Eur. J. Biochem. 209, 163-169 (1992) (9FEBS 1992

The structure of the mammalian antibacterial peptide cecropin Pl in solution, determined by proton-NMR David SIPOS

'. Mats ANDERSSON'

and Anders EHRENBERG

Department of Biophysics, Stockholm University, Arrhenius Laboratory, Sweden Department of Biochemistry 11, Karolinska Institute, Stockholm, Sweden (Received March 13/June 17, 1992) - EJB 92 0350

Cecropins are peptides with antibacterial activity originally found in insects. Recently a cecropintype peptide was isolated from pig intestine. This peptide, porcine cecropin P1, which has 31 amino acid residues and is not ainiddted in the C-terminus, has been synthesized, purified, and investigated by CD and two-dimensional 'H-NMR at pH 5.0 in aqueous solution with 30% (by vol.) 1,1,1,3,3,3hexafluoro-2-propanol. All proton resonances have been assigned except for the N-terminal serine. Using constraints derived from NOE connectivities and 3JN,,-coupling constants, three-dimensional structures have been calculated by means of a distance-geometry program. Some of these structures have been refined by energy minimization and restrained molecular dynamics. The structures reveal an a-helix of approximately seven turns along nearly the full length of the peptide. The central part of the helix is very well defined by the NMR constraints. Also the chemical shifts of the a protons and the results of CD measuremcnts are in accord with this structure, which is different from the helix-hinge-helix structure earlier found in cecropin A and related peptides. In the a-helix of cecropin PI there is a long amphipathic section, of 4 - 5 turns, and a short hydrophobic section of one to two turns, with an intervening Glu-Gly sequence, which is a potential bend-forming section. The helix can easily span a lipid membrane.

Until rccently cecropins were known as a family of small polypeptides with strong antibacterial activity produced exclusively by insects (Boman and Hultmark, 1987). The cecropins have the capacity to kill bacteria by making their cell membranes leaky (Steiner et al., 1981), and they also lyse liposomes (Steiner et al., 1988). Unlike several other bacteriolytic polypeptides produced by insects that can also damage mammalian cell membranes, the cecropins have no such significant effects. This biological compatibility between the cecropins and mammalian cells strongly suggested the possibility of the occurrence of cecropin-like peptides in mammalian organisms. Indeed, the first mammalian cecropin was found and isolated from pig small intestine (Lee et al., 1989). The peptide, cecropin P1 or porcine cecropin, has 31 amino acid residues and presents extensive homology with the insect cecropins. From comparison with synthetic peptides it was concluded that the isolated mammalian cecropin, in contrast to the insect cecropins, is not amidated in the C-terminus. In Scheme 1, the sequence of cecropin P1 is presented and compared with two insect cecropins. In Table 1 the antibacterial activities of cecropins A and B from the cecropia moth are compared with those of cecropin P1. It is seen from the table that cecropin P1, whether amidated or not, has a more Correspondence to A. Ehrenberg, Department of Biophysics, Stockholm University, Arrhenius Laboratory, S-106 91 Stockholm, Sweden Fax: f468155597. Abbreviations. DQF, two-dimensional double-quantum-filtered; COSY, correlated spectroscopy; NOESY, two-dimensional NOE spectroscopy; TOCSY, two-dimensional total correlated spectroscopy.

narrow antibacterial spectrum than the two insect peptides. The amidated cecropin P1 shows somewhat stronger antibacterial activity but is also more hemolytic than the nonamidated form. We preferred in this study to investigate the non-amidated peptide, which is the form extracted and purified from the biological material (Lee et al., 1989). The similarities and differences in sequences and antibacterial activities between the mammalian and the insect peptides raised the questions of to what extent their three-dimensional structures in solution compare, which structural details are conserved and which are different. The structures in solution ofcecropin A (Holak et al., 1988) and of a hybrid of cecropin A and mellitin (Sipos et al., 1991) have earlier been determined in the mixed solvents of H 2 0 / l , l , l ,3,3,3-hexafluoro-2-propano1 17:3 and 7:3 (by vol.), respectively. In the present study the conformation of the mammalian cecropin P1 in the same type of solvent is studied by two-dimensional proton-NMR methods (Ernst et al., 1987). After that assignments of the resonances had been made, distance constraints were derived from the two-dimensional NOE spectroscopy (NOESY) cross-peaks and dihedral angle constraints from spin-spincoupling constants measured in correlated spectroscopy (COSY) cross-peaks. Structures compatible with these constraints were obtained by distance-geometry calculations and refined by energy minimization and molecular-dynamics calculations.

MATERIALS AND METHODS

In order to determine the C-terminal structure of cecropin P1, Lee et al. (1989) synthesized both the amidated and the

164 Hyulophora A Drosophila A Porcine P1

KW--KLFKKI GWTJKKIGEKI SWLSKTAKKL

EKVGQNIRDG ERVGQHTRDA ENSAK-KR--

IIKAGPAVAV TI-QGLGIAQ -1SEGIAIAI

VCQATQIAK QAANVAATAR QGGPR

CONH, CONHZ COOH

Scheme 1. Comparison of amino acid sequences of two insect cecropins and the mammalian cecropin P1 from porcine small intestine. The sequences are for cecropins Hyalophoru A, cecropin A from Hyalophoru cecropia (Boman & Hultmark, 1987), Drosophila A, cecropin A from Drasophilu mrlunoguster (Kylsten et al., 1990) and porcine P1, cecropin PI from porcine small intestine (Lee et al., 1989). Conserved amino acids are in bold with gaps denoted by hyphens.

Table 1. Comparison of the antibacterial activities of the two insect cecropins, A and B from H . cecropiu and the mammalian cecropin PI against four bacteria. Data for Hyalophoru cecropins A and B are from Roman and Hultmark (1987) and for porcine cecropin PI from Lee et a]. (3989). PI-OH, non-amidated cecropin P I ; P1-NH2, amidated cecropin P1. Microorganism

Escherichiu roli, D21 P.seudomonas aeruginosa, OT97 Acinetohactercalcoaceticus, Ac 1 1 Bacillus meguteriurn, Bm 11

Lethal concentration of cecropin A

B

PIOH

P1NH2

0.3 2.6 0.3 0.6

0.3 1.9 0.7 0.4

0.4 13 0.5 5.3

0.3 5.9 0.2 1.8

non-amidated forms using the solid-phase methodology. The same batch of non-amidated peptide was used as starting material in this study and was purified by semipreparative reverse-phase C1 HPLC as earlier described (Andersson, 1990). An additional step of purification was made by cationexchange HPLC (TSK-ODS CM, 7.8 mm x 150 mm) using a linear gradient of 1 M ammonium acetate (20 -45% in 30 min at 1 ml/min) in 0.2 M acetic acid. The purity was ascertained by reverse-phase CI8analytical HPLC and mass spectrometry. Chemicals used for sample preparations for the CD and NMR studies were as reported in our recent study on a hybrid peptide of cecropin A and melittin (Sipos et al., 1991). CD spectra were recorded on a JASCO 5-500 spectropolarimeter at three temperatures; 15”C, 25°C and 40”C, and four concentrations of 1,1,1,3,3,3-hexafluoro-2-propanol, 0%, 20%, 30% and 40% (by vol.). Ellipticities at 222 nm were analyzed using the mean-residue contributions at this wavelength of - 35 700 deg cm2 dmol - I for a-helix and 3900 deg cm2 dmolfor random coil (Greenfield and I;asman, 1969). The sample used for the NMR structure determination contained 5.9 mM cecropin P1 in a solution of H 2 0with 10% ’H20 and 30% (by vol.) deuterated 3 ,I ,I ,3,3,3-hexafluoro-2propanol. The pH was adjusted to 5.0 by addition of deuterated acetic acid. An internal reference signal of proper intensity was provided by 0.4mM of the sodium salt of 3trimethylsilyl-[3,3,2,2-”H]propionic acid. Two-dimensional double-quantum-filtered (DQF) COSY (Piantini et al., 1982) and NOESY (Jeener et al., 1979) were applied at the proton frequency 400 MHz on a JEOL (3x400 NMR spectrometer over a spectral width of 5 kHz. In the DQF-COSY experiment the number of points in tl was 256, for each of which 2048 complex data points in t 2 were collected. In the NOESY experiments, with mixing times of 100 ms and 250 ms, the number of t l values was 512, for each of which a t2 data set with 512 complex data points was

acquired. In both types of experiments each t2 data set was the sum of 64 free induction decays. A total correlation spectroscopy (TOCSY) experiment (Braunschweiler and Ernst, 1983; Bax and Davis, 1985) at the proton frequency 500 MHz was made using a Varian Unity 500 spectrometer. In this experiment the mixing time was 70 ms and the spectral width 6 kHz, the number of tl values was 512, and for each of these, 32 free induction decays with 1024 complex data points in i 2 were accumulated. Zero filling was made in both t , and t2 data sets when required in order to optimize readability of the cross-peaks in the contour plots and to make full use of the resolution dictated by the number of data points originally collected. For the same reason shifted sine-bell filter functions were applied to the time functions of both tl and t 2 before Fourier transformation. The DQF-COSY experiment was also used to evaluate the spin-spin coupling constants 3JNHa.For this purpose, zero fillings were made to obtain thc digital rcsolution 2.44 Hz/ point in both dimensions. The three most central cross-sections along F2 through each antiphase cross peak pair that should be evaluated were added, and the peak splitting was determined as described by Kim and Prestegard (1989), eliminating the distortion caused by the overlap between the antiphase peaks. The solvent resonance was in all NMR experiments saturated at all times, except during data acquisition. In the twodimensional experiments the pulse delay was 1.5 s. When freeinduction decays for one-dimensional spectra were accumulated, longer delays were used so that integrated peak intensities should correspond to the number of protons. All final NMR experiments were made at 35°C. Experimental NMR-data sets were processed on a Convex C-2 minisuper computer using the FTNMR software from Hare Inc. A personal computer with a serial connection to the Convex was used as a graphics terminal. The distance geometry calculations were done on a Silicon Graphics Personal Iris 4D25TG workstation, using the program DIANA (Giintert et al., 1991; Guntert and Wuthrich, 1991). Structures computed by DIANA were energy minimized utilizing the software packages INSIGHT I1 and DISCOVER from Biosym Technologies Inc., with up to 1000 steps of steepest-descent iterations, until the maximal derivative values were near unity. Structure refinements by restrained molecular-dynamics computations using NOE constraints were carried out with the same software. Initial velocities were assigned according to a Maxwell distribution corresponding to 300 K. A time step of 1 fs was employed and integrations were performed with a Verlet algorithm. The velocities were rescaled automatically, corresponding to a heat bath coupled to the system (Berendsen et al., 1984). The AMBER forcefield was employed, including a distance-dependent dielectric constant aimed to compensate for the absence of the solvent environment (Weiner et al., 1986). The restrained molecular-dynamics simulations were performed on selected

165 Table 2. Amino acid residues of cecropin P1 in a-helix conformation at different temperatures and concentrations of 1,1,1,3,3,3-hexafluoro-Zpropanol (F6PrOH) estimated from CD measurements. Helix contributions were evaluatcd from the ellipticity at 222 nrn using the mean residue contributions - 35700 deg cmz dmol- for cc-helix and 3900 deg ern' dmol for random coil (Greenfield and Fassman, 1969).

'

FbPrOH

0 20 30 40

Amino acids in cl-helix conformation at 15°C

25°C

40"C

4.4 21.6 25.3 22.6

4.7 20.1

4.7 18.6 21.6 21.1

23.8 21.6

DIANA-calculated structures, i. e. those with lowest targetfunction values. Runs were extended over lo4 steps, corresponding to a 10-ps duration.

RESULTS The CD measurements on cecropin PI were of importance both for preparing a sample suitable for NMR measurements and for preliminary conclusions about the existence of helical structural elements in the peptide molecule. All recorded CD spectra (data not shown) are compatible with contributions from amino acid residues in a-helix conformation and random-coil type configuration. The estimated a-helix contributions are given in Table 2. As judged from the spectral amplitude at 222 nm the cecropin P1 has a largely randomcoil conformation in water. However, in a mixture of water and 1,1,1,3,3,3-hexafluoro-2-propanol, a solvent with reduced water activity presenting a hydrophobic membrane-like environment, the cecropin P1 adopts a mainly helical conformation. The CD measurements indicated maximum a-helix formation in a solution with 30% (by vol.) 1,1,1,3,3,3-hexafluoro-2-propanol. The helix content decreased slightly with increasing temperature in the range 15 - 40 "C. Preliminary NOESY experiments at 25 "C and 35 "C showed that the best resolution was obtained at the latter temperature. Hence this solvent mixture and temperature were employed in the NMR experiments. Using these conditions, we obtained data (Table 2) suggesting that 22 or 23 of the amino acid residues of cecropin P1 should be engaged in a-helix formation. Assignments of the proton resonances were made in the usual manner from the DQF-COSY, TOCSY and NOESY data. In the NH-NH region of the NOE spectrum (Fig. 1) practically all sequential cross-peaks could be identified. The peaks for Glu20-Gly21 and Ilc24-Ala25 were hidden in the diagonal, and those for Lys5-Thr6 and Lysl6-Argl7 are barely seen as extensions from the diagonal. The two outermost connectivities in the N-terminus could not be detected, probably because of dynamic effects. All these sequential assignments fitted well with the side-chain connectivity patterns obtained from the DQF-COSY and TOCSY data. However, no resonances could be assigned to Serl. The chemical shifts of all the assigned protons of the peptide are compiled in Table 3. From direct comparison of NOESY regions, representing interactions between NH and CctH, CPH and CyH, with the corresponding regions of the DQF-COSY and TOCSY spectra, it was possible to sort out the NOESY cross-peaks

I '

,

I

I

8.8

8.6

,

1

8.4

8.2

F:!

(PPm)

I

k.0

Im

km'

Fig. 1. Part of the proton NOE spectrum of cecropin P1 showing NH(t3NH(I+ 1) cross-peaks. Two NH-NH cross-peaks are outside the region shown. Mixing time 250 ms, 5.9 mM cecropin P1, 30% (by vol.) 1,1,1,3,3,3-hexafluoro-2-propanol, pH 5 at 35 "C.

belonging to intra-residue spin systems. When this had been done it was easy to identify the remaining cross-peaks with NOE interactions between protons in different amino acid residues. The NOESY-cross-peak intensities were estimated from the spectrum obtained with a 250-ms mixing time. For each peak the number of linear contour levels above the noise was counted using a suitable contour spacing. Intensities were grouped in three classes with either weak, medium and strong intensity. In the NOESY region shown in Fig. 1 the overlap cross-peak at 8.49 ppm x 8.37 ppm has such high intensity that each of the two overlapping peaks must be ascribed to have medium strength. In most other cases, when nearly coinciding cross-peaks could be resolved considering the form of the common contours, the two peaks had to be classified as weak. The inter-residue cross-peaks identified in this study and their intensities are presented in Fig. 2. The spin-spin coupling constants 3JNHafor the amide proton-to-a-proton coupling were determined for the cross-peaks in the fingerprint region of the DQF-COSY spectrum. Values for 20 of these were found to be in the range 4 - 6 Hz. Information about these coupling constants is also included in Fig. 2. The chemical shifts of the CaH resonances are known to be sensitive to the secondary-structure elements of which the residues are components (Clayden and Williams, 1982; Pastore and Saudek, 1990). In order to evaluate such influences the differences between the experimental values given in Table 1 and the corresponding random-coil chemical shifts (Wiithrich, 1986) were calculated. In order to eliminate local effects for each residue, data were smoothed by taking the shift-difference average over the own value and those of the nearest neighbours (Pastore and Saudek, 1990). Such smoothed values for cecropin PI arc shown in Fig. 3. The exchange rates of the backbone NH protons were studied by dissolving the lyophilized protonated sample in a

Table 3. Chemical shifts of the assigned 'H-NMR signals of the mammalian cecropin peptide cecropin P1 recorded in a water solution with 10% 'H20 and 30% (by vol.) 1,1,1,3,3,3-hexafluoro-2-propanolat pH 5 and 35°C. Rcsidue numbcr

Amino acid

1 2

S W

3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31

L S K T

Chemical shift for protons

A

K K L E N

S A K K

R I S E

G I A

I A

I

Q

C G P R

NH

c(

B

others

7.71

4.36

3.27

7.48 7.86 7.98 7.90 8.42 7.77 7.89 8.37 8.49 8.36 8.23 8.51 8.29 8.00 7.91 8.72 8.21 8.05 8.07 8.31 7.88 8.03 8.00 8.61 7.92 8.12 8.01

3.94 4.16 4.16 3.89 4.08 4.19 4.1 5 4.17 4.00 4.55 4.32 4.17 4.01 4.07 4.21 3.84 4.19 4.18 3.91 3.85 4.08 3.78 4.12 3.95 4.31 4.04; 4.14 4.00; 4.21 4.53 4.32

1.50 3.92 1.97 3.71 1.51 1.50 2.09 1.93 2.26 2.86; 3.05 4.25 1.57 2.02 2.04 2.08 1.91 4.03 2.18; 2.29

2H 7.39; 4H 7.55; 5H 7.19; 6H 7.29; 7H 7.52; NH 7.9 y 1.29; 6 0.82; 0.87

7.57

1

S

W

L

dNN( i, 1+1)

7 1.53; 6 1.67; E 3.02 7 1.22

y 2.63; 6 2.02; E 3.00 p 1.54; 6 2.07; E 3.03 y 1.73; 6 0.92 y 2.51

y 1.46; S 1.79; E 3.01 7 1.49; 6 1.78; E 3.02 y 1.82 7 CH2 1.15; y C H j 0.98; 6 0.88 y 2.42; 2.62

1.98 1.58 2.04 1.61 2.06 2.25; 2.35

y CHB 1.18; y CH3 0.98; 6 0.88

3.40; 3.50 1.98; 1.76

y 2.88; 3.05 y CHZ 1.65; 6 CH2 3.25

7 CH, 1.24; y C H j 0.91; 6 0.90

7 CH2 1.35; y CH3 0.98; 6 0.88 ?

-- - --

S

5 K

T

A

K

K

10 L E

N

S

A

15 K K

25

20

R

I

S

E

G

da6 ( i ,i f 3 1

~.

(i,i+4)

I

Q

30 G

G

P K

~~~~~

~~

~~

~

~~~~~~~~

_.___

3

~

A

~~~~~~~- ,

~~

~

I

~

.~

J

A

._--_

~~~

LXN

I

I -

d aN (i,i+I)

d

2.47; 2.60

a

A A

A A A A

A

A A A A A A A A

~

--- -- - . .

A A A A A

Fig. 2. Summary of observed inter-residual NOE connectivities and 35NH, spin-spin coupling constants obtained for cecropin P1. Experimental conditions as in Fig. 1. Thickness of lines connecting two residue positions shows the intensity of the corresponding NOESY cross-peak. Thrcc intensities are indicated: strong, medium and weak. Non-resolved ovcrlapping peaks are shown as broken lines. (A)Coupling constants 3 J N , 4 a determined to be in the range 4-6 Hz.

167

L

I

0 2 1

6

11

16

21

26

31

R P s i d i i e number

Fig.3. The ( n & 1) averaged upfield chemical shifts of the CaH of cecropin P1 relative to the corresponding random-coil shift values. Conditions for cecropin P1 shifts as in Fig. 1 . Chemical shifts of CaH from Table 3 and random-coil shift values from Wuthrich (1986).

solvent with deuterated components. One-dimensional spectra were recorded at 35 "C. The first spectrum was taken 12 min after dissolving the peptide, at which time many of the resonance amplitudes had decreased considerably. After 1 h: only resonances Ilel8-Gln27 showed remaining amplitudes, and after 5 h, only traces of the resonances Ile24-Gln27 remained. No significant signal intensities persisted after 24 h. The results shown in Fig. 2 were used for setting the input parameters for distance-geometry calculations of structures by means of the DIANA program (Guntert et al., 1991; Giintert and Wuthrich, 1991).The upper distance limits corresponding to the NOE connectivities with strong, medium and weak cross-peak intensities were set to 0.27,0.33 and 0.50 nm, respectively (Nilges et al. 1988). The spin-spin coupling constants 'JNHa depend on the dihedral angle @. For coupling constants measured to be in the range 4 - 6 Hz, Q, was given the limits -90" and -40". 85 inter-residual upper-limit distance contraints and 20 intra-residual angular constraints were used as input parameters for the DIANA calculations. Among the distance constraints deduced from the orH(i)-NH(i+ 1) cross-peaks with low intensity all, except the one between Glu20 and Gly21, were discarded as meaningless by the primary distance checks made by the program. Also the distance constraint due to the corresponding medium-intensity peak between the Gly28 and Gly29 was discarded, These constraints were not further used in the calculations. The DIANA calculations were made in 31 steps with steep van der Waals' constraints only in the last two steps (Giintert et al., 1991). Out of ten primary structures, four had about equally low target-function values. These four structures were subjected to refinement by energy minimization and restrained molecular dynamics retaining the distance constraints from the DIANA calculations. Fig. 4 shows the superimposition of the four final backbone structures, while Fig. 5 presents one of them with side chains included.

DISCUSSION Already the CD results suggest that a major part of cecropin P1 has an or-helical structure under the solvent conditions used. All the NMR results confirm and emphasize this conclusion. It is seen from Fig. 3 that the resonances of the CorH are strongly shifted upfield in the whole sequence Leu 3 - Ile26.

Fig. 4. Stereo view of the superimposition of four structures of cecropin P1. These are the four structures with lowest target function out of ten structures calculated by DIANA using constraints deduced from the results shown in Fig. 2 . The structures were refined by energy minimization and molecular dynamics. Only the three backbone atoms N , Ca and C of each amino acid residue are shown. One structure was chosen as target and the others were superimposed on this by optimizing the fit for residues 10-20. Bottom, N-terminus; top, C-terminus.

Fig. 5. Stereo view of one of the structures of cecropin P1 with all atoms shown. Bottom, N-lerminus; top, C-terminus.

This propcrty indicates a long and stable helix over this sequence (Clayden and Williams, 1982; Pastore and Saudek, 1990). The only point with lowered upfield shift is at residues Glu20 and Gly21. This glycine residue is conserved in most insect cecropins (Lee et al., 1989) and is part of the hinge region in cecropin A (Holak et al., 1988). All the NOE connectivities and all coupling constants compiled in Fig. 2 are compatible with an extended ahelix structure over nearly the whole length of the molecule. Tt should, however, be noted that only one strong NOE connectivity bridges residues Glu20 and Gly21. The calculated three-dimensional structures depicted in Figs 4 and 5 show the or-helix structure in great detail. In Fig. 4 the pairwise superimposition of structures is optimized for the backbone atoms N, Ca and C of each amino acid residue. Variation of the chain length, over which optimization of superimposition is made, shows that the structure of the central backbone region is well determined by the NMR data.

168 For the backbone range of residues 10-20 the root-mean- hclix and above Gly21 there is a short region of strongly square distance for a fittcd pair of structures is approximately hydrophobic helix. 0.06 nm. If the range of the fitted region is made longer the The various antibacterial peptides discussed here, insect root-mean-square distance rapidly increases. The conclusion cecropins, melittin, hybrids between a cecropin and melittin, is that a central region of the molecule with 13 or 14 residues and the mammalian cecropin P1, have been suggested to form forms a well characterized and stable a-helix. On either side voltage-dependent pores which cross the cell membrane they of this region there is still a rather well characterized a-helix attack (Christensen et al., 1988). If this is the way they funcstructure but it may be curved in various directions. The first tion, the structures of the peptides might well be similar to 2 -4 residues in the N-terminus and the last 3 - 5 residues in each other in the membrane, with elongated helices spanning the C-terminus are not structurally well determined and are from one side to the other. Pores could be formed by interaclikely to be flexible. For this 31-residue peptide the total length tion between three or more such helices. We can now try to of the a-helical structure involves 22 - 26 of the residues. Ths model such interacting oligomers from the known structures is slightly longer than the helix length of 22 or 23 amino acid in solution of the cecropin-type peptides and make compariresidues suggested by the CD measurements. Calculations on son with models of pore-forming structures in proteins (Oiki the structure of Fig. 5 show that there are, on average, 3.6 et al., 1990). For a peptide from such a pore-forming protein, rcsidues in each turn of helix; the step along the helix axis for the 23-residue M26 helical segment of the nicotinic each residue is 0.15 nm, and the length of the hydrogen bonds acetylcholine receptor, it has been shown that in a lipid bilayer CO(i) - NH(i+ 4) along the helix is 0.29 nm. These values fit it arranges itself with the helix axis perpendicular to the plane well with what is known from protein structures obtained by of the bilayer (Bechinger et al., 1991). In porcine cecropin P1 the amphipathic portion of the a-helix is by itself just long X-ray crystallography. enough to traverse the hydrophobic core of a membrane lipid Fig. 5 shows the general appearance of the peptide with side chains included. Since no NOE cross-peaks could be bilayer. found between side chains, their orientations are not well We thank Professor Hans G. Boman, Department of Microdefined. Analysis of Fig. 5 and similar structures suggests biology, Stockholm University, for valuable discussions and for critiseveral possibilities of interactions between charged groups. cally rcading the manuscript, Dr. Torleif Hard at the Center for Perhaps most significant is that the 6 carboxylate group of Structural Biochemistry, Karolinska Institute, Novum, Huddinge, Glull is close to the E amino group of LysS and oriented so Sweden. for generously making the TOCSY experiment on the that a hydrogen bond or salt link may be formed. The side 500 MHz spectrometer in his laboratory and Professor Bengt Norden, Chalmers University of Technology, Goteborg, Sweden, for generous chain ii carboxylate of Glu20 is situated between the &-amino permission to use the CD spectrometer. We also wish to thank Mr. group of Lyslb and the guanidinium group of Argl7 showing Torbjorn Astlind for skilled and professional dealing with computers the possibility of salt links or hydrogen bonds to both of them. and programs and Ms. Britt-Marie Olsson for skilled assistance in the Also the y carbonyl o f h s n l 2 and the E amino group of LyslS chemical laboratory. The work was supported by grants from the may interact in a similar manner. Swedish Natural Science Research Council, the Magnus Bergvall Foundation, the Carl Trygger Foundation for Scientific Research and Most of the charged and polar groups are concentrated towards the front of the peptide structure shown in Fig. 5. the Erna and Victor Hasselblad Foundation. The only charged group towards the back of this structure is that of Glul 1 which may be easily neutralized by Lys8 as just described. Most of the hydrophobic side chains are presented REFERENCES on the back face of the helix. Thus the sequence 1 --20 forms Andersson, M. (1990) J . Protein Chem. 9, 359. a strongly amphipathic a-helix section. A hydrophobic patch Bax, A. & Davis, D. G.(1985) J . Magn. Reson. 65, 355-366. round the whole a-helix is formed by the sequence Ile22Bazzo, R., Tappin, M. J., Pastore, A., Harvey, S., Carver, J. A. & Campbell, 1. D. (1988) Eur. J . Biochem. 173, 139-146. Ile26. The amide protons in this hydrophobic region show the slowest exchange rates. The hydrophobic side chains, and Bechinger, B., Kim, Y., Chirlian, L. E., Gesell, J., Neumann, J.-M., Montal, M., Tomich, J., Zasloff, M. & Opella, S. J. (1991) J . possibly also their solvation by 1,I ,1,3,3,3-hexafluoro-2-proBiornol. N M R I , 167-173. panol, decreases water accessibility to the backbone of this Berendsen, H. J. C., Postma, J. P. M., van Gunsteren, W. F., DiNola, section of the a-helix. A. & Haak, J. R. (1984) J. Chern. Phys. 81,3684-3690. The solution structure of porcine cecropin P1 is remark- Boman, H. & Hultmark, K. (1987) Annu. Rev. Microbid. 41, 103ably different from the structures of Hyalophora cecropin A 126. (Holak et al., 1988), melittin (Bazzo et al., 1988) and a hybrid Braunschweiler, L. & Ernst, R. R. (1983) J . Magn. Reson. 53, 521 558. between these two (Sipos et al., 1991). All these peptides have antibacterial and membranolytic effects, though to different Christensen, B., Fink, J., Merrifield, R . B. & Mauzerall, D (1988) Proc. ,Vat1 Acad. Sci. C7SA 85, 5072 - 5076. degrees and with different specificities. Whereas cecropin P1 Clayden, N. J. & Williams, R. J. P. (1982) J . Magn. Reson. 49,383shows a continuous amphipathic a-hclix over nearly the whole 389. length of the peptide, the other three peptides show structures Ernst, R. R., Bodenhausen, G. & Wokaun, A. (1987) Principles of with two helices and a bending hinge in between. In these bent nuclear magnetic resonance in one and two dimensions, Clarendon two-helix structures one of the helices is amphipatkc and the Press, Oxford. Greenfield, N. & Fasman, G. D. (1969) Biochemistry 8, 4108-4116. other is essentially hydrophobic. The continuous a-helix in cecropin P l seems to suggest Giintert, P., Braun, W. & Wuthrich, K (1991) J . Mol. Biol. 217.517530. that a bent or hinged structure in itself might not be essential Giintert, P. & Wiithrich, K. (1991) J . Biomol. N M R 1 , 447-456. for the funtion. However, it should be noted that the region Holak, T. A,, Engstrom, A,, Kraulis, P. J., Lindeberg, G., Bennich, of Glu20 and Gly21 of cecropin P1 might be a site where a H., Jones, T. A., Gronenborn, A. M. & Clore, G. M.(1988) Biobend could be initiated. This could depend on the charge chemistry 27, 7620 - 1629. interaction of Glu20, which might be quite different at a Jcener, I., Meier, B. H., Bachman, P. & Ernst, R. R . (1979) J . Ckern. Phys. 71,4546-4553. membrane surface. Below Glu20 there is a long amphipathic

169 Kim, Y. & Prestegard, J. H. (1989) J. Mugn. Reson. 84, 9- 13. Kylsten, P., Samakovlis, C . & Hultmark, D. (1990) EMBU J. 9,217224. Lce, J.-Y., Bornan, A,, Sun, C., Andersson, M., Jornvall, H., Mutt, V. & Boman, H. (1989) Proc. Natl Acad. Sci. USA 86, 91599162. Nilges, M., Clore, G. M. & Gronenborn, A. M. (1988) FEBS Lett. 239, 129-136. Oiki, S., Madison, V. & Montal, M. (1990) Proteins 8, 226-236. Pastore, A. & Saudek, V (1990) J . Mugn. Reson. YO, 165-176. Pianthi, U., Sorensen, 0. W. & Ernst, R. R. (1982) 1.Am. Chem. SOC.lO4,6800-6801.

Sipos, D., Chandrasekhar, K., Arvidsson, K., Engstrom, A. & Ehrenberg, A . (1991) Eur J. Biochem. 199, 285-291. Steiner, H., Andreu, D. & Mcrrifield, K.B. (1988) Biochim. Biophy.). Acta 939, 260-266. Steiner, H., Hultmark, D., Engstrom, Bennich, H. & Boman, H. G. (1981) Nature 292,246-2248. Wcincr, S. J., Kollman, P. A., Nguyen, D. T., & Casc, D. A. (1986) J . Comput. C'hem. 7, 230-237. Wiithrich, K . (1986) NMR ofprateins andnucleic acids, Wiley & Sons, New York.

a.,