Total Synthesis of Nominal Cyclocinamide B and Investigation into the Identity of the Cyclocinamides

Abstract

The total synthesis of nominal cyclocinamide B, a cyclic peptide marine natural product, is reported together with an isomer of nominal cyclocinamide A. Initial attempts at the synthesis of the title compounds by inclusion of a turn inducer failed. However, direct synthesis succeeded in formation of the 14-membered cyclic peptide structure. Comparison of the data from all synthetic cyclocinamide A and B compounds with those of the natural products leads to the conclusion that the two natural products possess the same relative stereochemistry and that the true structures have not been defined.

Macrocycles, defined as containing a ring of at least 12 atoms, make up about 3% of the more than 100,000 known natural product secondary metabolites.¹ Many of these compounds enjoy potent biological activity derived from built-in structural pre-organization that may or may not possess absolute rigidity. Yet few macrocyclic compounds have been developed into pharmaceutical agents, perhaps due to the larger molecular weight and increased heteroatom presence that constitute deviations from the ‘rule of five’.² Cyclic peptides exist as a subset of natural macrocycles. The most celebrated cyclic peptide may well be the immune repressor cyclosporine, an undecapeptide that presents multiple hydrogen bonding units yet is nonetheless membrane permeable and highly biologically active.

Nearly two decades have passed since Crews and coworkers reported the 2D and partial 3D structure of cyclocinamide A (1a, Fig. 1) and claimed solid tumor activity for the unusual 14-membered ring hexapeptide structure.³ Some years later cyclocinamide B (2) was reported⁴ with an almost identical 2D structure but alternate absolute stereochemistry to 1a. A proposal for the complete structure of cyclocinamide A appeared in 2008⁵ with the assignment of the absolute stereochemistry of the two β-amino acid residues. Synthetic work to define the absolute structure of cyclocinamide A (1a) has appeared^6,7 but, to date, the identity of this cyclic peptide has not been verified. No synthetic work has been reported for 2.

Figure 1.

Cyclocinamide structures

Recently we completed the total synthesis of nominal cyclocinamide A 1a and its 11R isomer 1b.⁸ Neither compound provided data supporting identity with the natural product. Given these results, the preparation of nominal cyclocinamide B 2 was undertaken. Herein we present our synthetic work, which also resulted in the production of the enantiomer of an additional stereoisomer of cyclocinamide A (ent-1c), together with a full examination of all data from our synthetic efforts on these 14-membered cyclic peptides.

Our published synthesis of nominal and 11R-cyclocinamide A arose from three dipeptide units, and the parallel effort toward 2 required residues 3-5 (Scheme 1). Key to success in our previous synthesis was the use of a (cyclo)Asn derivative similar to 3 as a turn inducer, allowing the macrocyclization event of the all S and 11R isomers at the C1-N2 bond to proceed in good yield. The synthesis of 3 arises from R-asparagine and S-diaminopropionic acid, and closely follows the published procedure (eq. 1).⁸ Similarly, fragment 4 arises from R-5-bromotryptophan and S-isoserine and is obtained through standard amide bond formation to afford 6, followed by ester hydrolysis (eq. 2). Finally, achiral 5 is prepared from N-methylpyrrole and glycine methyl ester (eq. 3).

Scheme 1.

Initial retrosynthesis and preparation of dipeptides

Unfortunately, a fateful decision was made to employ pivalaldehyde for preparation of 3 rather than the aromatic aldehyde used successfully in the synthesis of 1a and 1b. Both the aromatic and aliphatic aldehyde-derived heterocycles had seen extensive utilization in early (cyclo)Asn studies in this laboratory with equivalent success.^9-11 However, in the present case synthetic intermediates containing dipeptide 3 proved almost uniformly unstable to storage and handling. For example, linear hexapeptide 9 was prepared by removing the –Troc protection on 3 to afford 7 and appending the glycine-pyrrole side chain 5 to give 8 (Scheme 2). After –Fmoc removal, coupling to 4 afforded the desired branched hexapeptide 9, with only the final cyclization and protection group removal remaining. However, compound 9 proved highly unstable and did not succumb to cyclization.

Scheme 2.

Failed attempt at synthesis of 2.

Recognition that the core ring of 2 is enantiomeric with 1c, which was previously synthesized by Grieco via a linear tetrapeptide that did not employ a turn inducer,⁶ led to the synthesis of both nominal cyclocinamide B 2 and ent-1c outlined in Scheme 3, adapted from Grieco's approach. Nitrogen protection was removed from 6 under standard conditions to afford free amine 10, which was coupled to R-Fmoc-Asn(Tr)-OH to afford 11 in good yield, provided the reaction time was kept short. Overnight coupling of 10 with the asparagine residue resulted in formation of a compound analogous to 11 but bearing a nitrile in place of the trityl-protected asparagine side chain amide. Although the formation of a nitrile from the side chain amide functionality of an asparagine residue is well precedented,¹² we are unaware of any former instance of a similar one-pot deprotection/dehydration transformation. Tripeptide 11 was N-deprotected and coupled with diaminopropionic acid derivative 12 over 48h to give linear tetrapeptide 13 in 52% yield for the two steps. As opposed to the previous coupling, no nitrile formation occurred over the 2 day reaction nor was any nitrile detected following any of the other coupling reactions in the sequence, some of which extended to 4 days. Removal of the –Fmoc and methyl ester protection groups proceeds only in this order; no product is isolated if the ester is deprotected first. Furthermore, production of the cyclization precursor was accompanied with some undesired loss of isoserine hydroxyl protection. Fortunately, the cyclization precursor was quite stable, in stark contrast to compound 9 in Scheme 2.

Scheme 3.

Synthesis of target macrocycles.

Ring formation, which again was accompanied by minor adventitious loss of C4 protection, afforded a combined 79% yield of cyclic material after extensive experimentation to define DEPBT¹³ as the optimal coupling reagent. Subsequent to separation, C4 protected material 14 was treated with dilute TFA/Et₃SiH to remove the –Boc group. Coupling with dipeptide fragment 5 was sluggish, requiring 4 days to provide a mixture of 15 and 16. Curiously, the formerly labile –OTBDPS group became robust upon macrocycle formation; 15 required two steps (TFA, then NaOH) for complete deprotection to desired final product 2. Compound 16 required only base treatment to yield nominal cyclocinamide B. Finally, ent-1c was produced via a parallel series of reactions on 14, with –Boc removal and coupling to fragment 17⁸ to afford 18. Removal of the trityl and –OTBDPS groups, the latter with TBAF, gave the target compound. NMR data obtained from the synthetic samples¹⁴ 1a, 1b, 2 and ent-1c are given in Tables S1-S4,¹⁵ respectively, and support the connectivity represented in Fig. 1. Our working hypothesis was that the macrocycle sp³ proton and carbon chemical shifts would be least sensitive to small impurities, as compared to the hydrogen bonding amide units, while maintaining high sensitivity to changes in absolute stereochemistry. This hypothesis could be tested with the synthetic samples. The spectra of ent-1c and 2, which differ only in the presence of the substituent at distal C36 (H vs. Cl), should be nearly identical in the macrocycle, whereas the spectra of 1a and 1b, identical structures except for the absolute stereochemistry at C11, should display distinct differences in the core.

Table 1 gives the data for both comparisons, and bear out the discussion above. Substantial differences exist between the ¹³C resonances at C11 when comparing 1a and 1b (3.9 ppm), and a correspondingly large difference (0.41 ppm) is seen in the ¹H resonances between the C10 positions of these two compounds. Conversely, ent-1c and 2 have very similar spectra when comparing the macrocycle sp³ resonances, with carbon differences at or below 0.4 ppm and proton differences at or below 0.08 ppm. These latter numbers, particularly the ¹³C data, can act as a threshold for further comparison between spectra. Reinforcement of this analysis is seen in the nearly identical optical rotation values, in both sign and magnitude, of ent-1c ([α]_D = −12.5) and 2 ([α]_D = -15.9) recorded in MeOH.

Table 1.

Comparison of Cyclic Peptide sp^{3 1}H and ¹³C resonances

	ent-1c		2		Δδ 1H	Δδ 13C	1a		1b		Δδ 1H	Δδ 13C
Atom number	1H	13C	1H	13C	(ent-1c)-(2)	(ent-1c)-(2)	1H		1H	13C	(1a)-(1b)	(1a)-(1b)
3 (CH2)	3.37	42.9	3.29	43.2	0.08	−0.3	2.81	42.4	2.85	42.3	−0.04	0.1
	3.56		3.51		0.05		3.59		3.56		0.03
4* (C-H,OH)	4.06	70.8	4.03	70.8	0.03	0	3.75	69.2	3.87	69.1	−0.12	0.1
7* (C-H,R)	4.64	52.1	4.61	52.4	0.03	−0.3	4.43	53.1	4.47	53.2	−0.04	−0.1
10 (CH2)	3.29	40.5	3.36	40.4	−0.07	0.1	2.81	40.3	3.22	40.4	−0.41	−0.1
	3.42		**				3.45		3.56		−0.11
11* (C-H,N)	4.34	54.3	4.34	54.0	0	0.3	4.17	51.6	4.19	55.5	−0.02	−3.9
14* (C-H,R)	4.55	49.7	4.54	50.1	0.01	−0.4	4.52	49.0	4.58	49.0	−0.06	0

The bar graph in Figure S1 gives a comparison of the ¹H NMR data for the core sp³ protons of natural cyclocinamide A¹⁶ (CC-A, blue) and synthetic cyclocinamide B (2, red) to natural cyclocinamide B (CC-B), and ent-1c to natural CC-A (green). Figure S2 gives the analogous data derived from the ¹³C spectra of the compounds. It should be noted that the chemical shifts used in these figures for natural CC-B result from an adjustment of the ¹H and ¹³C spectra references to 2.50 ppm and 39.52 ppm, respectively, as these were the references used in this laboratory for all synthetic samples.

It is apparent from these figures that the two natural samples, once they are set to the same reference peaks, are within the threshold set by the comparison of ent-1c and 2 for compounds that have the same array of stereocenters. With the exception of the C10 ¹³C value, which shows a 0.5 ppm difference, all ¹H and ¹³C Δδ values reside within the threshold limits. The carbon spectra are particularly close, with three of the six sp³ centers identical. The relative stereochemical identity of CC-A and CC-B is an attractive conclusion, suggesting that these two natural products arise from the same biosynthetic machinery. The reported optical rotations bear the same sign but differ in magnitude: [α]_D = 9.6 for natural CC-B and [α]_D = 29 for the initial isolation of CC-A.

Comparison of synthetic CC-B (2) to its natural counterpart affords a less clear-cut conclusion. The proton comparison in Figure S1 shows two centers with differences greater than the 0.08 ppm benchmark, and several centers in the carbon comparison fall beyond the 0.4 ppm limit. Furthermore, the sign of the optical rotation for the synthetic 2 (vide supra) is opposite that of the natural material. Likewise, the chemical shifts of the ent-1c and natural CC-A cores, while nicely aligned in the proton comparison, deviate significantly in the C4 and C7 carbon comparison.

In conclusion, spectral data reexamination supports the notion of identical relative stereochemistry among the four chiral centers of cyclocinamide A and B. However, there exists no compelling data to suggest that either of these cyclic peptide natural products has been prepared in the laboratory. Further studies to determine the true 3D structures of cyclocinamide A and B are continuing and will be disclosed in due course.