Corrigendum to “Synthesis and Stability Studies of a Simplified, Thiazole-Containing Macrocycle of the Anticancer Agent Salarin C

In our published manuscript describing the synthesis of a highly simplified version of an oxazole-containing macrocycle, previously synthesized by Lindel,⁹ and the corresponding thiazole-containing macrocycle of salarin C, due to a series of unfortunate oversights, it has become clear that the thiazole macrocycle was never synthesized but in fact only the same oxazole, previously synthesized was prepared a second time since sulfur was not introduced during an attempted conversion of the amide to thioamide using Lawesson’s reagent by a second student. When attempting to repeat the synthesis of a thiazole fragment recently in our efforts toward salarin C, through the reported 3-step sequence proceeding from amide 10, involving thiolation by Lawesson’s or Belleau’s reagent, we could only obtain traces of the corresponding thioamide. Thus, a simplified thiazole-containing macrocycle 3a was never in fact synthesized. Direct comparisons of similar O- vs S-containing intermediates clearly show expected differences and a direct comparison of the two macrocycles shows distinct but subtle differences in chemical shifts (see Figure S1). To date, we do not have a rationale for how we found differential rates of Wasserman rearrangement other than the presence of different impurities in the two samples (as seen in the published NMR data in the Supporting Information). To rectify this error, we now report (i) the synthesis of the targeted thiazole-containing macrocycle 3a through a completely different synthetic strategy (ii) resynthesis of the oxazole-containing macrocycle through a slightly modified strategy which avoids the previously uncharacterized regioisomeric Suzuki coupling product thought to be a minor alkene isomer (iii) an updated, side-by-side stability study of these two macrocycles as monitored by both ¹H NMR and LC-MS over 13 days. The principal finding remains the same, namely that there is a definite difference in behavior of the two macrocycles when exposed to light and air. Namely, the oxazole-containing macrocycle 3b undergoes both Wasserman rearrangement and alkene isomerization however, the thiazole-containing macrocycle 3a in fact does not undergo Wasserman rearrangement, but rather only undergoes alkene isomerization at rates similar to the Wasserman rearrangement of the oxazole-containing macrocycle 3b.

The updated sections below should replace paragraphs 5–10, Schemes 2–4, and Table 1 in our orginal publication.

Scheme 2.

Simplification of SalaC and Retrosynthesis of Thiazole and Oxazole-SalaC Macrocycles 3a and 3b

Scheme 4.

Synthesis of a Simplified Thiazole-Containing SalaC Macrocylce 3a

Our proposed syntheses of the two, simplified macrocycles 3a and 3b employed similar disconnections for simplicity with an eye toward scalable routes that could potentially be modified to access derivatives of thia-SalaC (Scheme 2). Key steps in our syntheses of the targeted, simplified macrocycles 3a, 3b ultimately included formation of the oxazole through cyclodehydration and oxidation; formation of the thiazole containing coupling partner from commercially available ethyl 2-bromothiazole-4-carboxylate; an sp²-sp³ Suzuki−Miyaura coupling of diene 7 and vinyl bromides 8a and 8b, successfully utilized by Altmann toward SalaC;¹ a cationic Heck reaction to install the required Z,E-dienoate; and a final macrolactonization.

Synthesis of the simplified thiazole macrocycle 3a (Scheme 3) commenced with a Sonogashira coupling between commercially available ethyl 2-bromothiazole-4-carboxylate 9 and trimethylsilyl-propyne to furnish the known thiazole 10 in 82% yield.^2,3 Methylation at the 5-position to afford thiazole 11 proceeded by lithiation with LiHMDS and excess MeI (50%).⁴ Rapid addition of LiHMDS was critical to reduce the addition of the resulting lithiated thiazole into the ethyl ester leading to dimerization. Subsequent treatment of the alkyne with LiBr/LiOAc in AcOH provided the desired (Z)-vinyl bromide (>19:1 Z/E)⁵ and direct semi-reduction of the ethyl ester employing DIBAl-H, gave aldehyde 8c in 88% yield over two steps. Interestingly, and in contrast to reduction of the corresponding oxazole ester, full reduction to the alcohol with DIBAl-H was not observed at −78 °C, eliminating the need for an undesirable redox manipulation. Installation of the terminal alkene utilizing standard Wittig olefination conditions resulted in significant elimination of the vinyl bromide and reformation of the alkyne. However, Julia-Kocienski olefination conditions using NaHMDS and 5-methanesulfonyl-1-phenyl-1H-tetrazole 12 at −78 °C furnished the desired thiazole coupling partner 8a in 78% yield (> 19:1 Z/E).⁶ Hydroboration-Suzuki-Miyaura coupling between thiazole 8a and the known diene 7, available in 3 steps from propionic acid,⁷ provided alkene 5a in 57% yield which was inseparable from a C12,C13 reduced by-product (not shown; see SI, p. S7, S2, <10%). This reduction presumably arises from a second hydroboration, either before or after coupling to the thiazole, followed by protodeborylation. This inseparable reduced by-product was carried through subsequent reactions and more readily removed following the final macrocyclization step (vide infra). Heck coupling of alkene 5a with (Z)-vinyl iodide 6 delivered ester 13 in 34% yield with high retention of stereochemistry for the C2,C4-(Z,E)-diene (>19:1).⁸ Desilylation of the primary TBDPS group with TBAF gave alcohol 4a (85% yield) and macrolactonization of the seco ester was completed with the use of TMSOK at high dilution in 15 min to yield the desired thiazole containing macrocycle 3a in 45% yield. Longer reaction times led to significantly reduced yields as eventual ring opening of the lactone led to the corresponding seco acid. An analytically pure sample for characterization and subsequent stability studies was prepared through purification by preparative reverse-phase HPLC.

Scheme 3.

Synthesis of a Simplified Thiazole-Containing SalaC Macrocycle 3a

For comparative stability studies, we also synthesized a simplified, oxazole-containing SalC macrocycle 3b, identical to that reported by Lindel and co-workers (Scheme 4).⁹ The synthesis mirrored that of the thiazole variant with a few notable changes, including a de-novo synthesis of the oxazole ring, as the oxazole variant was not commercially available, as well as changing the side chain of the coupling partner to the benzyl ester instead of a terminal alkene, as we observed decreased stability of the alkene. A new palladium source was also required for the Suzuki-Miyaura coupling to prevent an undesired regioisomer not previously seen with in the synthesis of thiazole 3a.

Synthesis of the corresponding oxazole-vinyl bromide fragment started with the hydrobromination of 2-butynoic acid 14¹⁰ with NaBr in AcOH followed by amidation with threonine benzyl ester, through in-situ generation of the acid chloride, to afford the desired amide 15 in 95% yield (>19:1 Z/E) over two steps. Subsequent treatment with DAST afforded the desired oxazoline 16 which was oxidized using radical oxidation conditions developed by Meyers based on the Kharasch-Sosnovsky reaction¹¹, delivering the desired oxazole-vinyl bromide 8b in 65% yield (>19:1 Z/E) over 2 steps. The resulting terminal alkene after reduction and olefination, was found to be highly unstable during subsequent coupling reactions so we proceeded forward with the benzyl ester in place, opting to install the olefin later. As a result of the benzyl ester being in place, and the increased inductive effects of the oxazole compared to the thiazole, an undesired regio isomer S6 was observed during the subsequent Suzuki-Miyaura coupling. This could potentially be explained by findings outlined by Zhang and co-workers¹² where more electron deficient aryl-bromides favor β-hydride elimination, after transmetallation, followed by hydropalladation to now couple at the undesired secondary carbon. Zhang found that the use of rac-BINAP significantly suppressed this undesired β-hydride elimination. We applied these conditions using Pd(rac-BINAP)Cl₂ and found we could completely eliminate the wrong regioisomer, giving the desired alkene 17 in 86% yield (C8,9 >19:1 Z/E). Full reduction of the benzyl ester to the corresponding alcohol proceeded smoothly with DIBAl-H in 68% yield followed by oxidation to the aldehyde in 98% yield with IBX in DMSO. Wittig olefination was used to install the terminal alkene 5b in 75% yield. Heck coupling using (Z)-vinyl iodide 6 afforded dienoate 18 in 50% yield maintaining a >19:1 Z/E isomeric ratio about the C8,9 alkene.⁸ Desilylation with TBAF afforded the desired seco ester 4b in 83% yield. TMSOK was used to hydrolyze the dienoate ester to the corresponding acid followed by a macrolactonzation to the desired macrolactone 3b with the use of PyBOP in 55% yield.¹³

Towards understanding the stability of these macrocycles under typical experimental conditions, stability studies were performed in an NMR tube on the benchtop using CDCl₃ as the solvent in the presence of air O₂ (air) to determine the rate of Wasserman rearrangement between the thiazole 3a and oxazole 3b bearing macrolactones. LC-MS and NMR data was collected to monitor both the Wasserman rearrangement and/or isomerization of the dienoate of both macrocycles to determine an approximate half-life, in days, via reverse-phase LC-MS chromatogram integrations. NMR experiments were performed on a 500 MHz Bruker instrument every ~24 hours monitoring the C_H4 proton. After 11 days, it was found the thiazole bearing lactone 3a does not undergo Wasserman rearrangement (Figure 3), however significant isomerization (t_1/2 = ~8.3 days via LC-MS, Figure S1) of the dienoate was observed. The shift of C_H4 from 7.7 ppm 8.3 ppm as well as the observed mass ([M=H]⁺ = 316.11) indicate this isomerization was taking place without the incorporation of O₂. This result suggests the thiazole is less reactive towards ¹O₂ but is sensitive to ambient light induced photo-isomerization when left in solution on the benchtop. The oxazole variant 3b however undergoes both isomerization and Wasserman rearrangement giving a mixture of multiple macrocycles. Comparing our spectra (Figure 4) with Lindels work⁹, where the Wasserman rearrangement was forced with the photosensitizer Rose Bengal, ¹HNMR indicates the major Wasserman product is (E,Z)-19b as indicated by the C_H4 proton at 8.1 ppm. LC-MS data indicates multiple rearrangement isomers however, that exist in concentrations undetectable by NMR. This conversion of 3b to varying isomers and rearrangement products (t_1/2 = 8.5 days via LC-MS, Figure S3) show the oxazole readily isomerizes, similar to that of the thiazole, but is far more reactive towards ambient ¹O₂.

Figure 3.

(a) Stability of the thiazole macrocycle 3b monitored over time by ¹H NMR (500 MHz) after being stored in CDCl₃, in a regular glass NMR tube, open to air on the benchtop. (b) expanded spectra tracking H4 (highlighted in red) for isomerization over time.

Figure 4.

(a) Stability of the oxazole macrocycle 3b monitored over time by ¹H NMR (500 MHz) after being stored in CDCl₃, in a regular glass NMR tube, open to air on the benchtop. (b) expanded spectra tracking C_H4 (highlighted in red) for isomerization and Wasserman rearrangement over time.

In conclusion, substitution of the oxazole for the thiazole in these simplified SalaC macrocycles eliminates the Wasserman rearrangement with ambient O₂ in CDCl₃ when placed on the benchtop. These results support the notion that replacing the oxazole with a thiazole in more elaborate SalaC derivatives will eliminate the Wasserman rearrangement, increasing stability during synthesis and biological studies. Conformational studies (DFT) suggest that introduction of sulfur does not perturb the macrocyclic conformation nor planarity around the thiazole and thus should not dramatically impact bioactivity. It was observed, however, the rate of isomerization of the thiazole macrocycle 3a is comparable to that of the combination of isomerization and Wasserman rearrangement found in the oxazole 3b. At this time, it is unknown if this isomerization effects biological activity but the studies disclosed herein also suggest ideal ways to store both synthetic samples following macrocyclization and samples for biological assays. A thia-salarin C analogue may provide greater aerobic stability, facilitating mode of action studies which are currently underway.