Total Synthesis and Stereochemical Assignment of Micrococcin P1

Abstract

Fifty years after the discovery of the thiopeptide antibiotic micrococcin P1, the constitutional and stereochemical uncertainties concerning its structure have been lifted with its total synthesis and absolute stereochemical assignment (see picture).

This paper describes the total synthesis of the thiopeptide antibiotic micrococcin P1 (MP1, 1; Figure 1),1 thereby establishing its constitution and its configuration. Compound 1 is the major component of “micrococcin P”, a cytotoxic extract isolated from Bacillus pumilus that consists of a roughly 7:1 mixture of 1 and of the corresponding ketone, 2, which is termed micrococcin P2 (MP2). MP1 binds tightly to ribosomes, thereby disrupting protein synthesis.2 The compound thus exerts a potent antibiotic activity toward microorganisms,3 including the malarial parasite Plasmodium falciparum.4

Figure 1. Actual structure of micrococcin P1 (1): R=iPr, R′=H; Z=OH, Z′=H. Actual structure of micrococcin P2 (2): R=iPr, R′=H; Z, Z′=O. Bycroft–Gowland structure of MP1 (4): R=H, R′=iPr; Z=OH, Z′=H. Synthetic “micrococcin P1” 13b,d (6): R=H, R′=iPr; Z=H, Z′=OH.

While MP15 is one of the structurally less complex thiopeptides,6 its precise structure has remained uncertain for over 50 years. The constitution of the central pyridine–thiazole cluster was firmly established by X-ray diffractometry.7 Important work by Walker and Mijovic ascertained that the 1-amino-2-propanol segment of 1 (see Figure 2, region c) has the D-(R) configuration8 and that MP1 incorporates a L-threonine unit.9 Walker et al. also advanced the hypothesis10 that the valine-derived thiazole in region a of the molecule had the (R) configuration, thus implying that the thiazole in question is a formal derivative of D-valine. This would make MP1 unique among thiopeptide antibiotics, all of which incorporate thiazole segments derived from L-amino acids. On the basis of these data and of a presumed similarity with other thiopeptides, in 1977 Walker and Lukacs proposed structure 3 for MP1 (Figure 2),11 but without the benefit of evidence in support of the alleged topography of the macrocycle. Shortly thereafter, new chemical evidence induced Bycroft and Gowland to promulgate the revised structure 4 (Figure 1).12 The latter authors were unable to assign the configuration of region b of the molecule, which, correctly, they left undefined. Moreover, they also left unresolved the issue of macrocycle topography. Errors possibly present in the Walker assignment thus propagated to the revised structure, which nonetheless gained tacit acceptance and gradually came to be consistently represented with the configuration shown.

Figure 2. Walker–Lukacs structure of MP1 (3): Z=OH, Z′=H. Synthetic “micrococcin P” 13a,c (5): Z=H, Z′=OH.

Remarkably, past synthetic work has been unable to resolve the structural uncertainties surrounding MP1. Indeed, synthetic epimers of the Walker–Lukacs (see compound 5, Figure 2)13a,c and of the Bycroft–Gowland (see 6, Figure 1)13b,d structures have both been stated to be identical to the natural product. Not only the two structures are mutually exclusive: they also possess the (S) configuration, instead of the secure (R) configuration, at c.14 Even more problematic is the fact that synthetic 4 (Figure 1) is not identical to natural MP1.15

Extensive NMR studies ultimately confirmed the Bycroft–Gowland constitution of MP1,16 and by default that of MP2, ruling out the possibility that MP1 may be 3 or 5, and implying that the difference between 4 and natural MP1 must be purely stereochemical. While spectroscopic methods failed to unravel the relative configuration of the natural product, incisive work by Bagley and Merritt17 led to the conclusion that MP1 is likely to be 1. Total synthesis now confirms this surmise.

The retrosynthetic logic that directed the construction of 1 is delineated in Scheme 1. Experience had revealed the necessity of minimizing chemical operations after macrocycle formation. Accordingly, MP1 would emerge upon the union of a pair of suitably COOH- and NH₂-protected segments, A and B. Past experience had also shown that macrocyclization was facile only if the order of bond formation was (a) first, then (b). In turn, each segment was accessible by means of the fusion of a triad of appropriately protected subunits: C–E for A; F–H for B.

Scheme 1. Retrosynthetic disconnection of micrococcin P1 (1) into fragments C–H. Pg=protecting group.

Building blocks 8, 10, and 11 were thus prepared from the known 75a, 15 and 918 as previously described (Scheme 2).15 A challenging aspect of the synthesis of 1 was the assembly of the central pyridine–thiazole cluster, an objective that is best attained through a Hantsch-type pyridine construction proceeding through the merger of 10 with 12.19 The proclivity of 12 to undergo base-promoted polymerization precluded the implementation of traditional procedures for the initial Michael reaction leading to intermediate 13 (Scheme 3). Thus, the union of 10 and 12 could be achieved only through the use of a heterogeneous catalytic system comprising powdered Li₂CO₃ in EtOAc (Scheme 3).18 The resultant 13 was converted into 14, and the latter was then advanced to the complete pyridine core of MP1, 21.

Scheme 2. Synthesis of fragments 8, 10, and 11. a) DCC, (R)-isoalaninol, CH₂Cl₂, RT, overnight; b) Ac₂O, DMAP, pyridine, 2 h, 85 % a–b; c) 4 n HCl in dioxane, 20 min, then addition of H₂O, 15 min, 100 %; d) 3 equiv of 2-(lithiomethyl)-4-(tert-butyldimethylsilyloxy)methylthiazole, THF, −78 °C, 81 %; e) Boc₂O, Et₃N, DMAP, 99 % (crude); f) LiOH, 50 % aq. THF, then acidification to pH 3 with NaH₂PO₄ sol., 95 % (crude). TBS=tert-butyldimethylsilyl, DCC=N,N′-dicyclohexylcarbodiimide, DMAP=4-dimethylaminopyridine, Boc=tert-butoxycarbonyl.

Scheme 3. Construction of the pyridine–thiazole cluster of MP1. a) 10, cat. Li₂CO₃, EtOAc, 92 %; b) NH₄OAc, EtOH then DDQ, toluene, 97 %; c) LiOH, H₂O, THF; d) Boc₂O, DMAP, Et₃N, DCM; e) 8, BOP-Cl, Et₃N, CH₃CN, 77 % over 3 steps (c–e); f) MsCl, Et₃N, then DBU, DCM; g) TBAF, THF; h) Dess–Martin periodinane, NaHCO₃, DCM, 88 % over three steps (f–h); i) NaClO₂, 2-methyl-2-butene, NaH₂PO₄, THF, H₂O, 84 %. DDQ=2,3-dichloro-5,6-dicyano-1,4-benzoquinone, DCM=dichloromethane, BOP-Cl=bis(2-oxo-3-oxazolindinyl)phosphinic chloride, MsCl=mesityl chloride, DBU=1,8-diazabicyclo[5.4.0]undec-7-ene, TBAF=tetrabutylammonium fluoride.

Parallel work reached 27 through the sequence outlined in Scheme 4. Owing to the propensity of valine-derived thiazole 2220 and of its derivatives to racemize/epimerize,21 the stereochemical integrity of each intermediate in this sequence was ascertained by ¹³C and ¹⁹F NMR scrutiny of Mosher derivatives. No racemization/epimerization occurred during subsequent transformations. This was also apparent from the ¹H NMR spectra of intermediates 23, 25, and 26, wherein a single diastereomer was discernible.

Scheme 4. Synthesis of segment 27 of MP1. a) 11, HOBt, DCC, CH₂Cl₂, 84 %; b) 4 M HCl in dioxane, 100 %; c) 7, DCC, CH₂Cl₂, 81 %; d) MsCl, Et₃N, CH₂Cl₂, then DBU, 93 %; e) 4 n HCl in dioxane, then H₂O, THF, 100 %. HOBt=1-hydroxy-1H-benzotriazole.

The final sequence of the synthesis (Scheme 5) commenced with the coupling of 21 and 27 to furnish 28, which subsequently underwent deblocking and macrocyclization (reaction with DPPA).22 This produced compound 1 contaminated with a byproduct of unknown structure and with similar chromatographic characteristics. This contaminant appeared to be present also in an aged sample of natural micrococcin P1.23 At this time, we believe that the unknown material is likely to be a product of dehydration of the threonine-derived thiazole segment comprising region b of the molecule (see Scheme 2). In any event, purification of synthetic 1 was accomplished by HPLC. Purified 1 was chromatographically (HPLC, TLC) indistinguishable from authentic micrococcin P1, and its optical rotation {=+68° (90 % aq. EtOH, c=0.45 g cm⁻³; lit. =+63.7° (c=1.19 g cm⁻³, 90 % aq. EtOH)^[24]} and ¹H and ¹³C NMR spectra are coincident with those of authentic MP1. This established the identity of 1 to the natural product.

Scheme 5. Total synthesis of micrococcin P1. a) BOP-Cl, Et₃N, 27, MeCN, 73 %; b) LiOH, THF/H₂O (1:1); c) 4 n HCl in dioxane; d) DPPA, Et₃N, DMF, 24 h, 41 % over 3 steps (b–d). DPPA=diphenylphosphoryl azide.

In summary, chemical synthesis has now settled the structural ambiguities that have surrounded micrococcin P1 during more than fifty years since its discovery. The methods detailed herein are applicable to a number of other synthetically appealing thiopeptide antibiotics, and developments in this domain will be the subject of future reports.