Towards a streamlined synthesis of peptides containing α,β-dehydroamino acids

Abstract

Investigation of a strategy to streamline the synthesis of peptides containing α,β-dehydroamino acids (ΔAAs) is reported. The key step involves generating the alkene moiety via elimination of a suitable precursor after it has been inserted into a peptide chain. This process obviates the need to prepare ΔAA-containing azlactone dipeptides to facilitate coupling of these residues. Z-dehydroaminobutyric acid (Z-ΔAbu) could be constructed most efficiently via EDC/CuCl-mediated dehydration of Thr. Formation of Z-ΔPhe by this or other dehydration methods was unsuccessful. Production of the bulky ΔVal residue could be accomplished by DAST-promoted dehydrations of β-OHVal or by DBU-triggered eliminations of sulfonium ions derived from penicillamine derivatives. However, competitive formation of an oxazoline byproduct remains problematic.

Graphical Abstract

Peptides are of great medicinal importance due to their ability to fill the space between small molecules and biologics.¹ However, their prospects as pharmaceuticals have been impeded by their short half-lives due to rapid proteolysis. As a result, several strategies have been devised for enhancing the proteolytic stability of peptides.^2-10 Our own contributions to this area were inspired by Stammer and co-workers, who demonstrated many years ago that α,β-dehydroamino acids (ΔAAs) such as ΔAla, Z-ΔPhe, and Z-ΔLeu are able to protect modified enkephalins from proteolysis.¹¹ The stabilizing effects of ΔPhe and ΔLeu are presumably a consequence of A_1,3 strain, which increases rigidity by reducing the number of low-energy conformations. This phenomenon amplifies the thermodynamic preference for folded states over unfolded random coils, the latter of which are cleaved more rapidly by proteases than the former (Fig. 1).¹²

Fig. 1.

Dehydroamino acids.

We have shown that incorporating a bulky ΔAA such as ΔVal into the turn region of a model β-hairpin peptide can significantly enhance its proteolytic stability without altering its secondary structure.¹³ Recently, we found that peptides containing either a bulky or a medium-sized ΔAA (i.e., a ΔAA possessing at least one allylic alkyl group) are inert even at elevated temperatures to conjugate additions by nucleophilic thiols.¹⁴ Together, these discoveries suggest that inserting ΔAAs at appropriate sites in bioactive peptides will increase their proteolytic stability without altering their shapes or rendering them susceptible to degradation via conjugate addition. Thus, ΔAAs represent a promising tool for boosting the viability of peptides as drugs.¹⁵

Our studies of ΔAA-containing peptides have primarily relied on oxazolone (azlactone) ring-openings for installation of the unsaturated residues.^13,14 The requisite azlactones can be readily generated from peptides containing a C-terminal β-hydroxy amino acid via tandem dehydration–cyclization¹⁰ (Scheme 1). Unfortunately, the yields of the ring-openings are often modest.¹⁴ Moreover, adaptation of this chemistry to solid-phase peptide synthesis (SPPS) requires an unwieldy solution-phase preparation of azlactone dipeptides.¹⁷ Other known methods of incorporating ΔAAs into peptides that bypass azlactones require cumbersome protecting group chemistry^18,19 and would also necessitate lengthy construction of dipeptide intermediates for SPPS applications. Inoue and co-workers have devised a traceless Staudinger ligation that allows installation of ΔAAs into peptides via SPPS.²⁰ This protocol avoids many of the drawbacks inherent to other ΔAA synthesis methods, but it requires the use of air-sensitive phosphinophenols and the construction of alkenyl azides. Thus, the discovery of an operationally simple process for integrating ΔAAs into peptides that employs readily accessible amino acid derivatives would be highly impactful.

Scheme 1.

Azlactone ring-openings for ΔAA installation.

In principle, an efficient approach to installing ΔAAs into peptides would entail coupling of a suitable ΔAA precursor (e.g, a β-hydroxy amino acid) via standard protocols followed by additional couplings and generation of the alkene after this residue is embedded within a peptide chain. By unveiling the ΔAA after rather than before peptide coupling, this strategy would eliminate the need to prepare azlactone dipeptides or other related ΔAA-containing dipeptide building blocks in solution, thereby streamlining the construction of ΔAA-containing peptides by SPPS. Whereas dehydrations of β-hydroxy acids or esters to afford C-terminal ΔAAs are facile,^16,21 dehydrations of β-hydroxy amides to forge ΔAAs embedded within peptides are quite challenging and, in our experience, have not been viable with residues larger than threonine.^14,21 Herein, we report studies targeting the synthesis of peptides containing medium-sized and bulky ΔAAs involving initial coupling of a ΔAA precursor and subsequent generation of the unsaturated residue. While this work was conducted in solution, it sets the stage for further studies using SPPS.

We targeted the formation of two common medium-sized ΔAAs (Z-dehydroaminobutyric acid or Z-ΔAbu and Z-ΔPhe) and one bulky ΔAA (ΔVal). We synthesized model tripeptides 5 and 8, which are required for studying dehydrations of Thr and β-OHPhe, as outlined in Scheme 2. Standard peptide coupling and deprotection chemistry was employed. The yields of some couplings were modest, but extensive optimization was eschewed in favor of focusing efforts on the dehydration reactions.

Scheme 2.

Synthesis of model tripeptides 5 and 8.

Two strategies were investigated for installing ΔVal: dehydrations of β-OHVal derivatives, and sulfonium²² or sulfide²³ eliminations of S-alkylated penicillamines (Pen). The requisite β-OHVal-containing tripeptide 11 and Pen-containing tripeptide 14 were constructed as shown in Scheme 3. Racemic β-OHVal-OEt (9) was employed due to its ready availability via OsO₄-catalyzed aminohydroxylation²⁴ and the fact that an enantiopure residue was not required for the dehydration studies. Saponification of Gly-β-OHVal dipeptide 10 was complicated by retro aldol scission of the β-hydroxy amino acid. This could be minimized by using Me₃SnOH²⁵ as the base, and subsequent coupling of Gly-OMe•HCl (1) furnished tripeptide 11. Fortunately, elaboration of Fmoc-Pen(Trt) (12) into tripeptide 14 was straightforward.

Scheme 3.

Synthesis of model tripeptides 11 and 14.

The dehydrations of Thr-containing peptide 5 are summarized in Table 1. Competition between the desired dehydration pathway affording Z-ΔAbu-containing 15 and an undesired cyclization pathway yielding oxazoline 16 was significant, but each reaction produced a single detectable product rather than mixtures of both products. For example, treatment of 5 with MsCl followed by DBU²⁶ furnished exclusively 16 when 1,2-dichloroethane (DCE), CH₃CN, and DMF were employed as solvents (Entries 1, 3, and 4). However, using a larger excess of MsCl and performing a solvent switch from DCE in the mesylation step to CH₃CN in the elimination step delivered only 15 in low yield (Entry 2). Exposure of 15 to the conditions of Entry 3 did not produce oxazoline 16, demonstrating that the ΔAA is not an intermediate leading to the cyclized byproduct. Reaction of 5 with Boc₂O and DMAP followed by 1,1,3,3-tetramethylguanidine (TMG)²⁷ also furnished 15 (Entries 5 and 6), but attempts to increase the modest yields by elevating the reaction temperature were unsuccessful (Entry 7). Given the success of others in employing EDC•HCl and CuCl for Thr dehydrations,^28,29 we examined this reagent combination extensively (Entries 8–12). The highest yield of 15 was obtained by enlisting CH₂Cl₂ as solvent with an extended reaction time (Entry 12). Switching from CuCl to Cu(OTf)₂³⁰ was not fruitful (Entry 13). Finally, treatment of 5 with PPh₃ and I₂³¹ afforded a modest yield of oxazoline 16 (Entry 14).

Table 1.

Dehydrations of Thr-containing peptide 5.

Entry	Conditions	Temp	Time (h)	Producta
1	MsCl/Et3N, then DBUb	rt, then reflux	17	16 (38)
2	MsCl/Et3N, then DBUc	rt, then reflux	17	15 (31)
3	MsCl/pyr, then DBUd	rt, then reflux	17	16 (70)
4	MsCl/Et3N, then DBUe	rt, then reflux	17	16 (20)
5	Boc2O/DMAP, then TMGd	0 ° C to rt	12	15 (18)
6	Boc2O/DMAP, then TMGd	0 ° C to rt	24	15 (44)
7	Boc2O/DMAP, then TMGd	0 ° C to 40 °C	12	ndf
8	EDC/CuCl, CH2Cl2	rt	16	16 (9)
9	EDC/CuCl, CH2Cl2–DMF	rt	16	15 (7)
10	EDC/CuCl, CH3CN	rt	16	16 (7)
11	EDC/CuCl, CH2Cl2	reflux	16	15 (55)
12	EDC/CuCl, CH2Cl2	rt	40	15 (84)
13	EDC/Cu(OTf)2, THF	60 °C	1.5	15 (6)
14	PPh3/I2, CH2Cl2	rt	4	16 (20)

^aIsolated yield in parentheses.

^bDCE as solvent.

^cDCE as solvent for mesylation, then CH₃CN for elimination; 4 equiv of MsCl used instead of 2 equiv employed in other reactions.

^dCH₃CN as solvent.

^eDMF as solvent.

^fNeither product was detected.

The most promising methods for dehydrating 5 were then applied to β-OHPhe-containing peptide 8. Surprisingly, treatment of 8 with EDC•HCl and CuCl afforded only trace amounts of the desired Z-ΔPhe-containing tripeptide. The use of Boc₂O and DMAP followed by TMG yielded comparable results. Attempted dehydrations of 8 employing DAST and pyridine³² were also unsuccessful. The increased bulk of the phenyl group relative to the methyl group present in Z-ΔAbu presumably impedes the dehydration of β-OHPhe-containing peptides such as 8.

The dehydrations of β-OHVal-containing peptide 11 were also challenging, presumably due to the substantial A_1,3 strain inherent in the tetrasubstituted alkene moiety of ΔVal (Table 2). Treatment of 11 with SOCl₂ and pyridine³³ in CH₂C1₂ as solvent furnished the desired ΔVal-containing peptide 17 in low yield (Entries 1 and 2). Switching the solvent to CH3CN afforded a modest yield of a separable mixture of 17 and oxazoline 18 (Entry 3). Exposure of 11 to the Burgess reagent³⁴ produced only trace amounts of 17 (Entry 4). The use of DAST and pyridine³² delivered 17 in an encouraging 38% yield (Entry 5). Unfortunately, all attempts to optimize this result were futile, largely resulting in mixtures of 17 and 18 or producing 18 exclusively (Entries 6–12). Subjection of 17 to the conditions of Entry 7 did not yield 18, once again indicating that the ΔAA is not a precursor to the oxazoline. The EDC•HCl/CuCl conditions that gave the best results for generating Z-ΔAbu were not examined for the dehydration of 11 because efforts to use these reagents to form ΔVal in the context of our total synthesis of yaku’amide A¹⁶ were unsuccessful.³⁵

Table 2.

Dehydrations of β-OHVal-containing peptide 11.

Entry	Conditions	Temp (°C)	Time (h)	Productsa
1	SOCl2/pyr, CH2Cl2	0	1	17(18)
2	SOCl2/pyr, CH2Cl2	0	2	17(14)
3	SOCl2/pyr, CH3CN	0	1	17(25), 18(15)
4	Burgess, CH3CN	70	12	17(5)
5	DAST/pyr, CH2Cl2	0	0.5	17(38)
6	DAST/pyr, CH2Cl2	−30	0.5	17(27), 18(50)
7	DAST/pyr, CH2Cl2b	0	0.5	17(13), 18(58)
8	DAST/DMAP, CH2Cl2	0	0.5	17(31), 18(21)
9	DAST/2,6-lut, CH2Cl2	0	0.5	17(5), 18(46)
10	DAST/DBU, CH2Cl2	0	0.5	18(40)
11	DAST/pyr, PhCH3	0	0.5	17(19), 18(30)
12	DAST/pyr, THF	0	0.5	17(8), 18(25)

^aIsolated yield in parentheses.

^b2 equiv of DAST used instead of 1 equiv employed in other reactions.

To improve on the above results, we examined the eliminations of sulfonium ions²² derived from Pen-containing peptide 14 (Table 3). Treatment of a solution of 14 in DMF with 1,4-diiodobutane and K₂CO₃ resulted in facile conversion to a mixture of 17 (24%) and 18 (54%, Entry 1). Extensive optimization efforts involving different bases, solvents, temperatures, and reaction times succeeded in raising the yield of 17 to 53% (Entry 5). However, substantial quantities of oxazoline 18 were still produced (46%). We also attempted Ag⁺-promoted elimination of the methyl sulfide derived from 14,²³ but neither 17 nor 18 were obtained (Entry 9).

Table 3.

Eliminations of Pen-containing peptide 14.

Entry	Conditions	Temp	Time (h)	Productsa
1	I(CH2)4I, K2CO3, DMF	rt	4	17(24), 18(51)
2	I(CH2)4I, Na2CO3, DMF	rt	4	17(31), 18(32)
3	I(CH2)4I, Cs2CO3, DMF	rt	4	17(15), 18(72)
4	I(CH2)4I, DBU, DMF	rt	4	17(40), 18(52)
5	I(CH2)4I, DBU, DMF	0 °C	4	17(53), 18(46)
6	I(CH2)4I, DBU, DMF	rt	16	17(21), 18(59)
7	I(CH2)4I, DMAP, DMF	rt	4	17(28), 18(5)
8	I(CH2)4I, DBU, CH3CN	rt	4	17(34), 18(29)
9	MeI/K2CO3, DMF; then AgNO3/DBN, CH3CN	rt	3	ndb

^aIsolated yield in parentheses.

^bNeither product was detected.

In conclusion, we have attempted to streamline the synthesis of ΔAA-containing peptides by investigating a strategy that entails incorporation of a ΔAA precursor into a peptide chain followed by generation of the alkene moiety via elimination after the key residue has been embedded within the peptide chain. We found that EDC•HCl and CuCl provide the best outcomes in the dehydration of Thr to forge Z-ΔAbu-containing peptide 15. Dehydrations of β-OH-Phe-containing peptide 8 were largely unsuccessful, revealing a gap in the methodology that is worthy of further study. While DAST-promoted dehydrations of 11 and DBU-mediated sulfonium eliminations of 14 delivered encouraging yields of ΔVal-containing peptide 17, further optimization including reduction of the amount of oxazoline 18 produced is necessary. In this regard, it is possible that the conformational flexibility of the Gly-containing substrates, which was intended to facilitate dehydration, may also be enabling oxazoline formation. Accordingly, substrates incorporating more hindered residues will be the subject of future studies, along with adaptation of the dehydration methods to SPPS.