A noncovalent approach to antiparallel β-sheet formation

Article abstract of DOI:10.1021/ja010701b

Four tripeptide chains, when attached to the same end of a hydrogen-bonded duplex (1.2) with the unsymmetrical, complementary sequences of ADAA/DADD, have been brought into proximity, leading to the formation of four hybrid duplexes, 1a·2a, 1a·2b, 1b·2a,

Full text of DOI:10.1021/ja010701b

Published on Web 02/27/2002
A Noncovalent Approach to Antiparallel â-Sheet Formation
Huaqiang Zeng,† Xiaowu Yang,† Robert A. Flowers, II,§ and Bing Gong*,†
Contribution from the Department of Chemistry, Natural Sciences Complex, State UniVersity of
New York, Buffalo, New York 14260, and Department of Chemistry and Biochemistry,
Texas Tech UniVersity, Box 41061, Lubbock, Texas 79409-1061.
Received March 16, 2001
Abstract: Four tripeptide chains, when attached to the same end of a hydrogen-bonded duplex (1‚2) with
the unsymmetrical, complementary sequences of ADAA/DADD, have been brought into proximity, leading
to the formation of four hybrid duplexes, 1a‚2a, 1a‚2b, 1b‚2a, and 1b‚2b, each of which contains a two-
stranded â-sheet segment. The extended conformations of the peptide chains were confirmed by 1D and
2D NMR. The peptide strands stay registered through hydrogen bonding and the â-sheets are stabilized
by side chain interactions. Two-dimensional NMR data also indicate that the duplex template prevents
further aggregation in the peptide segment. When the peptide chains are attached to the two different
termini of the duplex template, NMR studies show the presence of a mixture with no clearly defined
conformations. In the absence of the duplex template, the tripeptides are found to associate randomly.
Finally, isothermal titration calorimetry studies revealed that the hybrid duplex 1a‚2a was more stable than
either the duplex template or the peptides alone.
Introduction
stability of the resulting â-hairpins and related model structures
is found to depend on hydrogen bonding, side chain interactions,
Understanding â-sheet formation is critical to many problems
and applications involving protein folding and design. Discern-
ing the factors affecting â-sheet structure and stability may
eventually lead to novel peptide antibiotics,1-3 and to treatments
for diseases in which â-sheet formation plays a key role.4,5
Although as common as the R-helical structure in proteins, the
â-sheet secondary structure is not well understood due to the
lack of well-defined â-sheets amenable to detailed biophysical
evaluation. Current efforts in developing model systems of
â-sheets are focused on the design of short peptides that fold
in solutions.6-17 Artificial systems consisting of â-strands linked
by unnatural templates have also been described.18-20 The
and the types of â-turns and turn mimetics. We report here the
nucleation of antiparallel â-sheet-like structures based on a
noncovalent, self-assembling approach.
We recently described hydrogen-bonded duplexes based on
linear oligoamide strands bearing arrays of hydrogen-bonded
donors (D) and acceptors (A).21-24 These molecular duplexes
are formed by pairing two single strands of complementary
hydrogen-bonding sequences. The formation of these duplexes
is highly sequence-specific: a single strand only pairs with
another strand of its complementary sequence. Due to the
absence of secondary electrostatic interactions25-27 in this
system, the stability of a duplex is sequence-independent and
is proportional to the number of H-bonds found in that duplex.
Both the H-bonding sequences and the number of H-bonding
sites in a duplex are easily adjustable. Since the single strands
of the H-bonded duplexes adopt an extended conformation
similar to that of â-strands, these single strands can be regarded
as â-strand mimics and the corresponding duplexes as two-
stranded â-sheet mimics. Indeed, the interstrand distance of a
* Address correspondence to this author. E-mail: bgong@chem.buffalo.edu.
† State University of New York.
§ Texas Tech University.
(1) Hancock, R. E. W.; Chapple, D. S. Antimicrob. Agents. Chemother. 1999,
43, 1317.
(2) Lehrer, R. I.; Ganz, T. Curr. Opin. Immunol. 1999, 11, 23.
(3) Lansbury, P. T., Jr. Acc. Chem. Res. 1996, 29, 317.
(4) Kelly, J. W. Curr. Opin. Struct. Biol. 1998, 8, 101.
(5) Soto, C. J. Mol. Med.-JMM 1999, 77, 412.
(6) Smith, C. K.; Regan, L. Acc. Chem. Res. 1997, 30, 153.
(7) Gellman, S. H. Curr. Opin. Chem. Biol. 1998, 2, 717.
(8) Lacroix, E.; Kortemme, T.; de la Paz, M. L.; Serrano, L. Curr. Opin. Struct.
Biol. 1999, 9, 487.
(9) Espinosa, J. F.; Gellman, S. H. Angew. Chem., Int. Ed. 2000, 39, 2330.
(10) Griffiths-Jones, S. R.; Searle, M. S. J. Am. Chem. Soc. 2000, 122, 8350.
(11) Russell, S. J.; Cochran, A. G J. Am. Chem. Soc. 2000, 122, 12600.
(12) Pozdnyakova, I.; Guidry, J.; Wittung-Stafshede, P. J. Am. Chem. Soc. 2000,
122, 6337.
(13) Koide, S.; Huang, X. L.; Link, K.; Koide, A.; Bu, Z. M.; Engelman, D. M.
Nature 2000, 403, 456.
(14) De Alba, E.; Santoro, J.; Rico, M.; Jimenez, M. A. Protein Sci. 1999, 8,
854.
(15) Schenck, H. L.; Gellman, S. H. J. Am. Chem. Soc. 1998, 120, 4869.
(16) Kortemme, T.; Ramirez-Alvarado, M.; Serrano, L. Science 1998, 281, 253.
(17) Das, C.; Raghothama, S.; Balaram, P. J. Am. Chem. Soc. 1998, 120, 5812.
(18) Nowick, J. S. Acc. Chem. Res. 1999, 32, 287.
(19) Chitnumsub, P.; Fiori, W. R.; Lashuel, H. A.; Diaz, H.; Kelly, J. W. Bioorg.
Med. Chem. 1999, 7, 39.
(20) (a) Nowick, J. S.; Abdi, M.; Bellamo, K. A.; Love, J. A.; Martinez, E. J.;
Noronha, G.; Smith, E. M.; Ziller, J. W. J. Am. Chem. Soc. 1995, 117, 89.
(b) Holmes, D. L.; Smith, E. M.; Nowick, J. S. J. Am. Chem. Soc. 1997,
119, 9, 7665-7669.
(21) Gong, B. Synlett 2001, 5, 582.
(22) Zeng, H.; Ickes, H.; Flowers, R. A., II; Gong, B. J. Org. Chem. 2001, 66,
3574.
(23) Zeng, H.; Miller, R.; Flowers, B., II; Gong, B. J. Am. Chem. Soc. 2000,
122, 2635.
(24) Gong, B.; Yan, Y.; Zeng, H.; Skrzypczak-Jankunn, E.; Kim, Y. W.; Zhu,
J.; Ickes, H. J. Am. Chem. Soc. 1999, 121, 5607.
(25) Pranata, J.; Wierschke, S. G.; Jorgensen, W. L. J. Am. Chem. Soc. 1991,
113, 2810-2819.
(26) Jorgensen, W. L.; Pranata, J. J. Am. Chem. Soc. 1990, 112, 2008-2010.
(27) Murray, T. J.; Zimmerman, S. C. J. Am. Chem. Soc. 1992, 114, 4010-
4011.
9
10.1021/ja010701b CCC: $22.00 © 2002 American Chemical Society
J. AM. CHEM. SOC. VOL. 124, NO. 12, 2002 2903

A R T I C L E S
Zeng et al.
Figure 1. When attached to a duplex template, two otherwise flexible peptide chains may be directed to form a stably folded â-sheet.
duplex is nearly the same as that between two natural â-strands.24
Therefore, these hydrogen-bonded duplexes may serve as a set
of specific, noncovalent templates for the nucleation of â-sheet
structures when attached to natural oligopeptides (Figure 1).
The stability of the resultant hybrid structure should be enhanced
if there is cooperativity along the strand direction.14
To evaluate this strategy, we designed four hybrid single
strands 1a, 1b, 2a, and 2b, each consisting of a template segment
and a natural tripeptide chain. The template is based on the
quadruply hydrogen-bonded duplex 1‚2, which contains the
unsymmetrical, complementary H-bonding sequences ADAA/
DADD. The template-directed pairing of 1a with 2a and 2b,
and 1b with 2a and 2b, may lead to four different hybrid
duplexes 1a‚2a, 1a‚2b, 1b‚2a, and 1b‚2b. Three closely related
questions need to be addressed: First, will the unsymmetrical
duplex template regiospecifically bring the peptide chains into
proximity? Second, will the templated peptide strands pair with
each other? Third, can the paired peptide strands adopt a double-
stranded, antiparallel â-sheet conformation? The amino acid
residues of the tripeptides are chosen based on (1) facilitating
assignment of NMR spectra, (2) propensity for â-sheet forma-
tion, and (3) our finding that the amino acid residues directly
attached to the duplex templates need to be glycine.28
To test whether the template can indeed direct the assembly
of the peptide chains in a regiospecific way, hybrid single strand
1c was designed by attaching a tripeptide chain to the “wrong”
end of 1. As controls, the hybrid duplexes 1c‚2a and 1c‚2b,
formed by pairing single strand 1c with 2a and 2b, should not
be able to have a â-sheet segment because the peptide chains
are attached to the two different termini of the duplex template.
Tripeptides 3, 4a, and 4b are also included for comparison with
the above hybrid strands and duplexes.
Results and Discussion
Synthesis. The synthesis is based on standard amide/peptide
coupling chemistry.29
Figure 2. The 1H NMR spectra of the 1:1 mixtures of (a) 1c + 2a, (b)
1c + 2b, (c) 1a + 2b, and (d) 1a + 2a.
Characterization of the Hybrid Duplexes Containing
Tripeptide Segments. The solubility of the 1:1 mixtures of
single strands 1a and 2a, 1a and 2b, 1b and 2a, and 1b and 2b
in CHCl3 provided initial evidence for the formation of the
corresponding duplexes. These 1:1 mixtures of single strands
completely dissolved into CHCl3 and led to stable solutions (>6
mM at room temperature). On the other hand, single strands
1a, 1b, and 1c were less soluble in CHCl3 (<4 mM at room
temperature), and single strands 2a and 2b (<0.5 mM at room
temperature) precipitated out upon cooling from 60 °C to room
temperature.
(1) One-Dimensional (1D) 1H NMR Spectroscopy. As
shown in Figure 2, each of the 1D 1H NMR spectra of duplexes
1a‚2a and 1a‚2b contains a set of sharp resonances that are
indicative of well-defined conformational species in solution.
The presence of sharp amide NH signals between 9.50 and 10.50
ppm is consistent with the existence of strong intermolecular
H-bonds. Similarly, duplexes 1b‚2a and 1b‚2b also give spectra
with sharp, well-defined peaks that are similar to those of 1a‚
1
2a and 1a‚2b.29 In contrast, the 1D H NMR spectra of 1c‚2a
and 1c‚2b contain broad, poorly defined resonances, indicating
slowly equilibrating mixtures with multiple associating modes.
The ΗR chemical shifts of duplexes 1a‚2a and 1a‚2b are
compared with those of duplexes 1c‚2a and 1c‚2b, single strands
(28) Unpublished data.
(29) For NMR spectra, synthetic procedures, other details, and analysis, see the
Supporting Information.
9
2904 J. AM. CHEM. SOC. VOL. 124, NO. 12, 2002

A Noncovalent Approach to Antiparallel â-Sheet Formation
A R T I C L E S
Chart 1
1a and 1c, tripeptides 4a and 4b, and 1:1 mixtures of tripeptides
3 and 4a, and 3 and 4b. The results are listed in Table 1.
Significant differences exist between the ΗR chemical shifts of
1a‚2a and 1a‚2b and those of the others (Table 1). Most of the
HR signals in 1a‚2a and 1a‚2b are shifted downfield as
compared to those of the single hybrid strands, the tripeptides,
and 1c‚2a and 1c‚2b whose peptide segments are not likely to
be paired. These data suggest that the R-protons in duplexes
1a‚2a and 1a‚2b are placed in a unique structural environment.
The observed changes of the HR chemical shifts of 1a‚2a and
1a‚2b are most likely the result of interstrand interactions in
â-sheet-like structures.30
(2) Two-Dimensional 1H NMR Spectroscopy.29 Two-
dimensional (2D) NMR (NOESY, 500 and 800 MHz, 4 mM in
CDCl3) studies were carried out on duplexes 1a‚2a, 1a‚2b, 1b‚
2a, and 1b‚2b. Numerous interstrand NOEs corresponding to
the template regions of all four duplexes were observed. These
contacts include those between protons 1 and 42, 1 and 41, 4
and 41, 6 and 40, 10 and 36, 10 and 35, and 10 and 33 (Figure
3). These NOEs indicate that the template strands are sequence-
specifically registered in the expected unsymmetrical fashion,
a prerequisite for the correct pairing of the two tripeptides
(30) Maynard, A. J.; Sharman, G. J.; Searle, M. S. J. Am. Chem. Soc. 1998,
120, 1996.
9
J. AM. CHEM. SOC. VOL. 124, NO. 12, 2002 2905

A R T I C L E S
Zeng et al.
Figure 3. Summary of cross-strand NOEs (NOESY, 500 MHz, mixing time 0.5 s, 4 mM in CDCl3) observed for the four hybrid duplexes 1a‚2a, 1a‚2b,
1b‚2a, and 1b‚2b. NOEs shared by all four pairs are indicated by solid arrows. NOEs identified in some of the pairs are indicated by dashed arrows.
Table 1. Chemical Shift Data for R-Protonsa
registered as expected (Figure 4b). In the same peptide chain,
the absence of any adjacent, interresidue NiH‚‚‚Ni+1H NOEs
Ala1
Ala2
Leu
Phe
Val
Gly1
Gly2
and the observation that strong NOEs exist for CiRH‚‚‚Ni+1
H
1a‚2a
1a‚2b
1c‚2a
1c‚2b
1a
4.74
4.75
<4.70
4.99
5.34
5.26
4.76
4.71
4.44
4.45
4.33
4.41
5.13
(Figure 4c) provide additional support for an extended confor-
mation for the peptide chains. These data are fully consistent
with a structure consisting of extended, antiparallel peptide
chains that are registered through hydrogen bonds.
<4.70 <4.70 <4.70 <4.70 <4.70
<4.80 <4.80
<4.80 <4.80 <4.80 <4.80
4.60
4.64
5.04
4.57
4.48
4.66
4.65
4.30
4.54
4.25
1c
3 + 4a
3 + 4b
4a
4.61
4.31
4.82
4.94
4.54
4.67
3.99
To confirm that the duplex template is indeed responsible
for the observed pairing of the tripeptide strands, a 1:1 mixture
of hybrid strand 1a and tripeptide 4a, which is equivalent to
removing one of the templates of duplex 1a‚2a, was examined
by 2D NMR. The major NOEs are shown in Figure 5a. In
contrast to 1a‚2a whose NOESY spectrum is consistent with a
duplex with the expected registration of residues through
H-bonding, the NOESY spectrum of mixture 1a + 4a reveals
strong cross-strand NOEs that do not exist in the spectrum of
1a‚2a. The NOEs include those between protons 32 and 4, 32
and 10, 32 and 12, and 29 and 7. These cross-strand contacts
indicate that in the absence of one of the template strands, the
tripeptides cannot specifically associate with each other. Other
noteworthy contacts are those between protons 7 and 21, and 7
and 20, indicating that some molecules of 1a either adopt a
folded conformation or the molecules self-associate in their
tripeptide segments. NOESY study on a 1:1 mixture of 1a and
3 revealed similar NOEs that indicate nonspecific cross-strand
contacts (Figure 5b). Contacts between protons of the peptide
terminus and the template in 1a are also strong. Furthermore,
NOEs corresponding to the self-association of 3 are clearly
identified.
5.05
4.82
4.53
4.10
3.93
4b
a NMR (500 MHz) measurements were carried out at 4 mM of sample
concentration in CDCl3.
chains. The fact that no NOEs are observed between the
hydrogens of the template segments and those of the tripeptide
segments supports the expected, regiospecific registration of the
single hybrid strands. The NOEs between protons 1 and 42,
and 21 and 22, further suggest that there is no fraying at even
the two termini of these hybrid duplexes. Strong cross-strand
NOEs in the peptide segments of each of the four duplexes are
identified, which are consistent with the existence of â-sheet
structures.
Results from NOESY studies on all four hybrid duplexes are
consistent with the regiospecific registration of the single strands
and the existence of antiparallel â-sheet segments.29 The
NOESY data (800 MHz, CDCl3) based on duplex 1a‚2a, whose
1D spectrum shows the best resolved signals, are discussed here
in detail. The most conclusive evidence for the existence of a
â-sheet-like structure in the tripeptide segment of 1a‚2a involves
the observation of numerous long-range, strong NOEs between
residues of the two tripeptide chains. These NOE contacts
include those between the backbone protons such as the strong
cross-strand NH‚‚‚NH interaction of the H-bonded amide pair
(Ala‚‚‚Phe). A strong interstrand HR‚‚‚HR interaction is detected
between the non-H-bonded pair (Val‚‚‚Leu). Other strong cross-
strand backbone interactions include those between protons 12
and 32, 13 and 32, 20 and 24, and 21 and 22 (Figure 4a). Strong
NOEs are also observed between the amino acid side chains
Ala‚‚‚Phe and Val‚‚‚Leu, suggesting that the peptide chains are
A 1:1 mixture of tripeptides 3 and 4a, corresponding to
removing both template strands in 1a‚2a, was examined by 2D
NMR (NOESY, 500 MHz, 4 mM in CDCl3). Cross-strand NOEs
corresponding to all possible combinations of the tripeptides,
i.e., self-association of 3 or 4a, and association of 3 and 4a,
were observed (Figure 6). Qualitatively, NOEs corresponding
to the self-association of 3 are the strongest; NOEs resulting
from the contacts of 3 and 4a are weaker, and the self-
association of 4a gives the weakest NOEs. Therefore the
observed NOEs are not consistent with an alignment shown by
9
2906 J. AM. CHEM. SOC. VOL. 124, NO. 12, 2002

A Noncovalent Approach to Antiparallel â-Sheet Formation
A R T I C L E S
spectra of the hybrid duplexes show NOEs that only correspond
to the specific pairing of the two different hybrid strands. NOEs
corresponding to further association of the hybrid duplexes are
not found in the NOESY spectra. Therefore the duplex template
not only directs the specific pairing of the tripeptide chains,
but also prevents the nonspecific association of peptide strands,
which carry extra hydrogen bond donors and acceptors. The
templates may have impeded the random association of the
peptide segments because of (1) parallel alignment of duplexes
through H-bonding interactions in the tripeptide region is
prevented by the steric hindrance from the duplex template, (2)
antiparallel alignment of the tripeptide region is interrupted due
to the incorrect registration of H-bond donors and acceptors,
which is cause by the presence of intramolecular H-bond
between the NH group of one of the two Gly residues with the
aryl ether O atom of the template (the S(6) type intramolecular
H-bonds act to pre-organize20,24 the template strands and, more
importantly, to establish the hydrogen bonding pattern between
the tripeptides), and (3) the significantly enhanced stability (see
below) of the hybrid duplexes as compared to the presumed
tripeptide dimers (or higher aggregates).
(3) Vapor Pressure Osmometry (VPO). The aggregation
of the templates, tripeptides, and the hybrid strands was
investigated by VPO measurements. The results are listed in
Table 2. Template strand 1 (2 had limited solubility) was
monomeric and showed no self-aggregation. The template
duplex 1‚2 existed in solution as a heterodimer, as is evidenced
by its aggregation number of ∼1. The hybrid single strand 1a
(2a had limited solubility) did not show any self-association.
As expected, the hybrid duplex 1a‚2a had an aggregation
number of 1 and thus should be a heterodimer. At 5 mM,
tripeptide 4a (3 had limited solubility) showed almost no
aggregation, which may be due to its weak self-association. An
aggregation number of 1.4 was obtained for 4a at a higher
concentration of 10 mM. Surprisingly, a 1:1 mixture of 3 and
4a gave an aggregation number of ∼1 at 4 mM of each
tripeptide, indicating either the existence of a heterodimer or
that there is an equilibrium between monomeric tripeptides and
higher aggregates.
(4) 1H NMR Binding Studies. The binding strength of
1‚2, the self-association of 4a, and the association of 3 and 4a
were investigated by diluting the corresponding solutions and
following the concentration-dependent changes of the chemical
shifts of the amide protons that were involved in intermolecular
H-bonding. For 1‚2, all four aniline NH signals showed similar
concentration-dependent changes in their chemical shifts in
CDCl3 containing 5% DMSO-d6, supporting their role in
intermolecular H-bonding. Depending on the specific NH signal
followed, nonlinear regression analysis31 of the NMR data led
to slightly different values of an association constant that was
at the lower end of 103 M-1 for 1‚2 (Table 3). Assuming a
simple dimerization mode, similar analysis found that the self-
association of tripeptide 4a was very weak even in pure CDCl3
(Table 3).
Figure 4. Long-range strong NOEs revealed by NOESY studies of 1a‚2a
(800 MHz, mixing time 0.5 s, 4 mM in CDCl3). (a) Backbone-backbone
interactions. (b) Side chain-side chain interactions. (c) Interresidue CiRH‚‚‚
Ni+1H interactions.
the tripeptide chains in 1a‚2a. Instead, these results are in
agreement with a scenario involving random association of
tripeptide chains through H-bonding and/or side-chain inter-
actions. NOESY study on another 1:1 mixture of tripeptides 3
and 4b revealed NOEs similar to those of 3 and 4a.29 Strong
NOEs consistent with the self-association of tripeptide 3 were
once again identified.
At the same concentration (4 mM), the behavior of duplexes
1a‚2a, 1a‚2b, 1b‚2a, and 1b‚2b is in contrast to that of the 1:1
mixtures of tripeptides 3 and 4a or 3 and 4b. The NOESY
Although the limited solubility of tripeptide 3 did not allow
the direct assessment of its self-association, a 1:1 mixture of 3
(31) Wilcox, C. S. In Frontiers in Supramolecular Organic Chemistry and
Photochemistry; Schneider, H.-J., Durr, H., Eds.; VCH: New York, 1991.
The dimerization constant was obtained by fitting the NMR data into a
modified dimerization equation with the program Kaleidagraph on a
Macintosh computer.
9
J. AM. CHEM. SOC. VOL. 124, NO. 12, 2002 2907

A R T I C L E S
Zeng et al.
Figure 5. Summary of long-range NOEs (NOESY, 500 MHz, mixing time 0.5 s, 4 mM in CDCl3) observed for the 1:1 mixtures of (a) 1a + 4a, and (b)
1a + 3.
and 4a was quite soluble in chloroform. The association constant
of 3 and 4a was determined by diluting a solution in CDCl3.
Surprisingly, the resultant values varied significantly and fell
into two different groups: (1) data based on the NH signals of
3 led to numbers in the 103 M-1 range and (2) data based on
the NH signals of 4a led to much smaller values in the 102
M-1 range. Such a result indicates the existence of a strong
heterodimer consisting of 3 and 4a, which requires that similar
values of the association constant being obtained based on the
NH signals of either strand is unlikely. Instead, the two sets of
different values are consistent with a strong self-dimer (3‚3)
that interacts weakly with 3, which is consistent with the
conclusion from the above NOESY studies on 3, 4a, and 4b,
and can also explain the VPO result of the mixture of 3 and
4a. In the presence of polar solvent such as DMSO, the
association between 3 and 4a and the self-association of 3 or 4
should be much weaker. However, analysis of the same (3 + 4a)
mixture in CDCl3 containing 5% DMSO-d6 was not successful
due to the precipitation of 3 out of this mixed solvent.
a solution of 1.1 mM of 1a with 8.4 mM of 2a in 5% DMSO/
CHCl3 gave an association constant of (2.4 ( 0.4) × 104 M-1
,
along with ∆G ) -5.9 ( 0.1 kcal/mol and ∆H ) -3.9 ( 0.3
kcal/mol. The corresponding thermogram is shown in Figure
7. Under the same conditions, ITC titration failed to detect
association between the template strands 1 and 2, and between
tripeptides 3 and 4a in 5% DMSO/CHCl3. Therefore the addition
of the tripeptide strands to 1 and 2 led to a hybrid duplex with
enhanced stability. Such an increase in stability may very likely
be due to the cooperative interaction between the duplex
template and the tripeptide segment. However, probing the
energetic nature of duplex 1a‚2a by using differential scanning
calorimetry (DSC) was not feasible due to the low boiling points
of the solvents (CHCl3 and CH2Cl2) that were required for
solubilizing this duplex.
Conclusion
Our study shows that the noncovalent template based on a
hydrogen-bonded duplex is highly directional and specific in
effecting the assembly of two different structural fragments. The
resulted hybrid duplexes adopt a well-defined conformation in
solution. A two-stranded, â-sheet conformation is formed, as
revealed by the changes in HR chemical shifts, the numerous
inter- and intrastrand NOEs, the VPO results, and the enhanced
(5) Isothermal Titration Calorimetry. Determining the
association constant of the hybrid duplex 1a‚2a in CDCl3 or in
CDCl3/DMSO-d6 turned out to be difficult due to the extensive
overlap of numerous NH signals. Instead, titrations using
isothermal titration calorimetry (ITC) were carried out. Titrating
9
2908 J. AM. CHEM. SOC. VOL. 124, NO. 12, 2002

A Noncovalent Approach to Antiparallel â-Sheet Formation
A R T I C L E S
Figure 6. Summary of long-range NOEs (NOESY, 500 MHz, mixing time 0.5 s, 4 mM in CDCl3) observed for the 1:1 mixtures of 3 and 4a. Strong
contacts are indicated by solid arrows; weak contacts are represented by dashed arrows.
stability of the hybrid duplex 1a‚2a. This noncovalent approach
has turned an otherwise intramolecular process, i.e., â-hairpin
(or two-stranded â-sheet) formation, into an intermolecular one,
which should facilitate detailed thermodynamic studies in the
future. Given the facility in programming the H-bonded
duplexes, a template-based, noncovalent combinatorial approach
should provide numerous â-sheet structures, which not only will
lead to invaluable insight into the factors affecting â-sheet
formation, but also should provide a platform for generating
covalently linked â-sheet structures based on an assembly-
crosslinking strategy. The assembly of multiple peptide strands
can also be envisioned. For example, template-directed assembly
of two â-hairpins should lead to four-stranded â-sheets. The
design of duplex templates with various side chains that are
9
J. AM. CHEM. SOC. VOL. 124, NO. 12, 2002 2909

A R T I C L E S
Zeng et al.
Table 2. Concentrations and Aggregation Numbers (VPO) in
the existence of â-sheet structures in all four hybrid duplexes
containing different amino acid residues. This, combined with
the convenience of simply mixing templated peptide strands into
duplexes, distinguishes our system from existing, covalent
templating strategies. The specificity of a water-soluble duplex
template, in combination with the stability of peptide strands
that contain hydrophobic clusters,9 may lead to the specific
assembly of sheet-like structures that are stable in an aqueous
environment. The corresponding results will be reported in due
course.
Chloroforma,b
concn
(mmolar)
concn
(mmolal)
aggregation
no.
substrate
1
4.1
4.6
4.9
4.0
4.3
4.9
2.8
3.2
3.3
2.7
2.9
3.3
0.9 ( 0.1
0.9 ( 0.1
1.1 ( 0.1
1.0 ( 0.1
1.0 ( 0.1
1.0 ( 0.1
1 + 2
4a
3 + 4a
1a
1a + 2a
a Aggregation number of GFL at 10 mM is 1.4 ( 0.1. For 1 + 2,
3 + 4a, or 1a + 2a, the two components have identical concentration
with the shown value. Some of the data may be 0.9 or 1.1 because of the
10% experimental error in this type of measurement. b Plus/minus values
are reported at the 95% confidence level.
Experimental Section
General Methods. All chemicals were purchased from Aldrich,
Fluka, and Sigma and used as received unless otherwise noted. The
organic phase from all liquid extractions was dried over Na2SO4 unless
specified otherwise. All products were detected as single spots by thin-
layer chromatography (precoated 0.25 mm silica plates from Aldrich).
All samples were purified either by recrystallization or by flash column
chromatography and dried completely under high vacuum before
characterized by 1H NMR (400 M), 13C NMR (100 M), and elemental
analysis. All 1H NMR and 13C NMR spectra were recorded on a Varian
VXR 400 spectrometer (400 M). NMR chemical shifts were reported
in ppm relative to TMS. All J values are reported in hertz. NOE
measurements were performed with the steady-state NOEDIF protocol
on degassed samples. The synthesis of a, 1p, 1v, 2c, and 2f was reported
previously by us.22,23
Vapor-Pressure Osmometry. The aggregation numbers for the
peptide, template, and peptide-template strands were determined by
employing a Wescor 5500-XR vapor pressure osmometer operating at
37 °C. Known concentrations of the individual strands or the comple-
mentary duplexes were prepared in chloroform. Calibration curves of
VPO reading vs molality were obtained by employing biphenyl as the
calibration standard. A least-squares analysis of 10 points provided a
correlation coefficient of 0.991. Solutions of the individual strands of
the complementary strands were placed in the osmometer and the
instrument reading was converted into concentration by using the
calibration curve. Each VPO datum was generated by making three
independent solutions of each strand at the same concentration and
measuring each solution a minimum of four times.
Table 3. Association Constants from NMR Binding Studiesa
c
1‚2b
4a
data based on
K (M-1
)
data based on
K (M-1
)
a
a
H41
H6
H36
H33
(0.67 ( 0.10) × 103
(1.25 ( 0.24) × 103
(1.14 ( 0.12) × 103
(0.36 ( 0.04) × 103
H23
H28
H31
H33
14.5 ( 2.5
22.6 ( 3.2
8.4 ( 1.9
18.3 ( 2.8
3 + 4ad
data based on NH of 3
data based on NH of 4a
H11
H13
H16
H20
(0.92 ( 0.18) × 103
(1.91 ( 0.63) × 103
(0.58 ( 0.14) × 103
(1.90 ( 0.60) × 103
H23
150 ( 14
190 ( 14
110 ( 13
167 ( 20
H28
H31
H33
a See Figure 3 for atom labeling. b NMR measurements were carried out
from 64 to 0.125 mM of each of the two strands in 5% DMSO-d6/CDCl3.
c NMR measurements were carried out from 51.2 to 0.1 mM in CDCl3. A
dimerization mode is assumed when fitting data. d NMR measurements were
carried out from 25.6 to 0.1 mM in CDCl3. A dimerization mode is assumed
when fitting data.
Binding Studies. The binding parameters were determined by
titrating a 1.1 mM solution of 1a with 8.4 mM 2a in 5% DMSO/CHCl3
in an Omega isothermal titration calorimeter (MicroCal, Northampton,
MA). The cell was thermostated to (0.1 °C with use of a circulating
bath. All of the experiments were performed at 25 °C. The enthalpy of
binding between the two strands was determined from heats of multiple
single injections. Injection volumes were 5 µL, with 3 min of
equilibration time allowed between injections. The heat of dilution of
2a into 5% DMSO/CHCl3 solvent was determined and the host-guest
titration heat was adjusted by this contribution.
The binding constants, K, and the number of binding sites, n, were
extracted from the calorimetric data employing the Origin data analysis
software supplied with the Omega titration calorimeter. A complete
description of the data analysis has been published by Brandts and co-
workers.32
NMR dilution experiments were performed at 25 °C. The solution
of a sample was diluted and the chemical shifts of the NH-protons
were followed in the 1H NMR spectra and analyzed on the basis of an
equation described before.31
Figure 7. Calorimetry binding isotherm for the titration of 1a with 2a in
5% DMSO/chloroform.
Supporting Information Available: NMR spectra and ex-
perimental details (PDF). This material is available free of
charge via the Internet at http://pubs.acs.org.
compatible to a variety of solvents is being pursued. The
resultant duplexes will facilitate characterization based on
techniques such as DSC and far-UV circular dichroism that
could not be employed here due to the limit of the solvents
involved. The generality of our approach is demonstrated by
JA010701B
(32) Wiseman, T.; Williston, S.; Brandts, J. F.; Lin, L.-N. Anal. Biochem. 1989,
179, 131.
9
2910 J. AM. CHEM. SOC. VOL. 124, NO. 12, 2002