An indirect chaotropic mechanism for the stabilization of helix conformation of peptides in aqueous trifluoroethanol and hexafluoro-2- propanol

Article abstract of DOI:10.1021/ja973552z

A revised indirect mechanism is proposed for the effect of 2,2,2- trifluoroethanol on peptide conformation (TFE effect) that suggest tighter solvent shells in pure water for helical states than random coil states. The alcoholic cosolvent stabilizes the he

Full text of DOI:10.1021/ja973552z

J. Am. Chem. Soc. 1998, 120, 5073-5079
5073
An Indirect Chaotropic Mechanism for the Stabilization of Helix
Conformation of Peptides in Aqueous Trifluoroethanol and
Hexafluoro-2-propanol
Richard Walgers, Tony C. Lee, and Arthur Cammers-Goodwin*
Contribution from the Department of Chemistry and the UniVersity of Kentucky Center for Membrane
Sciences, UniVersity of Kentucky, Lexington, Kentucky 40506
ReceiVed October 10, 1997
Abstract: A revised indirect mechanism is proposed for the effect of 2,2,2-trifluoroethanol on peptide
conformation (TFE effect) that suggests tighter solvent shells in pure water for helical states than random coil
states. The alcoholic cosolvent stabilizes the helical state preferentially by disrupting the solvent shell, which
causes unfavorable enthalpic and favorable entropic contributions to the free energy of helix formation. This
revised mechanism was adopted because it best explained the solvent-dependent thermodynamic behavior of
the coil/helix transition. To define the TFE effect, solvent-dependent physicochemical behaviors of two
molecular probes for solvent character were monitored and compared with the solvent dependence of peptide
helix formation. The rate of decarboxylation of 6-nitro-3-carboxybenzisoxazole was determined in aqueous
i
mixtures as a function of concentration for DMSO, EtOH, MeOH, PrOH, HFIP, and TFE. To relate these
rate studies to the cosolvent-dependent thermodynamics of helix formation, ∆Hq and ∆Sq as a function of
concentration for EtOH and TFE were determined and interpreted. The mixed solvent dependence of the UV
spectrum of a solvatochromic ketone was also monitored to correlate the behavior of the mixed solvent systems
with a microscopic polarity index.
Goodman and co-workers in the early sixties discovered that
2,2,2-trifluoroethanol (TFE) coxed certain medium length
peptides to adopt helical conformation (the TFE effect).1
Although less dramatic than TFE, helix-inducing effects have
been observed for other alcohols.2 These examples of selective
stabilization of polypeptide conformations by alcoholic cosol-
vents bore significance for scientists working in biophysical and
biochemical areas. Medium length peptides tend to adopt ran-
dom coil conformations in water,3 impeding the study of poly-
peptide conformation outside the complex context of proteins.
Low concentrations of TFE in water enable comparisons of
conformational stability to be made between these peptides
presumably by magnifying latent conformational biases.4,5
Though persistent â-sheet6 conformations have been observed
in aqueous TFE, selective TFE-mediated stabilization within
the protein context has been observed more often for R-helical
conformations, sometimes in regions with native â-sheet prefer-
ence.7,8 These observations cast doubt on the hypothesis
positing simple magnification of natural conformational tenden-
cies in peptides by aqueous TFE.9
unclear. Since the TFE effect is so striking, much stands to be
learned from mechanistic studies. If the mechanism of the TFE
effect were known, back calculation from helix propensities in
TFE/water mixtures to native helix propensity in peptide chains
might be possible.
Proposed Mechanisms
Two sets of hypotheses for the conformational behavior of
peptides in varying concentrations of aqueous TFE have been
proposed. Direct mechanisms involve preferential binding of
TFE to the helical conformer of peptides. Indirect mechanisms
suggest that TFE-mediated changes in the aqueous solvent shell
around polypeptides account for the solvent-induced stabilization
of helical states.
In a detailed study of five peptide homologues, TFE-de-
pendent conformational behavior was modeled as a two-state
system in which TFE preferentially associated with the helical
conformer.5 Similar elements constitute another direct hypoth-
esis for the interaction of aqueous TFE with peptides.10 The
authors reasoned that the fluorocarbon terminus of TFE should
interact favorably with hydrophobic side chains, and the hydroxy
terminus should preferentially interact with the amide carbonyls.
A direct mechanism involving selective stabilization of intramo-
lecular hydrogen bonds by TFE has also appeared in the recent
literature.11 The decisive assay involved monitoring ∆pKa
between salicylic acid and 4-hydroxybenzoic acid as a function
Though investigators in the peptide sciences commonly use
aqueous mixtures of TFE to study structural propensities, the
kinship between TFE-induced states and aqueous states is
(1) Goodman, M.; Listowsky, I.; Masuda, Y. F. B. Biopolymers 1963,
1, 33-42.
(2) Conio, G.; Patrone, E.; Brighetti, S. J. Biol. Chem. 1970, 245, 3335-
3340.
(3) Mun˜oz, V.; Serrano, L. Nature Struct. Biol. 1994, 1, 399-409.
(4) Nelson, J. W.; Kallenbach, N. R. Proteins: Struct., Funct., Genet.
1986, 1, 211-217.
(8) Shiraki, K.; Nishikawa, K.; Goto, Y. J. Mol. Biol. 1995, 245, 180-
194.
(9) Chuang, W.-J.; Chitrananda, A.; Gittis, A. G.; Pederson, P. L.;
Mildvan, A. S. Arch. Biochem. Biophys. 1995, 319, 110-122.
(10) Rajan, R.; Balaram, P. Int. J. Peptide Protein Res. 1996, 48, 328-
336.
(5) Jasanoff, A.; Fersht, A. Biochemistry 1994, 33, 2129-2135.
(6) Schonbrunner, N.; Wey, J.; Kiefhaber, T. J. Mol. Biol. 1996, 260,
432-445.
(7) Jayaraman, G.; Kumar, T. K. S.; Yu, C. Biochem. Biophys. Res.
Commun. 1996, 222, 33-37.
(11) Luo, P.; Baldwin, R. L. Biochemistry 1997, 36, 8413-8421.
S0002-7863(97)03552-X CCC: $15.00 © 1998 American Chemical Society
Published on Web 05/06/1998

5074 J. Am. Chem. Soc., Vol. 120, No. 20, 1998
Walgers et al.
TFE concentration. The curves describing ∆pKa and helicity
were superimposed.
Opposite trends in both ∆H and ∆S have been observed for
MeOH, EtOH, iPrOH, BuOH,2 and TFE.11 Furthermore, d∆S/
d[HFIP] > 0 and d∆H/d[HFIP] > 0 of helix formation has been
implicated by cold denaturation in ala-rich icosomers by low
concentrations of HFIP (2-5 mol %).13 Overwhelming evi-
dence supports entropically controlled R-helix formation as the
concentration of alcoholic cosolvents increases. Except for
Andersen’s explanation of HFIP-induced cold denaturation,13
the current direct and indirect mechanisms predict exactly the
opposite of what is observed for the cosolvent-dependent ∆H
and ∆S of helix formation.
An hypothesis for the mechanism of the TFE effect involving
the perturbation of aqueous solvent properties alone was the
first explanation for the helix-inducing effects of low aqueous
concentrations of alcohols.2 In a study of helix-coil transitions
of (L-Orn)n and (L-Glu)n it was proposed that alcoholic cosol-
vents decrease the extent to which backbone amide functions
are solvated, and thus, selectively destabilize the random coil
state relative to closed, internally hydrogen bonded confor-
mations. Interpretations of NMR studies have corroborated
these findings. TFE-induced changes in chemical shifts of a
disulfide linked peptide consisting of two helical domains
failed to show evidence of binding to TFE.12 1,1,1,3,3,3-
Hexafluoro-2-propanol (HFIP)-promoted cold denaturation of
peptide helices has been explained by an indirect mechanism
that focused on chaotropic, differential solvation of the nonpolar
surface.13
Revised Indirect Mechanism
Since the TFE effect has been observed with simple amides
and a proline ester, we reasoned that backbone solvation is the
main element in a minimalist argument, and we focused this
revised indirect mechanism on the solvation of the backbone
in two states. The need to include the observed solvent
dependencies of ∆H and ∆S of helix formation leads to the
premise that the helical state, not the random coil state, is most
solvated and thus most perturbed by changes in solvent
environment. In this revised mechanism, helix formation is
impeded by the entropic cost of assembling the aqueous solvent
shell around the helix in pure water.
The random coil was chosen as the most solvated state in
previous formulations of indirect mechanisms because the
solvent supposedly makes more contacts with the amide
carbonyls in the random coil state than in the R-helix state.
However, the R-helix state has a greater per residue dipole
moment18 and should interact electrostatically with water better
than most other solvents. Random coil states are structurally
similar to â-sheet conformations on average, and peptides that
favor â-sheet conformation tend to precipitate from solution.
Furthermore, R-helices present curved surfaces to water solvent
and formation of helices reduces solvent accessible surface.
Thus, the coil to helix transition should multiply and enhance
water/water interactions in the solvent shells around polypep-
tides. Since the R-helix should promote water/water interac-
tions, it should be more sensitive to changes in aqueous solvent
environment. Focus of this mechanism on the cohesive nature
of water is congruent with the growing recognition that the
properties of water determine the chemical behavior of biologi-
cal molecules.19
The diagram allows comparison of the former indirect
mechanism with our revised indirect mechanism on a conceptual
level. For indirect mechanisms one assumes that the enthalpic
and entropic effects of solvation vastly outweigh thermodynamic
contributions from conformational changes in the peptide
backbone. Since fluoro alcohols at low concentration have such
a striking effect on peptide conformation, this assumption seems
sound. The CS and HS states are hypothetical coil and helix
states that are preferentially solvated by water whereas the C
and H states are comparatively unsolvated. Only three out of
the four states presented in the diagram need to be considered
for each mechanism. Previous indirect hypotheses involved the
pure aqueous solvent system CS h H transforming to C h H
upon addition of alcoholic cosolvent, which produces favorable
enthalpic and unfavorable entropic contributions to helix forma-
tion as Xcosolvent increases. The revised indirect mechanism
proposed here involves C h HS progressing toward C h H as
Xcosolvent increases, which causes unfavorable enthalpic and
In another study, strikingly similar TFE-induced phenomena
in simple amidic and polyamidic molecules best supported an
indirect mechanism for the TFE effect.14 Increasing aqueous
mole percent TFE (XTFE) increased the amount of helical
conformation in peptide conjugates covalently bound to a helix-
initiating template.15 This model effectively separated the TFE
effect on helix initiation from that of helix propagation, an
ambiguity present in all previous studies. The general shape
and coincidence of helicity with increasing XTFE for the template-
bound peptide conjugates matched the acceleration of cis-trans
peptide bond isomerization of N-acetylproline methyl ester as
14
a function of XTFE
.
Since the proline model lacked the helical
motif, selective, direct interaction between cosolvent and helical
conformations was ruled out.
Relevance of the proline model as a solvent probe can be
found in other studies. From studies of pure solvents and the
kinetic barrier to cis-trans amide bond isomerization, ∆H was
found to contribute the most to the stabilization of the cis and
trans ground states.16 Later investigations led the same authors
to conclude that hydrogen bonding by the solvent also contrib-
uted much to the stability of the ground states.17
Inadequacies in Proposed Mechanisms
All the indirect mechanisms for the action of TFE have
assumed decreasing stabilization of the random coil state with
increasing XTFE. Direct mechanisms have focused on favorable
hydrogen bonding or hydrophobic interactions between cosol-
vent molecules and helix conformations. Both mechanisms
predict favorable enthalpic contributions as Xcosolvent increases.
Since discrete interactions in pure water tie up the helical state
in the direct mechanism and liberate the random coil state in
the indirect mechanism, both models predict increasingly
unfavorable entropic contributions to the free energy of helix
formation as Xcosolvent increases. Current direct and indirect
mechanistic constructions of the TFE effect are inadequate
because they predict d∆H/d[TFE] < 0 and d∆S/d[TFE] < 0.
(12) Storrs, R. W.; Truckses, D.; Wemmer, D. E. Biopolymers 1992,
32, 1695-1702.
(13) Andersen, N. H.; Cort, J. R.; Liu, Z.; Sjoberg, S. J.; Tong, H. J.
Am. Chem. Soc. 1996, 118, 10309-10310.
(14) Cammers-Goodwin, A.; Allen, T. J.; Oslick, S. L.; McClure, K. F.;
Lee, J. H.; Kemp, D. S. J. Am. Chem. Soc. 1996, 118, 3081-3090.
(15) Kemp, D. S.; Boyd, J. G.; Muendel, C. C. Nature 1991, 352, 451-
454.
(16) Eberhardt, E. S.; Raines, R. T. Tetrahedron Lett. 1993, 34, 3055-
3056.
(17) Eberhardt, E. S.; Raines, R. T. J. Am. Chem. Soc. 1994, 116, 2149-
2150.
(18) Tsong, T. Y.; Astumian, R. D. Bioelectrochem. Bioenerg. 1986,
15, 457-476.
(19) Finney, J. L. Faraday Discuss. 1996, 103, 1-18.

Stabilization of Helix Conformation of Peptides
J. Am. Chem. Soc., Vol. 120, No. 20, 1998 5075
favorable entropic contributions to helix formation as the alco-
holic cosolvent is added.
of 0-60 mol % cosolvent. Decarboxylation of 1 has been
shown to proceed through a late transition state without inter-
mediates22 on the reaction pathway and to depend on hydrogen
bonding with rates spanning approximately 8 orders of mag-
nitude.24 Recently, the solvent-dependent rate of the decar-
boxylation of 1 has been characterized and touted as a molecular
probe for biologically relevant solvents.25
Two factors mainly determine the rate of decarboxylation of
1 with tetramethylguanidinium as the counterion. The reaction
runs fastest in polar aprotic solvents.24 As the polarity of the
solvent system decreases, greater contact between ion pairs in
solution decreases the rate of the reaction. As the hydrogen
bond acidity of polar solvents increases, the rate of decarboxy-
lation of 1 decreases. Changes in the rate of the reaction in
this study should be ascribable to changes in the hydrogen bond
donor ability of the solvent for a few reasons. Reaction rate
deceleration due to ion contact factors less than the deceleration
due to hydrogen bond acidity. Also the properties of the solvent
mixtures should reflect small perturbations in water because this
work focused on aqueous mixtures of polar solvents in which
the major component was water. Furthermore, these results
were obtained with potassium, a counterion incapable of forming
hydrogen bonds as the contact ion pair. We felt that the
applicability of the model outweighed gains in organic solubility
conferred by the tetramethylguanidinium counterion.
Increasing Xcosolvent perturbs the ground state 4 more than
transition state 5 because 4 is more solvated; 4 is analogous to
the HS state. Structures C and HS, when compared to 4 and
5, show that the TFE effect on ∆H and ∆S of the coil/helix
transition should be opposite to the TFE effect on the ∆Hq and
∆Sq for the decarboxylation of 1. As 4 progresses toward 5,
the extent of solvation decreases due to localized cross-
conjugated negative charge in 4 diffusing to conjugated negative
charge spread over the aromatic transition state, 5. The common
characteristic of both systems is differential solvation.
Herein two well-characterized molecular probes for solvent
properties (1 and 3) were employed and the observables were
titrated with TFE in water. When possible, MeOH, EtOH,
DMSO, and HFIP were also included in the titrations to compare
the overall magnitude of the solvent effect. These studies
approached mechanistic consensus by two routes.
The helix/coil transition is a complex system consisting of
many conformations with hypothetically many local minima
complicated by the presence of side chains. Fortunately, ∆Hq
and ∆Sq of the decarboxylation of 1 should behave predictably
in the mostly aqueous systems studied here. If increasing
Xcosolvent breaks ground-state solvation, ∆Hq should decrease
because the energy of the ground state with respect to the
transition state decreases (∆H of the ground state increases more
than ∆H of the transition state). Likewise ∆Sq should also
become less positive because the ground state will possess more
entropy with respect to the transition state with increasing
Chemical Models for the Helix/Coil Transition
One assay was designed to probe the hydrogen bonding
character of the media as a function of cosolvent concentration.
For this purpose the rate of decarboxylation of one of Kemp’s
other acids,20-23 6-nitro-3-carboxybenzisoxazole (1), to 2-hy-
droxy-4-nitrobenzonitrile (2) was used in the concentration range
Xcosolvent and decreasing solvation. For simplicity, entropic and
enthalpic changes in the transition state are assumed to be small
(23) Kemp, D. S.; Cox, D. D.; Paul, K. G. J. Am. Chem. Soc. 1975, 97,
7312-7318.
(24) Ferris, D. C.; Drago, R. S. J. Am. Chem. Soc. 1994, 116, 7509-7514.
(25) Lewis, C.; Kra¨mer, T.; Robinson, S.; Hilvert, D. Science 1991, 253,
1019-1022.
(20) Zipse, H.; Apaydin, G.; Houk, K. N. J. Am. Chem. Soc. 1995, 117,
8608-8617.
(21) Gao, J. J. Am. Chem. Soc. 1995, 117, 8600-8607.
(22) Kemp, D. S.; Paul, K. G. J. Am. Chem. Soc. 1975, 97, 7305-7311.

5076 J. Am. Chem. Soc., Vol. 120, No. 20, 1998
Walgers et al.
compared with solvent-dependent thermodynamic perturbations
in the ground state. The solvent effect should produce trends
in ∆Hq and ∆Sq for the decarboxylation of 1 with XTFE that
oppose each other energetically. When interactions between
water and the carboxylate are broken, lowering ∆Hq , there
should be corresponding decreases in ∆Sq .
a
Aqueous solutions of another molecular probe of solvation
were titrated with cosolvents. In this case, λmax of 4-[(1′-
pentylthio)acetyl](4′-chlorobenzyl)pyridinium chloride (3) re-
ports microscopic solvent polarity, not dielectric constant. The
dielectric constant as a function of XTFE is known26 and the
response of λmax of neither 3a nor 3c correlate with it. The
cationic and zwitterionic forms, 3a and 3c, produce shifts in
λmax as a function of solvent that correlate with the solvent
polarity indices of Snyder27 and Reichardt,28 respectively. λmax
has been shown to depend on the equilibrium 3a h 3b. In
studies with pure solvents, λmax as a function of solvent does
not correlate at all with the rate for the decarboxylation of 1.
Therefore hydrogen bond donation from the solvent should be
a minor factor for λmax. Most striking is the difference between
k of decarboxylation of 1 and λmax of 3 in response to the pure
solvents MeOH and DMSO. These two solvents are found at
opposite extremes of the rate of decarboxylation of 1, but they
coincide on the polarity indices generated by both 3a and 3c.
Both probes of solvation should respond similarly to the dis-
ruption of the aqueous solvent shell.
b
Results and Discussion
TFE-induced changes in the rate of decarboxylation of 1 and
on the λmax of 3a and 3c were not dramatic when compared
with the behavior of the other solvents tested; see Figure 1a-
c. In retrospect, a response in the opposite direction produced
by the other solvents might have been expected. A negative
response (i.e., k and λmax lower than pure water) should have
resulted if aqueous TFE associated to the basic sites in 1, 3a,
or 3c through hydrogen bonding. A few conclusions can be
drawn from these data. TFE/water mixtures conserved the
properties of aqueous solvent according to these probes for
solvent character. Similarities in microscopic polarity between
TFE and water have been observed previously.29,30 Furthermore,
TFE and HFIP showed poor hydrogen bond acceptor (electron
donor) properties. Increased solvent polarity and hydrogen bond
donation from the solvent stabilizes the ketone 3a, and decreases
c
λmax, whereas decreased solvent polarity and electron donor
character in the solvent stabilize the enol 3b.31 Similar trends
were observed for both 3a and 3c, which laid aside all claims
of direct mechanisms with TFE donating a hydrogen bond to
amide carbonyl functions in peptide chains at cosolvent
concentrations relevant to the TFE effect. Enolate 3c should
be a better electron donor than the amide carbonyl oxygen atom.
At a glance, changes in the solvent character brought about by
increasing XTFE appeared inconsequential for both solvent
probes. However, closer examination of the shapes of these
curves reveals that the behavior of the solvent probes changed
in a defined, nonlinear fashion with XTFE. In Figure 2, as XTFE
increased in the presence of K2CO3 and 1 at 30 °C, an initial
Figure 1. Nonlinear changes in k (s-1) of decarboxylation of 1 in the
low concentration region of XTFE and XHFIP
.
nonlinear increase in the rate of decarboxylation ensued,
followed by a decrease in the rate at XTFE > 20 mol %. Rate
depression at XTFE > 20 mol % was probably a function of
hydrogen bond donation to the carboxylate, which stabilized
the starting material relative to the transition state. This
conclusion was corroborated by rate depression at lower XHFIP
than XTFE. Complete proton transfer is not possible from the
fluoro alcohols to any of these molecular probes for solvation
(26) Murto, J.; Heino, E.-L. Suom. Kemistil. 1966, B39, 263-266.
(27) Snyder, L. R. J. Chromatogr. 1974, 92, 223-230.
(28) Reichardt, C.; Harbusch-Go¨rnert, E. Liebigs Ann. Chem. 1983, 721-
743.
(29) Hagen, P. A.; Heilbronner, E.; Straub, P. A. HelV. Chim. Acta 1967,
50, 2504-2520.
(30) Reichardt, C. SolVents and SolVent Effects in Organic Chemistry,
2nd ed.; Verlag Chemie: Weinheim, 1988; p 361.
(31) Biellmann, J.-F.; Holler, M.; Burger, A. J. Am. Chem. Soc. 1996,
118, 2153-2159.
31
because the pKa of 1, 3a, 3b, TFE, and HFIP are 1.6,22 7.9,
31
7.4, 12.4,10 and 9.3,10 respectively.

Stabilization of Helix Conformation of Peptides
J. Am. Chem. Soc., Vol. 120, No. 20, 1998 5077
Figure 4. ∆Hq and -T∆Sq (in kcal/mol) for the decarboxylation of 1
Figure 2. Nonlinear rise in λmax (nm) for 3a and 3c in the low
concentration region of XTFE and XHFIP. Studies with 3c and HFIP were
omitted because partial protonation of 3c was observed in the UV
spectrum.
plotted as a function of XEtOH
.
and 15 mol % TFE has been observed for helicity in peptide
chains.14 Above 20 mol % TFE, contributions to helicity from
X
TFE diminish. The coincidence of solvent probe response and
R-helix stabilization on the XTFE axis indicates that the TFE
effect at XTFE < 20% is a result of TFE-induced changes in
water structure instead of preferential interaction between TFE
and polypeptide.
The HFIP dependence of the λmax of 3a corresponded to the
cosolvent concentration at which cold denaturation of peptide
helices has been observed.13 Furthermore, the TFE dependence
of λmax of 3c approximated the HFIP dependence of the λmax of
3a. Increased charge on the carbonyl oxygen of 3c versus 3a
tightens the water structure33,34 in the solvent shell of the
chromophore and thereby increases the solvent effect. These
facts implicate the destruction of liquid water structure in the
solvent shell as the primary factor in the TFE effect at low
concentration (<20 mol %) and argue for generality in the
alcoholic cosolvent effect on peptide conformation.
Figure 3. ∆Hq and -∆TSq (in kcal/mol) for the decarboxylation of 1
Applicability of the TFE-dependent rate of decarboxylation
of 1 as a thermodynamic model for TFE-dependent helix/coil
transitions in peptides was further questioned by studying the
TFE dependence of the activation parameters for decarboxyla-
tion. Rates of decarboxylation of 1 were determined at five
temperatures for XTFE values between 0 and 55 mol %. Al-
though the general trends for both ∆Hq and T∆Sq were similar,
there were small relative changes between the two values that
were apparent when the two graphs were compared; see Figure
4. Furthermore, the contributions to ∆Gq of ∆Hq and ∆Sq
opposed each other. Hence, the net change in ∆Gq was small.
Helix stabilization has been observed for TFE and other
alcohols. However, stabilization by alkanol cosolvents versus
TFE pales comparatively. Ethanol should enact similar changes
in the activation parameters albeit at higher XEtOH. Figure 5
shows this to be true. The activation parameters for the EtOH
titration reached a minimum at ∼40 mol % EtOH instead of
∼13 mol % with TFE. Here again, entropic contributions of 4
to ∆Gq opposed enthalpic contributions of 4 to ∆Gq due to
decreased solvation of 4 as a function of XEtOH. Carrying this
study past 55 mol % resulted in precipitation of 4. Analogous
curve shapes as a function of XEtOH have been discovered by
Winstein for ∆Hq and ∆Sq for SN1 reactions.35
plotted as a function of XTFE
.
Analogous changes in λmax in the same ranges of XTFE
observed for k of 1 were observed for both 3a and 3c, except
λmax did not decrease with increasing XTFE for 3a; these results
are graphically displayed in Figure 3. λmax holds steady with
increasing XTFE past 15 mol % due to the lack of electron donor
character of the fluoro alcohols and the similarity between TFE
and water in terms of microscopic polarity. The inability of
TFE to act as a hydrogen bond acceptor has been noted
previously.32 Since the hydrogen bond acidity of both fluoro
alcohol cosolvents is greater than that of water, the changes in
signal observed for both probes at low concentrations of
cosolvent were anomalous. Observables for both solvent probes
deviated in the wrong direction for solvents with increased
hydrogen bond acidity and decreased electron pair donor ability
compared to water. These changes suggested cosolvent-
dependent restructuring of the aqueous solvent shell at low
concentration.
Analogous nonstoichiometric effects from 5 to 20 mol % TFE
and 2-5 mol % HFIP have been observed for the conforma-
tional behavior of peptides as a function of Xcosolvent as discussed
above. The TFE effect on peptide helicity and the observables
of these probes for solvent character coincided on the Xcosolvent
axis. In particular, the sharp nonstoichiometric rise between 5
(33) Pross, A. Theoretical & Physical Principles of Organic ReactiVity;
John Wiley and Sons: New York, 1995; p 208.
(34) Increased water structure with increased dipole moment is not
unanimously accepted, see: Haymet, A. D. J.; Silverstein, A. T.; Dill, K.
A. Faraday Discuss. 1996, 103, 117-124.
(32) Pitner, T. P.; Urry, D. W. J. Am. Chem. Soc. 1972, 94, 1399-1400.

5078 J. Am. Chem. Soc., Vol. 120, No. 20, 1998
Walgers et al.
addition of cosolvent, trends in ∆H and ∆S for helix formation
as a function of XTFE make sense. Perhaps due to the striking
nonstoichiometric establishment of helical structure with incre-
ments in XTFE, a few investigators have framed the TFE effect
as an associative mechanism. The behavior of molecular probes
for solvation studied herein approximates the TFE-dependent
conformational behavior of peptides because the observables
in all these cases are a function of the structural integrity of
water. Solvent dependence in λmax of 3 should be a function
of microscopic polarity and the decarboxylation of 1 has been
shown to be dependent on hydrogen bonding parameters of the
solvent. Both of these observables should vary in the same
direction with perturbations in liquid water structure.
This indirect model for peptide solvation as a function of
TFE offers an explanation for peptides that show â-sheet
conformational preferences in water and switch to R-helical
conformations upon addition of TFE. In shorter peptides,
â-sheet conformations outside of the protein context that are
not extensively stabilized by packing interactions are loosely
ordered and random coil-like.37 Addition of TFE entropically
drives the peptide toward helicity by loosening the solvent shell
around the helical conformations.
Figure 5. The graph of the observables with respect to Xcosolvent for
(a) the rate constant of decarboxylation, k (s-1) of 1. (b) λmax (nm) of
cation 3a. (c) λmax (nm) of the zwitterion of 3c.
Conclusion
The solvent probes studied offer the following evidence for
the revised indirect mechanism for the TFE effect on helix
propensity in oligopeptides. (1) Aqueous TFE at 5-15 mol %
produces similar results for a variety of physicochemical
phenomena, including helicity in polypeptides.14 This work
correlated the TFE effect to two different chemical phenomena
that do not involve polyamides or amidic structure; the generality
of the TFE effect on various physicochemical phenomena speaks
against a mechanism involving selective direct interaction
between cosolvent and peptide conformers or solvent probe.
(2) The rate of decarboxylation of 1 in aqueous TFE at 5-15
mol % is larger than the rate at XTFE > 30 mol %. This suggests
two mechanisms of interaction as XTFE increases. The direct
mechanism should be the one that occurs at high XTFE and
appears to involve hydrogen bond donation to 1. Hydrogen
bond donation to 3c at increased XTFE was also observed. (3)
Trends observed for the solvatochromic probe also suggest no
direct interaction by hydrogen bond donation. These results
were surprising because TFE is more hydrogen bond acidic than
water.36 If aqueous TFE interacted directly with 3, the ketone
form, 3a, should have been stabilized versus the enol 3b.
Furthermore, at XTFE < 20%, there is no evidence for direct
interaction with the enolate 3c. This result was significant since
the zwitterionic enolate 3c should hydrogen bond more strongly
than the peptide amide carbonyl. (4) Trends in both ∆Hq and
∆Sq for the decarboxylation of 1 as a function of XTFE indicate
TFE-mediated changes in solvent structure. Perturbations in
∆Hq detected breakage of hydrogen bonds from solvent to
carboxylate with increasing XTFE and ∆Sq detected the con-
comitant unfettering of solute structure. (5) Attenuated effects
For purposes of back calculation of helix propensity in TFE
to native helix propensity, solvation of the random coil states
cannot be ignored even though the helix is perturbed more by
the cosolvent. A few pertinent questions come to mind about
the nature of solvent accessibility to the backbone donor sites
in the random coil state. Are all the basic sites equally ac-
cessible for all random coils and is this independent of peptide
composition? If so, peptide chemists can still discuss native
helical preference as a function of TFE and correlate this pref-
erence to the purely aqueous state. If random coil states do
not randomize solvent access to the hydrogen bond acceptor
sites on the peptide backbone in a manner independent of pep-
tide composition, then helix propensity initiated by TFE is a
function of complex solvent contributions to the random coil
state. Recent calculations indicate the composition-dependent
solvation of the random coil factors to some extent because
solvent access to amide carbonyls in the random coil state de-
pends on peptide composition.38 Regardless, native helix pro-
pensity from conformational studies in TFE could be assigned
between families of peptides related by primary structure.
Our model for helix stabilization by cosolvents should be
viewed as a first approximation to the TFE effect. There are
complexities brought about by the complex nature of polypep-
tides that are beyond the scope of this model. For example,
helicity in peptides bearing lysine residues has shown strong
context dependence along with a tendency to melt.39 Further-
more, our studies say nothing about changes in side chain
entropy upon helix formation as a function of XTFE. Side chain
entropy has recently attracted the attention of investigators as a
primary factor in helix stability.40
on ∆Hq and ∆Sq were noted for aqueous EtOH at low XEtOH
.
Later in the titration EtOH eventually broke the aqueous
solvation of the carboxylate. Again, analogy to the helix/coil
transition is evident because helix induction with EtOH occurs
later in the titration than with TFE. (6) The similarity between
the dependence of λmax of 3a on XHFIP and the dependence of
λmax of 3c on XTFE also points toward a general solvent shell
perturbation by the cosolvent. The increased charge of the
enolate 3c tightens the water solvent shell and thereby increases
the cosolvent effect of TFE.
Experimental Section
Anhydrous solvents were prepared by refluxing reagent grade
commercial material over a suitable drying agent for at least 0.5 h
followed by distillation under a nitrogen atmosphere.
(36) Abraham, M. H.; Duce, P. P.; Grellier, P. L.; Prior, D. V.; Morris,
J. J.; Taylor, P. J. Tetrahedron Lett. 1988, 29, 1587-90.
(37) Schneider, J. P.; Kelly, J. W. Chem. ReV. 1995, 95, 2169-87.
(38) Creamer, T. P.; Srinivasan, R.; Rose, G. D. Biochemistry 1997, 36,
2832-2835.
(39) Renold, P.; Tsang, K. Y.; Kemp, D. S. J. Am. Chem. Soc. 1996,
118, 12234-12235.
(40) Aurora, R.; Creamer, T. P.; Srinivasan, R.; Rose, R. D. J. Biol.
Chem. 1997, 272, 1413-1416.
If the helical state is identified as more water solvated than
the random coil state and therefore the most perturbed by
(35) Winstein, S.; Fainberg, A. H. J. Am. Chem. Soc. 1957, 79, 5937.

Stabilization of Helix Conformation of Peptides
J. Am. Chem. Soc., Vol. 120, No. 20, 1998 5079
1H and 13C NMR were recorded on a Varian 200 Gemini and a
Varian VXR 300 spectrometer. UV-vis spectra were recorded with a
Shimadzu UV-3101PC UV-vis-NIR scanning spectrometer with a
temperature-controlled cell holder. Reaction mixtures for measurements
with probe 1 where kept at a constant temperature in a Fisher Scientific
7305 thermostated circulation bath.
of 70% sulfuric acid; the resulting solution was warmed to 80 °C for
4 h. Slowly water was added to the solution at 25 °C until the solution
was over saturated. This solution was left standing for 8 h. The
resulting precipitate was washed with water and dried under high
vacuum. The dried product 1 was obtained as off-white crystals in
67% yield.41
4-[(1′-Pentylthio)acetyl](4′-chlorobenzyl)pyridinium chloride (3).
These compounds were synthesized according to the method used by
Holler et al.31 For 4-[(1′-pentylthio)acetyl](4′-chlorobenzyl)pyridinium
chloride (3), the purification was changed to the following: After
removal of the remaining thiol by distillation, the orange oil was diluted
with 4 mL of anhydrous pyridine while heating at 35 °C. Ethyl acetate
was added until material precipitated from solution. The mixture stood
at 0 °C for 5 h. The precipitate was filtered under Schlenk conditions
and washed with a 10% anhydrous pyridine solution in ethyl acetate
(100 mL) and with ethyl acetate until no pyridine was present. After
the mixture was dried under vacuum, compound 3 was obtained in
50% yield. The material was spectroscopically identical with the
material reported in the literature.
Measurements of Solvent Character with Probe (3). For mea-
surements with the cationic form of 3, 50 µL of a 1.38 mg/mL 4-[(1′-
pentylthio)acetyl](4′-chlorobenzyl)pyridinium chloride solution in water
was added to 2.95 mL of a water/organic solvent mixture. The
absorption spectrum was then measured in the 350-550 nm range.
The wavelength of the absorption maximum (λmax) was recorded. For
observations of the zwitterionic form of 3, solutions with 0.1 mg/mL
of K2CO3 in both water and organic solvent were used to make up a
2.95 mL water/organic solvent mixture. The total concentration of the
cationic probe was the same as for the zwitterionic probe (3). All
spectra were taken at 30 °C.
Measurements of the Decarboxylation Rate of 1. An aqueous
solution of 1 (150 µL of 1 mg/mL) was added to 2.95 mL of water/
organic solvent mixture. The absorbance was measured at an appropri-
ate sampling rate at 396 nm with a slit width of 3 nm, until a significant
amount of the total expected maximum absorbance had developed.
During the measurements, the solution was kept at a constant
temperature in a sealed cuvette. All measurements were taken at 30
°C unless stated otherwise. For the determination of ∆Hq and ∆Sq of
the decarboxylation as a function of XTFE, the rate was measured at 3,
15, 30, 40, and 50 °C. Wynne-Jones and Eyring analysis42 was applied
to the rate vs temperature data to obtain the activation parameters. Error
in the plots of the Eyring analysis was estimated by assuming that the
third through the eighth points in ∆Hq for EtOH were the same value;
the standard deviation in this value was computed and used for the
error for T∆Sq and ∆Hq for EtOH and TFE.
Acknowledgment. These investigations were supported by
an NSF career award to Arthur Cammers-Goodwin, CHE-
9702287. Many thanks go to professors Trevor Creamer and
Mark Meier for collegiate discussions.
JA973552Z
(41) Lindemann, H.; Cissee, H. Liebigs Ann. Chem. 1929, 469, 44-
57.
6-Nitro-3-carboxybenzisoxazole (1). A flask was charged with 1
g (4.5 mmol) of methyl 6-nitrobenzisoxazole-3-carboxylate and 20 mL
(42) Laider, K. J. Chemical Kinetics; Harper & Row: New York, 1987;
p 188.