Proc. Natil. Acad. Sci. USAVol. 85, pp. 1384-1388, March 1988Biochemistry

Solid-phase peptide synthesis and solid-state NMR spectroscopy of[Ala3-_5N][Val']gramicidin A

(gramicidin A/oriented lipid bilayer)

GREGG B. FIELDS, CYNTHIA G. FIELDS, JONATHAN PETEFISH, HAROLD E. VAN WART, AND T. A. CROSS*Department of Chemistry and Institute of Molecular Biophysics, Florida State University, Tallahassee, FL 32306-3006

Communicated by Michael Kasha, October 26, 1987

ABSTRACT [Ala3-5NJ[Val']Gramicidin A has been pre-pared by solid-phase peptide synthesis and studied by solid-state '5N nuclear magnetic resonance spectroscopy. The syn-thesis of desformyl[Ala3-'5N][Val'lgramicidin A employed N-hydroxysuccinimide esters of 9-fluorenylmethoxycarbonyl-N-amino acids and completely avoided the use of acid. Sincedeblocking was done with piperidine and the peptide wasremoved from the resin by treatment with ethanolamine, thissynthetic protocol prevented oxidation of the indole rings ofthis tryptophan-rich peptide and reduced truncations pro-duced by acid hydrolysis. After formylation and purificationby anion-exchange and high-pressure liquid chromatography,the peptide was obtained in an overall yield of 30%. Solid-state'5N nuclear magnetic resonance spectra of this peptide anduniformly labeled ['5N]gramicidin A' oriented in hydratedlipid bilayers have been obtained, allowing unambiguousassignment of the ['5N]Ala3 resonance in the latter. The solid-state '5N nuclear magnetic resonance experiments provideevidence that [Val']gramicidin A is rotating about an axis thatis perpendicular to the plane of the lipid bilayer and that theN-H axis is nearly parallel with the rotational axis. Thisstudy demonstrates that site-specifically labeled ['5Njgramici-din A analogs prepared by solid-phase peptide synthesis arevaluable tools in the study of the solid-state nuclear magneticresonance spectra of samples in oriented lipid bilayers.

Gramicidin A' is a mixture of linear pentadecapeptides thatis produced by Bacillus brevis during the early stages ofsporulation. The mixture is composed of gramicidins A, B,and C, which have Trp, Phe, and Tyr, respectively, atposition 11. The residue at position 1 can be Val or Ile foreach of these gramicidins. The most abundant of the grami-cidins is [Val1]gramicidin A, an extremely hydrophobicpeptide that contains no formal charges:

1 5HCO-Val-Gly-Ala-D-Leu-Ala-D-Val-Val-D-Val-Trp-

10 15D-Leu-Trp-D-Leu-Trp-D-Leu-Trp-NHCH2CH20H

A gramicidin dimer forms a monovalent-cation-selectivechannel in membranes.The structural and dynamic properties of these gramici-

dins that enable them to serve as ion channels are not wellunderstood. A model for the conformation of gramicidin A ina lipid bilayer has been proposed on the basis of conforma-tional analysis; in the model the peptide backbone has ahelical structure with a central pore through which a mono-valent cation can pass (1). The structure of gramicidin A,however, has not been determined experimentally by anatomic-resolution structural method. Crystallographic anal-

ysis has not been achieved for any gramicidin in a lipidenvironment because of the difficulty of forming crystalsthat diffract to high resolution (2). The dynamics of theinteraction of gramicidin A with cations is also of greatinterest. The cations are largely stripped of their hydrationsphere upon entry into the channel and presumably havetheir coordination sphere occupied by the carbonyl oxygenatoms of the peptide groups. Molecular dynamics calcula-tions indicate that this process may require a rotation of thepeptide bonds of 5-50° toward the center of the pore (3, 4).Solid-state NMR spectroscopy is being used in this labora-tory to provide information at atomic resolution about thedynamic and structural aspects of this important biologicalsystem.The NMR spectra of proteins contain a wealth of potential

structural and dynamic information. To extract this informa-tion, however, one must be able to observe resonances thatarise from single atoms in a uniform environment. The NMRspectra of most proteins are complex, and often this criterionis not met. This is particularly true for solid-state NMRspectroscopy, which has inherently lower resolution. Thus,it is essential to devise means to observe and study single-site resonances. In this study, the preparation of the site-specifically labeled [Ala3-15N][Vall]gramicidin A by solid-phase peptide synthesis is described and its use in solid-state15N NMR studies of samples in oriented lipid bilayers isdemonstrated.

MATERIALS AND METHODSMaterials. Tryptophan with the amino group protected by

9-fluorenylmethoxycarbonyl (Fmoc) and p-alkyloxybenzylalcohol resin were obtained from Bachem Fine Chemicals(Torrance, CA). All other Fmoc-amino acids were pur-chased from Cambridge Research Biochemicals (AtlanticBeach, NY). ['5N]Alanine (99% isotope purity) was pur-chased from Cambridge Isotope Laboratories (Woburn,MA). 1,3-Dicyclohexylcarbodiimide, 4-dimethylaminopyri-dine, trifluoroacetic acid, piperidine and 9-fluorenyl-methylchloroformate were purchased from Aldrich. N-Hydroxysuccinimide, diisopropylethylamine, and 1,2-dithioethane were obtained from Sigma. AG 50W-X2 resinwas obtained from Bio-Rad, HPLC-grade methanol wasfrom American Scientific Products (Stone Mountain, GA),and constant-boiling HCI was from Pierce. Gramicidin A'was purchased from Sigma, and uniformly labeled ["NI-gramicidin A' was prepared biosynthetically as describedearlier (5).

Abbreviations: [Ala3-15N1[Val']gramicidin A, valine gramicidin Awith [15N]Ala at residue 3; Fmoc, 9-fluorenylmethoxycarbonyl;Fmoc-atnino acid denotes N-Fmoc; DMF, N,N-dimethylform-amide; CSA, chemical shift anisotropy; Boc, tert-butoxycarbonyl;all amino acids are of the L configuration unless otherwise noted.*To whom reprint requests should be addressed.

1384

The publication costs of this article were defrayed in part by page chargepayment. This article must therefore be hereby marked "advertisement"in accordance with 18 U.S.C. §1734 solely to indicate this fact.

Dow

nloa

ded

by g

uest

on

Apr

il 10

, 202

1

Proc. Natl. Acad. Sci. USA 85 (1988) 1385

Synthetic Methods. Fmoc-fWNJalanine was prepared from["5N]alanine (0.138 g, 1.53 mmol) by the method of Chang etal. (6) (yield 0.357 g, 1.14 mmol, 75%). TLC analysis inCHCI3/CH3OH/CH3COOH (95:20:3, vol/vol) gave a singlespot (RF = 0.63). Fmoc-tryptophan was coupled to thep-alkyloxybenzyl alcohol resin by a variation of the methodof Wang (7). Resin (10 g) was suspended in 160 ml of CH2CI2at 40C and dicyclohexylcarbodiimide (5.4 g, 26 mmol),4-dimethylaminopyridine (3.3 g, 26 mmol), and Fmoc-tryptophan (11.3 g, 26 mmol) were added. The mixture wasstirred for 30 min at 40C and 9 hr at 250C, after which theresin was filtered, washed several times with CH2Cl2, andresuspended in 160 ml of CH2CJ2 at 40C. It was treated with6.4 ml of pyridine and 9.2 ml of benzoyl chloride at 40C for15 min. The resin was collected by filtration, washed withCH2CJ2, N,N-dimethylformamide (DMF), and 2-propanol,and lyophilized. The Fmoc-Trp content was determined tobe 0.22 mmol per g of resin by spectrometric analysis of (i)the N-(9-fluorenylmethyl)piperidine concentration by amethod described previously (8) and (ii) the Trp concentra-tion. For the latter determination, the resin (200 mg) wasdeprotected with 20 ml of 50% (vol/vol) piperidine inCH2Cl2, filtered, and washed with CH2Cl2. It was thenstirred with 7 ml of 55% trifluoroacetic acid in CH2CJ2, 1.4ml of anisole, and 0.07 ml of 1,2-dithioethane in a nitrogenatmosphere for 3 hr. The suspension was filtered and theresin was washed with 9.5 ml of CH2Cl2. Analysis wasperformed on the combined filtrate and washing, using 6280= 5560 M - '-cm1 for tryptophan (9).The peptide was synthesized with a Vega model 100

peptide synthesizer using the solid-phase method withFmoc-blocked amino acids according to the procedure de-scribed by Chang et al. (10) with several modifications. Onecoupling cycle consisted of the following operations, utiliz-ing 30 ml of solvent per g of resin, except where noted: (i)CH2Cl2, three times for 1 min; (ii) DMF, three times for 1min; (iii) 20% piperidine in DMF for 1 min; (iv) 20%piperidine in DMF for 10 min; (v) DMF, three times for 1min; (vi) CH2Cl2, three times for 1 min; (vii) coupling with 4equivalents each of the Fmoc-amino acid, N-hydroxysuc-cinimide, and dicyclohexylcarbodiimide in 20 ml of CH2Cl2for 2-4 hr (1 equivalent of diisopropylethylamine is addedafter 15 min); (viii) CH2Cl2, three times for 1 min; (ix) 1equivalent of diisopropylethylamine in 20 ml of CH2Cl2 for20 min (after 5 min, 4 equivalents of acetic anhydride wasadded); (x) 2-propanol, three times for 1 min; and (xi)CH2CI2, three times for 1 min.

Analytical Techniques. Analytical HPLC was performedby using a Beckman instrument equipped with a RaininMicrosorb C18 5-,um-particle reverse-phase column (4.6 x250 mm), while a semipreparative Altex Ultrasphere ODS5-,um-particle reverse-phase column (10 x 250 mm) wasused for HPLC purification. Peptides were eluted isocrati-cally with 15% (vol/vol) water/methanol (Baker, HPLCgrade) while monitoring the absorbance at 280 nm. Aminoacid compositions (except for Trp) were determined with aDionex model D-300 analyzer after hydrolysis in constant-boiling HCl at 100°C for 22 hr. Prior to hydrolysis, thesamples were repeatedly freeze-thawed and degassed underhigh vacuum. Trp content was determined spectrophotome-trically before acid hydrolysis with a Varian model 219spectrophotometer. The amino acid compositions werewithin experimental error of the theoretical values. Fluores-cence spectra were obtained with a Perkin-Elmer modelLS-5 fluorometer and optical spectra with a Varian model219 spectrophotometer.NMR Spectroscopy and Sample Preparation. The 15N

NMR spectra were obtained with a modified IBM/BrukerWP200 SY spectrometer equipped with a solids package thatoperates at 200 MHz for 'H. NMR spectra of solution-phase

samples were obtained in completely deuterated dimethylsulfoxide with `5NH4NO3 as an external chemical shiftreference assigned to zero ppm. These spectra were re-corded with an INEPT polarization transfer sequence forsensitivity enhancement (11). Solid-state NMR experimentswere performed on samples of gramicidin A that had beencosolubilized in methanol with dimyristoyl phosphatidyl-choline (molar ratio, 1:8). The samples were dried underreduced pressure and hydrated with approximately 70%(wt/wt) water. Some of these lipid bilayer preparations werealigned between glass microscope coverslips and sealed in ashort segment of square glass tubing. The solid-state NMRspectra were obtained by cross-polarization with radio fre-quency fields generated by 4.6-ps 900 pulses, a mixing periodof 1 ms, and an increased 'H decoupling field of 2.0 mTduring data acquisition. The details of the sample prepara-tion and spectrometer modifications have been describedearlier (5).

RESULTSSynthesis and Characterization of [Ala3-'5N][VaI'JGrami-

cidin A. The desformyl peptide was synthesized starting withFmoc-Trp-resin (1.19 g, calculated from a final cleavageweight of dry resin), using the protocol given in Materialsand Methods. For the addition of Fmoc-valine and Fmoc-D-valine, step (vii) was extended to 12 hr. After the addition ofthe remaining 14 amino acids, the peptide-bound resin wassuspended in 33% (vol/vol) ethanolamine in methanol andshaken at 60'C for 40 hr in a sealed flask. The resin wasfiltered and washed with methanol, which was removedunder reduced pressure. The peptide was precipitated by theaddition of water, washed with water, and dried to yield 391mg (80% yield).The cleavage product (381 mg) was dissolved in 25 ml of

methanol and applied to a 40-ml column ofAG 5OW-X2 (H +form) equilibrated with methanol. The column was firsteluted with 400 ml of methanol at 40C to remove theacetylated products. Elution was then continued with meth-anolic ammonia (800 ml of methanol plus 160 ml of concen-trated ammonium hydroxide) to obtain 274 mg (58%) ofcrude desformyl[Ala3-'5N][Val']gramicidin A. A 17.6-mgsample of this material was purified by reverse-phase HPLCto yield 16.4 mg, which indicates that pure desformyl[Ala3-"5N][Val']gramicidin A has been obtained with an overallyield of 54%. Chromatography of the acetylated by-productsover a semipreparative HPLC column resolved four majorpeptides with the following compositions: Leu3Trp4,-Leu3Trp4Val, Leu4Trp4Val3Ala, and Leu4Trp4Val3Ala2Gly.The formylation reaction was carried out by first dissolv-

ing formic acid (0.94 ml, 25 mmol) in 14.5 ml of dry diethylether at 0C and adding 2.57 g of dicyclohexylcarbodiimide(12.5 mmol). After 4 hr, the solution was filtered to removedicyclohexylurea and reduced to a volume of 2 ml. Thissolution was added to the crude desformyl[Ala3-_5N]-[Vall]gramicidin A (118 mg) dissolved in 5 ml of dry DMFand 20 ILI of diisopropylethylamine (0.11 mmol) at 40C. Thesolution was stirred overnight and the solvent was removedunder reduced pressure. A portion of the resulting product(93 mg) was then dissolved in 5 ml of methanol and againpassed over the AG 5OW-X2 column. Elution with methanolyielded 51 mg of formylated product, which represents acumulative yield of 32%. The optical spectrum of the prod-uct in the 250- to 310-nm region clearly showed that noformylation of the side chain of Trp had occurred (12, 13).This conclusion is also supported by the finding that the Trpfluorescence spectrum (Aex = 280 nm) was identical to thatfor authentic gramicidin A'. A sample of this material (2.8mg) was purified by reverse-phase HPLC to yield 2.6 mg, foran overall cumulative yield of 30%. The final concentration

Biochemistry: Fields et al.

Dow

nloa

ded

by g

uest

on

Apr

il 10

, 202

1

Proc. Natl. Acad. Sci. USA 85 (1988)

was determined spectrometrically, using 6282 = 22,500M - 'cm-1 for gramicidin A (14).The analytical HPLC profile of natural gramicidin A'

obtained by using a solvent system of 15% (vol/vol)water/methanol is illustrated in trace A of Fig. 1 and showsthe expected distribution of gramicidin species, the largestpeak corresponding to [Val']gramicidin A (15). The profilefor the synthetic species prepared here (trace B of Fig. 1) hasa single peak with the identical retention time as this species.The natural-abundance 15N NMR spectrum of gramicidin A'in deuterated dimethyl sulfoxide is presented in Fig. 2 andcompared with the "5N NMR spectrum of [Ala3-_5N]-[Val1]gramicidin A. The resonances for gramicidin A' (spec-trum A of Fig. 2) have been assigned previously by Hawkeset al. (16), who reported that Ala3 and Ala5 have resonancesat 100.0 and 99.4 ppm, respectively. The chemical shift ofthe resonance observed for [Ala3-'5N][Vall]gramicidin A(spectrum B of Fig. 2) at 100.0 ppm confirms this assignmentfor Ala3. It is possible that there is a small high-field shoulderon this resonance, which could indicate that a small percent-age (-5%) of the gramicidin exists in an alternative confor-mation in dimethyl sulfoxide.

Solid-State NMR Spectroscopy of [AlaW-`5N[Val'JGramici-din A. Solid-state 15N NMR spectra of uniformly labeled['5N]gramicidin A' and [Ala3-15N][Val']gramicidin A in hy-drated lipid bilayers are shown in Fig. 3. The spectrum ofunoriented, uniformly labeled [15N]gramicidin A' is shown inspectrum A of Fig. 3 and is the superposition of 20 axiallysymmetric 15N powder patterns (16 peptide bonds and 4indole rings in gramicidin A). The axial symmetry is caused

A

A_ ,

B

110I10-100 90ppm

FIG. 2. Solution-phase '5N NMR spectra recorded at 20 MHz ina 15-mm probe by using INEPT polarization transfer for sensitivityenhancement. Spectrum A, natural-abundance spectrum of grami-cidin A' (800 mg) obtained with 40,000 acquisitions. Spectrum B,spectrum of [Ala3-15N][Val1]gramicidin A (14 mg) obtained with17,000 acquisitions.

by the rotation of the molecule about an axis parallel with thebilayer normal. The average static chemical shift anisotropy(CSA) tensor obtained from uniformly '5N-labeled gramici-din A is characterized by a frequency for each of the threeorthogonal axes: o11 = 40, 0-22 = 58, and 0J33 = 212 ppm (5).Each of the backbone amides has similar tensor elementfrequencies and consequently the isotropic chemical shift,Ois,, is similar:

[1]Oljso = (O' + O-22 + O-33]/3.

However, the motional averaging of the powder patternvaries, depending on the orientation of the atom-ic site with

A

B

10

Time (min)

B

C

20

FIG. 1. HPLC profiles of biosynthetic gramicidin A' (trace A),which is a mixture of linear gramicidins, and the purified syntheticproduct (trace B), [Ala3-'5N][Val1]gramicidin A. The gramicidinswere eluted isocratically with 15% water in methanol at a flow rateof 0.7 ml/min.

200 Ippm

FIG. 3. Solid-state '5N NMR spectra of hydrated lipid bilayerscontaining '5N-labeled gramicidins. The spectra were recorded at 20MHz under conditions of cross-polarization and high-power protondecoupling at 27°C. Spectrum A, spectrum of unoriented uniformlylabeled ["5N]gramicidin A' (50 mg) obtained with 4900 acquisitions.Spectrum B, spectrum of oriented uniformly labeled [15N]gramici-din A' (16 mg) obtained with 12,200 acquisitions. Spectrum C, spec-trum of oriented [Ala3-13NI[VaI']gramicidin A (14 mg) obtained with36,520 acquisitions.

1386 Biochemistry: Fields et al.

Dow

nloa

ded

by g

uest

on

Apr

il 10

, 202

1

Proc. Natl. Acad. Sci. USA 85 (1988) 1387

respect to the motional axis. The variables 611, 622, and 033are the angles between the motional axis and the coordinatesystem of the atomic site given by o-, O22, and 033Consequently, the values of the rotationally averaged and,hence, axially symmetric tensor elements all and o,, arehighly variable, where:

'Ill = r11cos2011 + 0r22coS2022 + o33cos2033 [2]

and

cr= (3oris - oll1]/2. [3]

Spectrum B of Fig. 3 shows the solid-state 15N NMRspectrum of an oriented sample of uniformly labeled [15N]-gramicidin A'. When the sample orientation is such that thebilayer normal is parallel with the magnetic field, a singlesharp resonance is expected for each nitrogen site at its -11frequency. This is readily seen by the alignment of the peaksat 198 and 146 ppm in spectrum B of Fig. 3, with thediscontinuities at the same frequencies in spectrum A of Fig.3. While the resolution is markedly improved in spectrum Brelative to spectrum A in Fig. 3, there are still manyoverlapping resonances with unknown assignments. Forexample, the resonance at 198 ppm in spectrum B arisesfrom the superposition of at least four single-site resonances.The single-site resonance observed for [Ala3-15N][Vall]gram-icidin A in spectrum C of Fig. 3 unambiguously assigns oneof these resonances to the Ala3 residue of gramicidin A.

If the orientation of the CSA tensor were known accu-rately with respect to the covalent bonds of the nitrogen sitevia a molecular symmetry axis (MSA) frame, then theobserved a-, frequency could be quantitatively interpreted.While the relative orientation of these two coordinate sys-tems is still uncertain (17, 18), it has been assumed in thepast that the CSA and MSA frames are collinear (19). Theprimary result of this assumption is that the N-H bond and033 are parallel. An observed frequency of 212 ppm wouldrepresent an orientation of the N-H bond parallel to themotional axis that is the bilayer normal (5). The angulardependence of the frequency in this motionally averagedsystem then becomes

0rll=,-233(3 cos2033 - 1) + 3/2corisin2033. [4]

From this relationship, the al frequency of 198 ppm indi-cates an angle of 170 between the N-H bond and the bilayernormal. The possibility of large-amplitude local motions ofthe peptide bond as a mechanism for averaging the CSApowder pattern has been considered. Temperature-depen-dence studies on the chemical shift powder pattern ofuniformly labeled [15N][Vall]gramicidin A' in hydrated bi-layers suggest that, in the absence of monovalent cations,there are no motions with an amplitude greater than 10° (5).

DISCUSSIONSolid-state NMR studies of 15N-labeled gramicidins in hy-drated, oriented lipid bilayers have the potential to providestructural and dynamic information about the monovalentcation channel. The solid-state NMR elucidation of thegramicidin structure is dependent upon interpreting theangular dependence of the nuclear spin interactions, such aschemical shifts and dipolar interactions, that reflect theorientation of molecular groups with respect to the magneticfield. To achieve this, bilayers containing gramicidin A havebeen uniformly oriented between glass plates such that thebilayer normal and channel axis are parallel with the mag-netic field (5). To interpret the resonance frequencies, themagnitude of the spin interaction must be known from

spectroscopic observations of powder pattern spectra ob-tained on unoriented samples. Accurate information on theinteraction magnitude is obtained only when the resonancesarise from single atomic sites. Because of the very largewidth of these resonances, the spectra must be recordedfrom samples with a single labeled site. Structural modelscan then be assembled from a determination of the orienta-tion of many molecular bonds with respect to a commonaxis. The dynamic analysis will be based on powder patternlineshape analyses obtained from single atomic sites andrelaxation time measurements. Thus, it is essential thatsite-specifically labeled samples be available for study.

Biosynthetic approaches for site-specific labeling of gram-icidin A based on growing Bacillus brevis on a mediumcontaining an "N-labeled amino acid have several limita-tions. Only amino acid types can be labeled (e.g., all four Trpresidues). "Scrambling" of the "5N by transaminases canresult in the labeling of other residue types and/or a reducedincorporation of label into the desired sites. Furthermore,because the bacteria produce a mixture of gramicidins,fractionation of the final product is necessary if observationsare to be made on a sample with uniform physical properties.To overcome these problems, solid-phase peptide synthesiscan be used to incorporate "N-labeled amino acids intospecific sites of any gramicidin so that site-specific reso-nances can be studied. A synthetic approach has beendeveloped here for preparing site-specifically labeled[Val']gramicidin A species that allows for both high yieldsand the incorporation of a wide range of isotopic labels.

Since [Val']gramicidin A contains several Trp residues,the Boc Nc-blocking group has been avoided because itsremoval with trifluoroacetic acid may lead to the oxidativedestruction of Trp (20, 21) as well as truncations due to thetrifluoroacetylation of the growing peptide (22). Repetitiveacidolysis also results in partial cleavage of the peptide fromthe resin (23, 24). The synthesis of [Ala3-15N1[Val1]-gramicidin A described here was designed to completelyavoid the use of acidic conditions. Since both the removal ofthe Fmoc group (by 20% piperidine in DMF) and cleavage ofthe peptide from the resin (by 33% ethanolamine in metha-nol) are nonacidic reactions, the oxidative destruction ofTrp, acid hydrolysis ofthe peptide-resin linkage, and trifluoro-acetylation of the growing peptide are minimized. Thecleavage of the peptide from the resin simultaneously blocksthe carboxyl terminus with ethanolamine (25) and removesthe last Fmoc group (6). After formylation and subsequentpurification, [Ala3-15N][Val']gramicidin A has been pre-pared with an overall yield of 30%. Analysis of the acetylat-ed by-products indicates considerable difficulty with cou-pling the Val' and D-Val8 residues, as noted previously (25).The coupling yield for two of the last three residues was alsorelatively low.The yield obtained by using this procedure represents a

substantial improvement over yields recently reported forlinear gramicidins by the Boc method (15, 21, 26) as well asthat for the predominantly Fmoc solid-phase synthesis re-ported for gramicidin B (25). Our use of N-hydroxysuccini-mide esters (27) of Fmoc-amino acids, instead of the bulkierpreformed symmetrical anhydrides (28), is the likely basisfor the increased coupling efficiencies. While truncations atpositions 7 and 8 were not completely eliminated, thecombination of coupling with active esters and the use ofextended coupling time reduced this problem significantly.

Purification of a small sample of the desformyl[Ala3-"5N][Vall]gramicidin A demonstrated that the synthetic pro-tocol presented here produced the peptide with an overallyield of 54%. The formylation reaction, as previously seen(21), is not very efficient, reducing the overall yield to 30%.Alternative methods of formylation (26) explored here wereeven less efficient (results not shown). A possible solution is

Biochemistry: Fields et al.

Dow

nloa

ded

by g

uest

on

Apr

il 10

, 202

1

Proc. Natl. Acad. Sci. USA 85 (1988)

the use of formylvaline instead of Fmoc-valine for the finalamino acid coupling. The use of formylvaline would, how-ever, prevent the separation of the desired product fromacetylated by-products by anion-exchange chromatography.Coupling efficiencies must therefore be increased to mini-mize acetylated by-products before formylvaline can be usedin this synthesis.While the solid-state "5N NMR spectrum reported here for

[Ala3-`5N][Val']gramicidin A in oriented lipid bilayers isonly the first of many needed to complete this study, certainpreliminary conclusions can already be drawn. First, one ofthe resonances at 198 ppm in the solid-state NMR spectrumof uniformly labeled ['5N]gramicidin A' is unambiguouslyassigned to Ala3. Second, the single resonance observed for[Ala3-_5N][Valligramicidin A establishes that the Ala3 sitesfor all of the [Vall]gramicidin A molecules in our prepara-tions have the same orientation with respect to the bilayernormal. This is strong evidence that all of the [Vall]gramici-din A molecules in our sample preparations have the sameconformation. Third, assuming that the CSA and molecularsymmetry axis frames of reference are collinear (19), theN-H bond orientation of the Ala3 site is 170 with respect tothe bilayer normal. Experiments in which dipolar interac-tions can be measured have the potential to check both theorientation estimated from the chemical shift data and thevalidity of the assumptions used here to estimate the molec-ular orientation.

The authors thank Richard Rosanske and Thomas Gedris for theirhelp in maintaining and modifying our NMR spectrometers. Thiswork was supported by National Institutes of Health Grant AI-23007and by National Science Foundation Grant DMB-8451876 andProcter and Gamble through a Presidential Young InvestigatorAward to T.A.C.; and by National Institutes of Health GrantGM-27939 and Research Career Development Award AM-01066 toH.E.V.W. The solid-state NMR spectrometer was purchased withthe assistance of National Science Foundation Grant DMB-8504250.

1. Urry, D. W. (1971) Proc. Natl. Acad. Sci. USA 68, 672-676.2. Wallace, B. A. (1986) Biophys. J. 49, 295-306.3. Fischer, W. & Brinkman, J. (1983) Biophys. Chem. 18, 323-

337.4. MacKay, D. H. J., Berens, P. H., Wilson, K. R. & Hagler,

A. T. (1984) Biophys. J. 46, 229-248.5. Nicholson, L. K., Moll, F., III, Mixon, T. E., LoGrasso,

P. V., Lay, J. C. & Cross, T. A. (1987) Biochemistry 26,6621-6626.

6. Chang, C. D., Waki, M., Ahmad, M., Meienhofer, J., Lundell,E. 0. & Hang, J. D. (1980) Int. J. Pept. Protein Res. 15,59-66.

7. Wang, S. S. (1973) J. Am. Chem. Soc. 95, 1328-1333.8. Meienhofer, J., Waki, M., Heimer, E. P., Lambros, T. J.,

Makofske, R. C. & Chang, C. (1979) Int. J. Pept. Protein Res.13, 35-42.

9. Mihalyi, E. (1969) J. Chem. Eng. Data 13, 179.10. Chang, C. D., Felix, A. M., Jimenez, M. H. & Meienhofer, J.

(1980) Int. J. Pept. Protein Res. 15, 485-494.11. Morris, G. A. (1980) J. Am. Chem. Soc. 102, 428-429.12. Previero, A., Coletti-Preverio, M. A. & Cavadore, J. C. (1967)

Biochim. Biophys. Acta 147, 453-461.13. Merrifield, R. B., Vizioli, L. D. & Boman, H. G. (1982) Bio-

chemistry 21, 5020-5031.14. Urry, D. W., Trapane, T. L., Romanowski, S., Bradley, R. J.

& Prasad, K. U. (1983) Int. J. Pept. Protein Res. 21, 16-23.15. Prasad, K. U., Trapane, T. L., Busath, D., Szabo, G. & Urry,

D. W. (1982) Int. J. Pept. Protein Res. 19, 162-171.16. Hawkes, G. E., Lian, L. Y. & Randall, E. W. (1984) J. Magn.

Reson. 56, 539-542.17. Cross, T. A. & Opella, S. J. (1985) J. Mol. Biol. 182, 367-381.18. Harbison, G., Jelinski, L. W., Stark, R. E., Torchia, D. A.,

Herzfeld, J. & Griffin, R. G. (1984) J. Magn. Reson. 60,79-82.

19. Cross, T. A. & Opella, S. J. (1983) J. Am. Chem. Soc. 105,306-308.

20. Noda, K. & Gross, E. (1972) in Chemistry and Biology ofPeptides, ed. Meienhofer, J. (Ann Arbor Scientific, Ann Ar-bor, MI), pp. 241-250.

21. Prasad, K. U., Alonso-Romanowski, S., Venkatachalam,C. M., Trapane, T. L. & Urry, D. W. (1986) Biochemistry 25,456-463.

22. Kent, S. B. H., Mitchell, A. R., Engelhard, M. & Merrifield,R. B. (1979) Proc. Natl. Acad. Sci. USA 76, 2180-2184.

23. Gutte, B. & Merrifield, R. B. (1971) J. Biol. Chem. 246,1922-1941.

24. Fankhauser, P., Schilling, B., Fries, P. & Brenner, M. (1973)in Peptides 1971, ed. Nesvadba, H. (North-Holland, Amster-dam), pp. 153-161.

25. Prasad, K. U., Trapane, T. L., Busath, D., Szabo, G. & Urry,D. W. (1982) J. Protein Chem. 1, 191-202.

26. Prasad, K. U., Trapane, T. L., Busath, D., Szabo, G. & Urry,D. W. (1983) Int. J. Pept. Protein Res. 22, 341-347.

27. Anderson, G. W., Zimmerman, J. E. & Callahan, F. M. (1964)J. Am. Chem. Soc. 86, 1839-1842.

28. Heimer, E. P., Chuang, C. D., Lambros, T. & Meienhofer, J.(1981) Int. J. Pept. Protein Res. 18, 237-241.

1388 Biochemistry: Fields et al.

Dow

nloa

ded

by g

uest

on

Apr

il 10

, 202

1