hi . J. Pepride Protein Res. 45, 1995, 209-216 Prinred in Belgium - all rights reserved

Coovriehr Q Munkseuurd 1995 l, Y

INTERNATIONAL JOURNAL OF PEPTIDE & PROTEIN RESEARCH

ISSN 0367-8377

Specificity and formation of unusual amino acids of an amide ligation strategy for unprotected peptides

JAMES P. TAM, CHANG RAO, CHUAN-FA LIU and JUN SHAO

Department of Microbiology and Immunology, Vanderbilt University, Nashville, Tennessee, USA

Received 21 December 1993, accepted for publication 14 May 1994

An important step in the recently developed ligation strategy known as domain ligation strategy to link un- protected peptide segments without activation is the ring formation between the C-terminal ester aldehyde and the N-terminal amino acid bearing a 8-thiol or 8-hydroxide. A new method was developed to define the specificity of this reaction using a dye-labeled alanyl ester aldehyde to react with libraries of 400 dipeptides which contained all dipeptide combinations of the 20 genetically coded amino acids. Three different ester aldehydes of the dye-labeled alanine: a-formylmethyl (FM), 8-formylethyl (FE), and 8,8,8-dimethyl and formylethyl esters (DFE), were examined. The D F E ester was overly hindered and reacted with N-terminal Cys dipeptides (Cys-X). Interestingly, it also reacted slowly with the sequences of X-Gly where Gly was the second amino acid and the X-Gly amide bond participated in the ring formation. Although the F E ester reacted similarly as the F M ester in the ring formation, the subsequent 0.N-acyl transfer was at least 30-fold slower than those of the FM-ester. The F M a-formyl methyl ester was the most suitable ester and was re- active with dipeptides of six N-terminal amino acids: Cys, Thr, Trp, Ser, His and Asn. The order and ex- tent of their reactivity were highly dependent on pH, solvent and neighboring participation by the adjacent amino acid. In general, they could be divided into three categories. (1) N-Terminal Cys and Thr were the most reactive. Cys reacted very rapidly and completely within 0.5 h to form thiazolidine in both aqueous and liigh content of water-miscible organic solvents. Thr reacted to form oxazolidine slowly in aqueous buffer ( i l l 2 > 300 h) but rapidly and completely within 20 h in organic-water solvents. (2) N-Terminal Trp, His and Ser were comparatively much less reactive than Cys or Thr. Trp reacted slowly and completely in aqueous buffer but significantly more slowly and incompletely in water-organic solvents. Both His and Ser reacted very slowly and incompletely in both solvent systems. (3) Finally, Asn reacted nearly insignificantly in both sol- vent systems. The significant rate enhancement by the water-miscible organic solvent on Thr was particularly important to allow the synthesis of disulfide-rich protein domains. Furthermore, the ring formation with N-terminal Trp, His and Asn provided a convenient route to prepare their bicyclic and unusual heterocyclic derivatives for structure-activity study. 0 Munksgaard 1995.

Key words: amide ligation; domain ligation strategy; methyl red dye; 0,N-acyl shift; peptide library; spot synthesis; un- protected peptide segment

The synthesis of peptides or proteins has become highly efficient with advances of the solid-phase peptide syn- thesis and the recombinant technology (1-5). Ideally, a chemical ligation method to form proteins utilizing the efficiency of the solid-phase method to generate specific segments and the availability of proteins from the re- combinant method would be desirable. In such a way, proteins can be engineered to contain unusual struc- tures or nongenetic coded amino acids by a specific ligation method. A strong impediment in this approach is a lack of an efficient method for their synthesis. In particular, there is no effective chemical method to

couple selectively two unprotected peptide segments to form an amide bond.

In general, protecting groups are necessary for the selective Cr-activation by a coupling reagent and the peptide bond formation with the Na-amino group with the second protected segment. Thus, the development of the various protecting group schemes has been the key for the conventional approach of ligating peptide segments (6-1 1). However, the conventional strategies have the disadvantages of being labor intensive for their preparation and unpredictable in their success, partly due to the solubility and coupling difficulties of pro-

209

J.P. Tam et al.

tected peptide segments. The problem of insolubility has been addressed with limited success in several ways, including the use of (1) partial protecting group strategy which masks all side chains except those of Ser, Thr and Tyr, (2) a minimal protecting group strategy which masks only thiol and amino side chains and (3) water- soluble protecting groups. Nevertheless, protecting groups are the root of the problem. Two strategies have developed in the past decade to use unprotected seg- ments for ligation (12-17). One of the ‘no side-chain protecting group’ strategies requires the use of enzymes in the reverse proteolysis process using high content of water-miscible solvents (14- 16). Although enzymatic synthesis has been successful with small peptides, it is more problematic with large peptides. In the case of large peptides, it is difficult to attain the stringent cri- teria of using high molar concentrations of peptide seg- ments and the rapid completion of the reverse pro- teolytic process without suffering hydrolysis or transpeptidation. Another strategy pioneered by Kemp and coworkers (17) uses a tricyclic aromatic template containing an aryl alcohol and a thiol to form ester and disulfide that bring the two unprotected segments into close proximity to effect an 0.N-acyl transfer reaction. These two strategies are similar in goals and aspirations to our recently developed chemical ligation method that allows the formation of a peptide bond between two unprotected peptide segments in an aqueous environ- ment without protecting groups and activation (18, 19).

The general concept underlying the new method of domain ligation strategy is outlined in Fig. 1. It consists of three steps: (1) aldehyde initiation, (2) ring formation and (3) 0,N-acyl rearrangement. The key to the car- boxyl component is an amino acid esterified to an

Y-Ala

Step 1 aldehyde inltiallon I

Y -Ala-OCH,-C H 0

Step 2

Ring formation

0

x=o, s Y=Z. Dpab

FIGURE 1 A schemc for the domain ligation strategy.

2 10

x-hydroxyl aldehyde to form an ester a-aldehyde. This ester aldehyde masked as an acetal is coupled to the carboxyl terminus of an unprotected peptide segment by reverse proteolysis. Liberation of the acetal to alde- hyde under acidic conditions would allow its reaction with the N-terminal but p-functionalized amino acid to form either thiazolidine (e.g. Cys) or oxazolidine (e.g. Thr). The net result is that carboxyl and amino com- ponents are brought together by a ring formation lead- ing to a well-positioned and facile intramolecular 0 , N - acyl rearrangement to form the desired amide bond. We have previously reported the aldehyde initiation step and the kinetics of ring formation using simple model compounds (18, 19). The present report describes the specificity of the ring formation using a newly devel- oped method to visualize the reactivity of different al- dehydes with a library of 400 dipeptides consisting of all possible dipeptide combinations of the 20 genetically coded amino acids.

EXPERIMENTAL PROCEDURES

Reagents Bromoacetaldehyde dimethyl acetal and bromopropi- onaldehyde dimethyl acetal were purchased from Ald- rich Chemical Company Inc. 2,2-Dimethyl-3- hydroxypropionaldehyde and Methyl Red were purchased from Fluka Chemical Corp. TLC systems:

(hexane:ethylacetate, 1: 1, v/v). CMA (CHC13:CH30H:CH3C02H, 6:2:1, v/v), HE

Z-Ala- x-(dimethylacetal)-methyl ester (1). Z-Ala (20 mmol, 4.46 g) was dissolved in methanol (40 mL) and the solution was adjusted to neutrality with cesium carbonate (0.5 M solution). The solution was evapo- rated to dryness several times with DMF. The cesium salt of Z-Ala was then allowed to react with bromo- acetaldehyde dimethyl acetal (40 mmol, 4.7 mL) in DMF (40 mL) at 60 “ C (20). The reaction was moni- tored by TLC (hexane:ethylacetate:acetic acid, 66:33: 1, v/v) and completed in 24-48 h. After the removal of DMF, the residue was dissolved in ethylacetate and washed sequentially by 1 M NaHC03 and H20 satu- rated with NaC1. The organic phase was dried over anhydrous MgS04, evaporated to dryness. Mass spec- trometry [ M + HI + (calcd./found, 3 12/3 12), ‘H NMR (CDCl3) in ppm 1.43 (d, 3H, J = 7.2 Hz), 3.38 (s, 6H), 4.17 (2q, 2H), 4.43 (dq, lH, J=7.2, 6.5 Hz), 4.56 (t, lH,J=5.3Hz),5.11(~,2H),5.31(bd, lH,J=6.5Hz) , 7.33 (s, 5H).

Methyl Red alanine ct-formyl-methyl ester (2b). Z-Ala-a- (dimethylaceta1)-methyl ester (2 mmol) was dissolved in THF (20 mL) and 5 % PdjC (0.4 g) was added. The reaction was performed on a Parr apparatus at 50- 60 psi for 3 h. After removal of THF, the product was obtained in > 95 % yield (Rf = 0.65 in CMA) and pro- ceeded without further purification to the coupling with

Ligation strategy of unprotected peptide

horizontal row represented the carboxyl terminal amino acid anchored to the cellulose paper while the vertical row represented N-terminal amino acid. The paper was washed in a glass tray with DMF (3 x 200 mL), metha- nol (3 x 200 mL) and dried with CH2C12 (2 x 200 mL). The paper was treated with 10% acetic anhydride in DMF for 1 h to cap the unreacted residue and fl-alanine. The washings were repeated. The procedures for the second cycle of deprotection, coupling capping and then deprotection of the Fmoc-dipeptides were similar to the first cycle. The side chains were removed with 82.5% CF3C02H in the presence of several scavengers ( 5 % thioanisole, 5 % phenol, 2.5 % ethanedithiol and 5 % H20) for 2 h. The paper was washed with CH2C12 (3x200mL), DMF (3x200mL) and dried with CH30H (3 x 200 mL). Usually, four identical dipeptide libraries were performed in each run and required four days for completion.

Methyl Red (2{ [4-(dimethyamino)phenyl]azo}benzoic acid, Dpab-OH), BOP (2 mmol, 0.884 g) and DIEA (3 mmol, 0.522 mL) in DMF:CH2CL ( l : l , v:v) for 3 h. After the removal of solvents in vacuo, the residue was dissolved in ethyl acetate and washed sequentially with 0.5 M NaHC03, 0.01 M HCl and saturated NaCl so- lution. The organic phase was dried over anhydrous Na2S04 and evaporated to dryness and purified by flash chromatography to obtain an oil (Rf= 0.43 in HE). Compound 2a was converted into the aldehyde by treatment with 30% CF3C02H in CH3CN for 0.5 h and used immediately after evaporation of solvent. 'H NMR in CDCl3 (ppm) 1.6 d, 3H (Ala CH3); 3.14 s, 6H (2CH3N); 3.37 S , 3H (OCH3); 3.39 S , 3H (OCH3); 4.20 dd ( J = 4.9 Hz), 2H (OCH3); 4.57 t ( J = 5.4 Hz), 1H (acetal); 4.93 m 1H (Alaa CH); 6.78 d and 7.94 d 4H (J= 9 Hz, aromatic); 7.53, 7.47, 8.81 and 8.38 m 4H (aromatic); 9.72 br d, 1H (NH). FAB mass spectro- metric analysis of 2b showed the correct M + H of 429 (calcd. 429). UV (max. at 452nm, E 2 . 8 7 ~ lo4 cm - 'M - in CHXN).

Methyl Red alanine P-formyl-ethyl ester (3bj and Methyl Red alanine fi,P-dimethyl-P-formylethyl ester (4b). The two title compounds were prepared similarly as de- scribed in the preparation of 2b.

Dipeptide library The procedure was adapted with modifications from those of Frank and Doring (21). A 18 x 18 cm sheet of cellulose paper (Whatman # 1 chromatography paper) was used for the 'spot-synthesis' of the 400 dipeptide library. The paper was first treated with the symmetri- cal anhydride of Fmoc-P-alanine (34 mmol) and 4-dimethylaminopyridine (3 mmol) in aminimal amount of DMF at room temperature for 3 days. The Fmoc- group was deprotected with 20% piperidine in DMF for 1 h and the piperidine was washed 5-8 times until the wash was bromophenol blue negative (to detect that all the piperidine was washed out and bromophenol blue was added to the washed solution). Stock solu- tions of each Fmoc-amino acid (1.2 mmol in 3 mL DMF) and 0.4 M HBTU and HOBt (24 mmol each and both in 60 mL DMF) were prepared. Aliquots (3 mL each) of the HBTU/HOBt solution were com- bined with the individual Fmoc amino acid in a trough (for multichannel pipetter application) and then DIEA (2.4 mmol, 0.42 mL) was added to each Fmoc amino acid. The first application of all 400 spots to the cellu- lose paper was 1 pL and the process was repeated 4 times with 1.5 pL an hour apart. The total coupling time was 24 h. The application of each activated Fmoc- amino acid in the spot synthesis was facilitated with the aid of a multiple pipetter (the Costar Octapette and Eflab Flow Laboratories Digital Multichannel Pipette) which applied eight spots of each amino acid in a ver- tical row for the first amino acid and then in a hori- zontal row for the second amino acid. In this way the

Reaction of the dipeptide library with Dpab-Ala-0-FM. The reaction of Dpab-Ala-0-FM with the dipeptide library was similar to those of the DFE and FE esters. A typical run is described below. The dipeptide library was incubated in a glass tray containing 150mL of 0.2 M buffer at various pH (pH 5-8) in different runs. Buffers for pH 5 and 6 were sodium acetate-acetic acid; pH 7 and 8, NaH2P04 and Na2HP04. Dpab- Ala-OFM (2 mg) dissolved in CH3CN (5-10 mL) was added to the tray. The pH was adjusted after the ad- dition of Dpab-Ala-OFM and during the reaction. Fresh Dpab-Ala-OFM (2 mg) was added every 24 h and the reaction was continued for 160 h. The color development was monitored in the first 6 h and every 24 h intervals by photography and, more conveniently, by directly xeroxing the paper (with a DCM and MeOH wash prior to the picture taking). In some runs, the concentrations of Dpab-Ala-OFM increased 10- 100- fold. In other runs, the library was incubated in 90% water-miscible solvents (isopropanol, DMF or DMSO) in 10% aqueous buffer at pH 7. After each run, the paper was washed sequentially with 10% HOAc (1 x 20 h), H20 ( 5 x 3 min), and CH30H (3 x 2 min). It was then treated with NaHC03 (pH 8.6) for 24 h to effect the O,N-acyl transfer reaction.

RESULTS AND DISCUSSION

Peptide library The effectiveness of the ring formation between two unprotected peptide segments depends on the high re- giospecificity of the C-terminal ester alkyl aldehyde to- wards the N-terminal amino acid, particularly those P-substituted hydroxyl or thiol amino acids (22, 23). Their productive encounter would lead to the formation of five-membered oxazolidine ring with Thr and Ser or thiazolidine ring with Cys (Table 1). At acidic pH, the N-terminal amino acids, side chains of lysine and the guanidino side chain of arginine are usually excluded

211

J.P. Tam et al.

TABLE 1

Proposed ring producrs from rhe reaction of N-terminal ammo acids with Dpa b-A le-0-FA4

N-termin a1 Ring products Reversible to amino acid natural amino

acid/reagent

Thr

S er

TrP

yes/electrophile

0

H O T 0 & yes,'acid

0

"7% 0

nT yeslacid

no

Asn

O.L\

ciently by changing to basic buffers or solvents under which the ester bond will be hydrolyzed and the 0 , N - acyl transfered product would be stable.

Dye-labeled amino ester aldehyde For the library to work efficiently, a reporter molecule on the amino acid alkyl ester aldehyde to provide de- tection of the bimolecular reaction between the alde- hyde and the dipeptides would be required. We con- sidered many existing reporter groups such as radiolabeling and fluorescence but found that the dye- labelling might be suitable because of its high sensitiv- ity for visible detection, stability under normal condi- tions, and ease of introduction to the amino acids or peptides. Thus, a dye-labeled amino acid derivative, 2{ [ 4-(dimethylamino)phenyl]azo}benzoic acid (Meth- yl Red, Dpab for abbreviation), was introduced to an amino acid derivative through amide bond that is stable toward acid and base treatments. Depending on the pH of the aqueous buffer, Dpab-amino acids possess in- tensive orange to red color and are visible on the paper at low concentration.

Amirro ester aldehydes Three different aldehydes were used: (1) a-formyl me- thyl ester (FM, 2b), (2) P-formyl methyl ester (FE, 3b)

no B FH3 R Dpab-Ala--O-C h-$-Cfu D pab-AIa-0-C y - C y-C H

CH, Dpab-Ala-OFE. 3a. 3b

Dpab-Ala-ODFE. 4a. 4b

from the reaction by protonation and their unlikely for- mation of Schiff bases are predicted to be reversible and unproductive.

In order to arrive at a conclusive determination re- garding the specificity of the ring formation, we devel- oped a method to study this reaction with all possible combinations of N-terminal amino acids and their side chain functionalities using a library of 400 dipeptides consisting of 20 genetically coded amino acids on cel- lulose paper (21) and allowed an alanyl ester aldehyde to react with each library under various conditions. The paper support used in these experiments served both as the solid support on which the 400 peptides were syn- thesized and as a monitoring device on which the ring formation could be observed.

The library was synthesized on a Whatman paper using the Fmoc chemistry. The arrangement of the li- brary was in a matrix system containing 400 spots, each representing a dipeptide. The reactivity of each dipep- tide towards the aldehyde could be visualized as either a horizontal row which showed the N-terminal amino acid was reactive (e.g. Cys-X, where X represents 20 amino acids) or a vertical column which indicated that the side chains of the carboxyl terminal amino acids (e.g. X-Cys) were reactive. Furthermore, the subsequent O,N-acyl rearrangement could also be observed effi-

212

3a, 4a R = (OCH,), 3b,4b R = O

and (3) P,P-dimethyl-P-formyl ethyl ester (DFE, 4b). Unlike the FM ester, the rearrangement reaction of the two formyl ethyl estcrs 3b, 4b would require a six- membered transition state (6 and 7), and according to Kirby (24) the rate would be 20-50-fold slower than those of the FM ester 2b, which would form a five- membered transition state 5. The DFE 4b represents

>a 0 &- - 0

5 6 R , = R z = H

the hindered aldehyde ester, and its interaction with the different N-terminal amino acids would be reflected in the reaction rate of the ring formation. The synthesis of

Ligation strategy of unprotected peptide

P G A V L I M F Y H w S T C D

1 ii - Iv

2: R I (0CHs)p

b R = O .. FIGURE 2 Synthesis of Dpab-Ala-OFM. (i) Cs + /DMF/60 "C, (ii) Hz, Pd/C, (iii) Dpab-OH, BOP/DIEA and (iv) 30"; CF~COZH/CH~CN.

the Dpab-Ala-OFM is shown in Fig. 2 and represents a general route to these derivatives 2b-4b.

Reactivity of N-terminal amino acids Six different amino acids (Cys, Thr, Ser, Trp, His and Asn) are known to form ring products with simple alkyl aldehydes (Table 1) (27, 28). In particular, ring forma- tion with formaldehyde has been used industrially for tanning and medically for inactivating toxins and bio- logical agents. Because the ester aldehyde 2-4 could be viewed as simple alkyl aldehydes, we would expect ring formation to various extent with these six amino acids.

Dipeptides with N-terminal Cys, Thr, and Ser (Cys-X, Thr-X and Ser-X, where X is any amino acid) are of major interest because their ring products can be re- verted to Cys, Thr or Ser (Table 1). Thus the thiazo- lidines and oxazolidines can be viewed as temporary protecting groups for these amino acids. Further, the relatively common occurrence of these amino acids in proteins makes them convenient points for ligation in our strategy. The two heterocyclic amino acids Trp and His bearing weakly acidic amines are known to react with alkyl aldehydes to form bicyclic compounds. The initial kinetic ring product with the heterocyclic amine would further rearrange to the stable product involving the C-2 carbon and the W-amine. These bicyclic ring products are not reversible to their natural amino acids. Similarly, the side chain amide of Asn is also known to participate to ring formation, usually under forcing con- ditions.

All reactions were performed at very dilute condi- tions ( 5 mg/150 mL) to simulate the condition of seg- ment synthesis which contains large molecular masses. Reactions were conducted using fresh reagents every 24 h, and the reaction was observed up to 160 h. The reactivity of each amino acid was compared with Cys-X based on the intensity and the size of the dot formation.

We have studied the ring formation in both aqueous buffered solutions at pH 5-8 and 90% water-miscible organic solvents at pH 7 (Figs. 3-9). However, the order of reactivity was significantly different in both systems. N-Terminal Cys reacted rapidly and com-

-

E N 0

FIGURE 3 A dipeptide library after incubation with Dpab-Ala-0-FM at a con- centration of 1 x M in aqueous buffer at pH 5.26 for 48 h prior to acid and base washes. The intensities of the spots indicate the extent of the reaction of the aldehyde with the dipeptides. For illus- trative purposes, the reaction of Dpab-Ala-0-FM to form the hy- droxymethyl thiol ether with the side chain Cys (vertical column under Cys) is shown. This reaction was reversible and the color would be eliminated after the acid and base washes. The color also diminished after 0.1 N base washes that hydrolyzed the ester prior to the rearrangement.

pletely with all three aldehydes at all pH ranges tested. With the unhindered FM and FE esters, the reaction was completed within 0.5 h, but required 2 h for comple- tion with the hindered DFE ester. In contrast, the re- activity of Thr was 100-fold slower. Ser was basically non-reactive.

In general, the reactivity of the N-terminal amino acids with Dpab-Ala-0-FM could be divided into three categories. First, Cys-X reacted exceptionally fast with Dpab-Ala-OFM in either aqueous or water-miscible organic and water mixtures. The reactions were com- pleted in 0.5 h at pH 5-8 even at a very dilute concen- tration of 1 x M. Second, Thr-X, Trp-X and His-X represented a category that reacted lo2- 103-fold slower than Cys-X. Their reactivity was highly depen- dent on concentrations of Dpab-Ala-0-FM, pH, and the neighboring amino acid. Trp-X formed a hetero- cyclic compound (28) in 30-50% in 160 h at the acidic pH. Furthermore, Trp-X reacted faster than Thr-X when the concentration of Dpab-Ala-0-FM was lower than 5 x 10- M, probably owing to the irreversibility of the Trp-X product (Fig. 4). In contrast, the oxazolidine ring and the Schiff base of Thr-X were not stable at the acidic range and only 5-30% of oxazolidines could be observed in 160 h. In contrast, at the neutral and basic pH, Thr-X reacted faster than Trp-X and 20-60% of

213

J.P. Tam et al.

, I & i @ # # . @ * l b I " .

I

4 . 4 a . 0 @ . * . 0 * 0 0 I@ I. 4 .

A H W S T C D E N Q K R

H W S T C D

B A dipeptidc library after incubation with Dpab-Ala-FM at a dilute concentration ( 5 x 10 - M) in aqueous buffer at pH 7 for 160 h. Only the bottom portion ofthe library is shown (see Fig. 3 for explanation).

P G A V L l M F V H W S T C D E N Q K R I * 1 H I . .

P G A V L I M F V H W S T C D E N Q K R

with alkyl aldehyde. In our model system we found that primary amides such as Asn, Leu-NHz and Alaz would react with Z-Ala-0-FM to form the heterocyclic com- pound. Interestingly, Gln-X, which would have formed a six-membered imidazolone ring, did not react with Dpab-Ma-0-FM, probably owing to the slow forma- tion of the six-membered ring. Similarly, the reaction

significant completion in the aqueous condition because

K R

. j FIGURE 4

A dipeptide library after incubation with Dpab-Ma-0-FM at a ver) dilute concentration (5 x 10- M) in aqueous buffer at pH 5 for 72 h (A) and 160 h (B). There is no reaction with other dipeptides (Pro-X.

bottom portion of the library containing the reactive dipeptides ih

shown. favored.

with Ser was much 'lower than Thr7 and never went to

the opened form of Schiff base and hydrolysis were G I ~ - X , Ala-x, ~ ~ 1 - x . I . ~ ~ - x . lle-x. M ~ ~ - x and Phc-X) and on ly the

P G A V L l M F Y H W S T C D E N Q K R

I

I 2

FIGCRE 5 A dipeptide library after incubation with Dpab-Ala-0-Fhl at a dilute concentration ( I x lo- ' M) in aqueous buffer at pH 6 for 160 h. When compared with Fig. 3, the intensity was stranger because a higher concentration of the aldehyde was used. On14 the bottom portion of the library is shown (see Fig. 7 for explanalion).

oxazolidines could be observed (Fig. 6). N-Terminal His formed a heterocyclic compound (28), but its for- mation was slow at pH 5-8 , and less than lo", of product was observable in 160 h. Third, the ring for- mation with Ser-X and Asn-X was essentially insignifi- cant in aqueous solution. Asn-X reacted verj slowly to form the heterocyclic compound (Table l) , and in aque- ous solution less than 5 " ; of the reaction product could be observed in 160 h. The participation of the P-carboxamide in the ring formation is known to occur

214

Neigliboririg group effects The neighboring amino acids exerted either a rate en- hancement or retardation in the ring formation. When the neighboring group amino acid is hydrophobic, such as X-Ile, X-Phe, X-Trp, X-Leu, X-Val and X-Tyr, the ring formation was accelerated when compared with X-Ala and X-Gly. This was particularly evident with those dipeptides containing N-terminal amino acids such as His and Ser (Figs. 2-6) with slow ring forma- tion. In contrast, when the neighboring amino acids are hydrophilic, and in particularly acidic such as Asp, Glu and Asn, the ring formation was retarded. A possible explanation for the observed result might be the par- ticipation of the side chains in assisting the hydrolysis of the ring form to the open form or the Schiff base to the starting material. The rate enhancement of the neighboring amino acids might be due to the hydro- phobic interaction of the Dpab. which contains two phenyl rings with the hydrophobic sequences. Interest- ingly, the hindered ester aldehyde ODFE reacted slowly with the X-Gly sequence. The participation of the gly- cine amide in the ring formation is similar to those of carboxamide of Ala and Leu.

A ccelemthg ring ,foriiiarion qf Thr-X and Ser-X using itwter-miscible organic solvents Except for Cys-X, the ring formation in 100% aqueous solutions and in a very dilute concentration of Dpab- /\la-0-FM was slow for Thr-X and Trp-X, and insig-

nificant for His-X, Ser-X and Asn-X. Equilibrium fa- vors the open forms of either the hydroxymethyl derivatives or the Schiff base, which are hydrolyzed by water to the starting materials. However, the equilib- rium would be predicted to favor the closed forms in the absence of water. To accelerate the ring formation, we explored the use of 90% water-miscible organic sol- vents such as hindered alcohol (isopropanol) and apro- tic polar solvents (DMF and DMSO). The use of these water-miscible organic solvents are necessary for re- verse proteolysis and compatible with our scheme using unprotected peptide segments. Furthermore, water- miscible organic solvents and water have been applied to effect the incorporation of the amino ester aldehyde in the first step of the domain ligation strategy.

When the reaction was performed in 90% water- miscible organic solvent buffered to pH 7 in 10% HzO, alcoholic solvents such as isopropanol did not improve the reaction rates. In contrast, the polar aprotic sol- vents D M F and DMSO greatly accelerated the reac- tion rates, particularly for Thr-X (except when X is Asp, Glu, or Asn) to give the oxazolidine ring forma- tion in 20 h (Figs. 8 and 9). The rate acceleration for Ser-X was difficult to quantify, since only about 25% of oxazolidine was observable in 30 h. Nevertheless, it represents an increase of about 10-fold when compared to the lOOo/, aqueous solution. Rate enhancement was also found in the ring formation of Trp-X, His-X and Asn-X, but apparently the formation of thiazolidines and oxazolidines was more favored.

H - w S T c D E N Q K R L

0- to N-Acy/ rearrangemenr The next key reaction is the base catalyzed intramo- lecular 0- to N-acyl rearrangement to form the amide

* - . 7 - 4

Q * ' * * @ * - @ ' + * ,

~ 0 ~ 0 0 0 ~ 0 ~ ~ 0 0 0 ~ ~ 0 ~ @ ~ 0 0 . e . 0 . *. 6 . 0 . 0 . 0 . C . O .

? b d Q

P I G A V

I M F Y H

S T c D E

Q K R

Ligation strategy of unprotected peptide

P G A V L I M F Y H W S T C D E N Q K R -~

E N Q

~

K R I

-

L <

w o

@ ~ ~ . . ~ . ~ . . 0 0 . * ~ ~ ~ em.

N *

6

FIGURE 8 A dipeptide library after incubation with Dpab-Ala-0-FM at a dilute concentration (5 x M) in 90;; D M F and lo", H:O buffered to pH 7 for 12 h. Only the bottom portion of the library is shown (see Fig. 3 for explanation).

bond. The facile interconversion between amide and ester through N- to 0-acyl or 0- to N-acyl transfer is commonly found in ,B-hydroxyl amino acids such as Ser and Thr and is strongly pH dependent (26-27). Under strong anhydrous acidic conditions such as HF, the 0,N-acyl transfer occurs to form the /?-ester linkage that is reversed nearly quantitatively by the subsequent aque- ous base treatment at pH 8-9. This reaction is particu- larly prone to occur with X-Ser/Thr sequences when X is an unhindered amino acid such as Gly and Ala. Thus, Ser- and Thr-containing peptides after H F depro- tection reactions are often routinely treated in base to reverse the 0,N-acyl transfer side reaction (28). The rearrangement was difficult to determine because in the library it was not possible to distinguish the ester and the amide form.

CONCLUSION

Two aspects of the present work need to be empha- sized: first, the specificity of the ring formation and second, the new route to incorporating unusual hetero-

215

J.P. Tam et al.

cyclic amino acids in peptides and which are useful for preparing peptide analogs and peptide libraries.

The ring formation between the C-segment ester al- dehyde and the library of 400 dipeptides is found to be specific and limited to six amino acids at the amino terminus of these 400 dipeptides. However, their reac- tivities are vastly different depending on pH, solvents, and the neighboring amino acids. In general, Cys and Thr would be most useful as they are characterized by a fast reaction rate and complete reaction. N-Terminal Cys, Thr and Ser are of particular interest because their ring products, thiazolidine and oxazolidine, are not only reversible but also are proline mimetics. If one takes the view that these rings are proline analogs, the combina- tion of Cys, Thr, Ser and Pro occurs very frequently in proteins and would not cause difficulty for the pro- posed break point for the domain ligation strategy. The formation of bicyclic analogs with Trp, His and Asn is not reversible but can provide a useful and convenient route to the formation of unusual cyclic analogs for structure-activity studies. A significant difference be- tween peptides and nucleotides is the absence of spe- cific enzymes to ligate unprotected peptides. The method described in this paper provided an approach to ligate chemically unprotected peptides via an amide bond.

REFERENCES

1. Offord, R.E. (1990) in Protein design and the Development of New Therapeurics (Hook, J.B. & Poste, G., eds.), Plenum, New York, pp, 253-282

2. Merrifield, R.B. (1963) J . Am. Chem. SOC. 85, 2149-2154 3. Gutte, B. & Merrifield, R.B. (1969)J. Am. Chem. SOC. 91, 501-

4. Tam, J.P. (1988) Proc. Natl. Acad. Sci. USA 85, 5409-5413 5. Du Vigneaud, V., Ressler, C., Swan, J.M., Roberts, C.W., Kas-

toyannis, P.G. & Gordon, S. (1953) J . Am. Chem. SOC. 75,

6. Hirschmann, R., Nutt, R.F., Veber, D.F., Vitali, R.A., Varga, S.L., Jacob, T.A., Holly, F.W. & Denkewalter, R.C. (1969) J . Am. Chem. SOC. 91, 507-508

7. Kiyama, S., Fujii, N., Yajima, H., Moriga, M. & Takagi, A. (1984) Int. J . Pepride Protein Res. 23, 174-186

8. Blake, J. & Li, C.H. (1981) Proc. Narl. Acad. Sci. USA 78,

502

4879-4880

4055-4058

9. Blake, J. (1986) Int. J . Peptide Protein Res. 27, 191-200 10. Yamashiro, D. & Li, C.H. (1988) Int. J . Peptide Protein Res. 31,

11. Hojo,H.&Aimoto,S.(t991)Bull. Chem.Soc. Jpn. 64,111-117 12. Bergmann, M. & Fraenkel-Conrat, H. (1937)J. Biol. Chem. 119,

13. Bergmann, M. & Fruton, J.S. (1938) J. Biol. Chem. 124, 321-

14. Homandberg, G.A., Mattis, J.A. & Laskowski, M., Jr. (1978)

15. Barbas, C.F., 111, Matos, J.R., West, J.B. & Wong, C.H. (1988)

16. Morihara, K., Oka, T. & Tsuzuki, H. (1979) Nature(London) 280,

17. Fotouhi, N., Galakatos, N.G. & Kemp, D.S. (1989) J. Org.

18. Liu, C.F. &Tam, J.P. (1994)J.Am. Chem. SOC. 116,4149-4153 19. Liu, C.F. & Tam, J.P. (1994) Proc. Natl. Acad. Sci. USA 91,

20. Wang, S.S., Gisin, B.F., Winter, D.P., Makofske, R., Kulesha, I.D., Tzougraki, C. & Meienhofer, J . (1977) J. Org. Chem. 42,

322-334

707-720

329

Biochemistry 17, 5220-5227

J . Am. Chem. SOC. 110, 5162-5166

412-413

Chem. 54, 2803-2817

6584-6588

1286-1290 21. Frank, R. & Doring, R. (1988) Tetrahedron 44, 6031-60 22. Ratner, S. & Clarke, H.T. (1937)J. Am. Chem. SOC. 59, 200 23. Tome, D. & Naulet, N. (1981) Int. J . Peptide Protein Res. 17,

24. Kirby, A.J. (1980) In Advances of Physical Organic Chemistry (Gold, V., Bethell, D., eds.), Academic Press, New York, pp.

25. Phillips, A.P. & Baltzly, R. (1947) 1. Am. Chem. SOC. 69, 200-

26. Winstein, S. & Boschan, R. (1950) J. Am. Chem. SOC. 72,4669-

27. Fodor, G. & Kiss, J. (1950) J . Am. Chem. SOC. 72, 3495-3497 28. Sakakibara, S., Shin, K.H., Schneider, W. & Hess, G.P. (1962)

501-507

183-278

204

4677

J . Am. Chem. SOC. 84, 4921-4928

Address:

Dr. James P. Tam Department of Microbiology and Immunology A51 19 Medical Center North Vanderbilt University Nashville, TN 37232-2363 USA Tel: (619-343-1463 Fax: (615)-343-1467

216