Intramolecular catalysis of amide isomerization: Kinetic consequences of the 5-NH- -N(a) hydrogen bond in prolyl peptides

Article abstract of DOI:10.1021/ja9815071

The presence of an intramolecular hydrogen bond has been proposed to play a key role in the catalysis of amide isomerization by peptidylprolyl isomerases (PPIases), which are highly conserved and ubiquitous rotamase enzymes that catalyze the cis-trans iso

Full text of DOI:10.1021/ja9815071

10660
J. Am. Chem. Soc. 1998, 120, 10660-10668
Intramolecular Catalysis of Amide Isomerization: Kinetic
Consequences of the 5-NH- -Na Hydrogen Bond in Prolyl Peptides
Christopher Cox and Thomas Lectka*
Contribution from the Department of Chemistry, Johns Hopkins UniVersity, Baltimore, Maryland, 21218
ReceiVed May 1, 1998
Abstract: The presence of an intramolecular hydrogen bond has been proposed to play a key role in the
catalysis of amide isomerization by peptidylprolyl isomerases (PPIases), which are highly conserved and
ubiquitous rotamase enzymes that catalyze the cis-trans isomerization of proline residues in peptides and
proteins. We present herein kinetic and spectroscopic evidence that indicates the existence of an intramolecular
hydrogen bond between the prolyl amide nitrogen and the adjacent amidic NH within a five-membered ring
(the 5-NH- -Na hydrogen bond) that is capable of catalyzing proline isomerization by up to 260-fold in model
prolyl peptides. Our results provide the first systematic study of intramolecular general-acid-catalyzed amide
isomerization.
Introduction
Scheme 1. Intramolecular Catalysis of Amide Isomerization
The catalysis of amide isomerization by strong Bro¨nsted acids
is a well-known process that proceeds through a putative
N-protonated intermediate present in low concentration.1 The
carbonyl oxygen of amides is universally believed to be the
preferred site of protonation in aqueous acidic media;2 however,
acid-catalyzed amide isomerization must proceed through the
disfavored N-protonated intermediate wherein we expect nearly
free rotation about the C-N bond due to complete loss of
conjugation between the nitrogen lone pair and the carbonyl
π-cloud.3 Only a very small percentage of amide molecules
need be N-protonated to effect a sizable overall barrier
lowering,2b,4 but this nevertheless requires exposure of the amide
to a solution of low pH. For example, the rate of amide
isomerization in DMA (dimethylacetamide) increases 130-fold
when the pH of the solution is lowered from 7.0 to 1.8.1b Given
the importance of the amide functional group in biological
systems, the question of whether N-protonation of amides plays
a role in biological processes, such as the catalysis of protein
folding, becomes a significant one.
donation to the amide nitrogen (Na) through a correctly aligned
cyclic intermediate replaces discrete intermolecular N-protona-
tion. Karplus has reported theoretical results that implicate this
interaction as a mechanism in the catalysis of cis-trans proline
isomerization by rotamase enzymes,6 and we have presented
the first experimental support of this interaction in a recent
Communication.7
The cis-trans isomerization of prolyl peptide bonds is an
inherently slow process with a free energy of activation (∆Gq)
of 20-22 kcal/mol in aqueous solution, and is known to be the
slow step in the folding of many proline-containing proteins.8
FKBP and cyclophilin are rotamase enzymes, or peptidylprolyl
isomerases (PPIases), that catalyze this isomerization, both for
model peptides and proteins in vitro1a,8a as well as in vivo.9
The PPIases10 have also been found to be the cellular targets of
immunosuppressive drugs used to prevent graft rejection in
humans;11 however, there appears to be no direct relationship
between the immunosuppressive function and isomerase activ-
ity.12 Although the full extent of the biological roles played
Nature, rather than operate with strong Bro¨nsted acidic media,
tends to avoid harsh, low-pH conditions by rendering processes
intramolecular, thereby gaining effective molarity through
pinpointed spatial proximity. Intramolecular general acid-
catalyzed amide isomerization of this type is pictured in Scheme
1 as a process in which hydrogen bond (H-bond)5 or proton
(1) (a) Stein, R. L. AdV. Protein Chem. 1993, 44, 1. (b) Gerig, J. T.
Biopolymers 1971, 10, 2435. (c) Steinberg, I. Z.; Harrington, W. F.; Berger,
A.; Sela, M.; Katchalski, E. J. Am. Chem. Soc. 1960, 82, 5263. Perrin
investigated the mechanism of acid-catalyzed proton exchange in amides;
see: (d) Perrin, C. L. Acc. Chem. Res. 1989, 22, 268.
(2) (a) Homer, R. B.; Johnson, C. D. In The Chemistry of Amides;
Zabicky, J., Ed.; Wiley: New York, 1970; Chapter 3. (b) The pKa of an
O-protonated amide is close to zero, whereas an N-protonated amide has a
pKa of about -7; see: Williams, A. J. Am. Chem. Soc. 1976, 98, 5645.
(3) On the other hand, hydrogen bonding to the carbonyl oxygen or
O-protonation will have a barrier-raising effect on amide isomerization;
see: (a) Scheiner, S.; Kern, C. W. J. Am. Chem. Soc. 1977, 99, 7042. (b)
Neuman, R. C., Jr.; Woolfenden, W. R.; Jonas, V. J. Phys. Chem. 1969,
73, 3177.
(5) List of abbreviations: IC, intramolecular catalysis; Na, prolyl amide
nitrogen; H-bond, hydrogen bond; ST, saturation transfer NMR.; DHFR,
dihydrofolate reductase; FKBP, binding protein for the immunosupressive
agent FK-506; PPIase, peptidylprolyl isomerase (generic term for cyclophilin
and/or FKBP); EDCI, 1-(3-dimethylaminopropyl)-3-ethylcarbodiimide hy-
drochloride, a water soluble carbodiimide used in peptide coupling.
(6) (a) Fischer, S.; Dunbrack, R. L., Jr.; Karplus, M. J. Am. Chem. Soc.
1994, 116, 11931. (b) Fischer, S.; Michnick, S.; Karplus, M. Biochemistry
1993, 32, 13830.
(7) Cox, C.; Young, V. G., Jr.; Lectka, T. J. Am. Chem. Soc. 1997, 119,
2307.
(8) (a) Schmid, F. X.; Mayr, L. M.; Mu¨cke, M.; Scho¨nbrunner, E. R.
AdV. Protein Chem. 1993, 44, 25. (b) Schmid, F. X. In Protein Folding;
Creighton, T. E., Ed.; Freeman: New York, 1992; Chapter 5. (c) Fisher,
G.; Schmid, F. X. Biochemistry 1990, 29, 2205. (d) Brandts, J. F.; Halvorson,
H. R.; Brennan, M. Biochemistry 1975, 14, 4953.
(4) (a) Martin, R. B.; Hutton, W. C. J. Am. Chem. Soc. 1973, 95, 4752.
(b) Martin, R. B. J. Chem. Soc., Chem. Commun. 1972, 793.
10.1021/ja9815071 CCC: $15.00 © 1998 American Chemical Society
Published on Web 10/01/1998

Intramolecular Catalysis of Amide Isomerization
J. Am. Chem. Soc., Vol. 120, No. 41, 1998 10661
by these novel enzymes has yet to be established, their
importance is supported by the observation that they are
ubiquitous and highly conserved in organisms ranging from E.
coli to humans.11
The detailed mechanisms by which the PPIases catalyze cis-
trans isomerization remain to be completely elucidated. Thor-
ough knowledge of these mechanisms is required to design better
inhibitors for studies of the immunosuppressive and protein
folding roles of the PPIases, and also to examine other postulated
roles of these enzymes in ViVo.13 One important aspect of the
PPIase mechanism that has been proposed based on theoretical
studies is intramolecular catalysis, whereby the prolyl nitrogen
becomes an H-bond acceptor in the rotational transition state
for amide isomerization (Figure 1).6,14 Intramolecular catalysis
is also believed to play a key role in the nonenzymatic folding
of some proteins including dihydrofolate reductase (DHFR,
Figure 1B).15 Substrates bound in the active site of FKBP are
postulated to adopt a type VIa proline turn,6 in which the amide
proton of the residue that follows the proline in primary
sequence sits directly above the ring and can donate an H-bond
to the prolyl Na (Figure 1A, the 5-NH- -Na H-bond); this
stabilizing interaction has been calculated to contribute 1.4 of
the 6.2 kcal/mol decrease in ∆Gq for FKBP-catalyzed proline
isomerization.6b On the other hand, cyclophilin binds its
substrates in a type VIb proline turn, in which the adjacent amide
Figure 1. Intramolecular catalysis in biological systems.
proton is not properly aligned to induce intramolecular catalysis.
However, by analogy to the proposed folding of DHFR,15 there
is an Arg residue close in tertiary structure within the active
site of cyclophilin (but not in the active site of FKBP) that may
act as the hydrogen bond donor during catalysis (Figure 1B).6b
Intramolecular hydrogen bonding between a prolyl Na and
nearby H-bond donors is quite common in structural protein
chemistry, yet its role in the folding and stabilization of proteins
is often ignored.16 Matthews examined H-bonding patterns in
the crystal structures of 42 proteins determined at high resolution
and found that eight have strong H-bond donors (Arg, His, Lys)
within 4 Å of a prolyl nitrogen; 12 more H-bonds were identified
in this same set of proteins if neutral donors were included.15
Additionally, there are a sizable number of cyclic peptidic
natural products which contain a 5-NH- -Na H-bond, as evi-
denced by crystal structures,17 as well as the spirocyclic
peptidomimetics synthesized by Johnson.18 It has yet to be
established conclusively whether H-bonds to prolyl nitrogens
are involved in intramolecular catalysis of protein folding, but
these interactions clearly occur with considerable frequency in
nature.
(9) There have only been a few reports of PPIases acting as folding
catalysts in vivo: (a) For the role of cyclophilin in the maturation of the
collagen triple helix, see: Steinmann, B.; Bruckner, P.; Superti-Furga, A.
J. Biol. Chem. 1991, 266, 1299. (b) For cyclophilin catalyzed folding of
the monomeric protein transferrin, see: Lodish, H. F.; Kong, N. J. Biol.
Chem. 1991, 266, 14835. (c) For the possible role of PPIases in protein
biogenesis in a eukaryotic cytosol, see: Kruse, M.; Brunke, A. E.; Escher,
A.; Szalay, A. A.; Tropschug, M.; Zimmermann, R. J. Biol. Chem. 1995,
270, 2588.
(10) The terminology used in the literature for rotamase enzymes can
be ambiguous. “PPIase” was originally synonymous with cyclophilin, but
the current usage, and the one we adopt here, is to use PPIase as a generic
term encompassing both cyclophilin and FKBP. When we wish to refer to
a specific enzyme, we will clearly differentiate between them.
(11) (a) Fruman, D. A.; Burakoff, S. J.; Bierer, B. E. FASEB J. 1994, 8,
391. (b) Belshaw, P. J.; Meyer, S. D.; Johnson, D. D.; Romo, D.; Ikeda,
Y.; Andrus, M.; Alberg, D. G.; Schultz, L. W.; Clardy, J.; Schreiber, S. L.
Synlett. 1994, 381.
(12) It is known that the catalysis of cis-trans amide isomerization is
not the direct function of the PPIases in immunosupression; rather, a
complex of the immunosupressive drug and the enzyme is believed to bind
to the protein phosphatase calcineurin, and thereby inhibit T-cell activation.
The immunosupressive drugs do, however, bind in the active site of
isomerase activity; see: Schreiber, S. L.; Crabtree, G. R. Immunol. Today
1992, 13, 136. It is interesting to speculate why these enzymes apparently
perform two unrelated tasks by utilizing the same active site.
(13) Other postulated roles of the PPIases in vivo include: (a) Acting
as chaperones, especially for escorting rhodopsin from the ER through the
secretory pathway to its cellular target, see: Baker, E. K.; Colley, N. J.;
Zuker, C. S. EMBO J. 1994, 13, 4886. (b) Effectors of Xaa-Pro bonds which
may act as “molecular switches” to trigger actions such as voltage-gated
ion channel opening; see: Suchyna, T. M.; Xu, L. X.; Gao, F.; Fourtner,
C. R.; Nicholoson, B. J. Nature 1993, 365, 847. (c) Modulation of calcium
release by interaction with calcium release channels; see: Jayaraman, T.;
Brillantes, A.-M.; Timerman, A. P.; Fleischer, S.; Erdjument-Bromage, H.;
Tempst, P.; Marks, A. R. J. Biol. Chem. 1992, 267, 9474. (d) Helping with
protein degradation by catalyzing unfolding and prevention of partially
unfolded proteins from precipitation, see: Andres, C. J.; Macdonald, T.
L.; Ocain, T. D.; Longhi, D. J. Org. Chem. 1993, 58, 6609. (e) Functioning
as an auxiliary enzyme in HIV-1 protease-mediated reactions, see: Vance,
J. E.; LeBlanc, D. A.; Wingfield, P.; London, R. E. J. Biol. Chem. 1997,
272, 15603.
In a preliminary report, we provided the first experimental
evidence for intramolecular catalysis in model systems (sub-
stituted prolines) attributed to the 5-NH- -Na H-bond depicted
in Figure 1A.7 We now describe a comprehensive experimental
study of this interaction in model systems, and have expanded
its scope to include the following: (1) correlation between the
Hammett σp values of remote substituents and the kinetics of
observed intramolecular catalysis; (2) extensive spectroscopic
characterization of the intramolecular 5-NH- -Na H-bond; (3)
intramolecular catalysis of prolyl carbamates in organic and
aqueous/organic solution; (4) intramolecular catalysis observed
in “polar” chlorocarbon solvents such as CH2Cl2 or CHCl3 is
less than that possible in a very nonpolar, non-hydrogen bonding
solvent such as CCl4; and (5) proof that intramolecular catalysis
is due to an H-bond and not to discrete N-protonation of the
prolyl Na.
(16) For earlier discussions of [NH- -Na] interactions, see: (a) Gieren,
A.; Dederer, B.; Schanda, F. Z. Naturforsch., C: Biosci. 1980, 35c, 741.
(b) Scarsdale, J. N.; Van Alsenoy, C.; Klimkowski, V. J.; Scha¨fer, L.;
Momany, F. A. J. Am. Chem. Soc. 1983, 105, 3438.
(17) See for example: (a) Shoham, G.; Lipscomb, W. N.; Wieland, T.
J. Am. Chem. Soc. 1989, 111, 4791. (b) Kopple, K. D.; Bhandary, K. K.;
Kartha, G.; Yang, Y.-S.; Parameswaran, K. N. J. Am. Chem. Soc. 1986,
108, 4637. (c) Springer, J. P.; Cole, R. J.; Dorner, J. W.; Cox, R. H.; Richard,
J. L.; Barnes, C. L.; van der Helm, D. J. Am. Chem. Soc. 1984, 106, 2388.
(d) Montelione, G. T.; Arnold, E.; Meinwald, Y. C.; Stimson, E. R.; Denton,
J. B.; Huang, S. G.; Clardy, J.; Scheraga, H. A. J. Am. Chem. Soc. 1984,
106, 6, 7946. (e) Karle, I. L. J. Am. Chem. Soc. 1979, 101, 181.
(18) Genin, M. J.; Ojala, W. H.; Gleason, W. B.; Johnson, R. L. J. Org.
Chem. 1993, 58, 2334.
(14) In his original reports, Karplus termed this interaction “autocatalysis”
because the hydrogen bond donor was attached to the proline system either
covalently or noncovalently, such as in an enzyme-substrate complex.
However, this terminology may cause confusion because it is commonly
used in the organic synthesis literature to mean catalysis of a reaction by
its product. We therefore term this interaction intramolecular catalysis, as
was proposed by Matthews.15.
(15) Texter, F. L.; Spencer, D. B.; Rosenstein, R.; Matthews, C. R.
Biochemistry 1992, 31, 5687.

10662 J. Am. Chem. Soc., Vol. 120, No. 41, 1998
Cox and Lectka
Results and Discussion
the strongly H-bond-accepting solvent molecules, so intramo-
lecular catalysis (IC) was defined as ∆(∆Gq) in the change from
aqueous solution to an organic solvent for model amides,
subtracted by the analogous ∆(∆Gq) for model esters (eq 2).7
However, there are alternative ways of defining intramolecular
catalysis, and one equation may not always be the most suitable
for every situation. For example, if the entropy of activation
(∆Sq) for amide isomerization is negligible, as precedent
suggests24 and our results confirm (vide infra), intramolecular
catalysis quantities defined in terms of either ∆Gq or ∆Hq would
be similar and eqs 2 and 3 can be used interchangeably. Thus,
we can compare the thermodynamically important quantity ∆Hq
in various systems without the errors commonly attendant with
its measurement.25 Additionally, eqs 2 and 3 presuppose that
intramolecular catalysis will always be completely washed out
in an aqueous environment, an assumption that we have found
is not always valid (vide infra); nevertheless, eq 2 represents a
safe lower limit on intramolecular catalysis. However, for
isosteric substrates, there is no reason to believe that the simpler
eqs 4 and 5 would not serve just as well, and can expand the
range of model systems amenable to investigation due to often
unfavorable separation of NMR resonances in aqueous/organic
solution. The kinetics of amide isomerization in this study were
measured by the highly reliable techniques of 1H or 19F
saturation transfer (ST) NMR.26
Intramolecular Catalysis in Prolyl Peptides. At the outset,
we reasoned that small peptides containing the correct structural
features should show intramolecular catalysis in an organic
medium that mimics the desolvated environment19 of the FKBP
active site, thus permitting clear-cut documentation of the
5-NH- -Na H-bond unobstructed by other effects. We chose to
compare activation barriers for two sterically similar prolines
in organic and in aqueous/organic solution; one proline contains
the catalytic NH general acid in the side chain, the other not,
while both side chains are nearly isosteric. Amides 1 and esters
2 fulfill these requirements.20 An A1,3-like strain between the
carboxamide side chain and the R substituent in cis prolines
(cis-1) can influence their conformations;21 to mitigate A1,3
-
strain, the cis proline may place its side chain pseudoaxially,
which would facilitate formation of the 5-NH- -Na H-bond, with
the carbonyl oxygen of the amide side chain oriented exo to
the proline ring.22 In nonpolar solution, we expect the cis form
of amides 1 to contain a 5-NH- -Na H-bond between the side
chain and the prolyl Na; this interaction should be strengthened
in the transition state for cis-to-trans amide isomerization as
Na becomes more basic (eq 1).6a The more abundant trans form
predominately consists of a γ-turn in organic solvents (trans-
1).23
Intramolecular catalysis in terms of ∆Gq (with aqueous
correction):
(1)
IC ) [∆Gqamide(aqueous) - ∆Gqamide(organic)] -
[∆Gqester(aqueous) - ∆Gqester(organic)] (2)
Intramolecular catalysis in terms of ∆Hq (with aqueous
correction):
IC ) [∆Hqamide(aqueous) - ∆Hqamide(organic)] -
[∆Hqester(aqueous) - ∆Hqester(organic)] (3)
Intramolecular catalysis in terms of ∆Gq:
IC ) [∆Gqester(organic) - ∆Gq
]
(4)
amide(organic)
Intramolecular catalysis in terms of ∆Hq:
IC ) [∆Hqester(organic) - ∆Hq
]
(5)
amide(organic)
We began our study by obtaining kinetic data for the cis-
trans isomerization of prolinamide 1a and control ester 2a by
19F saturation transfer at various temperatures. Figure 2 shows
the results of a typical ST experiment on 1a, and Figure 3
contains the Eyring plots produced from both substrates in
organic and aqueous/organic solutions.27 The results of the
kinetic and thermodynamic analyses are collected in Table 1.
In 60% MeOH/D2O,28 the barriers to amide isomerization (∆Gq)
We initially expected that in aqueous/organic solution in-
tramolecular catalysis would be eliminated by competition from
(19) Liang, G.-B.; Rito, C. J.; Gellman, S. H. J. Am. Chem. Soc. 1992,
114, 4440.
(20) Many proline derivatives show poor cis-trans ratios in nonpolar
solvents and/or have unresolved cis-trans rotameric signals. Our test
substrates were chosen in part because sufficient cis form could be detected
with baseline resolution to facilitate NMR analysis.
(21) Zhang, R.; Brownwell, F.; Madalengoitia, J. S. J. Am. Chem. Soc.
1998, 120, 3894.
(24) (a) Eberhardt, E. S.; Loh, S. N.; Hinck, A. P.; Raines, R. T. J. Am.
Chem. Soc. 1992, 114, 5437. (b) Drakenberg, T.; Dahlqvist, K.-I.; Forsen,
S. J. Phys. Chem. 1972, 76, 2178.
(22) We define the carbonyl as exo when it is pointed away from the
proline ring, whereas endo indicates that the carbonyl oxygen is directly
over top of the ring. The exo carbonyl and pseudoaxially positioned side
chain are clearly evident in the crystal structure of a cyclic peptide.17b.
(23) It is generally accepted that the γ-turn is the predominate form
present in organic solvents; see: (a) Madison, V.; Kopple, K. D. J. Am.
Chem. Soc. 1980, 102, 4855. (b) Higashijima, T.; Tasumi, M.; Miyazawa,
T. Biopolymers 1977, 16, 1259. It is unclear how much of the trans form
is in this conformation, as opposed to a nonintramolecularly H-bonded
conformation; see: (c) Liang, G.-B.; Rito, C. J.; Gellamn, S. H. Biopolymers
1992, 32, 293.
(25) It is generally more accurate to determine ∆Gq at a single
temperature than to measure ∆Hq by Eyring analysis over a limited
temperature range; see: Martin, M. L.; Mabon, F.; Trierweiler, M. J. Phys.
Chem. 1981, 85, 76.
(26) For applications of ST to amide isomerization, see: (a) Perrin, C.
L.; Thoburn, J. D.; Kresge, J. J. Am. Chem. Soc. 1992, 114, 8800. We
have used 19F ST NMR to take advantage of the broad chemical shift range
and generally favorable peak separations of the 19F nucleus; see: (b) Cox,
C.; Ferraris, D.; Murthy, N. N.; Lectka, T. J. Am. Chem. Soc. 1996, 118,
5332.

Intramolecular Catalysis of Amide Isomerization
J. Am. Chem. Soc., Vol. 120, No. 41, 1998 10663
Table 1. Activation Parameters and IC Values for Proline 1a and
Control Ester 2aa
proline
solvent
60% MeOH/D2O 19.2 19.1 18.5 -2.4 1.2
CDCl3 16.8 15.5 16.0 -3.0 9.8 1.4/2.6
60% MeOH/D2O 19.2 18.7 18.6 -2.2 2.2
CDCl3 18.2 17.7 17.9 -1.1 2.5
∆Gq b ∆Gq c ∆Hq d ∆Sq e Kf
ICg
1a
1a
2a
2a
a Kinetics measured by 19F ST NMR, 15 mg/mL. See Experimental
Section for details. b Trans-cis, (0.2 kcal/mol, 25 °C. c Cis-trans,
(0.2 kcal/mol, 25 °C. d Trans-cis, (0.3 kcal/mol. e Trans-cis, (4
cal/(mol K). f K ) [trans]/[cis]. g IC ) intramolecular catalysis in kcal/
mol via eq 2; first number is trans-cis, second is cis-trans.
Table 2. Electronic Effects on ∆Gq a
proline
∆Gq b
∆Gq c
K d
1b
2b
1c
2c
1d
2d
1e
2e
16.8
18.7
17.0
18.8
17.1
18.8
16.5
18.6
15.0
17.8
15.3
17.8
15.6
17.8
14.2
17.5
25.0
4.7
21.1
4.8
14.0
4.8
65.1
5.6
Figure 2. Result of 19F saturation transfer experiment on proline 1a.
Arrows represent the location of homonuclear decoupling.
1
a Measured by H ST NMR, 6 mg/mL in CD2Cl2, 25 °C. b Trans-
cis, (0.2 kcal/mol. c Cis-trans, (0.2 kcal/mol. d K ) [trans]/[cis].
that we attribute to intramolecular catalysis from the 5-NH- -
Na H-bond. As expected,24 negligible ∆Sq values were found
in all cases, so that values for intramolecular catalysis defined
in terms of either ∆Hq or ∆Gq are similar at 25 °C for these
simple amides.
Prolines 1b-1e with anilide side chains of different acidities
and the requisite controls 2b-2e were also analyzed kinetically
as shown in Table 2.30 Amide 1b31 affords intramolecular
catalysis of 2.8 kcal/mol (cis-trans) at 25 °C in CD2Cl2.32
Electron-donating substituents (1c, p-OMe and 1d, p-NMe2)
remotely placed on the aryl group show less catalysis (2.5 and
2.2 kcal/mol, cis-trans), whereas a remote electron-withdrawing
substituent (1e, p-COOMe) exhibits the greatest degree of
catalysis (3.3 kcal/mol, cis-trans). This latter result represents
a 260-fold rate enhancement of amide isomerization over control
ester 2e. Figure 4 shows a Hammett plot of the ln(relative rate
of amide isomerization) versus σp.33 A direct relationship is
seen between the degree of catalysis and the strength of the
H-bond donor, and the positive F value of 2.2 indicates that
electron-withdrawing groups accelerate the reaction, as expected.
This result provides kinetic evidence that intramolecular catalysis
due to the 5-NH- -Na H-bond is indeed responsible for the
enhanced rate of amide isomerization in these prolines.
Spectroscopic Characterization. We examined the pro-
posed 5-NH- -Na H-bond by IR and NMR spectroscopy to
characterize the interaction more fully. However, complications
were expected due to the predominance of the γ-turn between
the side chain NH and the carbonyl oxygen (trans-1). The top
trace in Figure 5 shows the high-field region of the 500 MHz
NMR spectrum of proline 1f at 10 mM in CD2Cl2. The presence
of the trans and cis forms in a ratio of 6:1 is apparent from the
Figure 3. Eyring plots of 1a in CDCl3 (O) and 60% MeOH/D2O (4)
and 2a in CDCl3 (X) and 60% MeOH/D2O (]).
of amide 1a and isosteric ester 2a were found to be identical,
as expected. The equilibrium constants (K ) [trans]/[cis]) were
also roughly equivalent.29 In CDCl3, however, ∆Gq in amide
1a dropped by 2.4 kcal/mol for trans-to-cis isomerization and
by 3.6 kcal/mol for cis-to-trans, whereas in ester 2a the
respective barrier lowerings were both 1.0 kcal/mol (in line with
a simple solvent effect).24 Employing eq 2 thus provides a
difference of 1.4 (trans-to-cis) and 2.6 kcal/mol (cis-to-trans)
(27) ST measurements on these substrates were made at 15 mM in the
solvent of choice. We found the degree of intramolecular catalysis to be
fairly insensitive to concentration; however, we always use equal concentra-
tion of substrates whenever direct comparisons are made. See the
Experimental Section for details of the NMR experiments, calculation of
rate constants, and formation of Eyring plots. For a discussion of activation
parameters, see: Carpenter, B. Determination of Organic Reaction Mech-
anisms; John Wiley: New York, 1984; p 123.
(28) A mixed solvent system (D2O/MeOH) was necessary due to lack
of solubility in pure water; in general we find that the barriers to rotation
of water soluble amides in pure D2O are not greatly different than those in
MeOH/D2O mixtures.
(29) An equilibrium constant K greater than 1 indicates that the energy
of the trans conformation is lower than that of the cis form. The ester controls
have a smaller K in organic solvents than the corresponding amides, due
largely to the absence of the stable γ-turn that is present in the amides;
thus, we find different values for ∆Gq of cis-to-trans and of trans-to-cis
isomerization. The former values are more significant in terms of importance
in enzymatic catalysis, whereas the latter values are the ones directly
measured by ST experiments (and are the ones reported throughout the
paper, unless otherwise noted). The cis-to-trans value is obtained simply
from K ) ktrans-to-cis/kcis-to-trans by plugging in the value of K.
(30) We found that the ∆Gq’s in aqueous/organic solution were identical
for 1b-1e and 2b-2e, and therefore eq 4 is sufficient to analyze the data.
(31) Suzuki, K.; Ikegawa, A.; Mukaiyama, T. Bull. Chem. Soc. Jpn. 1982,
55, 3277.
(32) Here we find it necessary to change from CDCl3 to CD2Cl2 due to
the extremely unfavorable cis-trans ratio of these prolines in the former
solvent. The larger dielectric constant of CH2Cl2 increases this ratio, but
we find that in general ∆Gq is uneffected by change between these two
similar chlorocarbon solvents.
(33) For a discussion of Hammett correlations and tabulated σp values,
see: March, J. AdVanced Organic Chemistry; John Wiley: New York, 1992;
pp 278-286.

10664 J. Am. Chem. Soc., Vol. 120, No. 41, 1998
Cox and Lectka
Figure 4. Hammett plot showing the correlation between the relative
rate of amide isomerization and the electronic properties of the aryl
ring.
Figure 6. Hydrogen bonding region of the IR spectra of proline 1f
(top) and locked amide 3a (bottom) at 10 mM in CH2Cl2.
Scheme 2. Synthesis of Locked Peptidomimetic 3a
Figure 5. High-field region of the 500 MHz NMR spectra of proline
1f (top) and locked amide 3a (bottom) at 10 mM in CD2Cl2.
synthesis of locked amide 3a began with CBz-proline methyl
ester 4, which was alkylated with allyl bromide after treatment
with lithium bis(trimethylsilyl)amide in THF (Scheme 2).
Hydroboration with disiamylborane and oxidation with hydrogen
peroxide afforded an intermediate alcohol that was subjected
to Jones oxidation to afford carboxylic acid 5. Removal of the
CBz protecting group followed by EDCI-promoted5 cyclization
provided 6 in 79% yield from 5. Titanium-promoted transes-
terification followed by benzyl ester hydrogenolysis and ami-
dation of the intermediate acid with EDCI provided locked
amide 3a (8 steps and 21% overall yield from 4).
two NH resonances at 7.4 and 6.5 ppm. The top trace in Figure
6 shows the H-bonding region of the IR spectrum under the
same conditions. The broad stretch centered at 3305 cm-1 is
assigned to the γ-turn, and the sharper stretch at 3430 cm-1 is
provisionally assigned to the 5-NH- -Na H-bond and that portion
of the trans form that is not intramolecularly H-bonded.34
The bottom traces in Figures 5 and 6 show the NMR and IR
spectra of 3a under conditions identical to those used for proline
1f. In the 1H NMR spectrum there is a single NH resonance at
6.8 ppm, and in the IR spectrum only one NH stretch is present
at 3419 cm-1
: this evidence is consistent with a single
For further support of the 5-NH- -Na H-bond, we desired a
model system free of complications from the trans form. To
this end, we synthesized proline peptidomimetics 3 which are
“locked” in the cis form. Our view was that 3 should faithfully
model the 5-NH- -Na unit of actual cis prolines without
interference from the trans isomer; for example, the side chains
of amides 3 in computer models are rigidly pseudoaxial, a
prerequisite for the assembly of the 5-NH- -Na interaction. Our
(34) This stretch has been identified in previous work as the overlapping
NH stretches of both the minor cis form and also a fraction of the trans
form that is not involved in a γ-turn (in the previous work, however, the
cis form was considered to be totally non H-bonded).23c The data we have
assembled herein are consistent with the belief that the stretch at higher
wavenumber is due predominately to the 5-NH- -Na H-bond. Previous
reports have noted that the intensity ratio of the two NH stretches is
independent of concentration over a limited range in CHCl3 and CCl4,23b
consistent with an intramolecular bond.

Intramolecular Catalysis of Amide Isomerization
J. Am. Chem. Soc., Vol. 120, No. 41, 1998 10665
Table 3. IR Data for Locked Amides 3 and Control Amides 7a
∆υb
υCdO
c
substrate
υ(NH- -Na)
υ(control)
3b
3c
3d
3e
3f
3392
3390
3385
3381
3376
3432
3433
3430
3427
3426
-40
-43
-45
-46
-50
1677
1682
1690
1694
1698
a Solutions were 10 mM in CH2Cl2, all values in cm-1
.
b Shift of
the NH- -Na bond in 3 relative to control 7. c Stretching frequency of
the endocyclic amide carbonyl.
conformation containing the 5-NH- -Na H-bond. Several pieces
of evidence support these assignments in locked amide 3a: (1)
The chemical shift of the NH resonance in the NMR spectrum
of 3a is independent of concentration below 15 mM, consistent
with an intramolecular H-bond. Above 25 mM this peak begins
to shift downfield with the onset of intermolecular H-bonding.
(2) IR spectra confirm that below 15 mM only one stretch exists
in the H-bonding region of 3a (at 3419 cm-1); however, at
higher concentrations a broad stretch centered at 3320 cm-1
begins to grow in, corresponding to the onset of intermolecular
H-bonding. This behavior contrasts that of 1f during dilution
studies, where the relative intensities of both H-bond stretches
shown in the top trace of Figure 6 are independent of
concentration over a 100 mM range (data not shown). (3)
Control amide 7a,35 which cannot engage in intramolecular
H-bonding, has a lone IR stretch at 3446 cm-1 at 10 mM in
CH2Cl2, consistent with reported values for non-hydrogen
bonded amide protons.36 (4) The locked geometry of 3a
precludes intramolecular H-bonding between NH and the
carbonyl oxygen due to geometric constraints, and therefore the
intramolecular H-bond must be between NH and the prolyl Na.
These data suggest that the 5-NH- -Na H-bond is the basis for
the shifts of -27 cm-1 in locked amide 3a and -16 cm-1 in
proline 1f (relative to control amide 7a), consistent with weak
H-bonding.
Figure 7. Crystal structure of 3d (50% ellipsoids). Selected bond
distances (Å): H(10A)-N(5A) 2.35(5); H(10A)-N(10A) 0.74(5);
N(5A)-N(10A) 2.790(6). Selected bond angle (deg): N(10A)-
H(10A)-N(5) 120(4).
so does the double bond character in the endocyclic amide Cd
O.37 This correlation provides further support for a 5-NH- -Na
H-bond in these cis proline peptidomimetics.
X-ray Crystallographic Studies. Additional evidence for
the 5-NH- -Na H-bond comes from X-ray crystallography of
3d,38 which reveals (1) a distance from the side chain hydrogen
[H(10A)] to the ring Na [N(5A)] of 2.35 ( 0.05 Å, (2) an N-N
distance of 2.790 ( 0.006 Å, and (3) an N(10A)-H(10A)-
N(5A) bond angle of 120 ( 4° (Figure 7). The position of
H(10A) was refined positionally and the isotropic displacement
parameter was held at 1.2 times the Uiso of the amide nitrogen.
These bond distances and angle are such as to permit the
observed 5-NH- -Na interaction to be considered a weak H-bond
based on the criteria of Jeffrey.39 The bond of 3d also qualifies
based on sums of van der Waals radii (N and N, 3.40 Å; N and
H, 2.70 Å)40 and Etter’s criteria for bent NH- -O H-bonding in
nitroaniline crystals.41
A packing diagram of 3d is shown in Figure 8, which
indicates that along with the 5-NH- -Na interaction, the side
chain NH also engages in a bifurcated H-bonding dimer with
the side chain carbonyl of the adjacent enantiomorph. The
5-NH- -Na H-bond distance in enantiomorph A is 2.35 Å (H- -
Na distance) and the intermolecular H-bond between this
hydrogen and the adjacent proline’s O(6) is 2.334 Å (N-O
distance). In enantiomorph B, the corresponding distances are
2.37 and 2.172 Å, respectively. It is well-documented that
bifurcation tends to lengthen the primary H- -A distance in
H-bonding systems,42 and these observations suggest that the
We also synthesized the cis locked amides 3b-3f with anilide
side chains to examine how remote functional groups effect the
5-NH- -Na H-bond. As the electron-withdrawing ability of the
side chain in 3 increases from 4-(dimethylamino)phenyl to
4-nitrophenyl, the catalytic H-bond stretch steadily shifts to
lower wavenumber, consistent with an increasingly stronger
H-bond. This shift is expected because the more strongly the
para substituent withdraws electron density from the ring, the
more acidic the side chain NH becomes. Table 3 summarizes
the data on this series of locked amides, along with the values
for the NH stretch of control amides 7b-7f. As illustrated in
column 4 of the table, the difference between the NH stretches
of the locked amides 3 and control amides 7 (∆υ) increases as
the acidity of the side chain NH increases. For instance, the
(37) The IR absorption discussed here, the amide I band, is primarily a
CdO stretch. Work by Raines has established a direct correlation between
∆Gq and υCdO; see: (a) ref 24a. Additionally, a relationship between Cd
O bond length and υCdO has been found; see: (b) Bennet, A. J.; Somayaji,
V.; Brown, R. S.; Santarsiero, B. D. J. Am. Chem. Soc. 1991, 113, 7563.
(38) Colorless plates of rac-3d (C14H15BrN2O2) were grown by slow
diffusion of hexane into a solution of benzene containing the amide. H[10A]
was refined positionally and the asymmetric unit consists of two enantio-
morphs of 3d and one-half molecule of benzene. One antipode is depicted
for simplicity. Data for 3d collected at 173 K: monoclinic, space group
P21/c, with a ) 11.8158(2) Å, b ) 20.2329(3) Å, c ) 13.2733(3) Å, R )
90°, â ) 104.4400(10)°, γ ) 90°, R1 ) 0.0558, Z ) 8, GOF ) 0.926.
(39) Jeffrey, G. A. An Introduction to Hydrogen Bonding; Oxford: New
York, 1997; Chapter 2.
locked amide 3f (R ) 4-nitrophenyl) shows a ∆υ of -50 cm-1
,
the strongest 5-NH- -Na H-bond characterized in this study.
As indicated in Table 3, we also find a trend in the IR
stretching frequencies of the endocyclic amide CdO. The data
indicate that as the strength of the 5-NH- -Na H-bond increases,
(35) Pavia, A. A.; Ung-Chhun, S. N.; Durand, J.-L. J. Org. Chem. 1981,
46, 3158.
(40) Bondi, A. J. Phys. Chem. 1964, 68, 441.
(36) Gellman has established hydrogen bound vs non-hydrogen bound
amide NH IR stretching frequencies, see: Gardner, R. R.; Liang, G.-B.;
Gellman, S. H. J. Am. Chem. Soc. 1995, 117, 3280 and references therein.
(41) Panunto, T. W.; Urba´ncyk-Lipkowska, Z.; Johnson, R.; Etter, M.
C. J. Am. Chem. Soc. 1987, 109, 7786.
(42) Taylor, R.; Kennard, O. Acc. Chem. Res. 1984, 17, 320.

10666 J. Am. Chem. Soc., Vol. 120, No. 41, 1998
Cox and Lectka
Table 4. Activation Free Energies of Prolyl Carbamates in
Organic Solvents
entrya
proline
solvent
∆Gq b
a
b
c
d
1g
2g
1h
2h
CDCl3
CDCl3
CD2Cl2
CD2Cl2
15.3c
17.5d
14.7e
17.5d
a Entries a, b by 19F ST; entries c, d by 1H ST; all measurements at
10 mg/mL. b Trans-cis, (0.2 kcal/mol. c -10 °C. d 25 °C. e -25 °C.
Table 5. Activation Energies of Prolyl Carbamates in Aqueous
Solvents
entrya
proline
solvent
∆Gq b
a
b
c
d
1g
2g
1h
2h
50% D2O/MeOH
50% D2O/MeOH
50% D2O/CD3CN
50% D2O/CD3CN
16.4c
17.5c
17.0d
17.6d
a Entries a, b by 19F ST; entries c, d by 1H ST; all measurements at
10 mg/mL. b Trans-cis, (0.2 kcal/mol. c 15 °C. d 25 °C.
of carbamate 1h provides intramolecular catalysis of 2.8 kcal/
mol as calculated by eq 4 (entries c, d), whereas the analogous
amide 1b produced only 1.9 kcal/mol. These values for
intramolecular catalysis in carbamates represent a lower limit
due to the fact that there is residual intramolecular catalysis in
aqueous/organic solution for these substrates (vide infra).
Eyring analyses of substrates 1g and 1h in CD2Cl2 were
performed to determine if there was a fundamental difference
in the activation parameters for isomerization in carbamates.
In accord with previous reports on the isomerization of
carbamates in organic solvents,25 the barrier is totally enthalpic
with a negligible entropy of activation. Thus, in line with our
expectations, the carbamate nitrogen seemingly engages in a
stronger and kinetically more significant 5-NH- -Na H-bond.
We also measured ∆Gq for substrates 1g, 1h and 2g, 2h in
aqueous/organic solution, as summarized in Table 5. The fact
that 2g and 2h have identical ∆Gq values in both organic and
aqueous/organic solvents (cf. entries b and d in Tables 4 and
5) is consistent with our previously reported observations that
∆Gq in carbamates is insensitive to solvent effects.44 It is
interesting to note that in these carbamates, as opposed to the
amides, the substrates with NH side chains have lower ∆Gq
values than the isosteric controls in aqueous/organic solution.
This observation is consistent with intramolecular catalysis in
aqueous/organic solution that accounts for the difference of 1.1
and 0.6 kcal/mol between 1g and 1h and their isosteric controls
2g and 2h.45 We then examined carbamates 1i and 1j in an
aqueous/organic mixture to determine if the distant substituent’s
electronic effect could influence ∆Gq, as was observed for prolyl
amides 1b-1e in organic solvents. However, ∆Gq values were
identical for both carbamates, indicating that the electronic
differences are not large enough to exert a measurable kinetic
difference in an aqueous/organic solution.
Figure 8. Packing diagram of 3d showing bifurcated hydrogen bonding
of the catalytic hydrogen in the solid state. The benzene molecule in
the unit cell has been excluded for clarity.
5-NH- -Na H-bond we observe in the solid state may be
weakened somewhat over that which we observe kinetically and
spectroscopically in dilute chlorocarbon solution.
Intramolecular Catalysis in Prolyl Carbamates. Due to
the influence of canonical form II in eq 6, the basicity of the
nitrogen in carbamates is enhanced relative to those in corre-
sponding amides. Previous experience with copper(II)-based
Lewis acid-catalyzed amide isomerization26b led us to believe
that, owing to the increased electron density at nitrogen, the
degree of intramolecular catalysis for prolyl carbamates would
be greater than that found in the simple amide cases discussed
above. We first examined the degree of intramolecular catalysis
in prolyl carbamates 1g and 1h43 in organic solvents with respect
to reference esters 2g and 2h, and the results are included in
Table 4. The solvent for each pair was chosen so as to
maximize peak separation and to facilitate NMR analysis.
Intramolecular Catalysis in Carbon Tetrachloride. Al-
though the chlorocarbon solvents used thus far are generally
considered nonpolar, the relatively high dielectric of CH2Cl2 (ꢀ
) 9.1) and the ability of CHCl3 (ꢀ ) 4.8) to act as a hydrogen
bond donor46 suggest that we may not be observing the
(6)
(44) Cox, C.; Lectka, T. J. Org. Chem. 1998, 63, 2426.
(45) We were not able to perform Eyring analyses on 1g and 1h in
aqueous/organic solution because of the poor peak separation and limited
temperature range available. Preliminary results on other prolyl carbamates
indicate that ∆Gq in aqueous solution is entirely enthalpic, analogous to
the case of amides, in contrast to the large negative ∆Sq we observed for
acyclic tertiary carbamates.44
The data in Table 4 in conjunction with eq 4 indicate 2.2
kcal/mol of intramolecular catalysis for prolyl carbamate 1g with
an alkyl amide side chain (entries a, b), compared to the 1.4
kcal/mol seen in the related amide 1a. The anilide side chain
(43) Mukaiyama, T. Tetrahedron 1981, 37, 4111.
(46) Dado, G. P.; Gellman, S. H. J. Am. Chem. Soc. 1993, 115, 4228.

Intramolecular Catalysis of Amide Isomerization
J. Am. Chem. Soc., Vol. 120, No. 41, 1998 10667
maximum intramolecular catalysis possible. However, the poor
cis-trans ratios of prolyl amides 1a-f in less polar solvents
such as CCl4 (ꢀ ) 2.0) prohibit kinetic measurements from being
performed. Prolyl carbamates, on the other hand, have more
favorable cis-trans ratios in organic solvents,47 and we were
able to investigate carbamate 1g and control 2g in CCl4. These
prolines, which were previously studied in CDCl3 (entries a,b
in Table 4), were examined in CCl4, and the ∆Gq for isomer-
ization was determined by ST. We found that for this system,
catalysis is enhanced by 0.3 and 0.7 kcal/mol for trans-to-cis
and cis-to-trans isomerization over that observed in CDCl3. This
result indicates that the values reported for intramolecular
catalysis of prolyl amides in CDCl3 and CD2Cl2 represent a
lower limit to the amount of catalysis possible in nonpolar
solvents.
Thermodynamics of Cis Versus Trans Prolines. We felt
that experimental values for the thermodynamics of cis versus
trans proline formation would prove helpful for future analyses
of proline conformational and turn-forming properties.48 To this
end, we examined the temperature dependence of the cis-trans
ratio in the fluorinated proline 1k in tetrachloroethane solvent.
The NH region of the IR of 1k shows two stretches as expected;
however, a cause for concern is intermolecular H-bonding, the
presence of which would be hidden under the stretch for the
γ-turn at 3310 cm-1 and would significantly alter the results of
our analysis. Therefore, as a control, we studied locked amide
3a at various concentrations in tetrachloroethane and found that
below 10 mM there was no evidence of intermolecular H-
bonding. A van’t Hoff plot of the cis-trans ratios of 1k in 5
mM tetrachloroethane was constructed over a temperature range
of 90 °C.49 From the plot, we determine that the trans
conformation is favored enthalpically by 1.7 kcal/mol, but the
cis form is favored entropically by 60 cal/(mol K).
Isotope Effects: Catalysis Through Hydrogen Bonding
or Discrete N-Protonation? An alternative to catalysis through
intramolecular H-bonding would be intramolecular N-protona-
tion through intermediate zwitterion 8 (eq 7). This possibility
would seem to be unlikely based on (a) the lessened degree of
intramolecular catalysis seen in polar solvents and (b) the large
pKa difference between 1 and 8 (ca. 22-25 units). However,
an amide in the transition state to isomerization (a twisted amide)
would be expected to have a pKa closer to that of an amine; for
example, Pracejus has found a pKa of 5.3 for the Na of a twisted
amide based on a tricyclo[2.2.2]octane skeleton.50 Additionally,
Kirby has recently synthesized a highly twisted amide that was
so labile under acidic conditions as to preclude pKa determi-
nation.51 The transition state for amide isomerization should
closely resemble Kirby’s system where the lone pair is almost
perfectly orthogonal to the carbonyl π-system, so that a pKa of
5.3 seems to be a safe lower limit for Na in the transition state
for isomerization. Thus, the difference in pKa values of the
side chain amide and Na in the transition state is reduced to a
more reasonable value (<12 units). To investigate the pos-
sibility of a zwitterionic transition state, we determined the
kinetic isotope effect upon replacement of the side chain amide
proton in 1a by deuterium. For identical concentrations of 1a
and deuterated-1a in CDCl3 at 15 °C, kH/kD ) 1.00. The
absence of an isotope effect in this system supports the
H-bonding mechanism rather than discrete proton transfer to
Na.
(7)
Conclusions
We have obtained evidence, both kinetic and spectroscopic,
for the existence of a 5-NH- -Na H-bond that is capable of
catalyzing amide isomerization by up to 260-fold in model prolyl
peptides. Support for the intramolecular catalysis mechanism
came from several experiments: (1) the amount of intramo-
lecular catalysis corresponds to the acidity of the side chain
NH, as evidenced by the Hammett correlation between the
relative rate of amide isomerization and σp values; (2) extensive
IR characterization clearly establishes the existence of an
intramolecular 5-NH- -Na H-bond in cis proline peptidomimetics
and also that the strength of this bond corresponds to the acidity
of the proton donor; (3) X-ray crystallographic proof of a
5-NH- -Na H-bond in a cis proline peptidomimetic was obtained,
as well as evidence that the H-bond observed in the solid state
may be weakened over that present in dilute chlorocarbon
solutions; (4) catalysis of amide isomerization is enhanced in
prolyl carbamates relative to the analogous amides, supporting
previous evidence that the carbamate nitrogen is a better H-bond
acceptor than the amide nitrogen; and (5) deuterium isotope
effect studies indicate that intramolecular catalysis arises from
H-bonding and not from discrete N-protonation of the amide
in the transition state. In addition to these results, we also
discussed evidence that the degree of intramolecular catalysis
observed in chlorocarbon solvents such as CHCl3 and CH2Cl2
is less than that in a more nonpolar solvent such as CCl4, and
that a significant amount of intramolecular catalysis takes place
in prolyl carbamates in aqueous solution.
We have thus verified experimentally the theoretical predic-
tion by Karplus that intramolecular catalysis may be responsible
for a significant fraction of the 6.2 kcal/mol by which the FKBP
rotamase enzyme reduces ∆Gq for cis-trans isomerization of
prolyl peptides.6b These findings also support the conclusion
that a 5-NH- -Na hydrogen bond, prevalent in protein and natural
product crystal structures, may also be involved in the intramo-
lecular catalysis of protein folding as predicted by Matthews.15
Our future work will focus on studies of the 5-NH- -Na hydrogen
bond in the active site of the FKBP rotamase enzyme, as well
as providing experimental support for the hypothesis that a
charged donor can supply the catalytic hydrogen bond required
for intramolecular catalysis, as is predicted for the Arg residue
in the active site of cyclophilin.
(47) The more favorable cis-trans ratio is likely due to the decreased
importance of the γ-turn as a result of the weaker H-bond accepting ability
of the carbamate carbonyl. A theoretical study has investigated the H-bond
accepting properties of carbamates, see: Bandekar, J.; Okuzumi, Y.
THEOCHEM 1993, 281, 113.
(48) Gellman has recently performed a somewhat similar analysis where
he differentiated between the thermodynamics of â-turn and γ-turn
conformations in a proline dipeptide; see ref 19.
(49) The line obtained from the van’t Hoff analysis had the form y )
860.8x - 6.062 with R2 ) 0.986. See the Supporting Information for the
plot.
(50) Pracejus, H. Chem. Ber. 1959, 92, 988.
(51) Kirby, A. J.; Komarov, I. V.; Wothers, P. D.; Feeder, N. Angew.
Chem., Int. Ed. Engl. 1998, 37, 785.
Experimental Section
General. Reactions were carried out in oven- or flame-dried
glassware under nitrogen, unless otherwise specified. All reagents used
are commercially available from Aldrich or Sigma. Solvents were
reagent grade and purified by standard techniques. EDCI is an
abbreviation for 1-(3-dimethylaminopropyl)-3-ethylcarbodiimide hy-
drochloride, a water-soluble carbodiimide used for peptide coupling.
Reactions were magnetically stirred and monitored by thin-layer
chromatography on Macherey-Nagel 0.25 mm precoated plastic silica
gel plates (Alltech). Flash chromatography was performed with EM

10668 J. Am. Chem. Soc., Vol. 120, No. 41, 1998
Cox and Lectka
Science silica gel 60 (particle size 0.040-0.063 mm). Yields reported
are for isolated, spectroscopically pure compounds unless otherwise
indicated. No reactions were optimized for yield. NMR solvents were
used as obtained, except that CDCl3 was dried over 4 Å molecular
sieves and run through basic alumina immediately prior to use, and
CCl4 was purified by filtration through silica gel. NMR spectra were
recorded on a Varian Unity Plus 400 or 500 MHz spectrometer. Proton
and carbon chemical shifts were referenced to residual solvent peaks
or internal TMS, and fluorine shifts were referenced to external CFCl3.
NMR samples run in nondeuterated solvents were locked on a standard
of either d6-acetone or d6-DMSO in a capillary pipet inserted into the
NMR tube. Cis-trans ratios were determined by the cut-and-weigh
method, which we find to be far superior to the computer integration
method in terms of both accuracy and reproducibility (especially when
dealing with compounds having very little cis isomer). As a result of
two conformations being present in many of the reported compounds
at room temperature, the 13C NMR spectra have many more lines than
The number of scans necessary for each sample was determined by
the signal-to-noise ratio, but was usually between 16 and 32. The
reliability of our rate measurements was determined by performing the
experiments in triplicate under identical conditions, which led to
reproducibility of at least (0.1 kcal/mol for individual ∆Gq measure-
ments. Ranges of (0.2 kcal/mol are reported in the tables to allow
for systematic errors. However, since we are interested only in ∆(∆Gq),
the absolute values are not important, and we assume that any systematic
errors mostly cancel.
19F Saturation Transfer. The most important precondition for the
ST experiment is that the cis and trans resonances of the amide be
1
well separated, so that selective irradiation can take place. In the H
NMR spectra of many substituted prolines, this situation is not the case.
Other (potentially) NMR active nuclei within the prolines pose problems
of sensitivity and experimental difficulty. Especially appealing,
however, is the abundance and sensitivity of the 19F nucleus.52
Moreover, it is well-known that 19F spectra are very sensitive to small
changes in chemical environment as a result of the large spectral range,53
and fluorine atoms can be placed at remote, noninterfering positions
within our prolines. ST experiments based on 19F have been used
previously to determine the kinetics of proline isomerization in
peptides.13e,54
Eyring Plots. Eyring plots were created by careful measurement
of the rate constants of isomerization for the same sample at a minimum
of five temperatures over a range of at least 20 °C. The number of
points obtained was determined by the range over which the ST method
provided meaningful results; at the high-temperature limit, the cis
resonance is completely saturated and sinks into the baseline, and at
the low-temperature limit there is no difference in the initial and final
height of the peak. We created the plots by graphing ln(k/T) vs 103/T
(K-1) and standard equations were solved to determine ∆Hq and ∆Sq.27
The correlation of the line was usually 0.99 or better. The listed errors
in ∆Hq were determined from least-squares analysis of the Eyring plot.
Questions have been raised about the reliability of ∆Sq values from
data obtained over a limited temperature range.25 We conservatively
report our error in ∆Sq as (4 cal/(mol K), even though least-squares
analysis of the Eyring data usually indicates (2 cal/(mol K). We have
performed in excess of 30 Eyring analyses on amides and carbamates,
and in no case (other than special ones where we now expect ∆Sq to
deviate from zero44) have we found ∆Sq to deviate from zero by more
than 3.3 cal/(mol K).
1
expected, and the H NMR spectra are often very complicated. Part
1
of the complicated nature of the H spectra involves the fact that a
peak may represent a fraction of one proton, and we account for this
by reporting the integration as a decimal value. Solid and neat IR
spectra were recorded on a Perkin-Elmer 1600 FTIR spectrometer, and
solution phase IR were recorded with a CaF2 flow-cell on a Bruker
IFS-55 FTIR spectrometer. High-resolution mass spectra were provided
by the Mass Spectral Laboratory at the University of Illinois, and
elemental analyses were provided by Atlantic Microlabs, Norcross GA.
Saturation Transfer (ST) Experiments.26a We followed rates of
amide isomerization by applying a saturating decoupling pulse to a
trans resonance in the 1H or 19F NMR spectra and measuring the transfer
of saturation to the cis resonance, as well as the associated apparent
spin-lattice relaxation rate. For example, the minimum power needed
to saturate fully the trans 19F resonance in proline 1a was applied, and
the height of the cis 19F peak was used to determine the intensity, I1
(see Figure 2). Peak heights were found to be in fairly close agreement
with those from the alternative integration method, which was used to
confirm the reliability of the results. A preacquisition delay of at least
7T1 was employed to allow for complete equilibration before the 90°
observation pulse was applied.
We determined the initial intensity of the cis resonance, I0, by
measuring the peak height with off-resonance decoupling a distance
of |δcis - δtrans| on the opposite side of the cis resonance to negate any
effects due to overlap of the decoupler’s frequency bandwidth. The
apparent spin-lattice relaxation time, T1, of the cis peak was then
measured by the inversion-recovery method while the trans site was
under irradiation. Homonuclear decoupling was applied during the
preacquistion delay, and during the τ delay of the T1 measurement, but
was switched off during acquisition. Apparent T1 values were obtained
from manual plots of the data from the inversion-recovery experiment.
This method supplies results in very good agreement with those
computed by the NMR software, unless the baseline is uneven, as is
the case when we are near the fast-exchange limit of detection. In
that case, we assume that our manual method is more accurate because
of the correction that can easily be applied for the uneven baseline by
visual inspection; this assumption is validated by the excellent
reproducibility we find in our manual methods versus the nonrepro-
ducible computer-generated ones.
Isotopic Exchange. To 100 mg of 1a was added 10 mL of CH3-
COOD, and the mixture was stirred for 1 h in a glovebox under N2.
The solvent was then removed in vacuo and the process repeated 5
times. The deuterium-enriched material in the glovebox was added to
an NMR tube (that had been rinsed several times with CH3COOD and
dried under vacuum) and then dissolved in freshly distilled CDCl3. The
1
isotopic purity was judged >90% by H NMR.
Acknowledgment. T.L. thanks the NIH (R29 GM54348)
and the American Cancer Society for support of this work. C.C.
thanks the Organic Division of the American Chemical Society
for a Graduate Fellowship (sponsored by Organic Reactions,
Inc.) and JHU for a Marks Fellowship. The authors thank
Professor John Toscano for use of his FTIR spectrometer and
Dr. Victor G. Young, Jr., director of the X-ray Crystallographic
Laboratory at the University of Minnesota, for solving the
structure of 3d. We also thank Professor Alex Nickon and a
referee for providing insightful comments on the manuscript.
The rate constant for isomerization, k, was then determined by eqs
8 and 9:
Supporting Information Available: X-ray crystallographic
data for 3d, van’t Hoff plot of 1k, and synthetic details and
characterization data for all new compounds (28 pages, print/
PDF). See any current masthead page for ordering information
and Web access instructions.
k ) % ∆/T1
(8)
(9)
% ∆ ) (I0 - I1)/I0
Once the rate constants were determined, the free energies of
activation (∆Gq) for the rotational process were readily available from
the Eyring equation, and ∆Hq and ∆Sq were available from a standard
Eyring plot. All samples were allowed to equilibrate at least 15 min
after probe temperature changes, and the actual probe temperature was
determined by peak separation of an ethylene glycol or methanol
sample. Probe temperature fluctuations during the acquisition of data
were e0.1 °C.
JA9815071
1
(52) 19F is a spin /2 nucleus with 100% natural abundance and is 83%
as sensitive as 1H; see: Abraham, R. J.; Fisher, J.; Loftus, P. Introduction
to NMR Spectroscopy; Wiley: New York, 1993; Chapter 1.
(53) Sievers, R. E.; Bayer, E.; Hunziker, P. Nature 1969, 223, 179.
(54) London, R. E.; Davis, D. G.; Vavrek, R. J.; Stewart, J. M.;
Handschumacher, R. E. Biochemistry 1990, 29, 10298.