Protein synthesis with conformationally constrained cyclic dipeptides

Abstract

We have synthesized several conformationally constrained dipeptide analogues as possible substrates for incorporation into proteins. These have included three cyclic dipeptides formed from Boc derivatives of 2,4-diaminobutyric acid, ornithine and lysine, having 5-, 6-, and 7-membered lactam rings, respectively. These dipeptides were used to activate a suppressor tRNA transcript, the latter of which had been prepared by in vitro transcription. Using modified E. coli ribosomes described previously, these activated suppressor tRNAs enabled the incorporation of the three cyclic dipeptides into dihydrofolate reductase (DHFR) at positions 18 and 49. The suppression yields increased with increasing lactam ring size and were found to proceed in suppression yields ranging from 3.4 to 8.9% at two different protein sites for the 5-, 6- and 7-membered lactam dipeptides. The greater facility of incorporation of the 7-membered lactam prompted us to prepare two 7-membered cyclic acylhydrazides (4 and 5) by 1-ethyl-3-(3-dimethylaminopropyl)carbodiimide (EDCI)-mediated cyclization of amino acids having selectively protected hydrazine functional groups in their side chains. In common with the lactam dipeptides, acylhydrazide dipeptides 4 and 5 could be used to activate the same suppressor tRNA transcript and to incorporate the cyclic dipeptides into DHFR. They were incorporated into the same two DHFR sites in suppression yields ranging from 8.3 to 11.2%.

1. Introduction

Nature has provided numerous examples of peptide hormones and other peptide-based mediators of biological processes.^1–5 In nature, it is not unusual for these to be employed under precise temporal and spatial control. In contrast, the use of such compounds to control biological processes experimentally, including during therapeutic intervention, requires consideration of the chemical and metabolic stability of the peptide analogues employed. Frequently, it also requires a consideration of peptide stability and conformation to optimize the desired properties of the peptides as agonists and antagonists.^6–11

Peptide stability has long been studied, e.g. by the replacement of proteinogenic amino acids in peptides with amino acid and peptide surrogates resistant to natural peptidases, or which can help to stabilize the native conformation.^12–14 Some of these structural surrogates can also confer useful conformational constraints, for example at peptide sites known to be conformationally labile or important for receptor recognition.¹⁵ In recent years, however, it has become increasingly common to control peptide conformation by creating peptides in which much of the structure has more limited conformational freedom, e.g. by the use of cyclic peptides. These can be made to restrain linear amino acid sequences, such as in cyclic peptides which mimic the RGD domain of fibronectin.^16–18 Modified ribosomal systems are currently also employed in chemoenzymatic protocols to elaborate libraries of modified cyclic peptides containing non-proteinogenic amino acids.^19–21

In comparison, the modification of proteins with non-proteinogenic amino acids to confer protein stability to cellular proteases is less common, although a number of examples may be cited.^22,23 Likewise, while many modifications undoubtedly alter protein conformation, the purposeful construction of structural cassettes designed to constrain specific regions of proteins to a subset of the conformations usually accessible has not been pursued as actively.

Recently, we have studied the use of modified bacterial ribosomes²⁴ to introduce dipeptides and dipeptidomimetic species into proteins.^25–28 Presently, we describe the preparation of several cyclic dipeptides that are constrained conformationally, but which can be incorporated into proteins from an activated suppressor tRNA using modified bacterial ribosomes. In particular, E. coli ribosomal clone 010328R2²⁵ containing a modified 23S rRNA was shown to incorporate five conformationally constrained cyclic dipeptides (1 – 5) (Figure 1) into two different positions of dihydrofolate reductase.

Figure 1.

Structures of five conformationally constrained cyclic dipeptides synthesized and incorporated into E. coli dihydrofolate reductase at positions 18 and 49.

2. Results and discussion

Compounds 1 – 3 were prepared for potential use to induce conformational constraints in peptides in a report by Freidinger et al.²⁹ Amblard et al.³⁰ subsequently incorporated them in lieu of Pro7-Phe8 in bradykinin and found that the resulting bradykinin analogues bound to the human bradykinin B₂ receptor. The preparation of dipeptide analogue 1, and the procedure used for its activation of a suppressor tRNA_CUA transcript obtained by in vitro transcription, is shown in Scheme 1. The routes used here for the preparation of compounds 1 – 3 were different from those reported by Freidinger et al.,²⁹ and there were no identical compounds due to differences in protecting groups and the order of steps, so no comparison of properties can be made.

Scheme 1.

General procedures for the synthesis of a suppressor tRNA_CUA transcript activated with five-membered lactam-constrained dipeptide analogue 1. Reagents and conditions: (a) BOP reagent, NaHCO₃,DMF, rt, 12 h; (b) NaH, benzyl bromoacetate, anh. THF, rt, 5 h; (c) 20% CF₃COOH, anh. CH Cl o2 2, 0 C, 1 h; (d) 10% Pd/C, H₂, 95% EtOH, rt, 1h; (e) 4-pentenoyloxysuccinimide, diisopropylethylamine, anh. DMF, rt, overnight; (f) chloroacetonitrile, Et₃N, anh. DMF, rt, overnight; (g) pdCpA TBA salt, 9:1 anh. DMF-Et₃N, sonication, rt, 12 h; (h) T4 RNA ligase, tRNA-C_OH, ATP, 37 °C, 2 h.

The synthesis of dipeptide 1 began with N^α-Boc (2S)-2,4-diaminobutyric acid, which was converted to lactam 6 in 54% yield by treatment with the BOP reagent ((benzotriazol-1-yloxy)tris(dimethylamino)phosphonium hexafluorophosphate). Treatment of lactam 6 with sodium hydride and benzyl bromoacetate in THF at room temperature afforded fully protected cyclic dipeptide 7 as a colorless oil in 60% yield after purification by silica gel chromatography. Following deprotection in quantitative yield by successive treatments with CF₃COOH and hydrogenolysis over 10% palladium on carbon, free dipeptide 1 was isolated as its colorless trifluoroacetic salt. The dipeptide was reprotected on N^α of the pyrrolidine moiety as its N-pentenoyl derivative 9, and then activated as cyanomethyl ester 10 in low overall yield (7%) following purification by column chromatography on silica gel.

In order to enable tRNA activation, activated ester 10 was combined with the tetrabutylammonium salt of the dinucleotide pdCpA, which is a sequence common to the 3′-end of all cytoplasmic tRNAs. The resulting pdCpA ester of 10 (compound 11) was formed in low yield, and purified by C₁₈ reversed phase HPLC, as were all of the pdCpA derivatives in this study. As shown in Scheme 1, this activated dinucleotide was attached to a suppressor tRNA in vitro transcript by enzyme mediated ligation with T4 RNA ligase.²⁴ The N-pentenoyl protecting group attached to N^α of the pyrrolidine moiety of 12 could be removed prior to the use of the activated dipeptidyl tRNA in protein synthesis by brief treatment with dilute aqueous iodine.^31,32

The synthesis of dipeptide 2 was carried out in a fashion closely analogous to the preparation of 1, except starting with N^α-Boc-protected ornithine rather than N^α-Boc 2,4-diaminobutyric acid (Schemes 1 and S1). The yields of the intermediates leading to the individual dipeptides 1 and 2 were also quite similar. Interestingly, while the two-step procedure leading to cyanomethyl ester 10 was very low, the analogous preparation of cyanomethyl ester 17 proceeded in 23% yield. Likewise, while pdCpA derivative 11 was formed in 9% yield (Scheme 1), the yield of 18 was 18% (Scheme S1). The synthesis of dipeptide 3 was also carried out as for 1 and 2, but starting with S-lysine. The yields of 3 were also quite similar to those realized in the steps leading to 1 and 2. For dipeptide 3, conversion to the fully protected cyanomethyl ester 24 proceeded in 21% yield and the pdCpA derivative was obtained in 38% yield (Scheme S2).

The synthesis of dipeptide 4 is outlined in Scheme 2. Ethyl (2S)-2-[(benzoxycarbonyl)(tert-butoxycarbonyl)amino]-5-oxopentanoate³³ was converted to hydrazone 27 in 68% yield by treatment with tert-butyl carbazate, and then reduced to the corresponding hydrazine derivative 28 by treatment with NaBH₃CN. Suitable positioning of protecting groups, and subsequent cyclization afforded a 1,2-diazepane with CBz protecting groups on N-1 and the 4-NH₂ substituent (31). Treatment with ethyl bromoacetate then introduced the C-atoms of glycine at the 2-position of the diazepane. Following removal of the CBz protecting groups, the 4-NH₂ substituent was protected by alkylation with 4-pentenoyloxysuccinimide, giving 34 in a yield of 62% for the two steps. Conversion of ethyl ester 34 to activated cyanomethyl ester 35 provided an intermediate capable of effecting esterification of dinucleotide pdCpA with the modified 1,2- diazepine, albeit in low yield. Dinucleotide 36 was ligated to the abbreviated tRNA_CUA-C_OH transcript, forming activated suppressor tRNA 37.

Scheme 2.

General procedures for the synthesis of a suppressor tRNA_CUA transcript activated with seven-membered cyclic acylhydrazide-constrained dipeptide analogue 4. Reagents and conditions: (a) tert-butyl carbazate, anh. THF, 0 °C, 18 h; (b) NaBH₃CN, HOAc, EtOH, 0 °C, 17 h; (c) benzyl chloroformate, diisopropylethylamine, anh. CH₂Cl₂, rt, overnight; (d) 37.5% CF COOH, anh. CH o 3 2Cl₂, 0 C, 3 h; (e) (1) 3:1 THF–MeOH, LiOH, 0 °C, 3 h; (2) HOBt, EDCI, diisopropylethylamine, CH₃CN, rt, 18 h; (f) NaH, ethyl bromoacetate, anh. THF, rt, overnight; (g) 10% Pd/C, H₂, 95% EtOH, rt, anh. MeOH, overnight; (h) 4-pentenoyl chloride, triethylamine, anh. CH₂Cl₂, rt, overnight; (i) (1) 3:1 THF–MeOH, LiOH, rt, 3 h; (2) chloroacetonitrile, Na₂CO₃, anh. DMF, rt, overnight; (j) pdCpA TBA salt, 9:1 anh. DMF–Et₃N, rt, 15 h; (k) T4 RNA ligase, tRNA-C_OH, ATP, 37 °C, 2 h.

Methyl (2S)-2-[(enzoxycarbonyl)(tert-butoxycarbonyl)amino]-5-(tert butoxycarbonylhydrazinyl)pentanoate (28), an intermediate in the synthesis of dipeptide 4, was also used for the preparation of dipeptide 5 (Scheme 3). Following methylation of the hydrazine moiety in 76% yield, compound 38 was treated with CF₃COOH to remove the Boc protecting groups and LiOH to hydrolyze the methyl ester. Cyclization by treatment with HOBt, diisopropylethylamine, and EDCI then effected cyclization, affording (4S)-4-(benzoxycarbonyl)amino-1-methyl--3-oxo-1,2-diazepane (40). Following treatment with ethyl bromoacetate to effect substitution at position 2 of the 1,2-diazepane in 92% yield, the CBz group at position 4 was replaced with an N-pentenoyl protecting group. Replacement of the ethyl ester at ring position 2 with a cyanomethyl ester then enabled esterification of the dinucleotide pdCpA with protected dipeptide 5.

Scheme 3.

General procedures for the synthesis of a suppressor tRNA_CUA transcript activated with seven-membered cyclic acylhydrazide-constrained dipeptide analogue 5. Reagents and conditions: (a) CH₃I, diisopropylethylamine, anh. DMF, rt, overnight; (b) (1) 50% CF COOH, anh. CH o3 2Cl₂, 0 C, 3 h; (2) LiOH, 16:5:1 THF–MeOH–H₂O, rt, 1 h; (c) 1:1 CH₃CN–DMF, HOBt, diisopropylethylamine, EDCI, rt, 18 h, (d) NaH, anh. THF, ethyl bromoacetate, rt, overnight; (e) 10% Pd/C, anh. MeOH, rt, overnight; (f) anh. CH₂Cl₂, diisopropylethylamine, 4-pentenoyl chloride, rt, overnight; (g)(1) LiOH, 3:1 THF–MeOH;(2) anh. DMF, chloroacetonitrile, Na₂CO₃, rt, overnight; (h) pdCpA TBA salt, 9:1 anh. DMF–Et₃N, rt, 15 h; (i) T4 RNA ligase, tRNA-C_OH, ATP, 37 °C, 2 h.

The cyclic peptidyl-pdCpA derivatives (compounds 11, 18, 25, 36 and 44) were all ligated to an abbreviated suppressor tRNA_CUA-C_OH prepared by in vitro transcription to afford the full length activated tRNAs by the use of T4 RNA ligase, as illustrated in Scheme 1 for the conversion of 11 → 12.^31,32 The ligation of the cyclic dipeptidyl-pdCpAs to the abbreviated tRNAs was verified by analysis on 8% polyacrylamine–7 M urea gels at pH 5.0 (Figure 2).³⁴ As noted in the Schemes, these tRNAs were protected with N-pentenoyl protecting groups, and were deprotected prior to their use in a protein synthesizing system by treatment with aqueous iodine at room temperature.^31,32

Figure 2.

Analysis of the activation of a suppressor tRNA transcript with cyclic dipeptides by polyacrylamide gel electrophoresis under acidic conditions. Lane 1, abbreviated tRNA_CUA-C_OH (74 nucleotides). Lanes 2 – 6, ligation products of tRNA_CUA-C_OH with pdCpA derivatives 11, 18, 25, 36 and 44, respectively (76 nucleotides). The lower bands in lanes 2, 3 and 4 comigrated with the starting material (tRNA-C_OH).

Because E. coli ribosomes normally incorporate only α-L-amino acids, we have devised a strategy to alter the 23S ribosomal RNA at key positions, and to select modified ribosomes capable of recognizing activated tRNAs containing non-proteinogenic amino acids.²⁴ These have included D-amino acids, β-amino acids, phosphorylated amino acids, dipeptides and dipeptidomimetics.³⁵ In the present case, we utilized ribosomal clone 010328R2, originally selected to recognize a puromycin derivative that contained the dipeptide phenylalanylglycine.²⁵ Because this clone was found to recognize several other dipeptides and dipeptidomimetic species, we used it in the present study to determine whether it would incorporate conformationally constrained dipeptides 1 – 5 into the model protein E. coli dihydrofolate reductase (DHFR).

Protein synthesis was carried out in triplicate independent experiments using an S-30 preparation with ribosomes from clone 010328R2. As shown in Figure 3, the five dipeptides were all incorporated, although the amounts varied from one analogue to another, and seemed to be incorporated somewhat more efficiently at position 18 than at position 49. The yields are summarized in Table 1. As shown in the Table, at both positions of DHFR, dipeptide analogue 3 (7-membered ring) was incorporated more efficiently than analogue 2 (six-membered ring), which was incorporated more efficiently than analogue 1 (five-membered ring). The suppression yields for 7-membered ring dipeptides 4 and 5 at position 18 (11.2 ± 0.9% and 10.2 ± 0.3%, respectively) were greater than for 7-membered ring 3 (8.9 ± 0.7%). The same relative efficiencies of incorporation were noted at position 49, although the differences between them was smaller (Table 1). Importantly, no significant incorporation of 1 – 5 was noted when an S-30 preparation made from wild-type ribosomes was employed (not shown). The choice of positions 18 and 49 in DHFR for replacement with non-proteinogenic amino acids was made to facilitate characterization as described.³⁶ Mass spectrometric data supporting the incorporation of 4 and 5 into DHFR at position 49 is illustrated in Table S1.

Figure 3.

SDS‒polyacrylamide gel electrophoretic analysis of the synthesis of proteins containing cyclic peptides 1 – 5 at DHFR positions 18 (panel A) or 49 (panel B). (A) Comparison of wild-type DHFR (lane 1) and proteins synthesized using a mRNA with a UAG codon at position 18 in the absence of any suppressor tRNA (lane 2) or in the presence of tRNA_CUA activated with dipeptide 1 (lane 3), dipeptide 2 (lane 4), dipeptide 3 (lane 5), dipeptide 4 (lane 6) and dipeptide 5 (lane 7). (B) Comparison of wild-type DHFR (lane 1) and proteins synthesized using a mRNA with a UAG codon at position 49 in the absence of any suppressor tRNA (lane 2) or in the presence of tRNA_CUA activated with dipeptide 1 (lane 3), dipeptide 2 (lane 4), dipeptide 3 (lane 5), dipeptide 4 (lane 6) and dipeptide 5 (lane 7). In panel B, the stronger lower bands are DHFR truncated at position 49 (no suppression). In panel A, the shorter truncated protein ran off the gel.

Table 1.

Characterization of yield of full length DHFR during in vitro translation from modified genes in the presence of different cyclic dipeptidyl-tRNA_CUAs

Cyclic dipeptides	Amount of full length proteins from modified genes, % wt
	TAG in position 18	TAG in position 49
-	0.6 ± 0.1	0.8 ± 0.2
1	4.2 ± 1.2	3.4 ± 0.8
2	6.3 ± 1.1	4.8 ± 0.7
3	8.9 ± 0.7	7.9 ± 2.5
4	11.2 ± 0.9	8.7 ± 0.8
5	10.2 ± 0.3	8.3 ± 1.2

The suppression yields in these experiments were less than what is sometimes observed for in vitro suppression using wild-type ribosomes. Nonetheless, earlier studies from our laboratory using DHFR^37–39 have explored a range of substitutions enabling several biochemical studies to be carried out, and the yields reported here will clearly allow such cyclic dipeptides to be used in support of mechanistically focused mutagenesis experiments. One further issue that needs to be addressed in future experiments is whether conformational constraint, which has been exceptionally useful in producing peptides with improved properties, can also be exploited productively for protein modifications. In this regard, it may be noted that nature already employs proline to control conformation. This is accomplished both through judicious placement of proline residues at protein sites that can benefit from conformational constraint,⁴⁰ and from cis-trans isomerization of the polypeptide backbone to effect reversible control of protein conformation.⁴¹ It is anticipated that the conformationally constrained dipeptides described here may find utility though the constraint of protein structure.

3. Conclusions

Five cyclic dipeptides have been prepared to enable the study of conformational constraint in proteins, a type of modification that has proven very useful in peptides. The modifications have included lactam-derived dipeptides having five-, six and seven-membered rings (1 – 3), and seven-membered acylhydrazides (diazepanes) 4 and 5. Each of these synthetic dipeptides was activated as a pdCpA ester and used to prepare a suppressor tRNA by the use of T4 RNA ligase. The activated cyclic dipeptidyl-tRNAs were used to introduce the five dipeptides into positions 18 and 49 of E. coli dihydrofolate reductase. The suppression yields obtained in replicate experiments were entirely sufficient to provide the amounts of protein required for routine biochemical experiments.

4. Experimental section

All chemical and biochemical methods are described in detail in the Supplementary Material sections.