Conformational preferences of oligopeptides rich in α-aminoisobutyric acid. I. Observation of a 310/α-helical transition upon sequence permutation

Conformational Preferences of Oligopeptides Rich in a-Aminoisobutyric Acid. I. Observation of a 310/a-Helical Transition upon Sequence Permutation GAUTAM BASU, KEN BAGCHI, and ATSUO KUKl Cornell University, Department of Chemistry, Baker Laboratory, Ithaca, New York 14853

SYNOPSIS

The solution conformation of peptides rich in the a,a-dialkylated amino acid Aib has proven to be a subtle problem, not because of helix/coil transitions, but rather because of a-heli~al/3~,,-helical competition.A special series of peptides containing 75% Aib has been synthesized that feature identical amino acid composition but differing sequences;they are sequence permutation isomers. Nuclear magnetic resonance hydrogen-bonding studies reveal that there is a sequence permutation induced transition between the two alternativehelical forms within this set. The implications for the design and conformational prediction of helical Aib-rich peptides are discussed.

INTRODUCT10N The presence of the a,a-dimethyl group in a-aminoisobutyric acid ( Aib ) induces severe restrictions onto the backbone conformation of polypeptides containing Aib.' This has motivated intense investigation directed toward a comprehensive understanding of the conformational states and energetics of such polypeptides using both e~perirnental'-~ and theoretical6-' probes. For all such polypeptides studied so far, an overall helical conformation for the peptide backbone has been unambiguously established; however, only pure Aib oligomers have been found to adopt exclusively the distinctive 310helical secondary structure.1°-12 Backbone conformations of sequence-specific oligopeptides with mixed composition ( Aib mixed with a-monoalkylated residues), on the other hand, may display either 310-, a-, or mixed 310/a-helical pattern depending upon composition, distribution, and the number of residues, and it has been demonstrated that the peptide environment can play an important role for peptides near a critical chain While quantitative understanding of the exact interplay of these factors is yet to come, empirical rules have been p r o p o ~ e d . ~In, ' ~this article, Biopolymers. Val. 31, 17fi3-1774 (1991) CCC ooo6-3525/91/141763-12$04.00 C) 1991 John Wiley & Sons, Inc.

we report our investigations on oligopeptides rich in Aib ( 75% ) ,and for the first time experimentally demonstrate a conformational transition ( 3101ahelical) upon sequence permutation of an Aib-rich short peptide backbone. The peptides reported were synthesized as a component of a parallel investigation where electronic interactions between chromophores incorporated into helical Aib-rich oligopeptides were ~tudied.'~ As a result, most of the sequences include two aromatic residues with systematically varied sequence positions. The two guest aromatic residues may be considered to replace two Aib units within a parent Aib sequence Ac- ( Aib)8-NHMe. The aromatic residues for three such peptides reported here--- ( 1'-napthyl) -L-alanine (Nap) and L-phenylalanine ( Phe ) -were positioned respectively at the 3,6 positions (peptide 3,6-NF), 3,5 positions (peptide 3,5-NF), or the 4 5 positions (peptide 4,5NF) of the parent sequence. Thus by design, the two aromatic groups within this series are spatially related to each other by a successive helical twist achieving exactly one turn separation in the peptide 3,6-NF, if one assumes a 310-helicalbackbone. Such a backbone conformation is established for an Aib homo-octamer." Two further variations of the peptide 3,6-NF are ( a ) an octamer where both the aromatic residues were replaced by L-alanines (peptide 3,6-AA), and (b) an octamer where the phenyl ring 1763

1764

BASU, BAGCHI, AND KUKI

Peptide

Aib

Ail

3,6-NF

Aib

i

Aib

Phe

Aib

A

F t O H

e

i, C H 3 C N ii) H +

3 SOCl2

CH3 C N

MeNH

F

H

OH

H-

OH

Ac20

F

I I

NHMe H+

t -OH

NHMe

F HNEt2

Ac20

NHMe

H t C H 3 C N / r e f lux

F

I

F

I

I

NHMe

I

HNEt2

2 NHMe

-OH H

F

II D Q

NHMe

F

HNEt2

3

NHMe

H

?/ref

IX

f Ac

c

c

c

4 NHMe

Figure 1. Synthetic scheme for the octamers 3,6-NF, 3,5-NF, and 4,5-NF. The 5 ( 4 H ) oxazolones, derived from the corresponding free acids by acetic anhydride treatment, are shown as “Ox,” and these function as C-activated Aib peptide blocks. The highly activated monomer Aib nucleophile, 2- ( dimethylamino) -3,3-dimethyl-azirine, is shown as “Az.” The N-protecting Fmoc group is represented as “F.” The common C-terminal Aib trimer in 3,5-NF and 4,5-NF was synthesized by the azirine method (extending F-Aib-Aib-OH) in a homologous extension of the 3,6-NF method. The Phe residue was introduced at the Nterminal of this trimer by an IIDQ coupling. See text for details of the individual reaction steps for the peptide 3,6-NF.

is bromo substituted at the para position (peptide 3,6-NF ’) . In addition, three pentamers (fragments of the 3,6 octamer series) containing one guest residue ( X ) substituted at position 3 of the parent peptide Ac- ( Aib)z-X-( Aib)z-NMe2 were also studied. The guest residues were Ala (peptide 3-A), Phe (peptide 3-F), or Nap (peptide 3-N). Solution phase conformations of these peptides as revealed by ‘Hnmr studies in DMSO-d, and CD3CN are presented in this paper. In addition, we comment on the avail-

able statistical models, and their ability to define and match the observed sequence permutation effect.

MATERIALS AND METHODS All peptides were synthesized by solution phase methods. Special methods are required to circumvent the inherent difficulty of coupling Aib residues,

PREFERENCES OF OLIGOPEPTIDES RICH IN AIB. I

Peptide Aib

Aib

Aib

Nap

1765

4,s-NF Phe

Aib

Aib

Aib

NHMe

CH $N/reflux IIDQ

NHMe

HNEt2 Ac

NHMe

IIDQ

P e p t i d e 3,5-NF

Aib

Aib

Nap

Aib

Phe

Aib

Aib

Aib NHMe

NHMe

NHMe

NHMe Ac

NHMe

Figure 1. (Continued from the preuious page.)

which arises due to the steric strain they impart on the resulting peptides. Specifically, we used a combination of the oxazolone, 11~15azirine, 16-18 acid chloride, l9 or the l-isobutoxy-carbonyl-2-isobutoxy-1,2dihydroquinoline (IIDQ)2o coupling methods-the exact combination being sequence dependent ( Figure 1).Aib dimer and trimer blocks were synthesized by reacting the highly N-activated 2- (dimethylamino ) -3,3-dimethyl-azirine with 9-flourenylmethyoxycarbonyl (Fmoc) protected Aib free acid. Subsequently, Fmoc protected guest amino acids were

reacted with the free N-termini of such dimers. Further Aib dimer blocks were then C-activated as oxazolones prior to coupling with the N-terminal of the growing peptide. The peptides, purified after each synthetic step by flash chromatography (5-10% CHBOHin CH2C12 on silica gel), were fully characterized by 'H-nmr along the synthetic route. The unique pattern of the required HC a and H2C resonances, side-chain aromatic resonances, and blocking groups were readily verified for each peptide fragment. The peptides @

1766

BASU, BAGCHI, AND KUKI

v

"1 'I"l"1 'l"l"l7

Ac-(NHCC0)-(NHCCO) (NHCHCO) (NHCCO) (NHCCO) (NHCHCO) (NHCCO) (NHCCO) NHMe

observedrnass

213.0

410.0

494.7

579.8

727.1

812.1

897.2

928.2

expected mass

2 13.1

410.2

495.3

580.3

727.4

812.4

897.5

928.5 (MH')

Figure 2. FAB/MS data for the peptide 3,6-NF. The fragmentation pattern, along with nmr data, confirms the amino acid composition as well as the sequence of the peptide (AcAib-Aib-Nap-Aib-Aib-Phe-Aib-Aib-NHMe). The fragmentation pattern arises from the type B peptide bond cleavage, where the peptide is sequentially fragmented one residue at a time from the C-terminal.

were finally purified by reversed-phase C18high performance liquid chromatography ( HPLC ) on a Waters Delta-Pak column with 5-20% HzO in CH,OH as eluent. The molecular mass and the amino acid sequence were checked by fast atom bombardment mass spectrometry ( FAB / MS; see Figure 2 and T a ble I ) . Synthesis of Peptides

The synthetic schemes for the three sequence isomeric octamers-peptide 3,6-NF, peptide 3,5-NF, and peptide 4,5-NF-are summarized in Figure 1 and the full details of the synthesis for a representative octamer, peptide 3,6-NF, are given here. All other peptides were synthesized according to combinations of reactions described below. The amino acids p- (1'-napthyl) -L-alanine and L-p-bromophenylalanine were obtained from Bachem Bioscience, Inc., and the intermediates 2- (dimethyl-

amino) -3,3-dimethyl-azirine (I),l8 Fmoc amino acids," Fmoc amino acid chloride^,'^ and the N-terminal dipeptide 2- ( 1-acetylamino-1-methyl ethyl) -4,4-dimethyl-5(4H ) -oxazolone (11)l5 weresynthesized according to literature procedures. The solvents for thin layer chromatography (TLC; silica), 5% CH3OH/CH2C12,7% CH,OH/CH,Cl,, and 10% CH30H/CH,Cl2 are abbreviated as A, B, and C, respectively. Fmoc-Aib-Aib-NMe2 (111). T o a solution of 6.1 g of azirine ( I )in 250 mL of dry CH3CN, 17.4 g of Fmoc( Aib) -OH was added. This was stirred under N, and within 20 min a solid precipitated. After 2 h the solid was collected on a Buchner funnel while the filtrate was concentrated and stirred under Nz overnight. Further precipitate formed overnight and was filtered, and the filtrate, after removing CH,CN, was redissolved in CHC13 and extracted with aqueous HCO, ( p H 8).The organic layer was dried ( MgS04)

Table I FAB/MS Data for all the Octamers' Observed Mass of Peptide Fragmentsb Expected Mass of Peptide Fragments

PEPTIDES 3,5-NF 4,5-NF 3,6-NF 3,6-NF 3,6-AA

213.1 213.1 213.1 213.1 213.0 213.1 213.0 213.1 213.1 213.1

410.3 410.2 298.2 298.2 410.0 410.2 410.0 410.2 284.1 284.2

495.5 495.2 495.5 495.2 494.7 495.2 495.0 495.2 369.2 369.2

642.4 642.3 642.4 642.3 579.8 580.3 580.0 580.3 454.2 454.3

727.8 727.4 727.8 727.4 727.1 727.4 805.0 805.3 525.2 525.3

812.9 812.4 812.9 812.4 812.1 812.4 890.1 890.4 610.3 610.3

897.6 897.5 897.6 897.5 897.2 897.5 975.1 975.4 695.3 695.4

928.5 928.5 928.5 928.5 928.2 928.5 1006.1 1006.5 726.4 726.4

" T h e fragmentation pattern, explained for a representative peptide (3,6-NF) in Figure l., was used to analyze the mass and the sequence for all the peptides. The octamer 3,6-NF contains Br and the calculated/observed mass is reported only for the fragments corresponding to the 79Br isotope. Fragments for the "Br isotope were observed as expected.

PREFERENCES OF OLIGOPEPTIDES RICH IN AIB. I

and evaporated to yield a white solid, which was identical in tlc pattern ( R f 0.4, solvent B; R f 0.67, solvent C ) with the solid collected earlier. This was identified a s the dimer (ZZZ)by nmr ( CDC13).Yield 22.57 g ( 96%). Fmoc-Aib-Aib-OH (IV). Two grams of Fmoc-AibAib-NMe2 (ZIZ)was dissolved in 10 mL of CH3CN/ H20/HC1 ( 16 : 1 : 4 ) and stirred for 1 h a t room temperature. T h e solvent was then removed under vacuum and the resultant solid ( R 0.35, solvent C ) was washed with water and collected on a Buchner funnel. The solid was pure, with no residual NMe2 resonance present in the nmr (DMSO-&) ,confirming complete deblocking. Yield 1.8 g (96%). Oxazolone from FMOC-Aib-Aib-OH (V). T h e amount of 1.19 g free acid ( Z V ) was added to 15 mL acetic anhydride that was kept a t 120°C for 20 min. The solvent was removed under vacuum. The residue was then repeatedly taken up in xylenes and evaporated to dryness. A viscous yellowish oil resulted, which contained the oxazolone and the free acid. The oxazolone was isolated by flash chromatography on a silica column (3% CHBOH/CHzC12)and solidified on storage. Th e solid ( R f 0.67, solvent C ) was confirmed to be the oxazolone by nmr and ir (1820 cm-'). Yield 0.91 g (80%). H-(Aib-Aib)-NHMe (VI). Six hundred eighty-five milligrams of the oxazolone ( V ) was dissolved in 15 mL of 33% CH3NH, in EtOH and the solution stirred for 2.5 h. The solvent was then removed under vacuum and the residue was dissolved in 20 mL CH2C1,; any insoluble material was then filtered off a t this stage. Th e CHzClz solution was then evaporated t o almost dryness and the free amine precipitated by adding hexanes/ether ( R 0.08, solvent C ) . Yield 500 mg (70% ) . Fmoc- (Phe-Aib-Aib)-NHMe (VII). Five hundred seventy milligrams of FMOC- ( Phe ) -C1 was placed in a separatory funnel and to it was added 255 mg of the free amine (VZ)dissolved in 20 mL CH2Clp; this was followed by the addition of 20 mL of CH2C12 and 40 mL 10% HCO; solution.* After shaking for 10 min, the organic phase was separated, dried over MgS04, and the trimer (VZI)isolated by flash chro-

* All other peptide couplings that were also performedvia the acid chloride route gave higher yields (70-85%).In this particular reaction, the acid chloride was not first prepared as a solution in CHzClzas was done with cases of higher yields.

1767

matography on a silica column (solvent A; R f0.27, solvent C ) . Yield 402 mg ( 55%) . Fmoc-(Aib-Aib-Phe-Aib-Aib)-NHMe (VIII). Two hundred fifty milligrams of the trimer (VZZ) was dissolved in a 10% solution'of HNEt, (in CHSCN). After 2.5 h, the solvent and HNEt2 were removed under vacuum and the residue dissolved in 5 mL dry CH3CN. T o this was added 175 mg of oxazolone ( V ) and the mixture refluxed for 8 h. The resulting solid was filtered off and most of it was redissolved in CH2Clzwith sonication and filtered again.+The filtrate was homogeneous on tlc ( R f0.3,solvent C ) a n d was identified as the desired pentamer by nmr ( D M S O - 4 ) . Yield 231 mg (71%) .

.

F M 0 C- ( N a p - A i b - P h e - A i b - A i b ) - N H M e (I X) Two hundred thirty milligrams of the pentamer (VZIZ) was dissolved in a 10% solution of HNEt, [in dimemethylformamide ( DMF ) 1. After 2 h, the solvent and HNEt, was removed under vacuum, and the residue dissolved in 10 mL of freshly distilled DMF. One hundred forty milligrams of Fmoc( Nap) -OH and 90 mg of IIDQ were added, and the reaction mixture was stirred a t room temperature for 24 h. Volatile components and solvent from the reaction mixture were then removed under vacuum, and the desired hexamer was isolated ( R 0.34, solvent C ) by flash chromatography on a silica column (solvent A ) . Yield 250 mg (84% ) .

Ac- (Aib-Aib-Nap-Aib-Aib- Phe-Aib-Aib) - N H M e (X). Two hundred fifty milligrams of the hexamer (ZX) was dissolved in a 10% solution of HNEt2 (in CH3CN). After 2.5 h, the solvent and HNEt, was removed under vacuum and the residue was washed with hexanes to remove dibenzofulvene (see daggered footnote on this page). The solid was then dissolved in 5 mL dry CH3CN with 65 mg of the oxazolone (ZI)and refluxed for 8 h. The desired octamer ( R f 0.38, solvent C ) was isolated by flash chromatography on a silica column (solvent A followed by solvent C ) . On a reversed-phase CIS column, the isolated octamer showed minor impurities (10% CHBOH in HzO, 220 nm, 280 n m) . The ocThe dibenzofulvene (DBF) generated during the deblocking of the Fmoc group may form an insoluble polymer during the course of the subsequent oxazolone coupling reaction, which may be removed by filtration. However, this strategy is not recommended instead the better and simple method of removing DBF is to wash the free amine with hexanes. When working with small quantities and with longer peptides, we found also that liquid/ liquid extraction of the free amine/DBF solution in CH,CN with hexanes is very efficient in removing the DBF.

1768

BASU, BAGCHI, AND KUKI

to the lack of a-hydrogens in Aib residues. Preliminary two-dimensional rotating frame nuclear Overhauser effect spectroscopy ( ROESY) 22 experiments performed on peptide 3,6-NF to pick up amide proton ( i + i 1) connectivities resulting from a helical backbone were complicated, possibly due to the inherent weak ROESY signals and other artifacts that accompany such experiment^.^^ However, enough cross peaks between the amide proton resonances were observed to support the helical backbone a ~ s i g n m e n t .The ~ ~ amide proton resonances for Phe or Nap residues were unambiguously assigned by decoupling studies. The observed amide resonances are represented as S, (singlets, Aib) ,D, (doublets), and Q , (quartet, which identifies the resonance for the terminal amide NHMe) where subscripts n refer to the order of appearance of all the resonances from the low field end of the spectrum in DMSO-&. For the present study with peptides containing two aromatic residues, D1 corresponds to the naphthyl amide resonance and D2corresponds to the phenyl (or bromophenyl) amide resonances, respectively.

tamer was confirmed by nmr in DMSO-& (all required amide resonances were present, and in particular the new amide resonance of the Nap residue J coupled to the Nap a-hydrogen). The sequence was readily confirmed by the fragmentation pattern in FAB/MS (Figure 2 ) . Yield 115 mg (46%).

+

'H-NMR Studies

All 'H-nmr conformational studies were performed on a Varian XL-400 spectrometer. The peptide concentrations were kept low to minimize aggregation; specifically, for peptides 3,6-NF, 3,5-NF, and 4,5N F it was 4 mM. Solvent titrations experiments were performed by adding measured aliquots of a peptide solution in DMSO-& to a peptide solution of identical concentration in CD3CN, thus maintaining a constant concentration of the peptides. These peptides were all purified by a final reversedphase HPLC run. Residual 'H resonances from the solvents were used as the internal reference.

RESULTS Assignment of Amide Protons

Temperature Perturbation

The assignment of all the peptide amide resonances could not be achieved by standard procedures21due

Temperature dependence of the amide proton chemical shifts in H-bonding solvents have often

Table I1 Chemical Shifts and Temperature Coefficients of Amide Resonances in DMSO of the Peptide Series" Chemical Shift, 6 (ppm) A6/AT (-ppb (deg)-') Pentamers Peptide A

S1

S2

S3

s 4

8.41

7.48

7.12

7.87

1.9

0.5

2.1

8.63

8.17 6.0 8.22

7.47

6.98

7.7

3.9

4.3

0.7

0.7

0.7

8.59

9.21

7.54

6.99

7.79

5.7

6.3

1.3

0.4

0.4

4.4

Peptide F Peptide N

Octamers Peptide 3,6-AA

s1

S2

s 3

D

s4

S5

Sli

D,

D2

Q 7.16

____

8.44

8.20

7.84

7.81

7.32

6.95

7.81

4.9

5.8

1.5

3.3

1.0

1.1

1.0

8.72

8.36

8.05

7.94

7.38

7.04

7.93

7.56 0.6 7.77

5.3

5.5

0.8

2.7

2.1

1.9

1.0

1.5

1.5

8.64

8.28

7.99

7.88

7.32

6.95

7.88

7.69

7.18

5.1

5.2

1.0

3.0

2.0

1.6

1.7

1.5

1.0

Peptide 3,5-NF

8.62

8.29

7.91

7.81

7.49

7.14

7.89

7.34

7.23

5.2

5.4

3.1

1.4

0.7

1.2

1.7

8.36

8.27

7.83

7.39

7.22

7.94

7.86

7.17

4.7

5.5

3.2 8.18 7.7

0.6

Peptide 4,5-NF

1.8

1.9

1.7

1.3

1.7

1.2

Peptide 3,6-NF Peptide 3,6-NF

a

The temperature dependence was studied over the range 20-50'C; all shifts are upfield with increasing temperature.

1.6

7.23

PREFERENCES OF OLIGOPEPTIDES RICH IN AIB. I

been used to delineate intra- from intermolecular H-bonded amide protons.25The results of the temperature dependence experiments performed on all the octamers and pentamers are summarized in Table 11. The temperature dependence of the amide resonances in the pentamer series in DMSO-& is presented in Figure 3. The number of intramolecular amide-carbonyl H bonds in the peptide depends on the type of backbone conformation and is illustrated in Figure 4. Two intramolecular H bonds (both singlets) characterize an a-helix while three (two singlets and one doublet) characterize a 3,0-helix. The intramolecular H bonds can be distinguished as their corresponding proton chemical shifts are less sensitive to changes in temperature than their counterparts, which are intermolecularly H bonded to the solvent. In the case of all three peptides (pentamers 3-N, 3-F, and 3-A),three amide protons (two singlets and one doublet) are in fact observed to be significantly less perturbed by temperature, supporting a 310H-bonding pattern. Note in particular the clear bimodal pattern of the temperature-sensitive resonances 2 4 ppb/deg, as distinct from the less temperature-sensitive resonances I2 ppb /deg.

8 Peptide 3-N Peptide 3-F PeDtide 3-A

6 h

0, Q,

?

0

0 Q. Y

4

Q,

Q.

-0

* 2

0

S

S,

D

S

amide protons Figure 3. Temperature dependence of the amide chemical shifts for the pentamer series in DMSO-4,.For all three peptides, two singlets, S , and Sz, are distinctly temperature sensitive, and the data is hence compatible with a 310H-bonding pattern.

1769

a) 3, , -helical hydrogen bonding

H

H

H

H

H

uck/

b) a -helical hydrogen bonding

W

uu H

H

H

H

Figure 4. The two possible helical backbone H-bonding pattern for the pentamer series: ( a ) 3,0-helical, i + i 3; ( b ) a-helical, i + i 4. For case ( a ) two amide hydrogens do not participate in the intrahelical H-bonding network and are available for H bonding to DMSO, while for case ( b ) three amide hydrogens are available. The nmr experiments (solvent titration and temperature-dependence series) enable one to distinguish the solvent-exposed vs solvent shielded amide protons, thus establishing the Hbonding pattern exhibited by the peptide in solution.

+

+

The same experiments performed on the octamers with conserved sites of guest residue substitution (sequence positions 3 and 6; peptides 3,6-NF, 3,6NF’, and 3,6-AA) yielded not only clearly temperature-sensitive and -insensitive amide protons, but also one amide proton resonance of intermediate nature ( S4,see Figure 5 ) . Two singlets were always sensitive to temperature, one singlet ( Sq) maintained intermediate sensitivity, while the rest ( two doublets, one quartet, and three singlets) were relatively temperature insensitive. An a-helical Hbonding pattern can be unambiguously ruled out since that requires a temperature-sensitive doublet (Nap amide resonance). A 310 H-bonding pattern can be justified only if one considers the intermediate amide signal to be actually insensitive to temperature, being below the conventional threshold value of 4 ppb /ppm. Although such justifications have

1770

BASU, BAGCHI, AND KUKI

peptide 3,6-NF peptide 3,6-NF'

0

S, S

3

S, S

peptide 3,6-AA

5

S , D, D 2 Q

amide protons Figure 5 . Temperature dependence of the amide chemical shifts for the 3,6 octamer series in DMSO-4. For all three octamers, two singlets, S , and S B ,are temperature sensitive, while one, S4, is moderately temperature sensitive. This pattern may be interpreted as a predominantly 3,,,-helical pattern with one weakened H bond (see text).

been used to delineate intra- from intermolecular H bonds in peptides,26we could not ignore this observation since this intermediate behavior was observed in all three peptides with guest residues as different as alanines replaced by a Phe and a Nap residue. Solvent Titration

Alternative perturbation experiments that provide a knowledge of the H-bonding pattern become more important when it is appreciated that they do not suffer from the inherent drawbacks of the temperature perturbation experiment, which is the assumption that the peptide's conformation remains unchanged throughout the temperature range, and further that the stability of all intramolecular H bonds against temperature increase are identi~al.'~ These alternative experiments are ( 1) line broadening induced by free radicals, ( 2 ) deuterium exchange rates of the amide protons, ar,d ( 3 ) monitoring the chemical shifts of each amide proton as a function of the solvent composition in a mixture of a hydrogen-bonding and a nonhydrogen-bonding

solvent. The first two experiments for the present set of peptides are complicated due to crowding of amide and aromatic proton resonances. The third experiment was performed for the octamer series 3,6-NF, 3,5-NF, and 4,5-NF. When the amide resonances of the peptide 3,6NF were monitored in CDBCNas a function of added DMSO-d,, two singlets showed dramatic perturbation when compared to the seven other amide resonances. The same observation was made for the peptide 3,5-NF. These results, summarized in Figure 6, do not exhibit the ambiguity of the temperaturedependence studies and definitively support a 310Hbonding pattern for the peptides 3,6-NF and 3,5-NF in CD3CN. Peptide 4,5-NF, on the other hand, exhibited three DMSO-d, sensitive singlets in the titration, and furthermore, maintained this distinctive pattern of three solvent-exposed singlets in the temperature-dependence study in pure DMSO-d, (Figure 7 ) . This clearly established that in CD&N peptide 4,5-NF adopts an a-helical H-bonding pattern, in stark contrast with the 310H-bonding pattern of peptides 3,6-NF and 3,5-NF. The latter differ from the former only in the sequence positions of the guest residues. In 100% DMSO-d,, we interpret the moderately temperature-sensitive singlet (along with two strongly sensitive singlets) observed in peptides 3,6NF, 3,6-NF ', and 3,6-AA to indicate an overall 310helical H-bonding scheme with the possible presence of one weakened intramolecular and partially solvent exposed H bond. By elimination, this weakened H bond must correspond to the amide of Aib no. 4, 5, 7, or 8 in the sequence (from the N-terminus). That the solvent titrations gave unambiguous results for all amide protons in the 3,6-octamer series suggests that this H bond (S,) may have been weakened in the transition between < 30% to 100% DMSOd,. Accordingly, we draw our primary conclusions from the solvent conditions of < 30% DMSO-4, and note that the conditions of heating to 50°C in pure DMSO in effect serves to probe quantitative questions about the relative strengths of intramolecular H bonds, and not just the qualitative pattern of intramolecular vs. solvent exposed.

DISCUSSION Current empirical that summarize the conformational preferences of Aib-rich peptides are based on a considerable number of measurements of peptide conformations in the solid state and in solution phase. A chain length of 8-10 residues is

PREFERENCES OF OLIGOPEPTIDES RICH IN AIB. I

1

Peptide 3,6-NF

/

sl

1771

Peptide 3,5-NF

8.0 h

5a

Y

v)

7.6

s2

v)

m .-V

5

r V

a

2

Em

7.2

s5

Q

‘6

6.8 10

30

20

6.8 I 0 10 20 30

Yo DMSO d6 in Acetonitrile-d3

Yo DMSO d6 in Acetonitrile-d3

Figure 6. Solvent titration curves in CD3CN for octamer 3,5-NF and octamer 3,6-NF. Both the peptides exhibit two DMSO-d, sensitive singlets and seven insensitive amide resonances, which is fully compatible with a 310H-bonding pattern.

Peptide 4,5-NF

Peptide 4 5 N F *

Dl

6-

6.8

“ S S S S S S D D Q 1

‘Yo DMSO-d6 in Acetonitrile-d3

2

3

4

5

6

1

2

amide protons

Figure 7. Solvent titration curves in CD3CN and the temperature dependence of the amide chemical shifts in DMSO-d, for the octamer 4,5-NF. Unlike all previous data, three amide singlets are sensitive to the perturbations in both the experiments ( S , , S2, S,) , clearly indicating three solvent-exposed amide protons characteristic of the a-helical Hbonding pattern.

1772

BASU, BAGCHI, AND KUKI

considered to be the 310/a transition length for peptides containing 50% Aib. Longer peptides favor the a-helix while shorter ones favor the 310-helix.Since the Aib residue is more stable in the 310-rather than the a-helical conformation, increasing the percent Aib content of the peptide further increases this transition length, while a decrease in the percent Aib content makes the 310/a transition possible a t shorter lengths. Percent Aib content and the length of the peptide are considered to be the major factors in predicting the final peptide conformation. By contrast, the nature of the monoalkylated residues present in the peptide or the exact sequence of the peptide is not considered within this rule in part due to the paucity of systematic experimental information in this regard. Studies that have sought to observe sequence effects on the backbone conformation of Aib-containing peptided include the characterization of repeated triads ( Ala-Leu-Aib), or tetrads (Val-Ala-Leu-Aib), containing 33 and 25% Aib, respectively. The total number of residues were 6 and 10 for the first and 7-16 for the second series. Selected Aib residues in these peptides were substituted, removed, or permuted to observe conformational changes, if any, induced by such a change. The x-ray diffraction structures did not in fact show any sequence-induced change in the result, which was an essentially ahelical backbone of all the peptides in the crystalline form. A failure to observe any conformational change in this study, however, does not necessarily imply an absence of sequence effects on conformation. These peptides with 2 5 3 3 % Aib content were typically much longer than the proposed critical length for the 310/atransition of peptides, and presumably succumbed to the a-helical conformation dictated by chain length and by Aib dilution. Under such circumstances, other effects, if present, become minor and unobservable. T o study only the effect of sequence changes on the 310/abackbone transition, not only should all other factors responsible for such a transition remain unchanged as the sequence is varied, but the chain length should also be in a critical range. The octamers we studied contain 75% Aib while the pentamers have 80% Aib. The empirical rule predicts the pentamers to be exclusively 310-helicalwhile the octamers fall in the neighborhood of the 310/atransition length. The three pentamers with different guest residues indeed possess a Slo-helicalbackbone, as evident from the temperature-dependence studies

*

This series of experiments were reported by Karle et al. and is best summarized in Ref. 2.

in DMSO-d,. T h e octamers, however, show marked variability among each other and in the two solvents studied, suggesting the presence of a fine balance between the competing factors responsible for the backbone conformational preference. In DMSO-& the experimental results for the octamer series with conserved sites of guest residue substitution-3,6NF, 3,6-AA, and 3,6-NF ’-were remarkably consistent, and suggest the presence of a predominant but not perfect 3,,-helical backbone. The consistency between the series reflects the presence of a definite backbone structure in which the intermediary value of the temperature coefficient for one amide proton, S4,may reflect the closeness of the peptides to the transition point between the two helical forms (“ambihelicity”). This ambihelicity of 3,6-NF (in DMSO-&) disappeared in CD3CN where a pure 310helical structure was observed. Solvent-induced conformational changes have been observed before and for the present case this appears to reflect the critical chain-length range of the octamers. T h e solvent titration nmr data for the sequence permutation series peptides 3,5-NF, 3,6-NF, and 4,5NF, in CD3CN showed no ambi helicity, but rather definite a- and 310-helicesfor the different sequence isomers. Peptides 3,5-NF and 3,6-NF exhibited a distinctive 310H-bonding pattern. When the guest residues occurred consecutively in the peptide 4,5NF, on the other hand, a dramatic change was observed, and the backbone was found to be cleanly a-helical in both CD3CN and in DMSO-&. By design, the guest residues within this sequence isomer series were always incorporated within the central four positions in the octamers to rule out any end chain effects, making this 310/a transition exclusively a sequence permutation induced transition. In terms of the linear sequence, a contiguity in the incorporation of non-Aib residues seems to play a n important role in the conformational transition. This hypothesis is supported by the conformational preferences of the decapeptides Boc- ( Aib-L-Val ),OMe and its sequence isomer Boc-Aib-L-Val- ( Aib)z(L-Val),-Aib-L-Val-Aib-OMe, reported in two separate studies by Balaram and co-workers. The former adopts a 310-helical conformation in both CDC13and DMSO-&, 25 whereas the latter maintains the 310conformation in CDC13 only26;in DMSO-d, it exhibits a n a-helical conformation or a partially unfolded 310-heli~.26 Further evidence supporting this ambi-helicity of the second peptide comes from the corresponding x-ray crystal structure data identifying both a n a-helix and mixed 310/a-helixas cocrystallized conformers.28The data clearly indicates a disruption of the 310-helix when the evenly dis-

PREFERENCES OF OLIGOPEPTIDES RICH I N AIB. I

tributed Aib and L-Val residues are permuted to produce a central L-Val triplet in the decapeptide. This contiguity effect, when observed in the peptide sequence isomer set we studied, becomes more dramatic in that a 310-helicalpattern changes completely to an a-helical pattern as we go from 3,6-NF and 3,5-NF to the 4,5-NF octamer. The contiguity effect proposed above cannot be explained by a noninteracting Zimm-Bragg modelz9 that is widely used to describe conformational transitions in biopolymers, as this model is insensitive to sequence positions. A nearest neighbor Ising model would clearly be capable of formally encapsulating the observed behavior, but here we wish to emphasize instead that two very distinct physical mechanisms can be proposed to give rise to the observed behavior. While both require interactions between residues, the first describes a nearest neighbor interaction between the guest residue side chains only, whereas in the second model the interaction is distributed over the peptide. In the first model, a strong local phenyl-naphthyl nonbonded side-chain interaction might be proposed to favor an a-helix over a 310-helix for the peptide 4,5-NF. But models readily reveal that the aromatic residues are also at comparable interacting distances in 3,6NFs where the equilibrium is shifted in the opposite direction. In fact, the aromatic y carbons are further away from each other in either the 310- or the ahelical forms of the peptide 4,5-NF than in either helical form of the peptide 3,6-NF. Information on the relative proximity of the aromatic groups in 3,6NF and 4,5-NF may also be gleaned from the observation that naphthalene fluorescence in both these octamers are quenched by similar orders of magnitude when the phenyl residue is replaced by the p-bromophenyl group.# The van der Waals interactions between the side-chain aromatic rings in the peptide 4,5-NF were also examined, but this weak attractive energy slightly favored the 310 form over the a-helical form.* * Thus direct intersideThe following data represents average interatomic distances (in Angstrom units) between aromatic side groups that were measured from several low-energy conformations generated with the Insight11 Molecular Modeling package. Interatomic distances are ( a ) 5.55 (Cs-Cs) and 7.90 (Ct-C7), and 5.98 (Ca-Cs) and 7.53 (C7-C'), for the peptide 4,5-NF in the a form and the 310 form, respectively; and ( b ) 5.79 ( C'-C8) and 6.43 ( C7-Ct), and 5.92 (Cs-Ca) and 6.27 ( C'-C7), for the peptide 3,6-NF in the a form and the 310 form, respectively. It is evident that the orientation of the side chains is such that they are basically pointing away from each other, out from the helical axis, in peptide 4,5-NF. # This fluorescence quenching, which in fact is greater for 3,6NF than 4,5-NF, is attributed to a remote heavy atom effe~t.'~,~' * * The 310and a-helical forms of the 4,5-NF and 3,6-NF oc-

1773

chain interactions in 4,5-NF are weak, do not bias the conformation to the a-helical, and cannot be the major factor in shifting the 310/a equilibrium for 4,5-NF. The equilibrium conformations of 3,6-NF and 4,5NF are apparently different due to the impact of the C,-monoalkylated guest residues upon the main chain. An alternative statistical model for the ( ~ / 3 ~ 0 transition that describes a mechanism for sequence dependence arising from such side-chain / mainchain interactions is considered in the following paper.3z The basic framework of the model is a Hbonding loop analysis, and the resulting proposal is that the observed contiguity effect arises from contiguous blocks of main-chain torsional angle preferences in the sequence. The nature of the side-chain controls the conformation energy functions for the main-chain torsional angles ($,$), and the cumulative effect of a contiguous block (two for the 310 form and three for the a-form) of such (4,$) energy functions in turn controls the H-bonding probabilities and strengths of the 310-helicalstructure relative to the alternative a-helical structure.

CONCLUSION In conclusion, we have developed a series of peptides where sequence permutation alone induces a backbone helical transition ( a/3l0) in Aib-rich peptide octamers. Energetically allowed torsional angles (4,+) for the Aib residue strongly favor the 310-helical conformation over the a-helical conformation. Yet the peptide 4,5-NF with an extremely high Aib content ( 75% ) nevertheless exhibit the a-helical conformation, which we refer to as a contiguity effect. It is therefore important in the prediction of the preferred conformation of Aib-containing short peptides to consider not only the percent Aib content and the peptide length, but also the sequence, and in particular whether the monoalkylated amino acids occur in contiguous sequence positions. From the tamers were generated systematically using the Insight11 Molecular Modeling package. Nine conformations of each helical type were created by rotating the XI angles of the aromatic side groups by 120' increments, followed by careful local minimization to relax bond lengths, angles, and steric contacts. Each rotamer conformation of the 4,5-NF octamer was then stripped of all atoms except for the two aromatic side groups. The p carbons were replaced with hydrogens a t a corrected bond length. Average van der Waals interactions between the aromatic rings in the nine rotamers were computed to be a type = -0.14 k 0.14 kcal/mole; 310type = -0.93 ? 1.34 kcal/mole. In a similar manner, the electrostatic interactions between the rings were computed using potentials from Ref. 31. Again, the interactions, on average, favored the 310 forms slightly.

1774

BASU, BAGCHI, AND KUKI

viewpoint of conformational energetics, while the sequence permutation did in fact switch the sign of the free energy difference between the t w o helical forms, it seems likely that the absolute magnitudes of such free energy changes are not much larger than three or four kbT. Along with experiments on specific peptides i n which solvent i n d ~ c e d ' ~o r, ~a~chain length induced34transitions have been observed, t h e sequence permutation induced transition that we report here provides revealing evidence to further our understanding of t h e subtle factors in the helical preferences of Aib-containing polypeptides. The support of this work by the NIHGMS First Program (R29-GM39576) is most gratefully acknowledged. Beth Secor synthesized Aib blocks via the azirine route. GB acknowledges Dave Fuller for assistance with the nmr spectrometers and AK wishes to thank the NSF Presidential Young Investigator program (CHEM-8958514) for valued support.

REFERENCES 1. Prasad, B. V. V. & Balaram, P. ( 1984) CRC Crit. Rev. Biochem. 16,307-348. 2. Karle, I. L. & Balaram, P. (1990) Biochemistry 2 9 , 6747-6756. 3. Marshall, G. R., Hodgkin, E. E., Langs, D. A., Smith, G. D., Zabrocki, J. & Leplawy, G. D. (1990) Proc. Natl. Acad. Sci. U S A 87, 487-491. 4. Toniolo, C. & Benedetti, E. (1988) IS1 Atlas Sci. Biochem. 1,225-230. 5. Jung, G., Bosch, R., Katz, E., Schmitt, H., Voges, K.-P. &Winter, W. (1983) Biopolyrners 22,241-246. 6. Paterson, Y., Rumsey, S. M., Benedetti, E., Nemethy, G. & Scheraga, H. A. (1981) J. Am. Chem. SOC.1 0 3 , 2947-2955. 7. Venkataram Prasad, B. V. & Sasisekharan, V. ( 1979) Macromolecules 1 2 , 1107-1110. 8. Barone, V., Lelj, F., Bavoso, A., Di Blasio, B., Grimaldi, P., Pavone, V. & Pedone, C. (1985) Biopolymers 24, 1759-1767. 9. Hogdkin, E. E., Clark, J . D., Miller, K. R. & Marshall, G. R. ( 1990) Biopolymers 30, 533-546. 10. Malcolm, B. R. (1983) Biopolymers 2 2 , 319-322. 11. Toniolo, C., Bonora, G. M., Bavoso, A., Benedetti, E., Di Blassio, B., Pavone, V. & Pedone, C. ( 1986) Mucromolecules 19,472-479. 12. Pavone, V., Di Blassio, B., Santini, A., Benedetti, E., Pedone, C., Toniolo, C., & Crisma, M. ( 1990) J. Mol. Biol. 214, 633-635.

13. Bavoso, A., Benedetti, E., Di Blasio, B., Pavone, V., Pedone, C., Toniolo, C., Bonora, G. M., Formaggio, F. & Crisma, M. (1988) J. Biomol. Struc. Dynam. 5 , 803-817. 14. Basu, G., Kubasik, M., Anglos, D., Secor, B. & Kuki, A. (1990) J. Am. Chem. SOC.1 1 2 , 9410-9411. 15. Levene, P. A. & Steiger, R. E. (1931) J . Biol. Chem. 9 3 , 581-604. 16. Vittorelli, P., Heimgartner, H., Schmid, H., Hoet, P., & Ghosez, L. (1974) Tetrahedron 30,3737-3740. 17. Obrecht, D. & Heimgartner, H. (1987) Helv. Chim. Acta. 7 0 , 102-115. 18. Haveaux, B., Dekoker, A., Rens, M., Sidani, A. R., Toye, J. & Ghosez, L. (1988) Organic Syntheses Collective Vol. VI, Noland, W. E., Ed., John Wiley & Sons, New York, pp. 282-289. 19. Carpino, L. A., Cohen, B. J., Stephens, K. E., Jr., Tien, S. Y. S. & Langridge, D. C. (1986) J . Org. Chem. 51, 3732-3734. 20. Bodanzky, M. ( 1984) Principles of Peptide Synthesis,

Springer-Verlag, New York. 21. Wuthrich, K. (1983) Biopolymers 2 2 , 131-138. 22. Bothner-By, A. A., Stephens, R. L., Lee, J.-M., Warren, C. D. & Jeanloz, R. W. (1984) J . Am. Chem. SOC. 106,811-813. 23. Marion, D. (1985) FEBS Lett. 192,99-103. 24. Wuthrich, K. ( 1986) N M R of Proteins and Nucleic Acids, John Wiley & Sons, New York. 25. Vijayakumar, E. K. S. & Balaram, P. (1983) Biopolymers 22, 2133-2140. 26. Balaram, H., Sukumar, M. & Balaram, P. (1986) Biopolymers 2 5 , 2209-2223. 27. Kopple, K. D. & Schamper, T. J. (1972) Chemistry and Biology of Peptides, Meinhoffer, J., Ed., Ann Arbor Science Publishers, Ann Arbor, MI, pp. 75-80. 28. Karle, I. L., Flippen-Anderson, J. L., Uma, K., Balaram, H. & Balaram, P. (1990) Proc. Natl. Acad. Sci. U S A 8 6 , 765-769. 29. Zimm, B. H. & Bragg, J. K. (1959) J. Chem. Phys. 3 1 , 526-525. 30. Basu, G., Kubasik, M. & Kuki, A., to be published. 31. Gruschus J. & Kuki, A. (1990) J. Comp. Chem. 1 1 , 982. 32. Basu, G. & Kuki, A. (1991) Biopolymers, following

paper. 33. Vijayakumar, E. K. S. & Balaram, P. (1983) Tetrahedron 39,2725-2731. 34. Pavone, V., Benedetti, E., DiBlasio, B., Pedone, C., Santini, A., Bavoso, A., Toniolo, C., Crisma, M. & Sartore, L. (1990) J. Biomol. Struct. Dynum. 7,13211331.

Received May 3, 1991 Accepted August 7, 1991