Oscar D. Monera, Nian E. Zhou, Cyril M. Kay, and Robert S

THE JOURNAL OF BIOLOGICAL CHEMISTRY Q 1993 by The American Society for Biochemistry and Molecular Biology, Inc.

Vol. 268, No. 26, lasue of September 15, pp. 1921&19227,1993 Printed in U.S.A.

DESIGN, SYNTHESIS, AND CHARACTERIZATION*

(Received for publication, May 4, 1993)

Oscar D. Monera, Nian E. Zhou, Cyril M. Kay, and Robert S. HodgesS From the Department of Biochemistry and the Protein Engineering Network of Centres of Excellence, University of Alberta, Edmonton, Alberta T6G 2H7, Canada

An antiparallel coiled-coil has been designed and characterized as a model for studying protein folding and assembly. This heterostranded antiparallel coiled- coil was formed by an interchain disulfide bond between cysteine residues at position 2 of one chain and at position 33 of the other chain. Each peptide chain has 35 residues which are composed of five heptad repeats of the sequence K-L-E-A-L-E-G with a single Leu + Ala substitution at position 16. Two homostranded parallel coiled-coils were also formed as co- products of the oxidation reaction to form the interchain disulfide bond. The CD spectra of the parallel and antiparallel peptides were very similar and their high molar ellipticities at 220 nm did not increase in the presence of 60% trifluoroethanol. These data suggest that, like the parallel peptides, the antiparallel peptide also exists in a coiled-coil structure. Urea and guanidine hydrochloride denaturation studies, in con- junction with molecular modeling studies, suggest that there are no physical restrictions to the packing of hydrophobic residues in an antiparallel coiled-coil. However, interchain electrostatic interactions can have positive or negative contributions to the overall stability of the disulfide-bridged coiled-coil. In addition, interchain electrostatic interactions appear to play a major role in protein folding by controlling the parallel or antiparallel alignment of the a-helical polypeptide chains. This study is also for the first time providing us with a new understanding of the information that can be obtained from urea and guanidine hydrochloride denaturation studies of proteins concerning the contributions of hydrophobic and electrostatic interactions on stability.

The coiled-coil motif, as proposed by Crick (1953) about 40 years ago based on the x-ray diffraction data of a-keratin, is characterized by two a-helical chains wrapping around each other to form a two-stranded rope with a left-handed super- coil. This prediction was recently confirmed by the first high resolution x-ray structure of a synthetic two-stranded coiled- coil representing the dimerization site of GCN4, a DNA- binding protein (O’Shea et al., 1991). However, during the intervening years some significant observations from the

* The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

$ To whom reprint requests should be addressed Dept. of Biochem- istry and the Protein Engineering Network of Centres of Excellence, University of Alberta, Edmonton, Alberta T6G 2H7, Canada. Tel.: 403-492-2758; Fax: 403-492-0095.

studies of tropomyosin have contributed to our understanding of the coiled-coil structure. Among these were the initial crystallization of tropomyosin (Caspar et al., 1969), the dis- covery of the 3-4 or 4-3 hydrophobic repeat from the amino acid sequence of the first two-stranded a-helical coiled-coil that is now characteristic of all coiled-coil structures (Hodges et al., 1972; Sodek et al., 1972), the parallel and in-register alignment of the two a-helical chains (Johnson and Smillie, 1975; Stewart, 1975; Lehrer, 1975; McLachlan and Stewart, 1975), the suggested role of ionic interactions on the stabilit of the coiled-coil structure (Stone et al., 1975), and the 15 K resolution structure of tropomyosin (Phillips et al., 1986). At present, more than 200 proteins are thought to contain the coiled-coil motif (Lupas et al., 1991), and many more are expected to be discovered.

Although in theory the alignment of the a-helical chains in a coiled-coil can be parallel or antiparallel, there seems to be a preference for the a-helical chains to align parallel in two- stranded a-helical coiled-coils. On the other hand, the a- helical chains in most globular proteins usually pack with an antiparallel alignment of a-helices (Richardson, 1981). Some globular proteins have the characteristic 3-4 hydrophobic repeat and therefore have coiled-coil domains (O’Shea et al., 1989a, 1991; Banner et al., 1987; Landschulz et al., 1988; Cusack et al., 1990; Gentz et al., 1989; De Francesco et al., 1991; Hu et al., 1990; Chakerian et al., 1991; Reddy et al., 1992). However, these coiled-coil motifs seem to be associated with a very specific function, as a result of dimerization, as opposed to just being part of the overall folded structure of the protein.

It is not unreasonable to believe that the collision of two (or more) structural elements, such as that between two a- helices, could become a nucleation site for protein folding. Thus, the alignment of the a-helical chains could determine the overall fold and, as a consequence, the function of the protein. Our goal therefore was to use de novo designed peptides to form parallel and antiparallel coiled-coils as model systems to study the factors that contribute to our understanding of protein folding and assembly.

The coiled-coil motif is an ideal model system to investigate protein folding because 1) it contains only one type of second- ary structure (the a-helix) which can be easily monitored by circular dichroism (CD) spectroscopy, 2) it contains two sub- units stabilized by both intrachain and interchain interactions, and 3) its small size reduces the potential complexity in the analysis and interpretation of results. A series of 35- residue synthetic peptides has been previously used to study the factors that contribute to the formation and stability of parallel coiled-coils. These studies included investigations of both heterodimeric and homodimeric hydrophobic interac-

19218

Antiparallel Coiled-coils 19219

tions and the contribution of disulfide bonds to protein stability (Hodges et al., 1990; Zhu et al., 1992, 1993; and Zhou et al., 1992a, 1992b, 1992c, 1993). In this study we have used similar peptide analogs to form an antiparallel coiled-coil in order to compare its structural characteristics with those of parallel coiled-coils.

EXPERIMENTAL PROCEDURES

Peptide Synthesis and Purification-The peptide analogs were synthesized by standard solid-phase methodology using an Applied Biosystems peptide synthesizer model 430A (Foster City, CA) using co-poly(styrene, 1% diviny1benzene)benzhydrylamine-hydrochloride resin. The peptides were cleaved from the resin by reaction with hydrogen fluoride (20 ml/g resin) containing 10% anisole and 2% 1,2- ethanedithiol for 1 h at -5 "C. After extraction with 30% acetic acid and lyophilization, the crude peptides were purified using a Varian 5000 HPLC' equipped with a semipreparative reversed-phase C18 column (Synchropak RP-P, 250 X 10 mm, inner diameter, 6.5-pm particle size, 300-8, pore size, SynChrom, Lafayette, IN). Typically, the samples were eluted at 2 ml/min with a linear AB gradient of 1% B/min for the first 15 min and 0.2% B/min thereafter, where solvent A was 0.05% trifluoroacetic acid in water and solvent B was 0.05% trifluoroacetic acid in acetonitrile. The fractions were verified for peptide purity using a Hewlett-Packard HP1090 liquid chromatograph equipped with an analytical reversed-phase C, column (Zorbax 300SB-C8,4.6 mm X 25 cm, 5-pm particle size, 300-A pore size) using a linear AB gradient of 2%B/min where solvents A and B are described above. The pure fractions were pooled and lyophilized. Peptide identity was confirmed by amino acid analysis and by mass spectrometry. For amino acid analysis, the purified peptides were hydrolyzed in 6 N HC1 containing 0.1% phenol at 160 "C for 1 h in an evacuated tube. HCl was then removed by drying over sodium hydroxide under reduced pressure. The samples were redissolved in 0.2 M sodium citrate, pH 2.2, and injected into a Beckman model 6300 amino acid analyzer equipped with a System Gold data process- ing system (Beckman, San Ramon, CA). Mass spectrometry was performed on a BIOION 20 Nordic plasma desorption time-of-flight mass spectrometer (Upsala, Sweden).

Air Oxidation-Oxidation of the peptides to form an antiparallel coiled-coil was carried out by dissolving together about 7.5 mg of each peptide in 3 ml of 100 mM NH4HC03, 6 M GdnHCl, pH 8.3, and stirring the solution magnetically in an open vial at room temperature (25 "C). Each parallel coiled-coil was formed by stirring about 5 mg of peptide in 1 ml of 100 mM NHIHCOB, pH 8.3, under similar conditions. The progress of oxidation was followed by analytical reversed-phase HPLC as described above. After approximately 18 h of stirring, the peptide solutions were neutralized with dilute acetic acid and the oxidized peptides were then purified by semipreparative reversed-phase HPLC as described above.

Circular Dichroism Spectroscopy-Stock peptide solutions (910 mg/ml) were prepared by dissolving 3 mg of peptide in 300 pl of 50 mM phosphate buffer containing 0.1 M KC1 at pH 7. Two sets of peptide solutions were prepared for CD scans: one was in benign buffer (10 pl of stock peptide solution was diluted with 50 pl of the same buffer), and the other was in 50% TFE (10 pl of stock peptide was diluted with 20 pl of buffer and 30 pl of TFE). The peptide solutions were then loaded into a 0.02-cm fused silica cell and their ellipticities were scanned from 190 to 255 nm.

Circular dichroism spectroscopy was performed at 20 "C on a Jasco J-5OOC spectropolarimeter (Jasco, Easton, MD) equipped with a Jasco IF-50011 interface connected to an IBM PS/2 model 30286 computer using Jasco DP-500/PS2 system, ver 1.33a software. A Lauda water bath (model RMS, Brinkmann Instruments, Rexdale, Ontario, Can- ada) was used to control the temperature of the cell. The instrument was calibrated daily with an aqueous solution of recrystallized d-10- (+I-camphorsulfonic acid at 290 nm. Ellipticity is reported as mean residue ellipticity ( 0 ) and the limit of error of measurements at 220 nm was & 300"/cm2/mol".

Denaturation Studies-A series of 10-p1 aliquots of the stock solutions were diluted with calculated volumes of 50 mM phosphate, 0.1 M KC], pH 7, buffer and of either 8 M GdnHCl, 10 M urea, or 6 M mixed denaturants (containing different molar ratios of GdnHCl and

The abbreviations used are: HPLC, high performance liquid chromatography; GdnHC1, guanidine hydrochloride; TFE, l,l,l-trifluoroethanol.

urea, both in the same phosphate buffer) to give the desired final denaturant concentrations for the CD measurements. The samples were allowed to equilibrate in the presence of different concentrations of denaturant(s) for a minimum of 3 h at 25 "C. Then, their ellipticities were measured at 220 nm and 20 "C under similar conditions described above. The fraction of the peptide in the folded state (fn) was calculated as fn = ( [ O ] - [O]u)/([O]n - [ O l d , where [el is the observed mean residue ellipticity at 220 nm at any denaturant concentration and [0]n and [0]u are the mean residue ellipticities at 220 nm of the native (folded) and the unfolded states, respectively.

For temperature denaturation studies, 20-4 aliquots of the stock peptide solutions were diluted with 220 pl of 50 mM phosphate, 0.1 M KC], pH 7.0, and loaded into a 0.05-cm cell. The ellipticities at 220 nm were measured after 10-min equilibration of the samples at different temperatures. The fraction of the peptide in the folded state was calculated as the ratio of the observed mean residue ellipticity at any temperature ([e],) to the mean residue ellipticity at 5 "C ([0]5,, at which temperature the peptide is presumed to be in the fully folded state.

Size Exclusion Chromatography-About 0.5 mg of pure oxidized C2A16/C33A16 peptide was dissolved in 0.5 ml of buffer containing 50 mM phosphate, 0.1 M KCl, 6 M GdnHCl, pH 7. The solution was transferred into a small dialysis tubing with molecular mass cut-off of 1000 daltons and dialyzed against 1 liter of buffer containing 50 mM phosphate, 1 M KCl, pH 7. A 25-p1 aliquot was injected into a Hewlett-Packard model H P 1090 liquid chromatograph equipped with a Superdex 75 size exclusion column (Pharmacia LKB Biotechnology, Uppsala, Sweden). The sample was eluted with 50 mM phosphate, 1 M KCl, pH 7, at 0.5 ml/min.

Similar batches of pure oxidized C2A16/C33A16 peptide were dissolved separately in 50 mM phosphate, 1 M KCl, 0.1 M GdnHCl and in 50 mM phosphate, 1 M KC], 0.2 M GdnHCl(1 mg/ml) and the solutions were incubated for 12 h at 25 "C. Then, 25-pl aliquots were analyzed by size-exclusion chromatography using a 50 mM phosphate, 1 M KC], pH 7, buffer for elution at 0.5 ml/min.

Molecular Graphics-The parallel and antiparallel coiled-coils were modeled on a Silicon Graphics Personal Iris with Insight I1 and Discover programs (Biosym Technologies, Inc., San Diego, CA). The consistent valence force field (CVFF) and a 12-A interatomic non- bond cut-off were used in all molecular dynamics and energy minimizations. For the parallel coiled-coil, the starting structure was built by superimposing two a-helices on the backbone of the x-ray structure of the GCN4 leucine zipper (O'Shea et al., 1991). For the antiparallel coiled-coil, the starting structure was built by simulta- neous superimposition of one a-helix on the backbone of residue 9- 26 and the other a-helix on the backbone of residue 35-52 of the ColEl Rop x-ray structure (Banner et al., 1987). The molecular dynamics (5000 steps with each step of 1 fs at 300 K) and energy minimizations (100 steps of steepest descents and 1000 steps of conjugate gradients) were then performed for the starting structures of both parallel and antiparallel coiled-coil using the same restraints and parameterizations. Distance restraints of 9.2-9.4 A between the two helical axes and 2.8-3.3 8, for intra-helical (i, i + 4) hydrogen bonds with a force constant of 50 as well as weak distance restraints of 4.8-5.6A for the distances of Ca atoms to the corresponding Ca atoms in the opposite helix at positions "a" and "d" with a force constant of 1 were used in the molecular dynamics and energy minimization calculations.

RESULTS

Fig. 1 shows the amino acid sequences of two synthetic peptides, designated as C2A16 and C33A16, and the three possible alignments of their a-helical polypeptide chains to form parallel or antiparallel disulfide-bridged two-stranded a- helical coiled-coils. In a mixture of dissimilar species that can react with themselves as well as each other, such as in the case of these two peptides, the statistical distribution of the products would be a 1:2:1 ratio for the C2A16/C2A16 homostranded, C2A16/C33A16 heterostranded, and C33A16/ C33A16 homostranded peptides, respectively, in the absence of any folding constraint. Fig. 2A shows that air oxidation of a mixture of the two reduced peptides in the presence of a denaturant (6 M GdnHCl) resulted in the appearance of the three products with the predicted 1:2:1 ratio. The peaks at 21.2 and 23.4 min coeluted with the independently oxidized

19220 Antiparallel Coiled-coils

C2A16C33A16 Heterostranded

Antiparallel

C2A16 Homostranded

Parallel

Resldue Number Sequence Number

Heptad

Ac Ac I

1 LYS I

LYS

2 p J " - S - I C y r l

3 Ala Ala H 1 I

4 GI u Glu

5 m o o o o o l L . u l

8

10 G p Glu 1 1 Ala

12 ~ o o o o o ~ H 2

13 Glu Glu

14 I

Gly G ~ Y I

25

26 27

28

29

30

31

32

33

34

35

Aia H 4 l ; ; ; l o 0 0 0 o ~

Number Sequence Residue Residue

Number

amide

1

2

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

32

33 34

35

a i d e

35' 34'

33

32' 31'

3 0

2 9

28'

27'

2 6

25'

24'

2 3

22'

21'

2 0

19 18'

17'

16'

1s'

14' 13' 12

11'

1 0

0 8' 7'

6

5' 4' 3

2 1'

C33A16 Homostranded

Parallel

Residue Number Sequence

Heptad Number

Ac Ac I

1 LYS I

LYS

2 . o o o . ~ 3 Ala Ala

I H 1

I 4

5

6

7 8

9 10 1 1

12 Aia H Z

13 Glu Glu

14 G!Y I I

GJY

l 5 16 &..oo.~ 9: 17 b Glu Glu b '

18 c Ala

19 d ~ o O O O o & z' I

20 e GIU Glu e '

21 I Gly G,Iy 1'

22

23 G o O O O o g

24 GIu GI u

25 Ala Aia H 4 26 @ o O O o o ~

27 Glu Glu

28 I

GIv Glv

I I

I

I

29

30 & o O O O o g

31 Glu Glu

32 Ala Ala t i 5 I I

33 @ - S - S - r n 34 Glu Glu

35 Gly Gly I

I

I

I amide amide

FIG. 1. Amino acid sequences of the parallel and anti-parallel coiled-coils. The a-amino group of each chain is acetylated and the terminal COOH group is amidated to prevent NHz-terminal and COOH-terminal charge interactions between chains. The five heptads in each 35-residue peptide chain are indicated as Hl-H5 and each amino acid residue in a heptad is indicated as g, a, b, c, d, e, and f, respectively. The corresponding amino acid residues in the other chain are indicated as g', a', b', c', d ' , e', and f', respectively. The 3-4 hydrophobic repeat consists of leucine residues (boxed) in all "a" and "d" positions except for position 16a which is Ala (circled). In the anti-parallel coiled-coil, the alignments of the hydrophobic residues between the two polypeptide chains are different from that of the parallel coiled-coils. The parallel coiled-coils contain an Ala-Ala interaction at position 16 and a Leu-Leu interaction at position 19. By comparison, the antiparallel coiled-coil contains 2 Leu-Ala interactions at positions 16-19' and 19-16'. Each parallel coiled-coil contains a single disulfide bond at either position 2-2' or 33-33'. The antiparallel coiled-coil contains a single disulfide bond between residue 2 of one chain and residue 33 of the other.

C2A16/C2A16 (C2A16 Ox) and C33A16/C33A16 (C33A16 Ox) homostranded peptides, respectively.

When the peptide in the peak at 21.9 min was reduced with dithiothreitol the reduced products coeluted with each other and had the same retention time as those of both reduced C2A16 and C33A16 (Fig. 2B). This confirms that the peak at 21.9 min is an oxidized product and, since no other oxidation product between these peptides is possible, the peak at 21.9 min corresponds to the C2A16/C33A16 heterostranded anti-

parallel peptide. This suggests that the fully denatured, reduced single-stranded peptides in 6 M GdnHCl react as inde- pendent species without structural prerequisites, as predicted, to form the disulfide-bridged two-stranded peptides. However, when air oxidation of a mixture of reduced C2A16 and C33A16 peptides was carried out in 100 mM NH4HCOs (without denaturant), this resulted in almost quantitative oxidation into homostranded parallel peptides C2A16/C2A16 and C33A16/ C33A16. Only about 10-15% of the C2A16/C33A16 hetero-


0.3 rA C E 6 1 C33A16

C2AlGC33416

OX

0

1.3

0

(u

: - a- g o D

m 2 0.5 n a

0 0.4

18 20 22 24 26

Retention tlme. min.

FIG. 2. A , HPLC chromatogram of the products of a mixture of C2A16 and C33A16 peptides after 24 h of air oxidation. The total peptide concentration was 1 mg/ml (0.5 mg of each peptide/ml) in 100 mM NH,HCO, containing 6 M GdnHCl, pH 8.3. B, HPLC chromatogram of the heterostranded peptide peak (C2A16/C33A16 ox) after 12 h reduction with a 10-fold molar excess of dithiothreitol. The reduced C2A16 and C33A16 peptides coelute. C, HPLC chromatogram of the products after 16 h of stirring the heterostranded peptide (C2A16/C33A16 ox, 1 mg/ml) in a redox buffer (250 p~ oxidized glutathione, 250 pM reduced glutathione, 0.2 M KCl, 0.1 M Tris, 1 mM EDTA, pH 8.7) (O’Shea et al., 1989a, 1989b). The arrow indicates the peak corresponding to the oxidized heterostranded C2A16/C33A16. Peaks a and b correspond to the mixed disulfides C2A16-glutathione and C33AlG-glutathione, respectively. D, HPLC chromatogram of the products of a similar redox mixture in C, except that 1 M GdnHCl was included in the redox buffer and stirring was extended to 24 h to ensure that product equilibrium was reached. The four chromatograms were obtained using an analytical reversed-phase C8 column (Zorbax 300SB-C8, 4.6 mm X 25 cm, 5-pm particle size, 300-A pore size) using a linear AB gradient of 2%B/min, where solvents A and B are 0.05% aqueous trifluoroacetic acid and 0.05% trifluoroacetic acid in acetonitrile, respectively, at a flow rate of 1 ml/ min.

stranded peptide was formed. These results indicate that the formation of the heterostranded antiparallel peptide was not favored under benign conditions.

In order to test if the disulfide-bridged heterostranded antiparallel peptide was stable under benign conditions, the oxidized C2A16/C33A16 peptide (peak at 21.9 min) was stirred for 24 h in a redox buffer containing 250 PM reduced glutathione, 250 pM oxidized glutathione, 0.2 M KC1, 0.1 M Tris, 1 mM EDTA, pH 8.7 (O’Shea et ~ l . , 1989a, 1989b). The HPLC profile of the redox mixture (Fig. 2C) showed that the heterostranded antiparallel peptide was almost totally converted to the two parallel homostranded peptides. The product profile was unchanged when 2 M KC1 was included in the redox buffer. However, the heterostranded antiparallel peptide became the dominant product when KC1 was substituted with 1 M GdnHCl in the redox buffer (Fig. 2 0 ) . This shows that KC1 and low concentrations of GdnHCl have different effects on the relative abundance of the redox products at equilibrium. As will be shown later in the text, 1 M GdnHCl

does not denature these coiled-coil structures and the difference in the effects of KC1 and GdnHCl probably lies on their differential abilities to mask interchain electrostatic interactions.

The CD spectra of the three disulfide-bridged peptides (Fig. 3) appear very similar to each other. Their high molar ellipticities (Table I) under benign conditions (50 mM phosphate, 0.1 M KCl, pH 7.0) indicate that their structures are predom- inantly a-helical. In addition, no increase in molar ellipticities at 220 nm ( [ B ] z z O ) was observed in the presence of a helix- inducing solvent, such as l,l,l-trifluoroethanol (TFE). In- stead, there were slight decreases in negative molar ellipticities at 220 nm in the presence of TFE, which was presumably due to the dissociation of the two strands of the coiled-coil to non-interacting a-helices. TFE is known to disrupt the ter- tiary and quaternary structures stabilized by hydrophobic interactions (Lau et d. , 1984) and stabilize single-stranded a- helical structures. These changes in CD spectra suggest that the interchain interactions in the coiled-coil in benign medium stabilize the a-helices to a greater extent than TFE can stabilize the single-stranded a-helix.

The CD spectra of the heterostranded antiparallel peptide, like the two homostranded parallel peptides, showed two minima at around 208 and 220 nm, and the ratio of [0]2z0/ [ B l Z o 8 was 1.02 (Table I). This is consistent with the value of 1.03 which was previously suggested (Lau et QL, 1984; Hodges et ~ l . , 1988, 1990; Zhou et d. , 1992c) to indicate the presence of a two-stranded a-helical coiled-coil in benign medium. In the presence of 50%, TFE the [O]2zo/[0]208 ratio decreased to 0.91, which is characteristic of single-stranded or non-interacting a-helices. These data, taken together, indicate that these oxidized peptides, including the C2A16/C33A16 antiparallel peptide, exist as coiled-coil structures in benign medium.

Since we now know that the disulfide-bridged antiparallel peptide exists in a coiled-coil structure, similar to the corresponding parallel coiled-coils, we assessed the stability of these coiled-coils using GdnHCl, urea, and temperature denaturations. In these analyses, the stability of the coiled-coils against these denaturants could be assessed in terms of their transition midpoints ([GdnHC1]1,2, [Urea]l/z, or t,, respectively), which are the concentration of denaturant or the temperature at which 50% of the peptide is unfolded.

The [GdnHC1]l/z values (Table I), obtained by non-linear curve-fitting program of the GdnHCl denaturation data (Fig. 4A), showed that the antiparallel coiled-coil was more stable than the parallel coiled-coils (3.7 M versus an average of 2.5 M, respectively). In contrast, the values (Table I and

0.0

-0.5 r

Q

i -‘.O

E -1.5

;

4

m Q a -2.0

- x -2.5 0 I

-3.0

-3.5 200 210 220 230 240 200 210 220 230 240

Wavelength, nm

in 50 mM phosphate, 100 mM KCl, pH 7.0, at 20 OC. 0, C2A16/ FIG. 3. Circular dichroism spectra of the three coiled coils

C2A16 parallel homostranded; ., C33A16/C33A16 parallel homostranded; A, C2A16/C33A16 antiparallel heterostranded peptides.


TABLE I Ellipticities and stabilities of the three coiled-coil DeDtide a m l o ~ s

~~~ ~

([8]220)" (deg/cm*/ Peptide Coiled-coil alignment dmol")

[ 8 1 ~ ~ / [ 8 1 ~ ~ ratio fGdnHClInb [Urealrh* t,'

Benign 50% TFE Benign 50% TFE M "C

C33A16/C33A16 Parallel homostranded -30,360 -27,056 1.01 C2A16/C2A16

0.93 2.4 3.0 68 Parallel homostranded -30,450 -29,060 1.04 0.90 2.6 3.1 71

C2A16/C33A16d Antiparallel hetero- -31,800 -31,030 1.02 0.91 3.7 1.7 56

C2A16/C33A16' Antiparallel hetero- -30,710 -29,920 1.05 0.94 3.7 4.4 75 stranded

stranded [#I220 is the calculated molar ellipticity of the coiled-coils at 220 nm.

t,,, is the temperature at which 50% of the peptide is unfolded. This antiparallel coiled-coil has interchain electrostatic repulsions (Fig. 5B).

e This antiparallel coiled-coil has interchain electrostatic attractions (Fig. 5C).

'The [GdnHCl]* and [Urea]% values represent the concentration of denaturant at which 50% of the peptide is unfolded.

0 1 2 3 4 5 6

[GdnHCI], M

I . . . . I . . . . . . . I . 1 . . . . I . . . I 0 1 2 3 4 5 6

[Urea], M

Temperature, C

1.5 0 10 20 30 40 50 6 0 7 0 8 0 90 100

YO GdnHCl

FIG. 4. Denaturation profiles of the parallel and antiparallel coiled-coils using GdnHCl (A), urea ( B ) , and temperature denaturations (C). D, dependence of [Denat~rant]~,~ on the molar percentage of GdnHCl in the mixture of GdnHCl and urea. 0, C2A16/ C2A16 parallel homostranded H, C33A16/C33A16 parallel homostranded A, C2A16/C33A16 antiparallel heterostranded; A, renaturation of urea-denatured C2A16/C33A16. The peptide concentrations in the final solutions for CD measurements ranged from 50 to 100 MM.

Fig. 4B) indicate that the two parallel coiled-coils were much more stable than the antiparallel coiled-coil (3.0 M uersus 1.7 M, respectively). The temperature denaturation data (Table I and Fig. 4C) were also consistent with those of the urea denaturations (an average t, of 70 "C for the parallel coiled- coils uersus 56 "C for the antiparallel coiled-coil). The apparent discrepancy between the GdnHCl denaturation results and those of urea and temperature denaturations may be due to the differences in interchain electrostatic interactions as a consequence of their opposite chain alignments, that is, attractions in parallel coiled-coils (Fig. 5A) and repulsions in antiparallel coiled-coil (Fig. 5B) . As will be discussed later, this contention is supported by the denaturation results of an antiparallel coiled-coil with interchain electrostatic attractions (Fig. 5C).

Interesting results were obtained when a mixture of GdnHCl and urea was employed as denaturant. It is clear that the [DenaturantIlp values changed as a function of the percentage of GdnHCl in the denaturant mixture for both the parallel and antiparallel coiled-coils (Fig. 40). The [Denatur- ant] value for the antiparallel coiled-coil increased gradually in the range of zero to 30% GdnHCl, followed by a rapid increase from 30-50% and leveled off beyond 50% GdnHCl in the denaturant mixture. On the other hand the [Denatur-

values for the parallel coiled-coils also increased in the range of 0-30% GdnHCl, followed by a gradual decrease from 30-100% GdnHCl in the denaturant mixture. Thus, the [De- n a t ~ r a n t ] ~ / ~ values for the parallel and antiparallel coiled- coils crossed-over in the 35-50% GdnHCl region.

Another interesting observation here is the significant dif-


Intercham Attraction A. PARALLEL

6. ANT!PARALLEL lnterchaln Repulsion

g ,/”“

Interchain Attraction C. ANTIPARALLEL

FIG. 5. Cross-sectional representation of the disulfide- bridged parallel and antiparallel coiled-coils. Electrostatic attractions are indicated as solid arrows and electrostatic repulsions are indicated as dashed arrows. In the parallel and antiparallel alignments, the directions of chain propagation are the same and opposite, respectively. The hydrophobic residues at positions “a” and “a”’ are boxed in the parallel coiled-coil. A , homostranded parallel coiled-coil with interchain electrostatic attractions as in oxidized C2A16 and C33A16. B, heterostranded antiparallel coiled-coil with interchain electrostatic repulsions as in oxidized C2A16/C33A16. C, a heterostranded antiparallel coiled-coil with interchain electrostatic attractions used for comparison with B. The hydrophobic packings of the two antiparallel coiled-coils are identical.

ference in stability of the two parallel coiled-coils when a mixture of GdnHCl and urea was employed as denaturant. In GdnHCl or urea alone, C2A16/C2A16 was only slightly more stable than C33A16/C33A16 (A[GdnHC1]1/2 or A[Urea]1/2 = 0.2 M). At about 25% GdnHCl in the denaturant mixture, the difference increased to 0.7 M. This may be a result of the differences in the type of charged amino acid residues in the “open” end of the coiled-coil (opposite end to the position of the disulfide bond) that is accessible to the denaturant, Glu (position 34) in the C2A16/C2A16 and Lys (position 1) in the C33A16/C33A16 homostranded parallel coiled-coils (Fig. 1). However, at this point this suggestion cannot yet be confirmed.

Fig. 6A shows that the heterostranded antiparallel coiled- coil appeared to be a mixture of four-stranded (D, for dimeric), six-stranded (T, for trimeric) and higher aggregates ( X ) when run on size-exclusion chromatography in benign medium (50 mM phosphate, 1 M KC1, pH 7). This observation inevitably raises the question of how much this aggregation affects the stability of the coiled-coil. In order to investigate this, we

. . . 0 10 30 40

Retention time, min.

2

FIG. 6. Size exclusion chromatography of C2A16/C33Al6 antiparallel coiled-coil. A , the peptide was dissolved in 50 mM phosphate, 0.1 M KC1, 6 M GdnHC1, pH 7 (0.5 mg/ml), and dialyzed for 30 h against 50 mM phosphate, 1 M KCl, pH 7. B, the peptide was dissolved in 50 mM phosphate, 1 M KC1,lOO mM GdnHCl, pH 7, and incubated at 25 “C for 12 h. C, the same as B except that the buffer contained 200 mM GdnHCl. The equilibration and elution buffer for the three chromatograms was 50 mM phosphate, 1 M KCl, pH 7 (see “Experimental Procedures”). The letters M , D, T, and X denote monomer (two-stranded disulfide-bridged coiled-coil), dimer, trimer, and higher aggregates, respectively.

incubated the antiparallel coiled-coil in 50 mM phosphate, 1 M KC1, pH 7, containing 0.1 and 0.2 M GdnHCl. Fig. 6B shows that in the presence of 0.1 M GdnHCl the amount of higher aggregates was significantly reduced and the peak corresponding to the two-stranded form ( M , for monomeric) increased dramatically. In the presence of 0.2 M GdnHCl, the peptide existed almost entirely in its two-stranded form (Fig. 6C). However, increasing the KC1 concentration to 2 M had little effect on the dissociation of the aggregates observed in Fig. 6A.

O’Shea et al. (1989a, 1989b) suggested that when GCN4 peptides were forced into an antiparallel alignment, they had a tendency to form higher, but definite, order of association or aggregation. Although some aggregation seemed to be present in our antiparallel coiled-coil, these aggregates were not stable and were easily dissociated into the two-stranded form in the presence of a low concentration of GdnHCl (0.2 M). The denaturation of the two-stranded coiled-coil to the unfolded two-stranded form did not begin until at least 2 M GdnHCl is reached (Fig. 4A). Therefore, this aggregation was probably due to nonspecific electrostatic interactions brought about by the “polarity” of the antiparallel coiled-coil structure, that is, one side of the antiparallel coiled-coil structure is positively charged (lysines in positions “g” and “g”’) while the other side is highly negatively charged (positions “b,” “e,” e , and “b”’) (Fig. 5B) . Since the aggregates were converted

into two-stranded coiled-coil species long before denaturation starts, the aggregation of the antiparallel coiled-coil is presumed not to involve major structural changes, such as in the packing of hydrophobic residues. Therefore, this antiparallel coiled-coil is a valid model to study protein folding and assembly.

“ , ,,

DISCUSSION

Most proteins are large complex molecules and it is often difficult to obtain information that can be easily interpreted


concerning the fundamental forces that govern protein folding and structure. The use of de novo designed model proteins provide a minimalistic approach to this problem. The peptide models that we have used in this study, or their analogs, have been previously used to investigate the factors that affect the formation and stability of parallel coiled-coils (Hodges et al., 1981, 1988, 1990; Lau et al., 1984; Zhou et al., 1992a, 1992b, 1992c, 1993; and Zhu et al., 1992,1993). In this study, we have designed, synthesized, and characterized a disulfide-bridged antiparallel coiled-coil and compared its characteristics to those of the parallel, disulfide-bridged homostranded coiled- coils that are produced from the same set of reduced peptides.

In designing parallel and antiparallel coiled-coils, there are fundamental similarities and differences that have to be taken into account. For example, the 4-3 or 3-4 hydrophobic repeat that was first identified by Hodges and co-workers in the parallel coiled-coil of tropomyosin (1972) also occurs in an antiparallel coiled-coil, such as the coiled-coil domain of the Escherichia coli seryl-tRNA synthetase (Cusack et al., 1990). In the antiparallel alignment of coiled-coils, the interactions between the two a-helices are stabilized by ionic and hydrophobic interactions (Cusack et al., 1990) like those observed by O’Shea et d . (1991) for the parallel GCN4 coiled-coil.

Conversely, there are also fundamental differences in their structures. In parallel coiled-coils, the hydrophobic residues in the “a” positions are paired and packed side-by-side (O’Shea et al., 1991) with the hydrophobic residues in the “a”’ positions of the opposite chain and similarly with “d” and “dl”, respectively (Fig. 5A) to form a dimer interface. In this packing the “a” and “d” positions are non-equivalent because the Ca-Cp vectors in position “a” are pointed away from each other while the Ca-Cp vectors in position “d” are pointed toward each other (Fig. 7A). Moreover, the Cy methyl groups in the “a” position of each chain are closer to each other than the Cy methyl groups in the “d” position of each chain. In the antiparallel coiled-coils, hydrophobic residues in the “a” positions are paired with hydrophobic residues in the “d”’ position of the opposite chain, and similarly with “d” and “a”’ (Fig. 5, B and C; Cusack et al. (1990)). Molecular modeling studies showed that this switch in the pairing of hydrophobic residues does not drastically change the geometry of hydrophobic packing. When the two a-helical chains are aligned in an antiparallel orientation and then energy minimized, only minimal rearrangements of the side chains occurred. The orientations of the Ca-CP vectors in the antiparallel coiled- coil are still similar to those in the parallel coiled-coil. How- ever, the mixed pairings (a-d’ and d-a’) in the antiparallel coiled-coil result in two pairs of identical hydrophobic packings, as opposed to a combination of non-equivalent “a-a’’’ and “d-d”’ packings in parallel coiled-coils. In addition, the Cy methyl groups in the “a” or “a”’ positions are closer to the Cp methyl groups in the “d” or “d”’ positions (Fig. 7B). These are most likely to result in changes in coiled-coil stability.

The interactions of charged residues are also different in parallel and antiparallel coiled-coils. When the a-helical chains in our model peptides are aligned in an antiparallel orientation, the residues at positions “e” and “e”’ would occur on one side of the coiled-coil and residues “g” and “g”’ would occur on the other side (Fig. 5B). This antiparallel alignment results in similar charges of residues at positions “e” and “e”’ and of residues at positions “g” and ‘‘g’,’’ which are structur- ally adjacent, which in turn would result in interchain electrostatic repulsions. In contrast, the same peptides aligned as a parallel coiled-coil would result in oppositely charged residues at positions “g” and “e”’ as well as “e” and “g”’ forming

A. PARALLEL

B. ANTIPARALLEL

FIG. 7. Computer modeling of the parallel and antiparallel coiled-coil structure to show the orientations of Leu side chains at positions a and d. Only side chain carbons of leucine at positions 16 ( a and a’ ) and 19 ( d and d ’ ) as well as seven main chain or-carbons around this leucine residue are displayed. The view is along the coiled-coil axis from the NH2 terminus for the parallel coiled-coil.

interchain electrostatic attractions (Fig. 5A). This is an important consideration because the potential formation of interchain electrostatic attractions of charge residues in the “e” and “g” positions could be responsible for controlling the type of alignment of the a-helices (parallel versu antiparallel) and the stabilization of the coiled-coil structure (Stone et al., 1975).

In this study, we have shown that the same peptide mono- mers that were used to prepare parallel coiled-coils could also form a stable antiparallel coiled-coil. The denaturation curves of the three disulfide-bridged coiled-coils in GdnHC1, in urea, and at different temperatures (Figs. 4, A-C, respectively) showed relatively smooth transition curves, suggesting that the transition intermediates, if any, are not stable enough to be detectable. The renaturation curve of the antiparallel coiled-coil in urea (Fig. 4B) indicates that the folding/unfolding pathway is a reversible process. These data suggest that, once disulfide-bonded, there are no physical restrictions for


the two a-helical chains to pack into an antiparallel coiled- coil.

But how do we interpret the apparent discrepancy in the stability values obtained from GdnHCl and urea denaturation studies? The [Urealll2 values indicate that the antiparallel coiled-coil is less stable than the parallel coiled-coils while the [GdnHC1]1/2 values indicate the opposite. Greene and Pace (1974) have compared urea and GdnHCl denaturations of a group of globular proteins and suggested that the mechanism of unfolding was the same in the presence of the two denaturants and, although the relative effectiveness of these denaturants depended on the particular protein, GdnHCl was more effective than urea in unfolding the proteins studied. For further discussion on the physical basis of protein denaturation by urea and GdnHC1, see Creighton (1993), which emphasizes that the mechanism of denaturation by GdnHCl and urea are not understood in detail. However, we believe that in our model system, the use of both urea and of GdnHCl denaturations can be useful in differentiating electrostatic and hydrophobic effects. For example, GdnHCl is a charged molecule and like most salts can potentially mask electrostatic interactions, increase hydrophobic interactions, and at higher concentrations becomes a denaturant. As a consequence, the stability values obtained from GdnHCl denaturation may be mainly indicative of the contributions of hydrophobic packing to the stability of the coiled-coil when hydrophobic interactions are the dominating interactions responsible for stability of the folded protein. On the other hand, since urea is un- charged and it cannot mask electrostatic interactions, the [Urea]1/2 values may be indicative of the total or net stability of the protein structure, that is, the sum of the hydrophobic and electrostatic contributions. Electrostatic interactions can be stabilizing (attractions) or destabilizing (repulsions) to the protein structure. Stability values obtained from urea denaturation studies should therefore be expected to be consistent with temperature denaturations.

The major consequences of aligning the two a-helical chains into an antiparallel coiled-coil include changes in the pairing of hydrophobic residues in the interface and, in the case of our peptides, a change from interchain electrostatic attractions in parallel coiled-coils to interchain electrostatic repulsions in the antiparallel coiled-coil. One can interpret these GdnHCl denaturation data to indicate that the hydrophobic packing in the antiparallel coiled-coil ([GdnHCl]l/2 = 3.7 M) makes a greater contribution to stability than the hydrophobic packing in the parallel coiled-coil (average [GdnHC1]1/2 = 2.5 M). However, the urea and temperature denaturation data indicate that the stability of the antiparallel coiled-coil is reduced by the destabilizing effect of 10 pairs of interchain electrostatic repulsions that result when the two a-helical chains are aligned in opposite directions (Fig. 5B). On the other hand, the slightly weaker hydrophobic packing in the parallel coiled-coil is supplemented by the stabilizing effect of 10 pairs of interchain electrostatic attractions (Fig. 5A) . In order to test this contention, we synthesized another antiparallel peptide that was similar in every respect to the C2A16/ C33A16 antiparallel coiled-coil, except that the 10 pairs of interchain electrostatic repulsions were changed to 10 pairs of attractions (Fig. 5C). This was accomplished by oxidizing a reduced C2A16 peptide with a reduced C33A16 peptide having a different amino sequence (Ac-E-L-A-E-L-K-G-E-L-

K-G-amide). This new antiparallel peptide also showed similar CD characteristics to the other parallel and antiparallel coiled-coils. Under benign conditions (50 mM PO4, 0.1 M KCl), the [Urea]1/2 of the antiparallel coiled-coil with interchain

A-E-L-K-G-E-A-A-E-L-K-G-E-L-A-E-L-K-G-E-L-A-E-C-

electrostatic attractions was 2.7 M higher than the antiparallel coiled-coil with interchain electrostatic repulsions (4.4 versus 1.7 M) (Table I, Fig. 8, A and B ) . In addition, the t, of the antiparallel coiled-coil with interchain electrostatic attractions was 19 “C higher than that with repulsions (75 uersus 56 “C) (Table I). These clearly indicate that the interchain electrostatic interactions, which are not masked in either the urea or temperature denaturations, have significant effects to the overall stability of the coiled-coil structure. The antiparallel coiled-coil with interchain electrostatic attractions also had a higher [Urea]1/2 and t, than those of the parallel coiled- coils (4.4 M and 75 “C versus an average of 3.0 M and about 70 “C, respectively). Since the interchain electrostatic attractions are similar in these parallel and antiparallel coiled-coils, this further indicates that the packing of hydrophobes in the antiparallel coiled-coil is better than those in the parallel coiled-coil, and this may be attributed to the change from “a- a”’ and “d-d”’ packings in a parallel coiled-coil to “a-d”’ and “d-a”’ in an antiparallel coiled-coil. It must be noted that the relative stability of the parallel uers’sus the antiparallel coiled- coils with regard to hydrophobic packing may be sequence dependent and thus may differ from the results reported in this study.

The GdnHCl and urea denaturation data in the presence of KC1 (Fig. 8) offer further insights into the differences in the actions of these denaturants. For example, the two antiparallel coiled-coils with identical hydrophobic packing but differing in their interchain electrostatic interactions (attractions or repulsions, Fig. 5, B and C) showed almost identical stabilities against GdnHCl denaturation, either in the presence of 0.1 M or 1.0 M KC1 (average [GdnHC1]l,2 value of 3.7 f 0.2 M, Fig. 8, C and D, Table 11). These data clearly show that electrostatic interactions have negligible effect in GdnHCl denaturations because such interactions are effec- tively masked by the charged GdnHCl molecules. In addition, these data also suggest that the effect of salt in promoting hydrophobic interactions had already been optimized by GdnHCl, again acting as a salt, since further addition of KC1 did not result in an additional increase in the [GdnHC1I1/, value. This is in contrast to the large increases in the

1 1 .o

0.8

0.6

.P 0.4

Interchain I

8

; 0.2

0.0

0 1 2 3 4 5 6 0 1 2 3 4 5 J 6

[Urea], M [GdnHCI], M

I I . . . ,

FIG. 8. Urea and GdnHCl denaturation profiles of two antiparallel coiled-coils in the presence of KC1 in 50 mM PO4 buffer, pH 7.0. A , urea denaturation curves of an antiparallel coiled- coil with interchain electrostatic repulsions. A, 0.1 M KC1; ., 1.0 M KCl; 0, 2.0 M KCl. B, urea denaturation curves of an antiparallel coiled-coil with interchain electrostatic attractions. 0, 0.1 M KCI; 0, 1.0 M KCl. C. GdnHCl denaturation curves of an antiparallel coiled- coil with interchain electrostatic repulsions. 0, 0.1 M KC1; 0, 1.0 M KCl. D, GdnHCl denaturation curves of an antiparallel coiled-coil with interchain electrostatic attractions. U, 0.1 M KC1; m, 1.0 M KC1.


TABLE I1 Stabilities of antiparallel coiled-coils determined from GdnHCl and

urea denaturation at varvina KC1 concentrations

Peptide electrostatic interactions 0.1 M~ 1.0 M 0.1 M 1.0 M 2.0 M

Interchain [GdnHCl],," [Urea],,"

C2A16/C33A16 Repulsions 3.7 3.5 1.7 3.2 4.5 C2AWC33A16 Attractions 3.7 3.6 4.4 6.0 ND' "The [GdnHCl]* and [Urea], values represent the concentration

of denaturant at which 50% of the peptide is unfolded. The concentration of KC1 in 50 mM phosphate buffer at pH 7.0.

Stock solutions of GdnHCl and urea were also prepared in the same buffer containing the desired KC1 concentration.

e Not determined because the coiled-coil was too stable and 6 M urea is the maximum concentration that can be prepared in the presence of 2.0 M KCl.

4.0

l.O[ ' ' ' I 0 0.5 1.0 1.5 2.0

0 1 2 3 4 5 6 7 [UREA], M

FIG. 9. Effect of KC1 on the stability of the antiparallel coiled-coil with interchain electrostatic repulsions. Depend- ence of the [Urea]l/n values ( A ) and AGHZo values ( B ) with the concentration of KC1 during urea denaturations. Panel B: 0, 0.1 M KC1; W, 1.0 M KCI; A, 2.0 M KCl. AG is the free energy of unfolding at each individual urea concentration and calculated according to the equation: AG = -RT In [(l-fn)/fn], where fn is the fraction folded at a particular urea concentration. The linear dependence of AG on the urea concentration is given by: AG = AGHzo - m[Urea], where m is the slope term and AGHzO is the free energy of unfolding in the absence of urea (Greene and Pace, 1974).

[Urea]l/z upon the addition of KC1 in the denaturation of these antiparallel coiled-coils. The [UreaI1l2 values increased from 1.7 to 3.2 to 4.5 M with increasing KC1 concentrations of 0.1, 1.0, and 2.0 M, respectively, for the antiparallel coiled- coils with electrostatic repulsions (Table I1 and Fig. 8A). There is a linear effect of KC1 in promoting the hydrophobic interactions, both in terms of the [Urea]l/z values (Fig. 9A) as well as the calculated AGHz0 values (Fig. 9B). Similarly, the [Urealllz values increased from 4.4 to 6.0 M with increasing KC1 concentration of 0.1-1.0 M, respectively, for the antiparallel coiled-coil with electrostatic attractions (Table I1 and Fig. 8B), and was too stable to completely denature in the presence of 2 M KC1. The difference in [Urealllz values with changing KC1 concentrations of about 1.0 M were approximately 1.5 M for both antiparallel coiled-coils. Taken together,

these observations strongly suggest tha t the [Gdr~HCll~ /~ values may be useful as relative indicators of the strength of hydrophobic interactions. On the other hand, urea denaturation studies become important when the total or net stability of the protein molecule is of interest, or if one is to assess the specific contributions of the ionic interactions to protein stability.

An important question that still remains to be answered is on the magnitude of the individual effects of salt (KC1 or GdnHCl) on the electrostatic interactions and in promoting hydrophobic interactions. Unfortunately, while the [GdnHC1I1/2 values seem to be good estimates of the relative strengths of the hydrophobic interactions when comparing different proteins, they cannot be directly compared with the [Urea]l/z values for the same protein because of the differences in their modes of denaturation. GdnHCl as a salt masks electrostatic interactions, promotes hydrophobic interactions, and at higher concentrations promotes denaturation. In contrast, KC1 is non-denaturing, promotes hydrophobic interactions, and should mask electrostatic interactions. However, the addition of 1 M KC1 resulted in roughly the same increases in stabilities of the antiparallel coiled-coil with interchain repulsions and that with attractions (Fig. 8, A and B). It was expected that the addition of KC1 to the antiparallel coiled- coil with interchain repulsions would have a stabilizing effect as a result of masking the electrostatic repulsions and a stabilizing effect as a result of promoting hydrophobic interactions. On the other hand, the addition of KC1 to the antiparallel coiled-coil with interchain electrostatic attractions would have a destabilizing effect as a result of masking the electrostatic attractions and a stabilizing effect as a result of promoting hydrophobic interactions. Since the same increase in stability was observed on the addition of KC1 to both coiled-coils (Table 11), this suggests that KC1 is unable to mask electrostatic interactions in this model system. These results are consistent with our observation that 0.1-0.2 M GdnHCl could disrupt aggregation in the antiparallel coiled- coil (Fig. 6 ) when 2 M KC1 could not. Similarly, 1 M GdnHCl in the redox buffer system was able to mask interchain electrostatic interactions and allow the formation of the antiparallel coiled-coil containing interchain electrostatic repulsions (Fig. 2 D ) , while the addition of 2 M KC1 could not and only parallel coiled-coils were formed. The reasons for the inability of KC1 to mask these interactions in this protein are not understood and are worthy of further investigation.

It is also interesting to note that although GdnHCl denaturation, which masks electrostatic effects, shows that the disulfide-bridged antiparallel coiled-coil with electrostatic repulsions is more stable than the parallel coiled-coils, it was not the favored product under benign conditions. In addition, similar analogs have been previously shown to form parallel homodimers in their reduced state (Hodges et al., 1990). This suggests that under benign conditions electrostatic forces are important in determining polypeptide chain alignment and stability. This is supported by the observation that the antiparallel coiled-coil with interchain electrostatic attractions formed spontaneously under benign conditions (100 mM NH4HC03, pH 8.3). In addition, Fig. 2 0 shows that the relative amount of the antiparallel product at equilibrium was significantly increased when the electrostatic repulsions were masked by addition of 1 M GdnHCl in the redox buffer, at which concentration the peptides remain in coiled-coil structures.

Therefore, results of these studies indicate that 1) there are no physical restrictions in packing the hydrophobic residues in either parallel or antiparallel fashion, 2 ) interchain electro-


static interactions appears to determine chain alignment, and 3) the overall stability of the coiled-coil is determined by the types of hydrophobic packing as well as the interchain electrostatic interactions. These are consistent with the observation that in naturally occurring proteins with coiled-coil domains, the alignment of &-helical chains occur either in parallel or antiparallel orientation. In globular proteins, the most simplistic way to bring two a-helices together is through a helix-turn-helix motif which aligns the a-helices in an antiparallel orientation. On the other hand, the parallel packing of &-helices in many coiled-coils such as in the DNA-binding proteins (O'Shea et al., 1989a, 1989b, 1991) and myosin (Parry, 1981) may be to modulate the stability but, as well, this alignment may represent the most simplistic way to bring the two functional binding regions together in register.

Acknowledgments-We thank Paul Semchuk for peptide synthesis, purification, and mass spectrometry, Kim Oikawa and Bob Luty for CD spectra measurements, the Alberta Peptide Institute for perform- ing the amino acid analyses, and Dr. Frank Soennichsen and Tim Jellard for their assistance in the use of the molecular graphics programs.

REFERENCES Banner, D. W., Kokkinidis, M., and Tsernoglou, D. (1987) J. Mol. Biol. 1 9 6 ,

Caspar,,D. L. D., Cohen, C., and Longley, W. (1969) J. Mol. Biol. 4 1 , 87-107 Bowie, J. U., and Sauer, R. T. (1989) Biochemistry 28, 7139-7143

Chakenan, A. E., Tesmer, V. M., Manly, S. P., Brackett, J. K., Lynch, M. J., Hoh, J. T., and Matthews, K. S. (1991) J. Biol. Chem. 266,1371;1374

Creighton, T. E. (1993) Proteins Structures and Molecular Propertws, 2nd ed., pp. 293-295, W. H. Freeman & Co., New York

Crick, F. H. C. (1953) Acta CrystaUogr. 6 , 689-698 Cusack, S., Berthet-Colominas C., Hartlein, M., Nassar, N., and Leberman, R.

De Francesco, R., Pastore, A., Vecchio, G., and Cortese, R. (1991) Biochemistry (1990) Nature 347,249-25$

Gentz, R., Rauscher, F. J., 111, Abate, C., and Curran, T. (1989) Science 2 4 3 , 30,143-147

1695-1699

657-6 15

Greene, R. F., Jr., and Pace, C. N. (1974) J. Biol. Chem. 249,5388-5393 Hodges, R. S., Sodek, J., Smillie, L. B., and Jurasek, J. (1972) Cold Spring

Hodges, R. S., Saund, A. K., Chong, P. C. S., St.-Pierre, S. A,, and Reid, R. E.

Hodees. R. S.. Semchuk. P. D.. Taneia. A. K.. Kav. C. M.. Parker. J. M. R..

Harbor Symp. Quant. Biol. 37,299-310

(1981) J. Biol. Chem. 2 6 6 , 1214-1224

a<d Mant, C. T. (1988) Pept.'Res. 1; 19-30 . " .

Hodges, R. S., Zhou, N. E., Kay, C. M., and Semchuk, P. D. (1990) Pept. Res. R 193-137

Hu, J. C., OShea, E. K., Kim, P. S., and Sauer, R. T. (1990) Science 2 6 0 ,

Johnson, P., and Smillie, L. B. (1975) Biochem. Biophys. Res. Commun. 6 4 ,

_,*" ". 1400-1403

1316-1322 Landschulz, W. H., Johnson, P. F., and McKnight, S. L. (1988) Science 2 4 0 ,

Lau, S. Y. M., Taneja, A. K., and Hodges, R. S. (1984) J. Biol. Chem. 2 5 9 ,

Lehrer, S. S. (1975) Proc. Natl. Acad. Sci., U. S. A. 72,3377-3381 Lupas, A., van Dyke, M., and Stock, J. (1991) Science 252,1162-1164 McLachlan, A. D., and Stewart, M. (1975) J. Mol. Biol. 9 8 , 293-304

O'Shea, E. K., Rutkowski, R., Stafford, W. F., 111, and Kim, P. S. (1989b) O'Shea, E. K., Rutkowski, R., and Kim, P. S. (1989a) Science 243,538-542

O'Shea, E. K., Klemm, J. D., Kim, P. S., and Alber, T. (1991) Science 264 ,

Parry, D. A. D. (1981) J. Mol. Biol. 163,459-464 Phillips, G. N., Jr., Fillers, J. P., and Cohen, C. (1986) J. Mol. Biol. 192 , 111-

Reddy, B. A., Etkin, L. D., and Freemont, P. S. (1992) Trends Biochem. Sci.

Richardson, J. S. (1981) Adu. Protein Chem. 34,167-339 Sodek, J., Hodges, R. S., Smillie, L. B., and Jurasek, J. (1972) Proc. Natl. Acad.

Stewart, M. (1975) FEES Lett. 63,5-7 Stone, D., Sodek, J., Johnson, P., and Smillie, L. B. (1975) in Proceedings of

Contractile Systems (Biro, E N . A, ed) Vol. 31, pp. 125-136, North-Hollanl the IX Federation of Euro an Biochemical Societies Meeting: Proteins o

Zhou, N. E., Zhu, B.-Y., Kay, C. M., and Hodges, R. S. (1992a) Biopolymers Amsterdam

Zhou, N. E., Kay, C. M., and Hodges, R. S. (1992b) Biochemistry 3 1 , 5739- 32,419-426

Zhou, N. E., Kay, C. M., and Hodges, R. S. (1992~) J. Bid. Chem. 267,2664- 5746

1759-1764

13253-13261

Science 246,646-648

539-544

131

17,344-345

Sci. U. S. A. 69,3800-3804

967n Zhou,h. E., Kay, C. M., and Hodges, R. S. (1993) Biochemist 32,3178 3187 Zhu, B.-Y., Zhou, N. E., Semchuk, P. D., Kay, C. M., and Hoges, R. S. (1992)

Zbu, B.-Y., Zhou, N. E., Kay, C. M., and Hodges, R. S. (1993) Protein Science Int. J. Pept. Protein Res. 4 0 , 171-179

2,383-394

Documents

Oscar D. Monera, Nian E. Zhou, Cyril M. Kay, and Robert S