Proc. Nati. Acad. Sci. USAVol. 76, No. 5, pp. 2180-2184, May 1979Biochemistry

Mechanisms and prevention of trifluoroacetylation in solid-phasepeptide synthesis

(termination/resin-bound functionalities/intersite reactions/phenylacetamidomethyl-resin)

STEPHEN B. H. KENT, ALEXANDER R. MITCHELL, MARTIN ENGELHARD, AND R. B. MERRIFIELDThe Rockefeller University, New York, New York 10021

Contributed by Bruce Merrifield, February 14, 1979

ABSTRACT A novel mechanism for trifluoroacetylationin solid-phase peptide synthesis, independent of the couplingstep, has been elucidated. It involves the presence of trifluo-roacetoxymethyl groups on the resin support, which react withresin-bound amines by an intersite nucleophilic reaction. Thetrifluoroacetoxymethyl groups are generated from preexistinghydi oxymethyl sites during treatment with trifluoroacetic acidin dichloromethane or by acidolysis of the benzyl ester bondbetween the peptide and the resin. The transfer of trifluoro-acetyl from hydroxyl to amine occurs during the subsequentneutralization with tertiary amine. The mechanism was firstelucidated by model studies with aminomethyl-resins. Then theexpected transfer of trifluoroacetyl groups from trifluoroace-toxymethyl-resin to the a-amino group of NE-benzyloxycar-bonyllysine benzyl ester in solution was demonstrated; k2, 6 X10- M-l sect. Lysine-resins were used to examine the extentof trifluoroacetylation under the conditions of solid-phasepeptide synthesis. After a series of acid/base cycles simulatingsynthetic conditions but without coupling, the poorly nucleo-philic a-amino group was approximately 1-2% trifluoroacety-lated per cycle when attached to resins already containing hy-droxymethyl groups. Standard benzyl ester resins withoutpreexisting hydroxymethyl groups gave comparable levels oftrifluoroacetylation after the first few synthetic cycles becauseof gradual acid cleavage of the ester and accumulation of tri-fluoroacetoxymethyl sites. Peptide chain termination resultingfrom trifluoroacetylation by this mechanism could be prevented(<0.02% per cycle) by the use of the aminoacyl-4(oxymethyl-phenylacetamidomethyl-resini support, which can be synthe-sized free from extraneous functionalities and which is stableto trifluoroacetic acid under the conditions of solid-phase pep-tide synthesis.

The majority of repo; ed solid-phase peptide syntheses (1-3)have involved the stepwise addition of tert-butoxycarbonyl(Boc) amino acids to a peptide chain that is anchored to thepolymeric support through the COOH-terminal amino acid.During the assembly of the protected peptide chain, the Bocgroup is removed at each step under conditions that leave otherprotecting groups unaffected, most commonly by the use of50% (vol/vol) trifluoroacetic acid in dichloromethane for 15-30min. Acylation of the a-amino group of the growing peptidechain, to give N't-trifluoroacetyl peptides, has been reportedon several occasions (4-7). This potentially general side reactionhas serious consequences because the N'"-trifluoroacetyl pep-tides formed are stable to the conditions of solid-phase peptidesynthesis, thus lowering the overall yield of the desired productand complicating its purification from the accompanying ter-minated peptides. ' ,gested explanations for trifluoroacety-lation include physical carryover of trifluoroacetic acid vaporsinto the coupling step (6), incomplete neutralization (4), thepresence of trifluoroacetate salts of quaternary sites on the resin(8), and physical entrapment of trifluoroacetic acid in com-

ponents of the apparatus or within the polystyrene beads usedas the support for peptide synthesis (9).

These proposals all assume that trifluoroacetic acid is acti-vated by dicyclohexylcarbodiimide in the coupling step andsubsequently reacts with the a-amino group. This has beenreported to occur in solution, where it was found that use ofcoupling conditions that minimized activation of trifluoroaceticacid substantially decreased trifluoroacetylation (10). A similarapproach has been used in solid-phase peptide synthesis withuse of preformed Boc-amino acid symmetric anhydrides forthe coupling reaction (6). This has decreased, but not entirelyeliminated, trifluoroacetylation (7). Up to now it has beenthought that this is a general side reaction which usually occursat a low level, although occasionally to a more serious extent,and that no method exists for preventing it.

Preliminary results in this laboratory showed that trifluo-roacetylation also occurred independently of the coupling step.An Nl-benzyloxycarbonyllysine-resin [Lys(Z)-resinl that hadbeen subjected to repeated alternate treatments with trifluo-roacetic acid and tertiary amine showed a low (<1%) but sig-nificant level of N't-trifluoroacetyllysine after HF cleavagefrom the resin. Similar experiments using N"-Z-Lys-resin gaveabout 10% N(-trifluoroacetylation. These totally unexpectedobservations were inconsistent with all explanations previouslyproposed and suggested that there must exist an additional,radically different, explanation for trifluoroacetylation.

In this paper, we report our investigations of trifluoro-acetylation that have led to the eltucidation of this novelmechanism. We discuss the levels of occurrence of trifluoro-acetylation under various conditions and show that it canreadily be prevented in solid-phase peptide synthesis.

EXPERIMENTALTrifluoroacetic acid was obtained from Halocarbon and usedwithout further purification. Dichloromethane (ACS reagentgrade) was obtained from Eastman and distilled from sodiumcarbonate before use. Other chemicals were reagent grade. Theresins used were copoly(styrene-1% divinylbenzene) beads,200-400 mesh. Chloromethyl-resins were from Pierce, LabSystems, and Bio-Rad; unsubstituted resin was Bio-Beads S-X1from Bio-Rad. All substitutions are given as mmol/g of un-substituted polystyrene resin. Infrared (IR) spectra of resinswere taken on KBr pellets.

Aminomethyl-resin was synthesized from unsubstituted resinby direct amidoalkylation followed by hydrazinolysis (11) orfrom chloromethyl-resin by a Gabriel synthesis (12). Hy-droxymethyl-resin was made from chloromethyl-resin by re-action with acetate followed by transesterification (13). Hy-

Abbreviations: iPr2EtN, N,N-diisopropylethylamine; IR, infrared;Lys(Z), N'-(benzyloxycarbonvl)lysine; -OCt 12-Pam, -4-(oxymethyl)-phenylacetamidomethyl; Boc, tert-btitoxycarbonyl; DCC, dicyclo-hexylcarbodiimide; DMF, N.N-dimethylformamide.

2180

The publication costs of this article were defrayed in part by pagecharge payment. This article must therefore be hereby marked "ad-vertisemtent in accordance with 18 U. S. C.§1734 solely to indicatethis fact.

Proc. Natl. Acad. Sci. USA 76 (1979) 2181

droxymethyl-resin was converted to the trifluoroacetoxy-methyl-resin by reaction with the p-nitropheny1 ester of tri-fluoroacetic acid in N,N-dimethylformamide in the preei'of N,N-diisopropylethylamine (iPr2EtN). This procedurewas used to determine hydroxymethyl sites on the resin: Theintensity of the 1785 cm-1 band relative to the 1601 cm-' bandof polystyrene divided by an empirical factor of 2.2 gave thesubstitution in mmol/g. The detection limit was <0.02 mmol/g.Hydroxymethyl-resin was converted to acetoxymethyl-resinby reaction with acetic anhydride in pyridine and to benzoyl-oxymethyl-resin by reaction with benzoyl chloride and pyri-dine in CH2C12 (14). Levels of free amine on the resin weredetermined by titration with picric acid (15) or imidazoliumpicrate (16).

Boc-Lys(Z)-OCH2-resin was made from chloromethyl-resinby the cesium salt procedure (8). Boc-Lys(Z)-4-(oxymethyl)-phenylacetamidomethyl-resin [Boc-Lys(Z)-4-OCH2-Pam-resin](12) was prepared from aminomethyl-resin and preformedBoc-Lys(Z)-OCH2-phenylacetic acid by dicyclohexylcarbo-diimide coupling (17). A simulated synthetic cycle withoutcoupling consisted of: CH2CI2, three times, 1 min each; 50%(vol/vol) trifluoroacetic acid/CH2CI2, 1 min; 50% trifluo-roacetic acid/CH2CI2, 30 min; CH2Cl2, six times, 1 min each;5% (vol/vol) iPr2EtN/CH2CI2, 5 min; CH2Cl2, three times, 1min each; 5% iPr2EtN/CH2CI2, 5 min; and CH2Cl2, threetimes, 1 min each. All volumes were 20 ml/g of starting resin.To exchange the salt to the hydrochloride, the trifluoroaceticacid step was followed by: CH2Cl2, six times, 1 min each; 0.13M tetraethylammonium chloride in CH2Cl2, 2 min; CH2CI2,three times, 1 min each; 0.13 M tetraethylammonium chloridein CH2Cl2, 2 min; CH2CI2, three times, 1 min each; then thefirst iPr2EtN, as above. To couple, the simulated synthetic cyclewas followed by: 4 equivalents of Boc-amino acid in CH2Cl2,5 min (do not filter); 4 equivalents of DCC in CH2CI2, 0.5-2 hr;CH2Cl2, three times, 1 min each.

Cleavage from the resin was performed in anhydrousHF/10% (vol/vol) anisole for 1 hr at room temperature (onlyfor these model studies). Na-Trifluoroacetyllysine was unaf-fected by these conditions. Amino acid analysis on a Beckman121 analyzer gave Na-trifluoroacetyllysine at the position ofleucine, with 22% of the color yield. Leu-Ala-Gly-Val-OCH2-resin and Leu-Ala-Gly-Val-OCH2-Pam-resin were preparedby standard-solid phase peptide synthesis as described (17).

RESULTS AND DISCUSSIONMechanism of trifluoroacetylation

Initial Observations. Aminomethyl-resin was prepared fromcommercial chloromethyl-resin (Bio-Rad) and subjected tosimulated synthetic cycles consisting of acid and base treat-ments, but without the coupling step. The resin was examinedby IR spectroscopy at the end of each cycle and showed highlevels of the trifluoroacetamide. Subsequently, samples of theresin from individual steps in the simulated synthetic cycle wereexamined. A species with a strong carbonyl absorption at 1785cm1 was present after trifluoroacetic acid treatment, in ad-dition to the trifluoroacetate salt at 1680 cm-'. On neutral-ization, the 1785 cm-' species decreased in proportion to theamount of trifluoroacetamide (1725 cm-') formed. From this,it was hypothesized that the 1785 cm-1 species was a resin-bound activated form of trifluoroacetic acid and was responsiblefor the trifluoroacetamide formation. Other evidence suggestedthat a likely candidate for the 1785 cm-' species was the tri-fluoroacetyl-OCH2-resin ester. The proposed mechanism oftrifluoroacetylation is shown in Fig. 1.

0

CF3C-OCH2aII - +

CF3C-O NH3CH2

o R3N

CF3C-OCH2Q

:NH2CH20

HOCH2CF3C-NHCH2-

FIG. 1. Mechanismoftrifluoroacetylation of resin-bound aminesduring the neutralization step of solid-phase peptide synthesis. Thetransfer of the trifluoroacetyl moiety occurs via direct intersitenucleophilic reaction within a resin bead. The same mechanism wasshown to occur with resin-bound a-amino groups of peptides. Thevertical wavy line represents the polystyrene backbone. The sub-stituents are not necessarily on adjacent styrene units but are locatedrandomly throughout the polystyrene network.

Generation of Trifluoroacetoxymethyl Sites. Hydroxy-methyl-resin was treated at room temperature with 50% (vol/vol) trifluoroacetic acid in dichloromethane. Resin samples werewithdrawn at intervals and rinsed with ethanollin dichloro-methane, and the amount of trifluoroacetyl ester formed (Fig.2A) was measured by IR spectroscopy. The reaction showedfirst-order kinetics with a rate constant of 8 X 10-4 sec- (tI/2,15 min). When aminomethyl sites were also present on the resin,the esterification was slightly slower (rate constant, 6 X 10-4sec-1; t1/2, 20 min).

Leu-Ala-Gly-Val-OCH2-resin (0.3 mmol/g) was shaken atroom temperature with 50% (vol/vol) trifluoroacetic acid indichloromethane. After 71 hr, a sample of the resin was rinsedand examined by IR spectroscopy. This showed that 61% of thetetrapeptide was cleaved and replaced by an equivalentnumber of trifluoroacetoxymethyl sites (Fig. 2B). The observedextent of cleavage was consistent with the reported rate (12) ofacidolysis in 50% trifluoroacetic acid/dichloromethane. Anidentical experiment using Leu-Ala-Gly-Val-OCH2-Pam-resin(0.54 mmol/g) containing the more acid-resistant 4-(carboxa-midomethyl)benzyl ester linkage showed no detectablecleavage (<2%) after 65 hr and no trifluoroacetoxymethylformation. The calculated extent of cleavage, based on the lit-erature cleavage rate (12), was only 1.35%.

Aminomethyl-Resin Model. Chloromethyl-resin containinghydroxymethyl sites was converted to aminomethyl-resin (0.7mmol/g) containing hydroxymethyl sites (0.53 mmol/g). Inorder to investigate the mechanism of the trifluoroacetylationof resin-bound amines, this resin was subjected to various con-ditions and samples were examined by IR spectroscopy andpicric acid titration (15, 16) (Table 1). After a simulated syn-thetic cycle without coupling, about 30% of the aminomethyl

Biochemistry: Kent et al.

Proc. Natl. Acad. Sci. USk 76 (1979)

A.

CF3COOH + HOCH2-resin ! CF3 -OCH2-resin + H20B.

CF 3COOHLeu-Ala-Gly-Val-OCH 2-resin - CF3 -OCH2-resin

Leu-Ala-Gly-ValFIG. 2. The formation of trifluoroacetoxymethyl-resin. (A) From the reaction of hydroxymethyl-resin with trifluoroacetic acid. (B) By

acidolysis of peptidyl-resin in trifluoroacetic acid.

groups had been trifluoroacetylated. No change was observedwhen the time of iPr2EtN treatment was varied. The total timeof neutralization and washing was sufficient for all possibletrifluoroacetylation to occur. When the trifluoroacetic acid stepwas followed by treatment with tetraethylammonium chloridein dichloromethane, which was shown to convert the initiallyformed trifluoroacetate salt to the hydrochloride, the same levelof trifluoroacetamide formation was observed after the iPr2EtNstep. This rules out the participation of the trifluoroacetate saltof the amine. A synthetic cycle omitting the neutralization stepsgave no trifluoroacetamide. When 5% benzylamine indichloromethane was used in the simulated synthetic cycle, theextent of trifluoroacetamidomethyl-resin formation was sub-stantially decreased. This is consistent with the lability of thetrifluoroacetyl ester on the resin toward primary amines insolution (k2, 0.15 M-1 sec-t for benzylamine).

Aminomethyl-resin containing no hydroxymethyl sites andhydroxymethyl-resin containing no aminomethyl sites weremixed together and subjected to a single simulated syntheticcycle without coupling. Although IR spectroscopy showed theformation of the 1785 cm-1 species after trifluoroacetic acidtreatment, no trifluoroacetamide was formed during the neu-tralizations (Table 1). This rules out the existence of a long-livedsoluble intermediate trifluoroacetylating species.

Occurrence and prevention of trifluoroacetylationScreening for Hydroxymethyl Sites. Examination of un-

substituted and chloromethylated resins from this laboratoryand the commercial sources listed above revealed only two lots

Table 1. Trifluoroacetylation of aminomethyl-resin in simulatedsynthetic cycles without coupling

Amount of trifluoroacetamide formedBy IR spectrum By picrate

Conditions Amount* % Amount* %

Simulated cyclet 0.21 30 0.19 28Simulated cycle,exchange to HCl saltt 0.19 28 0.22 33

Simulated cycle,no neutralizationt 0 <2 0 <2

Simulated cycle,mixed resinst 0 <5

Simulated cycle,repeated 20 times§ 0 <0.15

per cycle* In mmol/g of polystyrene.t Aminomethyl-resin prepared from commercial chloromethyl-resincontaining hydroxymethyl sites.Equal amounts of aminomethyl-resin without hydroxymethyl sitesand hydroxymethyl-resin without aminomethyl sites.

§ Aminomethyl-resin (0.54 mmol/g) prepared from unsubstitutedpolystyrene by direct amidoalkylation.

that contained hydroxymethyl groups: chloromethylated Bio-Beads S-X1 (Lot 13668, 0.78 mmol/g by total chloride analysis)contained 0.36 mmol of hydroxymethyl per g; Bio-Beads S-X1(lot 14391, 0.84 mmol/g by total chloride analysis) contained0.33 mmol of hydroxymethyl per g.

Various Boc-aminoacyl-OCH2-resins, prepared from chlo-romethyl-resins by the cesium salt method (8), were found tocontain no hydroxymethyl sites. Aminomethyl-resins preparedby the early procedure (12) from chloromethyl-resins in somecases showed additional hydroxymethyl sites that had beengenerated during the Gabriel synthesis. In contrast, screeningof a large number of aminomethyl-resins with up to 1.8 mmol/gsubstitution that had been prepared from unsubstituted Bio-Beads S-X1 by direct amidoalkylation (11) showed no detect-able hydroxymethyl groups in any instance.

Blocked Hydroxymethyl Sites. Samples of hydroxy-methyl-resin were blocked as acetoxymethyl-resin and ben-zoyloxymethyl-resin and were treated with 50% (vol/vol) tri-fluoroacetic acid in dichloromethane at room temperature. Therate of cleavage was determined by IR spectroscopy of resinsamples withdrawn at intervals. Pseudo-first-order rate con-stants were 50 X 10-7 sec-t for acetoxymethyl-resin and 40 X10-7 sec- for benzoyloxymethyl-resin. These compare with58 X 10-7 sec- (t1/2, 33 hr) for Leu-Ala-Gly-Val-OCH2-resinand 0.58 X 10-7 sec- (t1/2, 3300 hr) for Leu-Ala-Gly-Val-OCH2-Pam-resin under the same conditions (12). Within thelimits of experimental uncertainty, all the cleaved sites formedtrifluoroacetoxymethyl groups on the resin.

Aminomethyl-Resin Model. Unsubstituted resin was con-verted to aminomethyl-resin by direct amidoalkylation fol-lowed by hydrazinolysis (1 1). This resin was subjected repeat-edly to a simulated synthetic cycle without coupling, for a totalof 20 cycles to amplify any low-level reaction. The result isshown in Table 1. No trifluoroacetamide formation (1725 cm-')was observed (<0.15% per cycle), consistent with the completeabsence of the 1785 cm-I species after treatment with trifluo-roacetic acid.

(Nf-Protected)Lysine-Resin Model. A number of differentlysine-resins were prepared and subjected to simulated syntheticcycles without coupling, repeated 5-10 times to amplify anylow-level reaction. After HF cleavage, which does not affectthe trifluoroacetamide group, amino acid analysis by standardion exchange methods was used to separate and quantitate ly-sine and Na-trifluoroacetyl-lysine (Table 2). As little as 0.01%NI''-trifluoroacetylation per cycle could be detected. A standardbenzyl ester-resin with no initial hydroxymethyl sites gave asignificant level of trifluoroacetylation (0.6% per cycle, aver-aged over five cycles), which increased to about 1.5% per cycleafter prior treatment of the loaded resin for 5 hr with trifluo-roacetic acid in dichloromethane. Both the standard benzylester-resin and the Pam-resin (12) gave levels of trifluoro-acetylation of about 1.5% per cycle when prepared from resins

2182 Biochemistry: Kent et al.

Proc. Natl. Acad. Sci. USA 76 (1979) 2183

Table 2. Trifluoroacetylation of lysine(Z)-resins in simulatedsynthetic cycles without coupling

Exterit,*v% per

Resin Conditions cycle

Lys(2ClZ)-OCH2-resint 5 stand. cycles 0.65-hr pretreatmentt, 1.4

5 stand. cycles

Lys(3ClZ)-OCH2-resin§ 10 stand. cycles 1.3

Lys(Z)-OCH2-Pam-resin§ 1st stand. cycle 2.610th stand. cycle 1.7

Lyx(Z)-OCH2-Pam-resin1 10 stand. cycles 0.02

Lys(2ClZ)-OCH2-Pam-resin11 22-hr pretreatment,$ 0.125 stand. cycles

* % trifluoroacetylation = ([Na-TfaLysJ/[total Lys]) X 100.t The original resin was Pierce chloromethyl-resin, 0.7 mmol/g, nohydroxymethyl sites.Pretreatment with 50% (vol/vol) trifluoroacetic acid in dichloro-methane at room temperature.

§ Original resin was Bio-Rad chloromethyl-resin, 0.7 mmol/g, con-taining 0.33 mmol hydroxymethyl/g.Aminomethyl-resin, 0.1 mmol/g, from direct amidoalkylation ofunsubstituted resin.

11 Aminomethyl-resin, 0.53 mmol/g, from direct amidoalkylation ofunsubstituted resin.

that already contained hydroxymethyl groups, and exchangeof the trifluoroacetate salt to the hydrochloride salt did notaffect the level of trifluoroacetylation. However, lysine-OCH2-Pam-resins, prepared by using aminomethyl-resin fromdirect amidoalkylation of unsubstituted resin containing nohydroxymethyl sites, gave an amount of trifluoroacetylation(0.02% per cycle) that was the same as background. Pretreat-ment with trifluoroacetic acid in dichloromethane for theequivalent of 65 deprotection cycles (20 min each) caused onlya slight increase, to 0.1% per cycle. This very low backgroundis thought to be due to the traces of trifluoroacetic acid anhy-dride detected by gas chromatography in the trifluoroaceticacid (18).The rate of reaction of trifluoroacetoxymethyl-resin with

NI-Z-lysine benzyl ester hydrochloride in solution in 10%iPr2EtN/dichloromethane was determined, giving a second-order rate constant of 6 X 10-4 M-1 sec-1. For the levels ofhydroxymethyl groups present in the commercial (Bio-Rad)resin, this gives an apparent first-order rate constant of 4 X 10-5sec- (tI/2, 5 hr) for the formation of Na-trifluoroacetyllysineon the resin. The neutralization steps of the standard syntheticcycle involve 10-min exposure to iPr2EtN in dichloromethaneand would lead to 2.5% Na-trifldoroacetyllysine on the basisof this rate constant. This is in agreement with the approxi-mately 2% trifluoroacetylation per cycle actually observed inseveral experiments using this resin (see Table 2). These dataindicate that the mechanism of trifluoroacetylation of resin-bound lysine is the same as that deduced for the amino-methyl-resin model, as shown in Fig. 1.

CONCLUSIONSMechanism. We have shown that trifluoroacetylation in

solid-phase peptide synthesis can occur by a mechanism unlikeany previously proposed. This novel mechanism is independentof the coupling step and involves the reaction of trifluoroace-toxymethyl groups on the resin support with the amino groupsof resin-bound peptides, by an intersite aminolysis within a resinbead. The nucleophilic transfer reaction was shown to occurduring the neutralization step of the synthetic cycle. This

mechanism is shown in Fig. 1. Authentic trifluoroacetoxy-methyl-resin has been shown to have properties consistent withthose required-for the proposed mechanism, and in particularit reacts with primary amines in solution in dichloromethaneat the necessary rates. The a-amino groups of peptides werefound to react, but at <1% of the rate of primary alkyl amines.The trifluoroacetylation was independent of the nature of theamine salt present and did not involve a long-lived soluble in-termediate.

Trifluoroacetoxymethyl sites on a resin support can arise fromexisting hydroxymethyl groups or can be generated by acido-lysis of the peptidyl-resin ester bond (Fig. 2). Both of theseprocesses occur during the acid deprotection step and are in-creased by longer exposure to the trifluoroacetic acid. Exami-nation of commercial chloromethyl-resins showed that Bio-Radresin (nominal loading, 0.7 mmol/g) contained 0.35 mmol ofhydroxymethyl groups per g. This particular resin is widelyused in solid-phase peptide syntheses. The hydroxymethylgroups react with 50% trifluoroacetic acid/dichloromethaneto form the ester with a half-time of about 20 min. The standardpeptidyl-resin benzyl ester bond is labile to trifluoroacetic acidin dichloromethane and generates about 1% trifluoroacetoxy-methyl sites per 30-min deprotection step.

Occurrence. The chloromethyl-resin already containingthese high levels of hydroxymethyl sites gives rise to 1-2%Na-trifluoroacetylation in each standard synthetic cycle. Aci-dolysis of the normal peptidyl-resin benzyl ester bond by tri-fluoroacetic acid produces trifluoroacetoxymethyl sites, whichaccumulate during a stepwise synthesis and lead to comparablelevels of trifluoroacetylation per cycle within 10 cycles (Table2).The extent of trifluoroacetylation in either of the above is

rate-limited by the poor nucleophilicity of the peptide a-aminogroup. Repeated or prolonged neutralization (7) will lead tomore trifluoroacetylation. The IR studies reported here showthat neutralization is complete within 30 sec (the shortest timeexamined), as expected for a proton transfer reaction. Prolongedneutralization is therefore unnecessary. Similarly, repeatedpicric acid titrations of such a peptidyl-resin will lead to a steadydecrease in titer. When proline is the NH2-terminal residue,much more extensive trifluoroacetylation can be expectedbecause of the greater nucleophilicity of the a-imino groupwhich has a pKa about 1 unit higher than an a-amino group.This is why trifluoroacetylation has been most often reportedto occur at proline residues (4, 5).

Trifluoroacetylation by the mechanism reported here is se-rious and ubiquitous with existing standard solid-phase meth-ods. Many of the apparent difficulties of solid-phase peptidesynthesis undoubtedly can be attributed to unrecognizedchemical problems such as this rather than to putative physicalproblems inherent to the polymer-supported nature of the re-actions, as has often been claimed (19-24). The extremely highconcentration of hydroxymethyl sites on batches of one of themost commonly used commercial resins, over a period of at leastseveral years, must have caused innumerable difficulties insolid-phase syntheses. Such resins are clearly unacceptable foruse in solid-phase peptide syntheses. Similarly, partial esteri-fication of hydroxymethyl-resin with the COOH-terminalresidue, a common route to loaded resins for use in synthesis (25,26), will lead to substantial termination. The residual hy-droxymethyl sites must be blocked before the synthesis is con-tinued. However, both acetoxymethyl- and benzoyloxy-methyl-resin are cleaved by trifluoroacetic acid at rates com-parable to the rate of cleavage of the peptidyl-resin benzyl esterbond, generating trifluoroacetoxymethyl sites. Blocking witheither of these commonly used groups will temporarily lower

Biochemistry: Kent et al.

Proc. Natl. Acad. Sci. USA 76 (1979)

the extent of trifluoroacetylation but will not eliminate this sidereaction even if repeated several times during the synthesis.

Prevention. On the basis of the studies reported here, we can

avoid trifluoroacetylation by the nucleophilic transfer mech-anism by using a resin that is initially free of hydroxymethylsites and that will not generate them during the synthesis. Thealternative, blocking these sites, has been discussed above. Wehave found that neutralization with primary amines preventstrifluoroacetylation by scavenging the trifluoroacetoxymethylgroups generated in each cycle of the synthesis much faster thanthey can react with the resin-bound peptide. Such primaryamine neutralization could be used to minimize trifluoro-acetylation in the synthesis of small peptides on standard resins,although this may be undesirable for other chemical reasons

not considered here. To avoid trifluoroacetylation by thetransfer mechanism, the resin must be derivatized and loadedby unambiguous chemical methods, giving a peptidyl-resinbond that is stable to trifluoroacetic acid in dichloromethaneand that does not generate trifluoroacetoxymethyl sites duringthe synthesis. One such system is the peptidyl-OCH2-Pam-resin,but only when synthesized by the recommended route (17)from preformed Boc-aminoacyl-OCH2-phenylacetic acid andaminomethyl-resin prepared from unsubstituted resin by directamidoalkylation (11). The loaded resin prepared in this waydoes not contain any detectable extraneous functionalities. Thepeptidyl-resin bond is 100 times more stable to acidolysis intrifluoroacetic acid than is the standard peptidyl-resin benzylester bond, and generation of trifluoroacetoxymethyl sites willoccur to an extent of less than 1% in 100 synthetic cycles. Thesuccessful use of this resin system to eliminate trifluoroacety-lation has been demonstrated in the model studies reported hereand in actual peptide syntheses (to be reported elsewhere).

We thank Edward J. Potter for expert technical assistance in thiswork. This work was supported in part by Grant AM 01260 from theU.S. Public Health Service and by funds from the Hoffmann-La RocheFoundation.

1. Merrifield, R. B. (1963) J. Am. Chem. Soc. 85,2149-2154.2. Merrifield, R. B. (1964) Biochemistry 3, 1385-1390.3. Erickson, B. W. & Merrifield, R. B. (1976) in The Proteins, eds.

Neurath, H. & Hill, R. L. (Academic, New York), 3rd Ed., Vol.2, pp. 255-527.

4. Ondetti, M. A., Williams, N. J., Sabo, E. F., Pluscec, J., Weaver,E. R. & Kocy, 0. (1971) Biochemistry 10, 4033-4039.

5. Bush, M. E., Alkan, S. S., Nitecki, D. E. & Goodman, J. W. (1972)J. Exp. Med. 136, 1478-1483.

6. Yamashiro, D. & Li, C. H. (1974) Proc. Nati. Acad. Sci. USA 71,4945-4949.

7. Noble, R. L., Yamashiro, D. & Li, C. H. (1977) Int. J. Pept.Protein Res. 10, 385-393.

8. Gisin, B. F. (1973) Helv. Chim. Acta 56, 1476-1482.9. Birr, C. (1978) Aspects of the Merrifield Peptide Synthesis,

(Springer, New York), p. 42.10. Fletcher, G. A., Low, M. & Young, G. (1973) J. Chem. Soc. Perkin

Trans. 1, 1162-1164.11. Mitchell, A. R., Kent, S. B. H., Erickson, B. W. & Merrifield, R.

B. (1976) Tetrahedron Lett., 3795-3798.12. Mitchell, A. R., Erickson, B. W., Ryabtsev, M. N., Hodges, R. S.

& Merrifield, R. B. (1976) J. Am. Chem. Soc. 98,7357-7362.13. Erickson, B. W. & Merrifield, R. B. (1973) J. Am. Chem. Soc. 95,

3757-3763.14. Chang, J. K., Shimizu, M. & Wang, S. S. (1976) J. Org. Chem. 41,

3255-3258.15. Gisin, B. F. (1972) Anal. Chim. Acta 58,248-249.16. Hodges, R. S. & Merrifield, R. B. (1975) Anal. Biochem. 65,

241-272.17. Mitchell, A. R., Kent, S. B. H., Engelhard, M. & Merrifield, R.

B. (1978) J. Org. Chem. 43,2845-2852.18. Weygand, F. & Geiger, R. (1956) Chem. Ber. 89,647-652.19. Meienhofer, J. (1973) in Hormonal Proteins and Peptides, ed.

Li, C. H. (Academic, New York), Vol. 2, pp. -45-267.20. Wunsch, E. (1971) Angew. Chem. Int. Ed. Engl. 10, 786-794.21. Sheppard, R. C. (1973) in Peptides 1971, ed. Nesvadba, H.

(North-Holland, Amsterdam), pp. 111-125.22. Bayer, E., Hagenmaier, H., Jung, G., Parr, W., Eckstein, H.,

Hunziker, P. & Sievers, R. E. (1971) in Peptides 1969, ed. Scof-fone, E. (North-Holland, Amsterdam), pp. 65-73.

23. Fankhauser, P. & Brenner, M. (1973) in The Chemistry ofPolypeptides, ed. Katsoyannis, P. G. (Plenum, New York), pp.389-411.

24. Hancock, W. S., Prescott, D. J., Vagelos, R. P. & Marshall, G. R.(1973) J. Org. Chem. 38,774-781.

25. Bodanszky, M. & Sheehan, J. T. (1966) Chem. & Ind. (London)1597-1598.

26. Bodanszky, M. & Fagan, D. T. (1977) Int. J. Pept. Protein Res.10,375-379.

2184 Biochemistry: Kent et al.