Eur. J. Biochem. 28, 174-182 (1972)

Angiotensin I1 Conformations Infrared and Raman Studies

Serge FERMANDJIAN, Pierre FROMAGEOT Service de Biochimie, DBpartement de Biologie, Centre #Etudes NuclBaires de Saclay

Anne-Marie TISTCHENKO, Jean-Pierre LEICKNAM

S.R.I.R.M.A., DBpartement Recherche et Analyse, Gif-sur-Yvette

and Marc LUTZ Service de Biophysique, DBpartement de Biologic, Centre d'Etudes NuclBaires de Saclay

(Received December 23, 1971/March 29, 1972)

Vibrational spectra of angiotensinamide 11, some constitutive peptides, and an analog, are presented. Comparison of data for the solid hormone with those for the subunits yielded amide vibrations of the former in the 500-1700 cm-l range. Angiotensinamide I1 was shown to adopt a preferential antiparallel ,!? conformation in both the solid state and concentrated aqueous solu- tions. Conformational information obtained by other methods are discussed together with the present results.

The data pertinent to the behavior of hormone specific receptor sites [l] delineate with increasing accuracy their mechanical character. They require an agonist of very precise structure, which by being locked in the receptor frame may be submitted to structural changes and induce conformational varia- tions. To define the structure of the hormone itself before and after such an event in the receptor is an exciting challenge and the more so when the hormone is a peptide. Therefore, a better knowledge of the conformational characteristics of a peptidic hormone placed in a variety of conditions seems to be a pre- liminary step towards the solution of the problem defined above.

In addition it is very likely that no additive energy besides that resulting from interactions with the receptor, the surrounding medium and within the molecule itself, is required to give the hormone its specific structure.

On these bases we undertook study of angioten- sin I1 conformations as this molecule seemed to afford an interesting example of the environmental in- fluence. As a matter of fact, this octapeptide of linear primary structure appears to adopt in dilute aqueous solution two main conformations detectable by di-

Abbreviations. [Ile - 5,Arg - 8]angiotensinamide, angio- tensinamide with Val-5 replaced by isoleucine and Pro-8 replaced by arginine.

alysis experiments [2-41 and by gel filtration (151 and unpublished results, S. Fermandjian). Organic solvents like trifluoroethanol, alkaline methanol or ethanol promoted a considerable structuration of the dissolved angiotensinamide 11 into a predominant form, which, by circular dichroism measurements, has been tentatively assigned as a cross /? structure [6] , defined according to Pysh [7] and Timasheff et al. [S]. The required folding of the molecule was assigned principally to the hydrophobic interactions between the 3 and 5 valine and the 8 phenylalanine side-chains and to the cis conformation of the histidine-proline peptide bond. Interestingly enough, dry films of angiotensin I1 yielded circular dichroism spectra also reminiscent of a pleated sheet structure. We wished to check these conclusions with independent techni- ques in order to assess the validity of the approach used.

The present paper is devoted to a study of angio- tensin I1 in the solid state and concentrated aqueous solutions by infrared and Raman spectroscopy.

Early infrared work had suggested empirical relations between the vibration bands of the peptide bonds and the structure of polypeptides [9,10]. The theory of Rliyazawa [ll, 121, despite some possible inadequacies [ 131, rationalized the amide bands conformation correlations for homopolypeptides [14], and also permitted fruitful investigations of protein

Vo1.28, No.2,1972 S. FERMANDJIAN e t al. 175

structure [8,12,15,16]. As far as short peptide chains are concerned, good agreement has been found, through Miyazawa's theory, between infrared spectra and conformation of short homopeptides, like for instance hexa-L-alanine [i7], octa-benzylglutamate [18,19]. Similarly, Smith etal. [20], Sutton and Koenig [17] were recently able to deduce valid structural conclusions from Raman spectra of glycine and alanin oligomers.

In the particular case of angiotensin 11, the ques- tion was whether predictions initially established on consideration of homopeptides of infinite length would still hold for a short heteropeptide. A second difficulty was to recognize on both infrared and Raman spectra the amide vibration bands among those arising from the variety of the side-chains present.

We therefore used a two-step method; the first was to compare the vibrational spectra of angiotcn- sin I1 with those of shorter constitutive peptides and those of the free amino acids ([21-271, and this work). From these comparisons we were able to iden- tify bands which were certainly related to backbone vibrations. In a second step, the observed amide frequencies were confronted with the current data. We thus were led to the double conclusion that the classical predictions most probably hold for angio- tensin 11, and that in the solid state as well as in concentrated aqueous solutions, angiotensin I1 adopt- ed preferentially an antiparallel structure.

MATERIAL AND METHODS AngiotensinamideII has the amino-acid sequence :

Asp (NH,)-Arg-Val-Tyr-Val-His-Pro-Phe. The pepti- des examined were (1-7), (1-4), 4-8), (5-8), (6-7) and (7-8); these and angiotensinamide 11, were provided by Dr Riniker (Ciba-Geigy, Basel). The [Ile-5, Ala-8langiotensinamide was a gift of Dr Bumpus (Cleveland Clinic, Cleveland). Crystalline samples of angiotensinamide I1 were prepared from solutions in trifluoroethanol and alkaline methanol as previously described [28]. Replacement by deute- rium of the labile hydrogen atoms was carried out by repeated exchange with 2H,0. The deuterated materi- al was handled in a dry nitrogen atmosphere.

Infrared Spectrometry A Perkin Elmer 225 spectrometer was used. Its

resolution was of the order of 1 cm-1 from 4000 to 600 cm-l and 2 cm-l from 600 to 200 cm-l. Band positions were estimated to be exact within -J= 2 cm-l. Solid peptide samples were examined in KC1 pellets, and, where sufficient amounts of product were avail- able, in oil mulls of Nujol or Voltalef. Inclusion and compression of angiotensin I1 in a pellet certainly affected its conformation, although slightly. Some variations in relative intensities were noted in the

chain vibration regions, between spectra of the same sample, when oil suspended or pelleted. These varia- tions did not affect general conclusions drawn from infrared examination of pelleted peptides. It should be mentioned that we were unable to obtain films of angiotensin I1 oriented well enough for infrared di- chroism investigation, probably because of its length and conformation. Aqueous solutions were not studied by infrared, the Raman effect being more appro- priate for such work.

Raman Spectrometry The spectra were run on a PHO type CODERG

Spectrometer : the excitation source was a He-Ne- Laser giving an average 70 mW power a t 632.8 nni. Detection was carried out with an 5.20 photocathode photomultiplier, cooled down to -5 "C, and equipped with a magnetic defocalization of noise electrons issuing from the peripheral zones of the photocathode ; the photomultiplier was associated with a direct- current amplification chain. A few hundred micro- grams of solid sample were gently packed a t the bevel-edged tip of a capillary tube, forming a flat diffusing surface. The latter was oriented a t about 60" to the detection optics axis. The excitating beam was oriented at either 90" or 180" to the same axis. The very intense diffusion of exciting light by white, finely powdered samples was attenuated by various combinations of short-cut filters.

Solutions were contained in capillary tubes of i7-mm optical path and of 13-pl capacity. Both ends were shut by glued microscope coverslips perpendic- ular to the tube axis; multipass of exciting light was thus allowed. Solutions were filtered on 0.45-pm Millipore filters. Sample concentration were varied from some 10 to some 100mM, without noticeable spectral changes.

Most of the samples studied were found to fluor- esce strongly even under the red 632.8-nm excita- tion. Long preirradiation of the fluorescent sample, however, under the exact conditions of spectrum recording, often reduced fluorescence intensity to tolerable levels. During such long irradiations (up to 30 h) of solid samples, a dry nitrogen flow was estab- lished in the sample compartment ofthe spectromcter.

Resolutions of 6 ,8 or 12 cm-l were used, and time constants up to 16 sec. Global uncertainty on band positions was estimated f 3 cm-l for sharp and strong bands, and up to & 5 cm-l for the weakest bands or shoulders observed. Spectrometer calibration was checked on benzene spectrum.

RESULTS Representative spectra of solid angiotensinamide

11, constitutive peptides and [Ile-5, Ala-8langioten- sinamide, from 1800 to 400 cm-l, are given in Fig.1

176 Angiotensin I1 Conformations Em. J. Biochem.

A

\

1800 1500 1000 v (an-’)

Fig. 1. Infrared spectra of solid angiotensinamide 11 ( A ) from H,O ( B ) from trifluoroethanol, (C) from 2H,0 solutions, in

RC1 pellets. (B’) the same as (B) but in Nujol mull

imidazole ring of histidine were relatively weak compared to the preceding bands.

The analysis of both infrared and Raman spectra of solid samples will be presented for each successive spectral region characteristic for the amide vibrations, the Raman spectra of solutions being considered separately.

Amide-I Region: 1600-1700 em-l Infrared spectra of solid angiotensin I1 yielded a

broad band containing a t least two components a t 1637 and 1690 cm-l (Fig.1 and 2 ) . On each of the 1-4, 5-8, and 4-8 subunits spectra (Fig.3), a very broad absorption occurred a t 1640 to 1645 cm-l, which was certainly multicomponent. A shoulder appeared a t 1600 cni-l in those three spectra, while a second one, located a t 1615 cm-l, was present only for the tyrosine-containing peptides. These last two vibrations were attributed to aromatic-ring quadrant stretching [21]. Thus, when going from the octa- and heptapeptides to any of the penta- and tetrapeptides studied, we observed that the 1685cm-l shoulder, if not totalIy disappeared, a t least considerably decreased in relative intensity, and that the 1637 cm-l band intensity also decreased relatively to recognized

I I I I ( I I I I

1800 400

Fig.2. Infrared of solid angwtensinarnide I I from H,O solution, Nujoll mull v (cm” )

to 8. We may first note that the side-chain vibrations represented a large proportion of the bands observed in both infrared and Raman spectral. I n the latter, the strongest vibrations were indeed easily attributed to ring vibrations of phenylalanine and tyrosine, or to superposition of vibrations of identical groupings present in different amino acids, e.g. the 1440-1480 cm-l cluster of bands, a superposition of CH, asym- metric deformation of valine, and of CH, scissoring of the 6 other amino acids. Contributions of the

A table of the Raman frequencies observed on angio- tensin I1 and related peptides spectra has been prepared, and may be obtained from M. L.

side-chain bands (Fig.4). These bands then most likely arose mainly from peptide-bond vibrations. In particular, any other possible origin, such as an

asymmetric vibration of the -C< grouping of

arginine [25], or an amino group deformation, was thus excluded as a preponderant contribution to the 1685-~m-~ band.

Crystalline angiotensinamide I1 yielded essentially the same spectral features as the amorphous powder. In Nujol mulls, however, crystalline hormone had an enhanced 1685 cm-1 shoulder, as compared to the amorphous material (Fig. 1 B, and 2).

N

N

Vo1.28, No.2, 1972 S. FERMANDJIAN et al. 177

W C Fig.3. Infrared spectra of ( A ) Asn-Arg- Val-Tyr, (B) Tyr- Val-His-Pro-Phe, ( G ) Val-His-Pro-Phe solid from 2H,0

solutions, KCl pellets

t

1800 1500 1000 Y (cm-')

Fig.4. Infrared spectra of ( A ) Asn-Arg-Val-Tyr, (B) Asn- Arg- Val-Tyr-Val-His-Pro, ( C ) angiotensinamide I I . Solid

from H,O solutions, KCI pellets

1 1800 1500 1000

Y (ern-')

Fig.5. Infrared spectrum of solid [Ile-5, Ala-8]angiotensin- amide from H,O solution, KCI pellet

1 ' 1 ' 1 ' 1

A

.rcr

L d - . L , # . # , \ . % . j , # , . I , < . / . >

1800 300 v (cm-')

Fig.6. Representative Ramun spectra of ( A ) angiotensinamide I I , (B) Asn-Arg- Val-Tyr- Val-His-Pro, ( C ) [Ile-5, Ala-81-

angiotensinamide solids from H,O solutions

The 1-7 constitutive peptide and the [Ile-5, Ala-81-angiotensinamide yielded infrared spectra es- sentially similar to those of angiotensin 11, all over the spectral range here studied (Fig. 4 and 5).

Raman spectra of angiotensin I1 in the solid state showed two groups of bands between 1600 and 1700 cm-l (Fig. 6). Resolution of these bands into 7 compo- nents was achieved for amorphous powdered angio- tensin I1 (Fig. 7). Crystalline preparations, due to strong fluorescence, have given poor quality spectra, on which it was only possible to note a close similar-

178 Angiotensin I1 Conformations Eur. J. Biochem.

2

1700 1600 v (cm-')

Fig.1. Raman spectra of anzide-I region, redrawn after back- ground subtraction, solid samples. Curve (l), angiotensin- amide I1 : (2) Asn-Arg-Val-Tyr-Val-His-Pro, (3) Asn-Arg- Val-Tyr; (4) Val-His-Pro-Phe; (5 ) addition of curves (3) and (4). Individual intensity scales were chosen for each of the curves (1) to (4) so that the heights of the 1614-cm-1 tyrosine peaks in curves (lo), (2) and ( 5 ) were equal. Approximate matching of the intensities of common side-chain vibrations

was thus obtained for all 5 curves

ity in the overall profiles with the amorphous angio- tensin I1 spectra. The 1-4 and 5-8 tetrapeptides (Fig.7) each yielded one weak and broad band containing two components around 1655- 1660 cm-1 and 1670- 1675 cm-l. Two intense diffusions observ- ed a t 1605 and 1614 cm-l on angiotensin I1 were attributed to ring vibrations of phenylalanine and tyrosine, respectively (quadrant stretching). The five remaining vibrations, 1635, 1649, 1660, 1668 and 1689 cm-l probably corresponded to the peptide bond, as they all were absent, or of much lower relative intensity, in the subunits spectra. Significant contributions of side-chain or end-group vibrations, such as ring vibrations of histidine [22,26] or amino group deformation, were thus excluded.

The 1-7 heptapeptide, in addition to the tyro- sine 1614-cm-1 band, exhibited five bands a t 1630, 1647, 1662, 1672 and 1685 cm-l homologues of those observed in the octapeptide, and similarly attributed to the peptide bonds (Fig.6 and 7). It should be pointed out that [Ile-5, Ala-Slangiotensinamide had a Raman spectrum very similar to that of the 1-7 heptapeptide, with one exception: a relative decrease of the 1660-cm-1 band. Such variations of the 1660- cm-l band intensity were also occasionally observed with different amorphous samples of angiotensinami- de 11.

Amide 11 Region: 1500-1550 cm-l Infrared spectra of solid angiotensinamide pres-

ented two bands centered a t 1515 and 1545 6111-1.

The first was observed only in the peptides contain- ing tyrosine and was not shifted on deuteration. It might thus be attributed to one of the components of the semi-circle stretching of para-&substituted benzene rings. The second band a t 1545cm-l was also observed in the heptapeptide and decreased in relative intensity when shorter peptides were examin- ed. This band appeared to be made of t,wo compo- nents located between 1530 and 1555 cm-l and might correspond to the peptide bond.

In the present frequency range, Raman spectra showed weak bands only. Three bands observed with angiotensinamide 11, centered a t 1507, 1520 and 1550cm-l were not detectable with any of the smaller peptides and were tentatively attributed to backbone vibrations.

Amide 111 Region: 2200-2350 cm-I Solid angiotensin I1 displayed, in the infrared,

a very broad absorption composed of a t least three bands, between 1235 and 1270 om-l. These bands were resolved on spectra obtained from nujol mulls, and were located at 1235, 1248 and 1270cm-l (Fig.2). A similar cluster of bands was observed for the 1-7 subunit, slightly displayed towards low frequencies (1225-1245 cm-l). Spectra of the 1-4 and 4-8 subunits had a band centered on 1250 cm-l very broad in the case of the 4-8 subunit. The 5-8 subunit had a weaker absorption in this region. The 1235- 1270-cm-1 absorption appeared thus a.s a superposition of backbone and side-chain vibrations, whose relative contributions were difficult to dis- criminate.

Raman spectra ofangiotensin I1 showed a medium intensity group of bands between 1220 and 1290 cm-l and a second one between 1320 and 1360 cm-l. The first group was resolved into five bands 1236, 1251, 1264, 1274 and 1282 cm-1 (Fig.8). Some side-chain vibrations are likely to bring significant contributions to those observed bands, namely in tyrosine (1251, 1264 cm-l), histidine (1274 cm-l), and proline (1236, 1264 cm-l). As a matter of fact, polarization spectra of angiotensinamide I1 in aH,O indicated two components centered a t 1257 and 1264 cm-l. As amide-I11 frequencies are expected to shift markedly on deuteration, one might conclude on the participa- tion of side-chain vibrations in the group of bands investigated. However, backbone contributions were also present, since the relative intensities of the 1236 and 1274 cm-l bands were higher in the octapeptide spectrum than in any of those of the subunits (Fig. 8). It was also noted that the vibrations of histidine (e .g . 1490 cm-l) and proline rings (e.g. 922 cm-l) appeared only weakly, Sat all, throughout the spectra of angiotensin 11. These remarks showed that a t least the 1236 and 1274-cm-1 bands originated, in part, from amide-I11 vibrations.

Vo1.28, No.2,1972 S. FERMANDJIBN et al.

A

2;\

179

I 1300 1200

v (m i ’ )

Fig. 8. Raman spectra of amide-III region, redrawn after background subtraction, solidsamples. Curves numbered and intensity matched as in Fig.7; the heights of the Tyr + Phe 1207-cm-1 peaks were made equal for curves (1) and (5 ) , and those of the

approx. 1340-cm-1 bands for curves (1) and (2)

The heptapeptide yielded vibrations a t 1237, 1256, 1270, 1278 and 1290 cm-l and the [Ile-5, Ala-81- angiotensinamide gave bands a t 1233, 1252, 1266, 1278 and 1290 cm-1. Thus again they exhibited great similarities with angiotensinamide 11.

The second group of bands between 1320 and 1360 cm-’ corresponded well to the superposition of the frequencies observed on the 1-4 and 5-8 peptides spectra (Fig. 8). They were not displaced on deuteration and had therefore to be assigned to side- chain vibrations.

Other Regions (below 850 cm-l)

In infrared as well a0 in Raman spectra of angio- tensin 11, a certain number of weak to medium bands appeared in this zone. Some of them could be attri- buted to side-chain vibrations, e.g., in Raman spec- tra,: thee,frequencies 624, 750, 821 cm-l were attri- buted to phenylalanine, and those observed a t 647, 715, 832 and 854cm-1 to tyrosine. On the other hand, only a few bands could be attributed to amide vibrations with reasonable probability, namely 579, 596 and 690 om-l on Raman spectra, and 700 cm-l on infrared spectra of angiotensin 11. The 700-cm-1 infrared band arose mainly from out-of-plane ring bending of phenylalanine, but was seen also on the 1-7 peptide spectra. Moreover, it decreased in rela- tive intensity on deuteriation of the octapeptide.

Aqueous Xolutions. Raman spectra have been recorded for angiotensinamide I1 only, as concentrat- ed solutions were required (Fig.9). The amide-I region, although perturbed by the H,O deformation vibration, yielded some valuable spectral features. The 1685-cm-1 shoulder was still present, and the main band, culminating at 1668 cm-l, probably

contained components a t 1635, 1650 and 1670 cm-l. A broad band, centered on 1655 cm-l was observed in 2H,0 solutions, and was decomposed into 1645 and 1667-cm-1 vibrations. A small shoulder was still observed a t 1680-cm-l. The amide-I1 region did not display diffusions intense enough for possible compa- rison between different spectra.

In the amide-Ill region, angiotensin-I1 yielded a spectrum similar to that of the solid, in respect to both by the number and the position of bands; in 2H20, two bands were distinguished, 1257 and 1264-cm-l. By subtraction, vibrations observed in H,O a t 1238, 1273, 1285 and 1312cm-l might be attributed to amide-I11 vibrations. In heavy water, a reinforcement of the 900-980-cm-1 region, with faint shoulders a t 918, 964 and 981 cm-l, not ob- served in H,O solutions, might be attributed to amide-111 vibrations. Below 850 cm-l, the same remarks were made as for solid spectra. In H,O, two weak bands, a t 608 and 660cm-l, absent on solid spectra, were tentatively ascribed to backbone vibra- tions.

DISCUSSION

Out of the complex vibrational spectra recorded for solid angiotensinI1 we identified a number of bands which were attributed to vibrations of the peptide bonds. This first goal was reached through comparison with the spectral characteristics of angio- tensin constitutive peptides and amino acids; no other kind of information wa0 used a t this stage. The next step was to compare the data so collected with those obtained from other peptides or predicted from semi-theoretical calculations 1191.

180 Angiotensin I1 Conformations Eur. J. Biochem.

Fig. 9. Raman spectra of ( A ) 2H,0 solution, and (B) H,O solution of angwtensinamide 11, wmcentratwn 200 rnM, pH 5.5

The first striking fact was the multiplicity of amide bands observed in each of the three amide I , I1 and I11 regions. Such a multiplicity suggested by itself a j 3 conformation (or a mixture of conforma- tions) since, from studies on polypeptides, it has been shown [11] that the splitting of amide bands is structure-dependent. More precisely the highest multiplicity, 4 components, is obtained for j 3 confor- mations.

Secondly, the frequencies, and the relative inten- sities of the amide bands observed for angiotensin I1 also favoured the type structure hypothesis. I n the amide-I region, the presence, on infrared spectra, of an intense band around 1620- 1635 cm-l, together with a less intense shoulder around 1685 cm-l, is considered characteristic of an antiparallel j 3 struc- ture, for long-chain [11,29] as well as for short-chain polypeptides [20,30], and for proteins [8, 151. Miya- zawa has established a now classical set of equations [19] which interconnect the four components of each of the amide-I and I1 bands, for a polypeptide under j 3 conformation, provided the hypothesis of localised vibrations is satisfied. From these we may extract :

v (0,O) + v (n,n) = v (0,n) + v (n,O) = ZY,. (1)

In the hypothesis of an antiparallel j 3 structure, v (O,O), a symmetrical vibration, is infrared inactive, and is the most intense among the four components which should be observed in Raman. On angiotensin I1 Raman spectra, both frequencies a t 1668 and 1660 cm-1 corresponded well to these criteria and might each represent the v ( 0 , O ) component; leaving for a moment this ambiguity, we attributed the 1685-cm-1

vibration to v (O,n), as the highest amide-I frequency observed on our spectra. The 1635-cm-1 vibration, intense in the infrared, very weak in the Raman, might be ascribed to the v (n,O) component. Intro- ducing then the values v (0,n) = 1685 cm-l, v (n,O) = 1635 cm-l and v (n,n) = 1649 om-l, in Eqn (1) we obtained v (0,O) = 1671 cm-l. This figure was close to the observed 1668-cm-1 frequency, which thus was assigned to the v (0,O) Vibration. Note that we departed here from the initial attribution of v (0,O) a t 1650 cm-l, originally made from infrared spectros- copy alone [15].

Also, we found yo = 1660 cm-1 for the theoretical frequency of the amide group still submitted to H- bonding but no longer to vibrational interactions. That frequency was close to the classical [12] and approximate [31] value of 1658 cm-1 deduced from infrared spectra of nylon 66 and polyglycine. Finally, we calculated the intrachain, D,, and interchain, Df1, coupling terms from Miyazawa's original relations. The values D, = 18 cm-l, and D,' = -7 cm-l were of the expected signs and relative absolute values [13]. Such coherence in those results brought further support for an antiparallel j 3 structure, with the corresponding set of amide-I frequencies located a t 1635, 1649, 1668 and 1685 cm-l.

The 1660-cm-1 amide frequency observed in the Raman thus remained unattributed. It would well correspond to an unordered fraction, as indicated by the following arguments: (a) it was close to the theoretical value [12] ; (b) the relative intensity of the observed band appeared to vary among different samples of solid angiotensin 11. (c) the 1-4 and 5-8 constitutive peptides spectra showed a shift of the

Vo1.28, "0.2, 1972 S. FERMANDJIAN et sl. 181

main amide-I band, together with the decrease of the 1685-cm-1 shoulder. This displacement was toward higher frequencies on the infrared spectra, while in Raman the maximum shifted down to 1655- 1660 cm-l ; such a phenomenon might reflect an increase of the unordered fraction in those tetra- peptides, as compared to angiotensin, and would then confirm the attribution of the 1660-cm-1 band.

Four probable amide-I1 frequencies were obtained from analysis of infrared and Raman spectra, around 1507, 1520, 1530 and 1545-1550 cm-l for angioten- sin 11. They could be the four components expected for a p structure, but were not discussed further, owing to their uncertain values.

In the amide-I11 region, besides their multipli- city, the frequency range in which the peptide-bond vibrations were observed, gave another piece of evidence for a t9 structure. The frequencies 1236 and 1274 cm-l, which were observed on Raman spectra of angiotensin 11, compared favourably to the fre- quencies 1231, 1250 and 1268cm-l observed for solid penta-alanine [17], 1235 cm-l for polyserine and 1231, 1276, 1291 om-l for polyvaline [32]. By contrast helical structures would have provided vibrations in the range 1310-1350 cm-l [26,32], while unordered forms would probably give bands above 1260 cm-l [17].

In the amide-IV to VI region, the expected complexity of modes rendered the few data gathered for angiotensin of little help in conformation deter- mination. As an example, the amide bands observed a t 690 cm-1 in Raman and 700 cm-l in the infrared did not necessarily arise from amide-V motion, and thus, taken alone, were not a criterion for a given type of structure. As a matter of fact, amide vibrations have been observed a t such frequencies in helical, in p and in unordered structures [17,33,34], as shown by the spectra published.

We finally note that all the preceding discussion is equally pertinent to the 1-7 subunit and [Ile-5, Ala-Slangiotensinamide, and yields essentially the same conclusion, namely the presence of a predomi- nant antiparallel p structure in the solid state.

The results obtained by Raman spectroscopy of concentrated aqueous peptide solution were in line with those reported above for dry material. The multiplicity of bands in the amide-I region, the shoul- der a t 1685 cm-l, the frequencies of the amide-111 vibration bands indicated a high proportion of B structure for 50 to 200 mM angiotensinamide I1 solutions. However the relative increase in intensity of the bands centered on 1660 cm-l and 660 cm-l, when compared to those observed on spectra of the dry material, suggested an increase in the proportion of unordered peptide. In addition spectra of aqueous solutions of angiotensinamide I1 yielded bands a t 1650 cm-1 (H,O)-I646 cm-l (zH20), a t 1312 cm-l

and 608 om-l (H,O) which could tentatively be inter- preted as a sign for a certain proportion of helical structure.

Taken together, the data reported from infrared and Raman spectra of angiotensinamide I1 demon- strated that this peptide adopted mainly an anti- parallel @ structure in the dry state, amorphous or crystalline, as well as in concentrated aqueous solu- tions. A small proportion of unordered conformation was also present, in amounts varying, for solid samp- les, with preparative and sampling techniques.

Interestingly enough, circular dichroic studies of angiotensinamideI1 in the dry state as well as in dilute organic solution also led to the conclusion of a spontaneous tendency of the hormone to adopt an antiparallel /3 structure. The present work offers an independent support to that conclusion.

In addition, direct electron micrographs of angio- tensinamide I1 crystal lattice [35] showed dense re- gions which very probably corresponded to the pep- tide backbone. Their spacings of 0.45 nm would also agree with an antiparallel /? structure.

It should be mentioned that the dry state, which increases inter or intrapeptide interactions, might promote arrangements which would be less stable in solution. An example is that of angiotensinamide I1 itself, which exhibited much more ,$ structure dry than in dilute aqueous solution. Another example is afforded by the [Ile-5, Ala-8]angiotensinamide, which, unordered in diluted aqueous solution, but mainly helical in organic solution [6] appeared pre- dominantly in an antiparallel p structure in the dry state. The polymorphism thus exhibited is not sur- prising for a linear peptide and underlines the import- ance of the surrounding conditions in defining its conformation.

The question arises whether the antiparallel B structure found for angiotensin I1 in concentrated aqueous solutions corresponded to the conformation of single molecules or to aggregated states. Prelimin- ary results (unpublished, Capitini et al.) obtained by Rayleigh diffusion indicated that angiotensin I1 was present as a monomer in the concentration range 5 to 25 mM.

We suggested previously that the antiparallel B structure for angiotensinamide I1 in dilute organic solution could be accomodated by a folding of the peptide chain, first between the two valines, and second as a consequence of the cis conformation of the histidyl-prolyl peptide bond. This would bring the valine and phenylalanine side-chains close enough together to allow hydrophobic interactions stabilizing the conformation. The same model [35] would thus hold for the hormone in aqueous solution. Direct evidence for these hydrophobic interactions was looked for in the infrared and Raman spectra. Un- fortunately the deviation expected in the bands of phenylalanine and valine side chains, as well as in the

Eur. J. Biochem. 182 S. FERMANDJUN et al.: Angiotensin I1 Conformations

ring of proline were not far off the precision range of the present measurements, and/or localised in spec- tral regions already laden with numerous other bands. We thus refrained from over-interpreting these spectra. More information on that possibility will be presented in the reports on our nuclear magnetic resonance investigations.

We wish to thank Dr Riniker (Ciba-Geigy) and Dr Bum- pus for generous gifts of samples, and Dr C. de Lozb for fruitful discussion. Two of us (S. F. and P. F.) wish to ac- knowledge material support from CNRS (RCP no 220).

REFERENCES 1. Rasmussen, H. (1971) Protein and Polypeptide Hormones

(Margoulies, M. & Greenwood. F. C., eds) Vol. I, p. 194, Excerpta Medica, Amsterdam.

2. Craig, L. C., Harfenist, E. J. & Paladini, A. C. (1964) Biochemistry, 3, 764.

3. Fernandez, M. T. F., Delius, A. E. & Paladini, A. C. (1968) Bwchim. Biophys. Acta, 154, 223.

4. Printz, M. P., Williams, H. P. & Craig, L. C. (1972) Proc. Natl. Acad. Sci. 69, 378.

5. Ferreira, T., Hampe, 0. G. & Paiva, C. M. (1969) Bio- chemistry, 8, 3483.

6. Fermandjian, S., Morgat, L. J. & Fromageot, P. (1971) Ear. J . Biochem. 24, 252.

7. Pysh, E. S. (1967) J . Mol. Biol. 23, 587. 8. Timasheff, S. N., Susi, H., Townend, R., Stevens, L.,

Gorbunoff, M. J. & Kumosinski, T. F. (1967) in Con- formation of Bwpolymers (Ramachandran, G. N., ed.) Vol. I, p. 173, Academic Press, New York.

9. Elliott, A. & Ambrose, E. J. (1950) Nature (London) 165, 921.

10. Blout, E. R., de Loz6, C., Bloom, S. M. & Fasman, G. D. (1960) J . Amer. Chem. SOC. 82, 3787.

11. Miyazawa, T. (1960) J . Chem. Phys. 32, 647. 12. Miyazawa, T. & Blout, E. R. (1961) J . Amer. Chem. SOC.

13. Bradbury, E. M. & Elliott, A. (1963) Polymer 4, 47. 14. Fraser, R. D. B., Harrap, B. S., McRae, T. P., Stewart,

F. H. C. & Suzuki, E. (1965) J . Mol. Biol. 12, 482. 15. Krimm, S. (1962) J . Mol. Biol. 4, 528. 16. Cotter, R. I. & Gratzer, W. B. (1969) Eur. J . Biochem. 8,

17. Sutton, P. L. & Koenig, J. L. (1970) Biopolymers 9, 615. 18. Tsuboi, M. & Wade, A. (1961) J . Mol. Biol. 3, 480.

83, 712.

352.

19. Miyazawa, T. (1967) in Poly-a-Amino Acids, Protein Models for Conformational Studies [C. D. Fasman & M. Dekl;er, eds)'l. 69, New York.

20. Smith, M. A., Walton, A. G. & Koenig, J. L. (1969) Bio- polymers 8, 29.

21. D u ~ u v , B. (1967) ThBse de Doctorat d'Etat, Bordeaux,

'

t%RS no'AO 1275.

J . Chim. Phys. 62, 265.

80, 3807.

80, 3818.

80, 3827.

22. Garrigou-Lagrange, C., Dupuy, B. & Josien, M. L. (1965)

23. Garfinkel, T. & Edsall, J. T. (1958) J . Amer. Chem. 8oc.

24. Garfinkel, T. & Edsall, J. T. (1958) J . Amer. Chem. Soc.

25. Garfinkel, T. & Edsall, J. T. (1958) J . Amer. Chem. SOC.

26. Lord, R. C. & Yu, N. T. (1970) J . Mol. Bhl . 50,509. 27. Rippon, W. B., Koenig, J. L. & Walton, A. G. (1970)

J . Amer. Chem. SOC. 22, 7455. 28. Fermandjian, S., Morgat, J. L., Fromageot, P., Legres-

SUB, C. & Maire, P. (1971) FEBS Lett. 16, 192. 29. Rippon, W. B., Anderson, J. M. & Walton, A. G. (1971)

J . Mol. Bwl. 56, 507. 30. Theoret, A., Grenie, Y. & Garrigou-Lagrange, C. (1969)

J . Chim. Phys. 66, 1196. 31. Elliott, A. & Bradbury, E. M. (1962) J . Mol. Biol. 5,574. 32. Koenig, J. L. & Sutton, P. L. (1971) Biopolymers, 10, 89. 33. Miyazawa, T., Fukushima, K., Sugano, S. & Masuda, Y.

(1967) in Conformation of Biopolymers, (Ramachandran G. N., ed.) vol. 11, p. 557, Academic Press, New York.

34. Itoh, K. & Shimanouchi, T. (1969) Biopolymers, 7, 649. 35. Fermandjian, S., Morgat, J. L., Fromageot, P., Lutz, M.

& Leicknam, J. P. (1972) in Protein and Polypeptde Hormones (M. Margoulies & F. C. Greenwood, eds) Excerpta Medica, Amsterdam, in press.

S. Fermandjian and P. Fromageot Service de Biochimie, DBpartement de Biologie Centre d'Etudes Nucleaires de Saclay Bofte postale 2, F-91 Gif-sur-Yvette, France

A.-M. Tistchenko and J.-P. Leicknam S.R.I.R.M.A., D6partement Recherche et Analyse Boite postale 2, F-91 Gif-sur-Yvette, France

M. Lutz Service de Biophysique, Dbpartement de Biologie Centre d'Etudes Nuclbaires de Saolay Boite postale 2, F-91 Gif-sur-Yvette, France