1966) 100, 614

Optical Rotatory Dispersion, Circular Dichroism andFar-Ultraviolet Spectra of Avidin and Streptavidin

By N. M. GREEN* AND M. D. MELAMED*Laboratory of Chemistry, National Institute of Arthritis and Metabolic Diseases, National InstitUtes of

Health, Bethesda, Md., U.S.A., Department of Chemical Pathology, St Mary's Hospital MedicalSchool, London, W. 2, and National Institute for Medical Research, London, N.W. 7

(Received 20 January 1966)

1. The optical-rotatory-dispersion and circular-dichroism curves of avidinshowed positive Cotton effects centred at 228m,u and 280m,u, close to the ultra-violet-absorption bands of tryptophan. These effects disappeared when avidinwas dissociated into sub-units in guanidine hydrochloride. 2. Binding of biotinhad only a small effect on the optical-rotatory-dispersion curve of avidin. 3. Theabsence of negative circular dichroism at wavelengths above 216m,u showed thatthere was little or no ac-helix present in avidin. This interpretation was confirmedby Moffitt-Yang plots of the partial rotation due to the peptide bonds in thevisible region of the spectrum. The calculated dispersion constants were remark-ably similar to those of y-globulin and suggested the presence of peptide conforma-tions other than ac-helix and random coil. 4. The far-ultraviolet spectrum was alsosimilar to that of y-globulin, the mean extinction coefficient of the peptidechromophore being much lower than the experimental value for a random-coilstructure. 5. Streptavidin resembled avidin in showing two positive Cotton effects,but the negative dichroism below 220m,u suggested the presence of more ac-helixthan was found in avidin. Formation of the complex with biotin was accompaniedby changes in rotation that were rather larger than those observed with avidin.

Spectroscopic studies of the reaction betweenavidin and biotin (Green, 1963a) showed it to beaccompanied by a marked change in the environ-ment of the tryptophan residues of avidin. Theybecame less exposed to solvent when the biotinwas bound. It was hoped that further informationabout this change could be obtained from a studyof the ORDt of the protein. Previous measure-ments (Green, 1963b) had shown combination withbiotin to have little effect on [a]D, but it wasthought that there might be specific effects in theneighbourhood of the tryptophan absorptionbands. However, the effects at 290m,u proved tobe only slightly greater than those at 589m,.Although the ORD curves of avidin and the

avidin-biotin complex were very similar to eachother they were strikingly different from those ofany other protein so far studied, in showing Cottoneffects accompanied by positive rotation in theneighbourhood of both tryptophan absorptionbands. To facilitate the interpretation of theseeffects, measurements of the circular dichroism

* Present address: National Institute for MedicalResearch, London, N.W. 7.

t Abbreviation: ORD, optical rotatory dispersion.

and far-ultraviolet spectra were also made. Sincecircular dichroism is quantitatively related toORD it does not in principle provide any additionalinformnation. However, it has the advantage thatthe dichroism is confined to the wavelength regionof the optically active absorption band. Theresulting bands are usually discrete and the spectraare more readily interpretable than the ORD. Thefar-ultraviolet absorption of the peptide bond isdependent on its conformational relation to neigh-bouring peptide bonds, and this has been used toobtain information on the amounts of helical andrandom-coil structure in a number of proteins(Rosenheck & Doty, 1961; McDiarmid, 1965).Although the absorption is less sensitive than theORD to changes in conformation the measurementsare relatively simple and more accurate. How-ever, interpretation is still hazardous.The recently discovered biotin-binding protein,

streptavidin, isolated by Chaiet & Wolf (1964)from culture filtrates of Streptomyces avidinii,differs considerably from avidin in its amino acidcomposition. However, like avidin, each moleculebinds four molecules of biotin and the binding isaccompanied by changes in absorption spectrumand reactivity of tryptophan similar to those

614

COTTON EFFECTS OF AVIDINobserved with avidin from egg white (N. M. Green,unpublished work). The ORD and circular-dichroism curves are therefore included for com-parison with those of avidin.

METHODS

The avidin sample was preparation D. 31 of Melamed &Green (1963). The concentrations of its solutions weredetermined from the E2go values. Streptavidin was kindlysupplied by Dr F. J. Wolf. Measurements of the biotin-binding capacity by spectrophotometric methods (Green,1963a) confirmed that 1 mole of biotin was bound by15000g. of protein (Chaiet & Wolf, 1964). The value ofE2so/mg./ml. was 3*4. Synthetic (+)-biotin was suppliedby Roche Products Ltd., Welwyn Garden City, Herts. Allother reagents used were of analytical grade.The ORD measurements were made with two different

instruments. Preliminary measurements were made witha Rudolph recording spectropolarimeter (model 250/655/850/810-614) equipped with a 150w xenon arc. Therotation measurements above 300m,u, from which ao, bo,and A, were calculated, were also made with this instrument.Measurements below 300 m, were made with a BendixPolarmatic 62 recording spectropolarimeter in the Labora-tory of Professor W. Klyne. To check the validity ofmeasurements in regions of high absorption, solutions ofvarious concentrations and cells of different path lengthsbetween 0-1 and 5cm. were used. All curves shown areaverages of duplicate runs.

Circular-dichroism measurements were kindly made byDr S. Beychok (Department of Biochemistry, ColumbiaUniversity) with a modified Jouan Dichrograph. Theperformance of the instrument is described by Beychok(1965) and Beychok & Fasman (1964). The partial rotations[bK], produced by the Kth dichroism band were calculatedby the procedure ofMoskowitz (1960), by assuming gaussianbands and using the Kronig-Kramers transform:

x

2[0%](,c. A[K= i eX jedx-2A+d

where [08] is the molar ellipticity at the maximum of theKth band, AR is the half band-width and x is (A-A%)/A%.The values of the integral were obtained from Mathe-matical Tables (Jahnke & Emde, 1952). The rotationalstrengths RX were calculated by using the relation:

RR 0 696 x 10-42.fj[00]Far-ultraviolet spectra were measured with a suitably

modified Beckman DK-2 recording spectrophotometer(Rosenheck & Doty, 1961) in the Laboratory of Dr W. B.Gratzer (Department of Biophysics, King's College,London). A stock solution of avidin (4mg./ml.) was diluted1:20 with either 0-105M-NaClO4 or 0-105N-HC104 in 1cm.cuvettes. The difference in E233 was followed. When therewas no further change in AE233, showing completion ofthe dissociation in HC104, the far-ultraviolet spectra weredetermined. A base line was determined with 0-2 cm.Suprasil cells (Unicam Instruments, Cambridge) containingsolvent (0-5ml. of 0-1M-NaCIO4 or 0-lx-HC104). Theappropriate avidin solution (0-2ml.) was then added to one

cell and solvent (0.2ml.) to the other by using calibratedCarlsberg pipettes (H. W. Pedersen, Copenhagen, Den-mark). The spectrum was measured and then, with avidinin NaClO4, 5,1t. of 0-4mm-biotin was added to both refer-ence and sample cells by using a lO,l. Hamilton syringe(Shandon Scientific Co., London, N.W. 10) and the spectrumwas redetermined. The mean extinction coefficients of thepeptide bond under the different conditions were calculatedby subtracting the extinction due to the amino acid sidechains from the total extinction, by using the data for theextinction coefficients given by McDiarmid (1965) and theamino acid composition given by Melamed & Green (1963).The method of calculation and the assumptions involvedare the same as given by Rosenheck & Doty (1961). Theabsorption of the acetamido groups of N-acetylglucosaminewas included as one ofthe terms in the side-chain absorption.

Sedimentation coefficients were determined in a Spincomodel E ultracentrifuge at 150 and at 59780 rev./min.

RESULTS

Since one of the main objects of the experimentswas to correlate the changes in environment of thetryptophan residues accompanying biotin-bindingwith changes in optical rotation, the rotation inthe ultraviolet region was first investigated. TheORD curves of avidin and the avidin-biotin com-plex are shown in Fig. 1 (curves A). Below 220m,the results obtained on the Bendix instrument weredifficult to reproduce. However, curves obtainedwith a Cary model 60 spectropolarimeter confirmedthe position of the extremum at 220m,u.

Superficial examination of the curves suggestedtwo positive Cotton effects centred at 278 and228m,u and this interpretation was confirmed bythe circular-dichroism measurements presentedbelow. These Cotton effects disappeared whenavidin was dissociated into sub-units in 6M-guanidine hydrochloride. In the curve shown(Fig. 1, curve C) a trace of an inflexion remains at280m,u. This was probably caused by the presenceof a small amount of undissociated protein since insome experiments it was absent. When OCA2 wasplotted against a (one-term Drude equation) theseresults gave a straight line with A,,= 216m,u,typical of the behaviour of a denatured protein orrandom-coil polypeptide (Urnes & Doty, 1961).With 0-1N-hydrochloric acid, which brings aboutpartial dissociation of the avidin (Green, 1963b,and Fig. 4 below), greatly diminished Cottoneffects were observed (Fig. 1, curve B). In contrastwith the effects of guanidine hydrochloride,formation of the avidin-biotin complex (Fig. 1,curve A) had little effect on the rotation, apartfrom a slight intensification of the 280m, Cottoneffect. The change in the other Cotton effect wasonly just outside the rather large limits of errorof the short-wavelength measurements. Since theeffect of binding of biotin was so small, little furtherwork was done on this aspect of the problem and

Vol. 100 615

N. M. GREEN AND M. D. MELAMED

220 260 300Wavelength (mpi)

Fig. 1. ORD curves of avidin and the avidin-bplex. Curve A: native avidin (0*86mg./ml.) in 0-05M-sodium phosphate buffer, pH6-8, for wavelengths above240m,; below this wavelength the concentration of avidinwas 0118mg./ml. in water; lcm. path length was usedthroughout; the broken curve is as for the continuouscurve A, but for the avidin-biotin complex. Curve B:avidin in 0.1N-HCl; above 240mu the concentration ofprotein was 0.88mg./ml., and below 240m,u it was 0.19mg./ml.; 1cm. path length was used throughout. Curve a:avidin after standing for 17hr. at room temperature in6M-guanidine hydrochloride and then dilution to 3M-guanidine hydrochloride; the protein concentration above240m,u was 10mg./ml., and below 240mz it was 1.2mg./ml.; 0 1 cm. path length was used throughout.

subsequent experiments were devoted to inter-preting the two unusual Cotton effects and toestimating the contribution of helical and random-coil conformations of the peptide bonds to therotation.

Circular dichroi8m. The curve for avidin (Fig. 2)showed two positive maxima. The short-wave-length band (228m,u) was symmetrical and was atthe same wavelength as the corresponding absorp-tion band of tryptophan, which, in avidin, islocated at about 225-226mju (Green, 1962). Theband at 274m,u was at a somewhat shorter wave-

length than the absorption maximum of avidin(282m,u). It is clear that there are two positiveCotton effects, confirming the conclusions drawn

2000

Ca)

1000

0 220 240 260 280 300

Wavelength (m,u)

340 Fig. 2. Circular-dichroism spectrum of avidin. Themeasurements were made in cells of 0.1 cm. and 1 cm. pathlength. Avidin (0-57mg./ml.) was dissolved in 0 IM-sodium

4otin com- phosphate buffer, pH6.9.

from the ORD curves. The absence of negativedichroism at wavelengths below 220m,u suggeststhat there is little or no right-handed a-helix inavidin.For the purposes ofthis section, it will be assumed

that both effects originate mainly from the absorp-tion bands of tryptophan. The arguments for thispoint of view are presented below. The circular-dichroism results shown were based on a meanresidue weight of 119 (Melamed & Green, 1963).However, if the dichroism is mainly caused bytryptophan residues it would be more informativeto calculate it per mole of tryptophan. The molarellipticity [0]228 becomes + 100 000° cm.2/decimole(the rotational strength R22s= 46 x 10-40 c.g.s.units) and [0]274 becomes +11 5000cm.2/decimole(1274= 9.3 x 10-40 c.g.s. units). These values areabout one order of magnitude greater than thosefor the Cotton effects of polytyrosine (Beychok &Fasman, 1964) or free tryptophan (Legrand &Viennet, 1965). Similar calculations may beapplied to the ORD Cotton effects, the values of[M] being multiplied by 31 to express them asrotations per mole of tryptophan. The amplitudesof the two effects [SM] become 150000° and130000 for the 228m,mu and 274m,u regions respec-tively. These values are less reliable than the

616 1966

COTTON EFFECTS OF AVIDIN

dichroism results as a guide to the rotational con-tribution of tryptophan since the peptide bondscontribute far more to the rotation than to thedichroism in this region.

Optical rotatory dispersion in the visible region. Itis now possible to make a more meaningful inter-pretation of the ORD data from the visible regionof the spectrum. These are plotted according tothe Moffitt-Yang equation (Urnes & Doty, 1961)in Fig. 3. The dispersion constants calculated fromthese plots are given in Table 1. Although thisprocedure is not strictly valid for native avidin,on account of the tryptophan Cotton effects, linearrelationships were observed down to 400m,u..Avidin dissociated into sub-units in 3M-guanidinehydrochloride gave ao= - 480 and bo= 0, as wouldbe expected for a denatured protein (Urnes & Doty,1961). However, native avidin in either water or3M-guanidine hydrochloride was unusual in givinga positive bo. Both this and the departure of theexperimental points from the straight line atshorter wavelengths (Fig. 3) were probably causedby the tryptophan Cotton effects. The partialrotations contributed by these effects were there-fore calculated from the circular dichroism by usingthe Kronig-Kramers transform (Moskowitz, 1960)and subtracted from the total rotation. Theresulting values should approximate more closelyto the true partial rotation of the peptide bond,although it should be remembered that they willinclude contributions from aromatic and other sidechains with absorption bands in the far ultraviolet.The Moffitt-Yang plot (Fig. 3) was linear down to313mmp and bo was now much lower (+67). Thisresult confirms the conclusion from the dichroism

measurements that there is little or no right-handedac-helix in avidin.

Far-ultraviolet absorption. The absorption spectraof avidin in 0- 1 M-sodium perchlorate, with andwithout biotin, and in 0- 1 N-perchloric acid areshown in Fig. 5. Perchloric acid, which is trans-parent below 200m,u, was used in place of guanidinehydrochloride to bring about partial dissociationof the avidin into sub-units. The extent of dis-sociation had previously been determined in 0-1 N-hydrochloric acid by sedimentation experiments(Fig. 4), which show that most of the avidin waspresent as sub-units (S20,,r 2- 1 s). The avidin-biotincomplex was not affected by the low pH.The differences between the three spectra were

very small. The presence of biotin had a negligibleeffect (less than 1%) below 225m,. Even in acid,where only about 30% of the 228m,u Cotton effectremained (Fig. 1), there was only a slight increasein the extinction. The shoulder at 220m,u is dueto tryptophan and the shifts in the 230m,u regionare in agreement with previously published dataon the difference spectra (Green, 1963a,b). Themean extinction coefficients of the peptide bondswere calculated by subtracting the contribution ofthe amino acid side chains. They are shown inFig. 5 in relation to the experimental values forac-helical and random-coil conformations of syn-thetic polypeptides (Rosenheck & Doty, 1961). Ifthe data are interpreted as suggested by Rosenheck& Doty (1961) one would conclude that between60 and 100% of the peptide bonds in avidin werein ac-helical conformation and that even in acid,where most of the aromatic Cotton effect has dis-appeared, about 60% of this conformation remainsintact. Such conclusions clearly conflict with the

0

p

pwo-

---- * 0 - 05 10 1 5 20

106/(A2-4)Fig. 3. Moffitt-Yang plots of ORD data of avidin. Themeasurements were made in 5cm. cells on solutions ofavidin (8mg./ml.) in 0-2M-sodium phosphate buffer,pH6-8. *, Avidin; A, avidin after subtraction of thearomatic contribution, as described in the text; *, avidin,dissociated into sub-units in guanidine hydrochloride (6M);the guanidine hydrochloride concentration was diluted to3 before the ORI0 was measured.

Fig. 4. Sedimentation of avidin and the avidin-biotincomplex in 0-1N-HCl. Upper curves: avidin (8mg./ml. in041 N-HCI), S20,W of fast component 4-55 s and that of slowcomponent 2-15s. Lower curves: avidin-biotin complex(7-4mg./ml. in 0-1N-HCl), S20,2, 4*55s. Photographs were

taken after (a) 76min. and (b) 140min.

0

-10o

--l

-5

0

-20

(a) (b)

Vol. 100 617

N. M. GREEN AND M. D. MELAMED10 0 ,_

0 8-0

6) 0

460

2-0O, . .

0 190 200 210 220 230 240

Wavelength (mu)

Fig. 5. Ultraviolet spectra of avidin and the avidin-biotincomplex. The upper pair of curves show the observedextinction coefficients, expressed as an average value per

peptide bond. The lower pair of curves show the extinctioncoefficients after subtraction of the contributions of theamino acid side chains. The upper reference lines at 190,197 and 205m,u give the extinction coefficients of thepeptide bond in a random-coil polypeptide, and thelower ones give the corresponding values for an xc-helix(Rosenheck & Doty, 1961). , Avidin and avidin-biotin complex in 0-1 M-NaClO4; * M., avidin-biotincomplex, when distinguishable from avidin; avidin

in 0-lN-HC104.

results of the ORD measurements. The discrepancyis discussed further below.

Near-ultraviolet absorption of avidin. It has beensuggested (Fasman, Bodenheimer & Lindblow,1964; Fasman, Landsberg & Buchwald, 1965) thatthe Cotton effects of the aromatic residues inpolytyrosine and polytryptophan may result fromelectronic interaction between neighbouring resi-dues. A further possible consequence of suchinteractions would be a decreased E280 value. Thepossibility that this might be so in avidin was

checked by following E28o when avidin was allowedto stand in 6M-guanidine hydrochloride. A 4.0%decrease in E280 was observed. This small effect issimilar in magnitude and direction to that observedon denaturation of many proteins. It can mostreadily be explained in terms of an effect of solvent(in this case the protein environment) on theextinction coefficients of the individual tryptophanresidues. For example, the extinction coefficient ofacetyltryptophan ethyl ester is increased by 10%on transfer from water to ethanol (G. H. Beaven,personal communication).

Streptavidin. The ORD curves of streptavidinand its complex with biotin (Fig. 6) show similarpositive Cotton effects in the neighbourhood of thetwo tryptophan absorption bands. The shorter-wavelength effect is somewhat further to the red

-20001

-4000

-6000

220 240 260 280

Wavelength (m,u)300

+5000

0

-1000 b

-2000

-3000

Fig. 6. ORD curves of streptavidin and the streptavidin-biotin complex. The streptavidin (0-31 mg./ml.) was

dissolved in 0-1M-NH4C04, pH8-0, for measurementsdown to 240m,u. Below this wavelength the solution was

diluted to 0.045mg./ml. All measurements were made incells of 1cm. path length. The left-hand ordinate scalerefers to measurements below 233m,u, and the right-handscale to measurements at higher wavelengths.Streptavidin; ----, streptavidin-biotin complex.

than with avidin and the negative extremum ismuch more marked.The effect of combination with biotin on the

short-wavelength Cotton effect is somewhat greaterthan with avidin, the rotation at the positiveextremum (232m,u) becoming considerably more

negative. As with avidin (Fig. 2), the circulardichroism of streptavidin (Fig. 7) shows twopositive bands, though the shorter-wavelength one

is shifted to the red relative to that in avidin. Incontrast with avidin the dichroism becomesnegative below 225m,u.

DISCUSSION

The small effect of biotin on the ORD curve ofavidin does not add any information to thatalready available from other spectroscopic studies,except in the negative sense that it excludes anychange in total amount of a-helix consequent on

the binding of biotin. The slight increase in

/I

618 1966

COTTON EFFECTS OF AVIDIN

amplitude of the 279m,u Cotton effect, like the redshift of the absorption spectrum, could be causedby an increase in the refractive index ofthe mediumsurrounding the tryptophan. The effect of biotinon the ORD of streptavidin is somewhat greaterand is considered below.The results provide no evidence for any right-

handed a-helix in avidin. There was no sign ofany negative contribution to the dichroism in the220m,u region, and, after correction for the con-tribution of the tryptophan Cotton effects, bo wasalmost zero. In spite of the low value of bo, it isdifficult to reconcile the results with a predomi-nantly random arrangement of peptide bands.Randomly coiled polypeptides show rotations ofabout - 3000° in the 230m,u region and there isno sign of such a strong negative effect in the

Q0

0-;cq

+2000

+ 1000

0

-1000

230 250 270 290

Wavelength (m,)

Fig. 7. Circular-dichroism spectrum of streptavidin. Themeasurements were performed in cells of 1 cm. of 0-1 cm.path length in 0-05M-NH4Cl adjusted to pH9-0 with aq.NH3. The streptavidin concentration was 0-05mg./ml.

avidin ORD curve. Both the Cotton effects arecentred on a flat, almost zero, background and noton a rapidly descending negative one. It appearsthat other influences must be present that in partneutralize the negative circular-dichroism band ofthe random coil in the far ultraviolet (Holzwarth &Doty, 1965). At least two factors may operate inthis fashion: (1) the absorption band of tryptophanin the far ultraviolet (e193 20000); (2) the arrange-ment of some ofthe peptide bonds in a ,-conforma-tion (Urnes & Doty, 1961; Imahori, 1963). It isalso possible that there is very little random coilpresent and that most of the peptide bonds are ineither a left- or a right-handed helical conformation.The absorption band of tryptophan in the farultraviolet almost certainly makes some contribu-tion to the rotation, though it is unlikely to belarge enough to counteract a random-coil peptidecontribution. The evidence for the presence of#-structure in globular proteins is slight, butImahori (1963), Callaghan & Martin (1963), Martin(1965) and Sarkar & Doty (1966) have producedsome evidence for its presence in y-globulin.Table 1 shows that the dispersion constants ao, boand A, calculated for the partial rotation of thepeptide bonds of avidin are remarkably similar tothose of y-globulin (Gould et al. 1964), which inturn differ markedly from those of most otherproteins (Umes & Doty, 1961). The far-ultravioletspectrum provides a further point of resemblancebetween avidin and y-globulin, since both proteinhave mean peptide extinction coefficients thatsuggest a high proportion of helical structure inthe molecule (Rosenheck & Doty, 1961), in conflictwith the evidence from their ORD.

There are, however, two observations that aredifficult to reconcile with the presence of any largeproportion of fl-structure. First, neither avidin(N. M. Green, unpublished work) nor y-globulin(Gould et al. 1964) shows any sign of a 1610cm.-lshoulder in the infrared spectrum, which would becharacteristic of the amide I absorption of peptidebonds in this conformation. Secondly, the extinc-tion coefficient of the peptide bond in a fl-structure(E197 7500) is higher than in random conformation(E197 6300) (Rosenheck & Doty, 1961), whereas

Table 1. Di&per8ion constants of avidin compared with those of rabbit y-globulin

[<X]D A aoAvidin -19.80 -148Avidin after subtraction of aromatic contribution -50-5 190 -325Avidin in guanidine hydrochloride (3M) -76-1 210 -480y-Globulin* (rabbit) (pH7-3) -58 195 -275y-Globulin* (rabbit) (pH 1-9) -76 -350

bo1466743510

* Gould, Gill & Doty (1964). Further data on human y-globulin and Bence-Jones proteins are given by Urnes& Doty (1961) and Callaghan & Martin (1963).

Vol. 100 619

N. M. GREEN AND M. D. MELAMED

avidin (E197 4800) and y-globulin (E197 5400) bothshow values that are unusually low for proteinscontaining little a-helix. The presence of equiva-lent amounts of left- and right-handed helix iny-globulin is unlikely, since the peptide hydrogenatoms exchange rapidly. One cannot exclude thispossibility in avidin, though it seems an unlikelyone.

Aromatic contribution to the Cotton effects. Inview of the coincidence of the dichroism bands ofavidin with the absorption bands of tryptophanthey can be assigned to this chromophore withsome certainty. The shift of the long-wavelengthmaximum to 274m,t (avidin) or 270m, (strept-avidin) appears to be within the normal range,since Legrand & Viennet (1965) have observedthe corresponding band of free tryptophan to be at266m. Small contributions from tyrosine, whichis responsible for 5% of the E280 value, and cystine(Beychok, 1965) cannot be excluded. A left-handed a-helix would produce positive dichroismin the 230m,u region, but by analogy with theright-handed helix (Holzwarth & Doty, 1965) onewould expect the band to be much broader and tobe located at shorter wavelengths than was in factobserved with avidin. However, this possibilityshould be kept in mind. The results given bystreptavidin are consistent with the presence of asmall amount of right-handed a-helix, whosenegative dichroism peak would be at 219mp(Holzwarth & Doty, 1965) and would partiallycancel the positive dichroism caused by tryptophan.This would account for the dichroism of strept-avidin becoming negative below 225mp, and forthe asymmetry of the 232m,u band. Similarly, itwould explain the differences between the ORDcurves of avidin and streptavidin. Both the strongnegative extremum of the streptavidin ORD at217m, and the apparent red shift of the positiveextremum to 238m, could be explained by thepresence of two overlapping Cotton effects ofopposite sign. The rotation at 238m,u was muchmore affected by biotin than was that of avidin,and its direction suggested a small increase in theamount of a-helix. This would have little effect inthe 280m,u region, where the ORD curves forhelix and coil lie close together (Blout, Schmier& Simmons, 1962).

Several reports of Cotton effects caused bytyrosine and tryptophan in proteins have appeared.The earliest examples were small inflexions in the280m,u region shown by tobacco-mosaic virus(Simmons & Blout, 1960) and carboxypeptidase(Fujioka & Imahori, 1962). Larger effects havebeen observed with carbonic anhydrase (Myers &Edsall, 1965), lysozyme (Glazer & Simmons, 1965)and cytochrome c (Urry & Doty, 1965). Theamplitude of the effects at 290m,u in lysozyme and

carbonic anhydrase was about 4000°/mole oftryptophan (avidin 130000°). So far no otherprotein has shown a positive Cotton effect in the230m,u region, presumably because any such effectwould be masked by the much stronger negativeCotton effect produced by a small proportion ofright-handed ac-helix. It is the absence of ac-helicalarrangement of peptide bonds that causes the twoavidin effects to stand out so clearly. Streptavidinprovides an intermediate case in which the twoopposing effects are of similar magnitude, but,being centred at slightly different wavelengths,only partially cancel each other.

In all these examples the effects disappear whenthe protein is denatured, showing that much ofthe dissymmetry is provided by the tertiarystructure of the protein, though not by c-helix inavidin. This is not surprising in view of the Cottoneffects observed in the visible region when manysymmetrical chromophores are bound to proteins(cf. Ulmer & Vallee, 1965); indeed, it is the weaknessor absence of aromatic Cotton effects in proteinsthat is remarkable. Tyrosine and tryptophan arehydrophobic residues, and are likely to be situatedin the interior of the protein in a rigid and asym-metric environment. It is possible that whenseveral such residues are present their partialrotations cancel, but this is unlikely to be general.The most likely alternative hypothesis is that onlycertain types of asymmetric pertubation canproduce strong Cotton effects. Fasman et al. (1964,1965) studied the ORD of solutions and films ofpolytryptophan, polytyrosine and of their co-polymers with glutamic acid. Tyrosine co-polymersshowed Cotton effects only when they containedmore than 20% of tyrosine. Fasman et al. (1964,1965) suggested that this was because the Cottoneffects were dependent on tyrosine-tyrosine inter-actions. Similar interactions between tryptophanresidues were invoked to explain the multipleCotton effects observed in the 280m, region andthe 10% hypochromism of polytryptophan. [Thislatter effect was not observed by Shifrin (1961)with polytryptophan samples of lower molecularweight.] If strong tryptophan-dependent Cottoneffects require interaction between two or moreclosely situated residues this could explain theirrarity. However, such interactions could lead bothto hypochromism and to a splitting of the aromaticdichroism bands into two components of oppositesign (cf. Tinoco, 1964; Van Holde, Brahms &Michelson, 1965). There is no indication of such asplitting of the dichroism bands of avidin, nor didthe slight decrease in E28o accompanying the dis-sociation of avidin into sub-units support thishypothesis. Although the present state of theoryis such that one cannot exclude electronic inter-actions between tryptophan residues on these

620 1966

Vol. 100 COTTON EFFECTS OF AVIDIN 621

grounds, it seems more likely that the Cottoneffects are due to some other perturbation of thearomatic chromophore.

The authors thank Dr F. J. Wolf for the sample ofstreptavidin, Dr W. S. Jennings for assistance with some ofthe ORD measurements, Dr S. Beychok for the dichroismmeasurements, Dr W. B. Gratzer for assistance with thefar-ultraviolet-absorption measurements, and Dr W. R.Carroll for the sedimentation-velocity measurement.N. M. G. also thanks the Wellcome Foundation for aResearch Travel Grant, and was a Visiting Scientist of theU.S. Public Health Service 1963-64.

REFERENCES

Beychok, S. (1965). Proc. nat. Acad. Sci., Wash., 53, 999.Beychok, S. & Fasman, G. D. (1964). Biochemistry, 3, 1675.Blout, E. R., Schmier, I. & Simmons, N. S. (1962). J.Amer. chem. Soc. 84, 3193.

Callaghan, P. & Martin, N. H. (1963). Biochem. J. 87, 220.Chaiet, L. & Wolf, F. J. (1964). Arch. Biochem. Biophys.

106, 1.Fasman, G. D., Bodenheimer, E. & Lindblow, C. (1964).

Biochemistry, 3, 1665.Fasman, G. D., Landsberg, M. & Buchwald, M. (1965).

Canad. J. Chem. 43, 1588.Fujioka, H. & Imahori, K. (1962). J. biol. Chem. 237, 2804.Glazer, A. N. & Simmons, N. S. (1965). J. Amer. chem.

Soc. 87, 2287.Gould, H., Gill, T. J. & Doty, P. (1964). J. biol. Chem.

239, 2842.

Green, N. M. (1962). Biochim. biophys. Acta, 59, 244.Green, N. M. (1963a). Biochem. J. 89, 599.Green, N. M. (1963b). Biochem. J. 89, 609.Holzwarth, G. & Doty, P. (1965). J. Amer. chem. Soc. 87,

218.Imahori, K. (1963). Biopolymers, 1, 563.Jahnke, E. & Emde, F. (1952). Table of Higher Functions,

p. 26. Leipzig: B. G. Teubner Verlag.Legrand, M. & Viennet, R. (1965). Bull. Soc. Chim. Fr.

p. 679.McDiarmid, R. F. (1965). Ph.D. Thesis: Harvard Univer-

sity.Martin, N. H. (1965). Proc. Roy. Soc. Med. 58, 41.Melamed, M. D. & Green, N. M. (1963). Biochem. J. 89,

591.Moskowitz, A. (1960). In Optical Rotatory Dispersion, p.

151. Ed. by Djerassi, C. New York: McGraw-HillBook Co.

Myers, D. V. & Edsall, J. T. (1965). Proc. nat. Acad. Sci.,Wash., 53, 169.

Rosenheck, K. & Doty, P. (1961). Proc. nat. Acad. Sci.,Wash., 47, 1775.

Sarkar, P. K. & Doty, P. (1966). Proc. nat. Acad. Sci.,Wash. 55, 981.

Shifrin, S. (1961). Nature, Lond., 192, 750.Simmons, N. S. & Blout, E. R. (1960). Biophys. J. 1, 55.Tinoco, I. (1964). J. Amer. chem. Soc. 86, 297.Ulmer, D. & Vallee, B. L. (1965). Advanc. Enzymol. 27, 37.Urnes, P. & Doty, P. (1961). Advanc. Protein Chem. 16,40.Urry, D. W. & Doty, P. (1965). J. Amer. chem. Soc. 87,

2756.Van Holde, K. E., Brahms, J. & Michelson, A. M. (1965).

J. molec. Biol. 12, 726.