@ Copyright 1972 by the American Chemical Society
Volume 11, Number 4
Spectroscopic Studies on the Conformation of
Gramicidin A’. Proton Magnetic Resonance
Assignments, Coupling Constants, and H-D Exchange*
J. D. Glickson, D. F. Mayers, J. M. Settine, and D. W. Urryt
ABSTRACT: Analysis of proton magnetic resonances of a
commercial preparation (Nutritional Biochemicals) of grami-
cidin, referred to herein as gramicidin A’ (GA’), shows the
preparation to contain gramicidin A (GA), gramicidin B
(GB), and gramicidin C (GC) in ratios of 72:9:19, respec-
tively. These gramicidins differ only in the amino acid at
position 11. The analysis is based on the relative intensities
of tryptophan indole CH resonances, the phenylalanine
phenyl CH resonance (of GB), and the tyrosine ortho CH
resonance (of GC). Resonances in the 220-MHz proton
magnetic resonance spectrum of GA’ in hexadeuteriodimethyl
sulfoxide (MezSO-d6) are assigned to specific hydrogens. An
ordered conformation is indicated by the following evidence,
all of which is consistent with the TL,D helices previously
proposed for the structure of GA’ in transmembrane channels
(Urry, D. W. (1971)) Proc. Nut. Acad. Sci. U. S. 68, 672;
Urry, D. W., Goodall, M. C., Glickson, J. D., and Mayers,
D. F. (1971a), Proc. Nut. Acud. Sci. U. S. 68, 1907). (1) Ob-
servation of distinct resonances from side-chain protons
that are chemically equivalent except for the sequence posi-
teriostatic agents active against gram-positive bacteria and
are therapeutically administered in combination with tyroci-
din as the broad-spectrum antibiotic mixture Tyrothricin
(Hunter and Schwartz, 1967; Korzybski et ul., 1967). These
linear pentadecapeptides should not be confused with the
cyclic decapeptide gramicidin S which, in accordance with the
suggestion of Goodall (1970), we refer to as tyrocidin S.
Commerical preparations of gramicidin,’ herein referred to
as GA‘, are composed of at least four structurally related com-
he gramicidins, isolated from Bacillus brevis, are bac-
* From the Division of Molecular Biophysics, Laboratory of Molec-
ular Biology, University of Alabama Medical Center, Birmingham,
Alabama 35233. Received April 19, 1971.
t To whom to address correspondence.
Abbreviations used are: GA, GB, GC, and GD, gramicidins A, B,
C, and D, respectively,
tions of their corresponding amino acids indicates that con-
formational constraints produce magnetically nonequivalent
local environments in different segments of the polypeptide
chain. In particular, as a result of proximity to the faces of
nearby tryptophan indole rings, the methyl resonances of one
valine and three leucines are shifted by ring-current inter-
actions to high field from the methyl resonances of the re-
maining three valines and one leucine. Elimination of these
ring-current effects by hydrogenation of the tryptophan indole
rings shifts the high-field leucine and valine methyl resonances
back to their “normal” position. (2) Rates of deuterium re-
placement of labile hydrogens of GA’ in 5 % DzO-MezSO-&
(v/v) indicate that the four tryptophan indole N H protons
and the ethanolamine OH proton are exposed to the solvent,
but most of the peptide hydrogens are internally hydrogen
bonded. (3) The peptide NH-aCH coupling constants are
significantly larger than values identified with random coil
and a-helical polypeptides but are comparable to values antic-
ipated for the TL,D helices.
ponents, gramicidin A (GA), gramicidin B (GB), gramicidin C
(GC), and gramicidin D (GD), which have been separated
by countercurrent distribution (Gregory and Craig 1948;
Craig et al., 1949; Ramachandran, 1963; Gross and Witkop,
1965). We demonstrate here that the composition of GA‘
can, for the most part, be determined from its proton magnetic
resonance (pmr) spectrum.
Sarges and Witkop (1965a) showed that GA has the follow-
ing sequence: HCO-L-Val-Gly-L-Ala-D-Leu-L-Ala-D-Val-L-
CHZCH~OH. GB and GC differ only in that the L-trypto-
phan in position 11 is replaced by L-phenylalanine and L-
tyrosine, respectively (Sarges and Witkop, 1965b,c). GA,
GB, and GC each contain 5-20% of a congener in which
L-isoleucine replaces the N-terminal valine (Sarges and Witkop
1965a-c). The structure of GD has not yet been determined,
but this component is usually present in only trace quantities
B I O C H E M I S T R Y ,
VOL. 11, NO. 4, 1 9 7 2 477
U R R Y e t al.
(Ramachandran, 1963). The basic structural characteristics
of the gramicidins are a linear, electrically neutral sequence
of hydrophobic amino acids with alternating L and D con-
figurations which begin and terminate with formyl and eth-
anolamine groups, respectively.
The gramicidins, like a number of other antibiotics which
transport cations across natural and synthetic membranes,
serve as models for biological transport systems (Pressman,
1965, 1968; Chappell and Crofts, 1965; Tosteson et a/.,
1968; Goodall, 1970). There is now considerable evidence
that, under given circumstances, gramicidin forms channels
through lipid bilayer membranes (Hladky and Haydon, 1970;
Krasne et al., 1971) and that two molecules of the antibiotic
associate to form each channel (Tosteson et al., 1968; Goodall,
1970; Urry et al., 1971). As models for proteins the gramicidins
manifest many of the physical properties of their complex
analogs, e.g., aggregation and interaction with lipids and metal
It would be highly desirable to explain the pharmacological,
transport, and physical properties of the gramicidins in terms
of their molecular conformations. On the basis of infrared
evidence Sarges and Witkop (1965a) suggested an antiparallel
fi structure for the gramicidin dimer in which the molecules
are cyclically joined head to tail (formyl end to hydroxy
end). Steric interactions of bulky side chains and the presence
of three cis peptide bonds per molecule are unfavorabIe fea-
tures of that conformation. A lipophilic, left-handed helical
structure termed the XL,D helix, which is both energetically
plausible and fulfills the requirements for channel formation,
has recently been proposed (Urry, 1971; Urry et al., 1971).
Two such helices coaxially joined head to head (formyl end
to formyl end) by hydrogen bonds form a channel through
their centers which, with the aid of local ion-induced con-
formational fluctuations, can accommodate the passage of
various cations. The dimensions of this channel are consistent
with estimates of membrane thickness and with the observed
ion specificity of gramicidin: NHa+ > K+ > Na+ (Urry
With this model as a working hypothesis, we have been
pursuing various experimental approaches with the purpose
of relating the conformation of GA' in solution to the mech-
anism of transmembrane cation transport. High-frequency
pmr spectroscopy has been used in elucidating the conforma-
tions of a number of peptide antibiotics and their metal
complexes, as well as conformations of peptide hormones in
solution (Urry and Ohnishi, 1970), e.g., tyrocidin S (Stern
et al., 1968; Ohnishi and Urry, 1969), valinomycin (Ivanov
et al., 1969; Ohnishi and Urry, 1970; Mayers and Urry,
1972), nonactin (Prestgard and Chan, 1969, 1970), enniatin
(Shemyakin et al., 1959), antamanide (Ivanov et al., 1971),
oxytocin (Johnson et a!., 1969; Urry et al., 1970; Urry and
Walter, 1971), actinomycin D (Victor et al., 1969), and lysine
vasopressin (von Dreele et al., 1971).
The extent of structural and dynamic information obtain-
able from pmr spectra depends ultimately on the successful
assignment of resonances to specific hydrogens. Here we
present the assignment of resonances in the 220-MHz pmr
spectrum of GA' in Me,SO-d6 solution. Deuterium-exchange
rates are used to monitor the exposure to the solvent of labile
hydrogens, and peptide NH-aCH coupling constants yield
estimates of backbone dihedral angles. In subsequent studies
pmr spectra are used to monitor chemical modifications
of GA' (Urry et al., 1971) and the effects of solvent variations
and chemical denaturants (Urry et al., 1972) on conforma-
B I O C H E M I S T R Y , V O L . 11, N O . 4, 1 9 7 2
Spectra were recorded using a Varian Associates HR-220
spectrometer. Sample concentrations were 10 % (w/v), and
the internal reference was tetramethylsilane. The probe tem-
perature was measured to within 1 2 " using the chemical
shifts of methanol or ethylene glycol samples. Proton homo-
nuclear spin decoupling was accomplished by the field-sweep
method using a side band generated by a Hewlett Packard
Model 5103-A frequency synthesizer.
Gramicidin (Nutritional Biochemicals Corp., Cleveland,
Ohio) was used without further purification. MezSO-d6 was
purchased from Diaprep Corp., Atlanta, Ga. (99.5% D),
and from Columbia Organic Chemical Co., Columbia, S. C.
(99.8% D). Deuterium oxide (99.7% D) and acetone-ds
were purchased from Merck Sharpe and Dohme, Montreal,
Can. N-Acetylethanolamine was prepared according to
D'Allelio and Reid (1937). Hydrogenation of gramicidin
was accomplished by the procedure of Ruttenberg et al.
(1966). This reaction was monitored by the 285-nm absorption
of aliquots periodically removed from the reaction mixture.
L-Alanyl-L-alanine diketopiperazine and the (y-methyl-L.
glutamy1)dryptophan polymer were prepared by Fox Chem-
icals, Inc., Los Angeles, Calif.
Results and Discussion
Assignments. The 220-MHz pmr spectrum of GA' in
MezSO-ds appears in Figure la. A trace of trifluoroacetic
acid was added to move the water resonance away
from overlapping gramicidin resonances. Addition of the
acid sufficiently catalyzed the proton-exchange rate of the
ethanolamine OH hydrogen so that its resonance, which in
the absence of trifluoroacetic acid occurred at 1043 Hz (inset
of Figure la), fused with the water resonance. As will be dis-
cussed below, the indicated assignments were accomplished
by combined studies of chemical shifts, resonant intensities,
deuterium replacement of labile protons, and spin-decoupling
Replacement of labile hydrogens (ie., N H and OH hy-
drogens) by deuterium yielded the spectrum in Figure lb.
Exchange was accomplished by heating a 10 % (w/v) solution
of gramicidin in 20% D20-Me2sO-d6 (v/v) solution for 1
hr at 57" (some samples required 100" for complete exchange).
Since spectral changes were observed upon addition of water,
it was necessary to minimize the water content of the exchanged
sample before meaningful comparisons could be made to
spectra of unexchanged GA' in MesSO-d6. Presumably these
spectral changes resulted from aggregation of the highly hy-
drophobic antibiotic. The solution of GA' in 20% D20-Me2-
Sod6 was therefore lyophilized and the deuterated sample
was redissolved in 2 z D20-Me2SO-d6. A small amount
of DaO was necessary to limit the extent of back-exchange
of labile deuterons with inevitable traces of HzO in the solvent.
The resonances eliminated by deuterium replacement were
assigned to N H and OH protons, the latter being readily
identified by its characteristic high-field position, single proton
intensity, extreme sensitivity to temperature changes, and
spin coupling pattern [a triplet coupled to the quadruplet
at 735 Hz, J = 5.4 Hz). This resonance could only be resolved
from neighboringaCH resonances In sufficiently wet MenSO-ds
(note the relatively strong intensity of the water resonance
in Figure la). The chemical shifts of the indole NH resonances
of a series of tryptophan analogs in MeeSO-d6 at 23" are 2358
Hz for skatole, 2452 Hz for P-3-indolacetonitrile, 2388 Hz
P M R O F G R A M I C I D I N A '
a 10% Gtomicidn A ' in DMSO. d,, 2 3 ' C
n - N .
Grwrcidin d In DMSO-%. 2 3 ' C
FIGURE 1: Pmr spectra of 10% (w/v) gramicidin A' in MeZSO-& at 23". (a) Spectrum showing assignments of various resonances. A few
drops of trifluoroacetic acid were added to shift the water resonance to low field of normally overlapping peaks. The inset shows the aCH
and CHIOH resonances before addition of trifluoroacetic acid. (b) Spectrum of gramicidin A' after all its labile hydrogens have been re-
placed by deuterium. The solution contained 2% D10. (c) Spectrum, calculated by the algorithm of McDonald and Phillips (1969), of a
random coil gramicidin A in neutral DzO. Resonances of all the labile hydrogens and all the aCH protons except glycine have been omitted.
(d) Spectrum of a sample of GA' whose indoles have been hydrogenated by the procedure of Ruttenberg et al. (1966).
B I O C H E M I S T R Y , VOL. 1 1 , NO. 4, 1 9 7 2 479
U R R Y e t al.
10% Gramicidin/DMSO-d6, 38'c
Amide N H
1 4 0 0 m
70 6 5
10% Gramicidin/Ace!one-d~ DMSO-d6 ( I I), 10°C
FIGURE 2: Expansion of the 1400- to 1900-Hz portion of the spectrum of 10% gramicidin A' in Me*SO-d6 at 23" (compare to Figure la).
for N-acetyltryptophan, and 2378 Hz for N-acetyltryptophan-
amide (J. D. Glickson and W. D. Phillips, unpublished data),
whereas peptide N H resonances generally occur at con-
siderably higher field. The resonance at 2368 Hz in Figure la,
therefore, originates from the indole N H hydrogens of the
four tryptophans of GA'. The GA' indole N H absorptions
consist of two partially overlapping resonances with relative
intensities of 3 : 1. The remaining labile proton resonances
between 1685 and 1860 Hz originate from peptide N H hydro-
gens, the lone formylated NH, and the ethanolamine NH.
Figure 2 shows the 1400- to 1900-Hz region of the spectrum
of GA' in MenSO-d6 on an expanded scale (section of Figure
la). Replacement of N H hydrogens by deuterium readily
distinguished NHresonances from CH resonances (Figure 1 b).
The formyl CH hydrogen produces the sharp spike at 1775
Hz, which survives deuterium replacement. Formyl CH ab-
sorptions of various compounds had similar chemical shifts,
e.g., dimethylformamide, 1764 Hz (Bhacca et al., 1962). As-
signment of the tryptophan indole CH resonances is readily
accomplished by comparison to the spectrum of the free amino
acid (McDonald and Phillips, 1967). Each of the four trypto-
phans of GA contributes a doublet originating from the C-4
(or C-7) indole CH proton to the triplet at 1661 Hz. Two
of the doublets coincide as do the remaining two. The re-
sultant two doublets (each of two proton intensity) have one
peak in common, and, therefore, appear as an apparent 1 :2 : 1
triplet (J = 8.8 Hz). All four C-7 (or C-4) tryptophan indole
CHdoublets coincide to produce the doublet at 1605 Hz (J =
8.6 Hz). The sharp singlet at 1558 Hz originates from four
coincident C-2 indole CH singlet absorptions, and the C-5
and C-6 triplets overlap to produce the complex multiplet
centered at 1535 Hz.
A number of weak aromatic CH absorptions were associated
with GB and GC, which contain a phenylalanine and a tyro-
sine, respectively, in place of Trp-11 of GA (Sarges and
Witkop, 1965a). In disordered proteins in DzO solution the
phenylalanine phenyl absorption occurs at 1598 Hz, and the
tyrosine aromatic absorptions at 1500 Hz (ortho to OH)
and at 1560 Hz (meta to OH) (McDonald and Phillips, 1969).
B I O C H E M I S T R Y , V O L . 11, NO. 4, 1 9 7 2
Consequently, it is reasonable to assign the 1681-Hz absorp-
tion of GA' to the GB phenyl hydrogens, and the 1451-Hz
doublet to the ortho CH hydrogens of the GC tyrosine. The
meta CH tyrosine doublet overlaps with tryptophan resonances
of C-5 and C-6 indole hydrogens (the ortho and meta hydro-
gens of tyrosine are expected to yield an AzBz pair of doublets).
From the relative intensities of the tryptophan, phenylalanine,
and tyrosine aromatic CH absorptions, we estimate that GA'
is composed of 72 % GA, 9 % GB, and 19 % GC. Other im-
purities appear to be present in negligible amounts (Rama-
chandran, 1963; Gross and Witkop, 1965) as may be seen
from further assignments of resonances (see below).
Using the algorithm of McDonald and Phillips (1969),
we computed the spectrum of the side-chain absorptions
of a hypothetical disordered GA in D20 (Figure IC). (GA'
is not soluble enough in water to measure its spectrum).
Table I compares the chemical shifts of side-chainhydrogens in
MezSO-d6 and DzO. The data in MezS0-d~ were obtained from
various compounds and may reflect effects of conformation
as well as solvent on chemical shifts. Within about 20 Hz,
the leucine, valine, and aromatic tryptophan CH chemical
shifts in MezSO-d6 agree with the values obtained in DtO.
More pronounced differences are observed for alanyl CH3,
glycyl CH2, and tryptophan CHz resonant frequencies. Despite
these differences, the spectrum in Figure IC serves as a useful
starting point for the assignment of aliphatic CH resonances.
Chemical shifts and line widths of formyl CH and ethanol-
amine CH2 absorptions were obtained from spectra of grami-
cidin and N-acetylethanolamine, respectively.
Homonuclear proton decoupling was very helpful in the
assignment of a number of resonances. Most of the decoupling
experiments were performed at 84", at which temperature
more distinct hyperfine structure is observed. A continuous,
but small, change in chemical shift and coupling constant
was observed as the temperature was raised. These changes
may reflect a thermally induced conformational change that
is rapid on the pmr time scale, i.e., the rates of transition
are much greater than the chemical-shift differences of reso-
nances of a given proton in the various conformations. Con-
P M R O F G R A M I C I D I N A ‘
sequently, it was possible to perform decoupling experiments
at higher temperatures and then to correlate coupled reso-
nances with their positions in the 23 O spectrum.
The leucine and valine methyl resonances are split into
two absorptions, each of 24 proton intensity (Figure la).
Spin decoupling from valine PCH resonances indicated that
three valines and one leucine contributed resonances to the
lower field peak at 180 Hz, which is associated with methyl
groups in a predominantly solvated environment because
of its coincidence with the “disordered” absorption (Figure
IC). The remaining leucine and valine methyl resonances
are high field shifted to 130 Hz by ring-current interactions
resulting from the close proximity of methyl groups to the
faces of indole rings. Hydrogenation of the gramicidin indoles
eliminates these ring currents and shifts the methyl resonances
to their “normal” position (Figure Id).
The presence of similar high-field ring-current-shifted
methyl resonances in pmr spectra of proteins has been as-
cribed to the close proximity of methyl groups and aromatic
side chains in the folded native conformation (McDonald
and Phillips, 1967; McDonald et al., 1971). Denaturation
of the protein exposes both the aromatic and methyl reso-
nances to an essentially equivalent solvated environment,
thereby eliminating these ring-current shifts. Because of their
extreme dependence on the relative orientation of methyl
and aromatic side chains (Johnson and Bovey, 1958), these
ring current shifts are valuable parameters for detecting
even very subtle conformational changes and for distinguish-
ing between disordered and ordered conformations. Observa-
tion of these high-field methyl ring-current shifts in spectra
of GA’ indicates that the molecules must be oriented in such a
manner as to bring the methyl groups of three leucines and
one valine in close proximity to the faces of tryptophan indole
rings. It is interesting to note that in the proposed helical
transmembrane conformation of GA (Urry, 1971 ; Urry
er al., 1971) the methyl groups of Val-8 and Leu-10, -12,
and -14 are indeed in very close proximity to the faces of
the indole rings of Trp-9, -11, -13, and -15, respectively
Both alanine residues exhibit magnetically equivalent methyl
and aCH hydrogens. Their methyl resonance (Figure 2a)
is the distinct doublet at 226 Hz (J = 3.4 Hz) which is coupled
to an CYCH 1:3:3:1 quartet at 936 f 2 Hz. Even though
the alanine aCH resonance is buried beneath a number of
other CYCH absorptions, the characteristic spin-coupling pat-
tern of this multiplet could be discerned on an expanded
scale once its chemical shifts and coupling constant had been
ascertained by spin decoupling.
Since the leucine PCH hydrogens are magnetically nonequi-
valent, the two PCH hydrogens of a given residue may yield
distinct resonances. The four proton resonances at 335 Hz
(Figure la) originated from four PCH hydrogens of either
two or four distinct leucine residues. This assignment was
supported by the chemical shift of this resonance (compare
to the “disordered” spectrum in Figure IC) and by its cou-
pling to CYCH resonances at 927 i 5 Hz (spin decoupling
was performed at 85”, at which temperature the 335-Hz
resonance was a quartet, J = 5.8 Hz-actually
of doublets due to &Hand YCHcoupling). The broad back-
ground upon which the alanine methyl doublet appears is
composed of yCH and PCH absorptions.
The assignment of the high-field ring-current-shifted methyl
resonances to one valine and three leucines is based on spin-
decoupling experiments performed on the two valine PCH
resonances at 449 and 411 Hz (relative intensities three and
TABLE I: Chemical Shifts of Amino Acid Residue Side-Chain
CHProton Resonances in MezSO-ds and DzO.
A 1 any 1 *
Chemical Shift (Hz) in
DMSO-ds Protein D20~
C-4 or -7
C-7 or -4
C-5 or -6
C-6 or -5
= Chemical shifts of random coil proteins in neutral DzO
(McDonald and Phillips, 1969). * From L-alanyl-L-alanine
diketopiperazine (Fox Chemicals, Inc.).
(Johnson et al., 1969). From tyrocidin S (Stern e? al., 1968).
e From (y-methyl-~-glutamyl)~tryptophan polymer (Fox
Chemicals, Inc.). f From valinomycin (Ohnishi and Urry,
c From oxytocin
one, respectively) in Figure la. Both resonances appear in
the region of the spectrum where valine PCH absorptions
are anticipated (compare to Figure IC). Furthermore, the
449-Hz three proton peak couples to an aCH at 930 f 5 Hz
and to the methyl resonance at 177 f 5 Hz (the nonring-
current-shifted methyl), and the 411-Hz single proton valine
PCH couples to an aCH at 911 f 10 Hz and to the ring-
current-shifted methyl absorption at 130 i 5 Hz. If we as-
sociate the 449-H and 177-Hz peaks with unperturbed PCH
and methyl resonances of valines, respectively, then the ring-
current shifts experienced by one of the four valines are
about 38 and 50 Hz for PCH and methyl hydrogens, respec-
tively. This suggests that the valine methyl group falls some-
what closer to the center and plane of the proximal indole
ring then does the PCH hydrogen.
The glycine CH2 resonance at 817 Hz in Figure la was
readily identified by its characteristically high-field chemical
shift for an aCH (Figure IC) and by its structure (a doublet
at 85”, J = 5.1 Hz, but a sharp singlet after deuterium ex-
change). At lower temperatures the glycine CH2 resonance
is too broad and complex to readily yield a coupling constant.
Consequently, the 5.1-Hz coupling constant may not be as-
sociated with the most stable conformation. The 758432
two-proton resonance, which coupled to the OH at 1043
Hz, emanates from the ethanolamine CHzOH protons. The
remaining ethanolamine NHCHz methylene resonance con-
tributes two protons to the six-proton resonance centered
at 700 Hz. This assignment is supported by the spectrum of
N-acetylethanolamine in MeSO-d6 in which the N-methylene
occurred 35 Hz to high field of the 0-methylene resonance.
An alternate assignment of the N-methylene to the four
proton resonance at 643 Hz was ruled out by failure of this
resonance to couple to the 0-methylene resonance (the 700-
Hz resonance was too close to the 0-methylene resonance
B I O C H E M I S T R Y , V O L . 11, NO. 4, 1 9 7 2 481
U R R Y e t al.
1 0 % Deformyl Grmicidm A'/Acetmt-d6.DMSO-~ (Ill), I0.C
I N O
FIGURE 3: The low-field region of 220-MHz pmr spectra of (a) 10% GA' (w/v) and (b) 10% deformyl-GA' (w/v) both in MezSO-ds-aCetOne
to permit spin-decoupling experiments). The 643- and
700-Hz resonances must be associated with tryptophan ab-
sorptions, either two PCHZ absorptions per peak or four
PCH absorptions per peak (the PCH hydrogens are magneti-
cally nonequivalent). Consistent with this assignment, the
four proton resonance at 997 Hz, which is associated with
the tryptophan aCH hydrogens, coupled to both tryptophan
TABLE 11: Chemical Shifts of aCH Resonances of Gramicidin
A' in MezSO-d6.
aCH Chemical Shift at:
Residue 23" (Hz)
1 Valine + 3 leucines
ring current shifted)
3 Valines + 1 leucine
not ring current
930 j = 5
@CH2 resonances at 643 and 700 Hz. The remaining absorp-
tions can now be assigned to four leucines, four valines,
and two alanine aCH hydrogens.
The assignment of aCH resonances is crucial to subsequent
spin-decoupling experiments involving peptide NH reso-
nances. Assignment of the 643- and 700-Hz resonances to
tryptophan PCHz hydrogens indicates that the 997-Hz reso-
nance to which they are coupled originates from the four
tryptophan aCH hydrogens. The aCH absorptions of (y-
methyl-L-glutamyl)3tryptophan polymer (Me2SO-d6, 23 ") glu-
tamic acid and tryptophan residues occurred at 935 and
990 Hz, respectively. An anomalous low-field position may,
therefore, be characteristic not only of the PCH2 resonances
but also of the aCH resonances of tryptophan. This is alto-
gether reasonable since both PCH, and aCH hydrogens are
close to the edges of indole rings, where the ring-current
field is expected to produce a low-field shift. Hydrogenation
of these indole rings eliminates the ring-current effects and
causes all the aCH absorptions, except that of glycine, to
collapse into the 937-Hz peak (Figure Id). This lends further
support to the assignment of tryptophan aCH hydrogens and
suggests that unless hydrogenation was accompanied by a
occur (Urry el al., 1972Fthe resonances previously at 911 Hz
must have originated from aCH hydrogens that were ring
current shifted to high field. Spin-decoupling experiments
indicate that one of these aCH resonances originates from a
infrared data suggest did not
B I O C H E M I S T R Y , V O L . 11, NO. 4, 1 9 7 2
P M R O F G R A M I C I D I N A '
b 90% ACETON-de
d 70% ACETWE-ds
e 50% ACETCNE-d6
f 375% ACETONE-d6
g 25% ACETONE-%
h 125% ACETaUE-dg
FIGURE 4: 220-MHz pmr spectra of 10% deformyl-GA' (w/v) at 10" in mixtures of acetonedo and Me2SO-de.
valine whose methyl resonance is ring current shifted to
high field. It is, therefore, reasonable to assign the remaining
aCH hydrogens contributing to the 911-Hz absorption to
three leucines whose methyl resonances are also ring current
shifted to high field. The complete assignment of aCH ab-
sorptions is summarized in Table 11.
Coupling Constants. Theoretically, the vicinal proton-pro-
ton-coupling constant J depends in the following manner
on the dihedral angle 8, which is defined by these protons
and the two atoms to which they are joined
J = A cos2 8 + B cos 8 + C sin2 0
B I O C H E M I S T R Y , VOL. 11, NO. 4, 1 9 7 2 483
U R R Y e t al.
TABLE III: Peptide NH-aCH Coupling Constants in Acetone4
MezSO-ds(1 :l,V/V), 10".
gramicidin A '
T L , D ~
973 f 5
954 f 5
967 f 5
930 f 5
960 f 5
1005 i 5
Q An a helix of alternating L and D residues would be ex-
pected to exhibit about half of its resonances with J = 2.4 Hz
and the other half with J = 6.0 Hz. This possibility appears
to be readily excluded even with the uncertainties which
presently must be associated with an empirical expression
for J. b From Tonelli and Bovey (1970).
where A, B, and C are constants which depend on electroneg-
ativities of adjoining atoms, their states of hybridization,
bond lengths, and bond angles (Karplus, 1959, 1963; Barfield
and Grant, 1965). While these constants have been calculated
for ethane, ethylene, and fluoroethylene (Karplus, 1959), it is
best for experimental purposes to empirically determine their
values from spectra of molecules similar to the system being
studied and having a well defined conformation. Such an
approach has proven particularly successful in studies of
organic compounds (Bothner-By, 1965) and has recently
been applied to the aCHNH fragment of peptides (Stern
et al., 1968; Bystrov et al., 1969; Ramachandran et al.,
1972).* In the present study, we use the most current values-
A = 7.9, B = -1.55, and C = 1.35 (Ramachandran et al.,
1972)-in applying eq 1 to the NH-aCHfragment of peptides.
Because the values of these constants may require further
refinement, dihedral angles determined by eq 1 are recognized
to be only approximate.
Overlap of N H resonances in the pmr spectrum of GA'
in MezSO-d6 (Figure 2) hampered reliable estimation of COU-
pling constants. Lowering the temperature, which necessitated
the addition of acetone to lower the solvent freezing point,
permitted resolution of a number of doublets which were
decoupled from aCH hydrogens by double resonance (Fig-
ure 3). Optimum resolution was obtained at 10"; at lower
temperatures resonances broadened out, presumably because
of aggregation and/or increased viscosity of the sample. Be-
We are indebted to I<. D. Kopple for sending us a preprint of this
manuscript for use before publication.
cause spectra of deformyl-GA' were somewhat better resolved
in the NH absorption region than spectra of GA', and be-
cause spectral similarity of these compounds attests to their
conformational resemblance, most of the coupling constants
were obtained from the deformyl derivative.
The effect of acetone on the pmr spectrum of deformyl-GA'
is illustrated in Figure 4. The broadness of resonances in the
pure acetone spectrum together with the observation that
acetone precipitates GA' suggests that this solvent promotes
extensive association of GA' and deformyl-GA'. A gradual
transition complete by about 40% Me?SO-ds occurs as acetone
is enriched with Me2sO-d~. Consequently, coupling constants
measured in acetone-d6-MezSO-de mixtures containing at
least 50 % MezSO-d6 (v/v) reasonably reflect the MezSO-ds
conformation. Acetone-d6 promotes the exchange of labile
hydrogens. The diminution of tryptophan indole N H reso-
nances is particularly apparent, even in spectra obtained
immediately after solution of the sample; exchange of peptide
hydrogens is considerably slower. This exchange reaction
may occur as a. result of proton exchange with water impurities
in the solvent, Le.
2CD3COCD3 + H20
D,O + NH If HDO + ND
2CDaCOCHDn + D20
Peptide NH-aCH coupling constants and frequencies of
corresponding aCH resonances, as determined by double
resonance, and estimated dihedral angles, are summarized
in Table 111. Rough association of coupling constants with
specific amino acids can be accomplished by referring to
Table 111. The dihedral angles in Table 111 fall within the
range in which eq 1 is single valued (J = >6.4; 140" < 0 5
180"). Consequently, there is no ambiguity in 0. Obviously
the coupling constants could be associated not with a unique
conformation, but could be averaged over a number of rapidly
interconverting conformations. Furthermore, only about one-
fourth of all the aCH-NH coupling constants could be
determined. Despite these reservations, and despite the re-
cognized uncertainty in the dihedral angle estimates, the
coupling constants appear to be well outside the range as-
sociated with a-helical and random coil polypeptides. Their
large magnitude is, however, consistent with coupling con-
stants anticipated for the T L , D ~ and n u 6 helices (T'able 111).
Deuterium-Exchange Kinetics. Pmr spectroscopy is uniquely
capable of monitoring the rates of deuterium replacement
of specific labile hydrogens of complex macromolecules when
the resonances of these hydrogens are well resolved. In this
manner the state of hydrogen bonding of the various peptide
NH hydrogens of tyrocidin S (Stern et al., 1968) and the
tryptophan indole NH hydrogens of lysozyme (Glickson
et al., 1969; 1971) has been determined. When more poorly
resolved spectra are obtained, the envelope of overlapping
resonances of labile hydrogens offers a means of nonspecifically
monitoring the number of exchanged hydrogens. Wishnia
and Saunders (1962) used this method to follow the exchange
of peptide NH hydrogens of ribonuclease.
We have employed a combination of these techniques to
study the accessibility to the solvent of the replaceable protons
of GA' and two of its derivatives-deformyl-GA'
malonyl-GA' (Mal-GA') which consists of two deformyl-GA'
molecules joined amino end to amino end by a malonamide
bridge (Urry et al., 1971). The temperature for the exchange
reaction was adjusted to produce measurable rates. Replace-
ment of OH hydrogens, as indicated by the complete absence
B I O C H E M I S T R Y , V O L . 11, NO. 4, 1 9 7 2
P M R O F G R A M I C I D I N A '
of the OH resonance, was complete within 16 min, the time
required to measure the first spectrum of all three compounds.
Consequently, their OH protons appear to be exposed to the
Overlap of peptide N H resonances permitted only nonspe-
cific monitoring of the exchange rates of all the peptide hy-
drogens (Figure 4 ) . Comparison of the intensities of the peptide
N H and tryptophan indole N H resonances to the intensity
of the tryptophan indole C H doublet (Figure 2) was used
to determine the number of nonexchanged hydrogens. In
a separate experiment the exchange of GA' was studied under
conditions identical with those used to obtain the data in
Figure 4, except that a fivefold molar excess of alanylalanine
diketopiperazine was included in the sample. By the time
the first spectrum was recorded (20 min after addition of DzO)
all the diketopiperazine peptide hydrogens had exchanged,
whereas hardly any of the peptide hydrogens and tryptophan
indole N H hydrogens of GA' had been replaced. Conse-
quently, if diketopiperazines are suitable models for solvated
peptides, then the much slower rate of exchange of most of
the peptide hydrogens of GA' and its derivatives reflects a
conformation which conceals most of the peptide hydrogens
from the solvent. Hydrogen bonding, indicated by the chemical
shifts of peptide "resonances,
responsible for this concealment of peptide protons. Chen
and Swenson's (1969) observation that cis and trans amide
N H hydrogens exchange at similar rates suggests that diketo-
piperazines whose amide hydrogens assume the cis configura-
tion may have similar exchange properties to disordered
polypeptides, whose amides are trans. For comparison of
exchange rates it was necessary to keep the GA' and reference
compound in the same solution, because exchange rates
in Me2S0 were not very reproducible (probably as a result
of variable trace amounts of catalytic impurities in the solvent
It is significant that all the hydrogens are replaceable by
deuterium. While this is not indicated in Figure 4, it was as-
certained that all the reactions there depicted did go to comple-
tion. Consequently, GA' cannot exist in a stable rigid con-
formation in which a significant number of its labile hydrogens
are inaccessible to the solvent. A plausible structural model
for this polypeptide must allow for disruptions of ordered
structure which expose peptide hydrogens to the solvent. A
rapid order disorder equilibrium heavily favoring the
ordered conformations would be consistent with the pmr
data. Alternately, the molecule may exist in a stable ordered
conformation, such as one of the RL,D helices, but local
disruptions of ordered structure associated with small rota-
tions about a few contiguous bonds could occur. In this regard
it is interesting to note that the proposed TL.D helical con-
formation allows for considerable flexibility about the back-
bone bonds (the dihedral angles specifying this structure, + 1 :
270°, 4 'v 60", for the T L , D ~ helix, occur in a low energy
basin of the potential energy-dihedral angle plot (Gibson
and Scheraga, 1966)). Tilting of the planes of the amide bonds,
such as is required to permit coordination of carbonyl oxygens
to the cation in the channel, would tend to expose the peptide
hydrogens to the solvent. Consequently, the exchangeability
of peptide hydrogens is entirely consistent with the TL,D
Exchange of the four tryptophan indole N H hydrogens
of GA' followed first-order kinetics (k = 1.88 X 10-24min-1
at 38") (see Figure 5). Kinetic equivalence of these protons
indicates approximately equal accessibility to the solvent.
In a separate experiment, the exchange rates of the indole
appears to be at least partially
A' AND DERIVATIVES
A MALONYL GA'
FIGURE 5: Kinetics of peptide proton exchange in 5
(v/v) of 2.0% (w/v) GA', Mal-GA', and deformyl-GA'. The
exchange of tryptophan indole NH hydrogens of the GA' sample
is also included. The logarithm of the number of unexchanged N H
protons (12) is plotted as a function of time (1).
NH hydrogens of the GA' tryptophans and of indole, both
contained in the same solution, were 1.78 X
min-I, respectively (2 z
MezSO-dE, 26"). The similarity of the rates of indole NH
proton exchange of GA' tryptophans and indole indicates
that the four tryptophan indole N H hydrogens are exposed
to the solvent. This is consistent with both the aL,D4and R L , D ~
conformations, which have their indole rings on the outside
of the helix.
and 3.81 X
GA', 0.8 z
indole, 5 z
The authors thank William J. Wakefield for technical
assistance in obtaining nuclear magnetic resonance spectra, B.
Starcher for carrying out the amino acid analysis, and the
Mental Health Board of Alabama for financial support.
Barfield, M., and Grant, D. M. (1965), Aduan. Magn. Res.
I , 149.
Bhacca, N. S., Johnson, L. F., and Shoolery, J. N. (1962),
NMR Spectra Catalog, Vol. 1, Palo Alto, Calif., National
Press, Varian Associates, Spectrum No. 39.
Bothner-By, A. (1965), Aduan. Magn. Res. I , 195.
Bystrov, V. F., Portnova, S. L., Tsetlin, V. I., Ivanov, V. T.,
and Ovchinnikov, Y . A. (1969), Tetrahedron 25,493.
Chappell, J. B., and Crofts, A. R. (1965), Biochem. J. 95,393.
Chen, C. Y. S., and Swenson, C. A. (1969), J. Amer. Chem.
Craig, L. C., Gregory, J. D., and Barry, G. T. (1949), Cold
Spring Harbor Symp. Quant. Biol. 14,24.
D'Allelio, G. F., and Reid, E. E. (1937), J. Amer. Chem. SOC.
Gibson, K. D., and Scheraga, H. A. (1966), Biopolymers 4,
Glickson, J. D., McDonald, C. C., and Phillips, W. D. (1969),
Biochem. Biophys. Res. Commun. 35,492.
B I O C H E M I S T R Y ,
V O L . 11, NO. 4, 1 9 7 2 485
U R R Y e t al. Download full-text
Glickson, J. D., Phillips, W. D., and Rupley, J. A. (1971),
J. Amer. Chem. Sac. 93,4031.
Goodall, M. C. (1970), Biochim. Biophys. Acta219,471.
Gregory, J. D., and Craig, L. C. (1948), J. Biol. Chem. 172,
Gross, E., and Witkop, B. (1965), Biochemistry 4,2491.
Hladky, S. B., and Haydon, D. A. (1970), Nature (London)
Hunter, F. E., and Schwartz, L. S. (1967), in Antibiotics,
Mechanism of Action, Vol. 1, Gottlieb, D., and Shaw,
P. D., Ed., New York, N. Y., Springer, p 648.
Ivanov, V. T., Laine, I. A., Abdulaev, N. D., Senyavina,
L. B., Popou, E. M., Ovcbinnikov, Yu. A., gnd Shemyakin,
M. M. (1969), Biochem. Biophys. Res. Commun. 34,803.
Ivanov, V. T., Miroshnikov, A. I., Abdulqev, N. D., Senya-
vina, L. B., Arshipova, S. F., Uvarova, N. N., Khalilulina,
K. Kh., Bystrov, V. F., and Ovchinnikov, Y. A. (1971),
Biophem. Biqphys. Res. Commun. 42,654.
Johnson, C. E., and Bovey, F. A. (1958), J. Chem. Phys. 29,
Johnson, L. F., Schwartz, I. L., and Walter, R. (1969),
Proc. Nat. Acad. Sci. U. S. 64,1269.
Karplus, M. (1959), J. Chem. Phys. 30,ll.
Karplus, M. (1963), J. Amer.Chem. SOC. 85,2870.
Korzybski, T., Kowsyzk-Gindifer, Z., and Kurylowicz, W.
(1967), Antibiotics, Origin, Nature and Properties, Paryski,
E., Translator, Oxford, Pergamon Press, p 60.
Krasne, S., Eisenman, G., and Szabo, G. (1971), Science
Mayers, D. F., and Urry, D. W. (1972), J. Amer. Chem. Sac.
McDonald, C. C., and Phillips, W. D. (1967), J. Amer. Chem.
McDonald, C. C., and Phillips, W. D. (1969), J. Amer. Chem.
SOC. 91, 1513.
McDonald, C. C.,
Phillips, W. D., and Glickson, J. D. (1971),
J. Amer. Chem. SOC. 93,235.
Ohnishi, M., and Urry, D. W. (1969), Biochem. Biophys. Res.
Ohnishi, M., and Urry, D. W. (1970), Science 168,109.
Pressman, B. C. (1965), Proc. Nut. Acad. Sci. U. S. 53,1077.
Pressman, B. C. (1968), Fed. Proc., Fed. Amer. SOC. Exp. Biol.
Prestgard, J. H., and Chan, S. I. (1969), Biochemistry 8,
Prestgard, J. H., and Chan, S. I. (1970), J. Amer. Chem. SOC.
Ramachandran, G. N., Chandrasekaran, R., and Kopple,
K. D. (1971), Biopolymers 10,2113.
Ramachandran, L. K. (1963), Biochemistry 2,1138.
Ruttenberg, M. A,, King, T. P., and Craig, L. C. (1966),
Sarges, R., and Witkop, B. (1965a), J. Amer. Chem. SOC. 87,
Sarges, R., and Witkop, B. (1965b), J. Amer. Chem. SOC. 87,
Shemyakin, M. M., Ovchinnikov, Y. A., Ivanov, V. T.,
Anotonov, V. K., Vinogradova, E. I., Shkrob, A. M.,
Malenkov, G. G., Evstratov, A. V., Laine, I. A,, Melnik,
E. I., and Ryabova, I. D. (1969), J. Membrane Biol. I , 402.
Stern, A., Gibbons, W. A., and Craig, L. C. (1968), Proc.
Nat. Acad. Sci. U. S. 61,734.
Tonelli, A. E., and Bovey, F. A. (1970), Macromolecules, 3,
Tosteson, D. C., Andreoli, T. E., Tiffenberg, M., and Cook,
P. (1968), J. Gen. Physiol. 51, 3738.
Urry, D. W. (1971)) Proc. Nut. Acad. Sci. U. S. 68,672.
Urry, D. W.
(1972a), Ann. N. Y. Acad. Sci. (in press).
Urry, D. W. (1972b), Biochim. Biophys. Acta Biomembrane
Rev. 1 (in press).
Urry, D. W., Glickson, J. D., Mayers, D. F., and Haider, J.
(1972), Biochemistry 11,487.
Urry, D. W., Goodall, M. C., Glickson, J. D., and Mayers,
D. F. (1971), Proc. Nat. Acad. Sei. U. S. 68,1907.
Urry, D. W., apd Ohnishi, M. (1970), in Spectroscopic
Approaches to Biomolecular Conformation, Urry, D. w.,
Ed., Chicago, Ill., American Medical Association Press,
Urry, D. W., Ohnishi, M., and Walter, R. (1970), Prac. Nut.
Acad. Sci. U. S. 66, 11 1.
Urry, D. W., and Walter, R. (1971), Proc. Nat. Acpd. Sci.
U. S. 68,956.
Victor, T. A,, Hruska, F. E., Bell, C. L., and Danyluk, 8. S.
(1 969), Tetrahedron Lett. 53,4721.
von Dreele, P. H., Brewster, A. I., Scheraga, H. A., Ferger,
M. F., and du Vigneaud, ; I .
U. S. 68,1028.
Wishnia, A., and Saunders, M. (1962), J. Amer. Chem. SOC.
and Witkop, B. (1965c), Biochemistry 4,2491.
(1971), Proc. Nut. Acad. Sci.
B I O C H E M I S T R Y , V O L . 11, NO. 4, 1 9 7 2