Macrocyclization approach by ring-closing metathesis

A widely used method for generating macrocycles is the ring-closing metathesis.^[78] Advantages are the high functional group tolerance and the availability of highly advanced catalytic systems like the Grubbs catalysts.^[79] One example can again be found in the Epothilone synthesis by Danishefsky (Scheme 42).

Scheme 42. Ring-closing metathesis in the synthesis of Epothilone analog 78.

Even with the unprotected alcohol 79, the desired product was obtained in 41% yield employing the second generation Grubbs catalyst.^[80]

Fürstner et al. accomplished a very impressive application of the ring-closing metathesis for their synthesis of Iejimalide B (Scheme 43).^[81] Starting from precursor 80, the very sensitive macrocycle 81 could be constructed in an excellent yield of 96%. This result is particularly interesting because the reaction takes place exclusively at the two terminal double bonds.

Despite the reaction time of two days, no interference with the internal double bonds was observed. This result encouraged us that a similar strategy would also work in our case.

Scheme 43. Cyclization of the complex macrolactone 81 using the second generation Grubbs catalyst.

A possible precursor for the ring-closing metathesis is given in scheme 44. Due to the lack of reactive functional groups, we envisioned compound 82 to be much more stable compared to the previous cyclization precursors, and we hoped that this would lead to much better control over the reaction. As with the previous strategies, the C8-C11 linker should be installed by a Nozaki-Hiyama reaction starting from alcohol 83 and the dienyl iodide 84. We envisioned that the dienyl iodide could be prepared from known alcohol 28 by an elimination reaction.

Fragment 83 should be accessible by coupling iodide 36 with an appropriate C2 unit, followed by oxidation of the C1 alcohol and esterification with known alcohol 67.

Scheme 44. Retrosynthetic analysis for the metathesis approach.

Starting from vinyl iodide 36, the desired diene 85 could be synthesized by a Stille coupling^[82]

using tributyl vinyl tin and Pd2(dba)3, albeit in a very low yield of only 12% (Table 12, entry 1). Gratifyingly, the vinyl residue could also be attached by a Suzuki reaction using commercially available vinylboronic acid pinacol ester (entry 2) or potassium vinyltrifluoroborate (entry 3) with much better yields.^[83]

Table 12. Installation of the vinyl group by different cross-coupling experiments. HF ‧ pyridine, and the alcohol was oxidized in a two-step protocol, delivering the carboxylic acid 86 in 33% yield over three steps (Scheme 45).

Scheme 45. Cleavage of the TES group followed by two-step oxidation to the carboxylic acid 86.

Esterification with the secondary alcohol 67 proceeded smoothly under known Yamaguchi conditions, and the primary TES group could be removed in 88% yield using HF ‧ pyridine (Scheme 46).

Scheme 46. Esterification of acid 86 with known alcohol 67.

For the synthesis of the C8-C11 linker, alcohol 28 was converted into the mesylate 89 by reaction with mesyl chloride (Scheme 47).^[84] Then, the mesylate was treated with potassium tert-butoxide, which cleanly furnished the desired elimination product 84.^[85]

Scheme 47. Preparation of sensitive dienyl iodide 84 by elimination of known alcohol 28.

Dienyl iodide 84 turned out to be very volatile and not stable on the column, which made isolation and purification difficult. Therefore, diethyl ether was chosen as the solvent for the

aqueous workup was done with pentane instead of diethyl ether to exclude polar side products.

By employing this protocol, the desired dienyl iodide 84 could be isolated in reasonably pure form without the need for further purification. Due to the volatility of the compound, it was routinely obtained in varying concentrations with diethyl ether (20-30%). Attempts to isolate the compound in a pure form usually led to a high loss of material. Furthermore, the compound rapidly decomposed if the solvent was removed completely. The residual solvent proved to have no significant impact on the subsequent Nozaki-Hiyama reaction, though.

The quality of the potassium tert-butanolate turned out to be of great importance for the reaction outcome. Older batches of the base resulted in sluggish conversion and required 1.5 – 2.0 equivalents to achieve complete elimination.

Scheme 48. Preparation of the metathesis precursors 90 and 91.

On the contrary, when employing fresh potassium tert-butanolate, even slight excess of reagent immediately led to decomposition of the product.

With the dienyl iodide in hand, we proceeded towards the Nozaki-Hiyama reaction with alcohol 88 (Scheme 48). After oxidation to the aldehyde, the standard conditions for the Nozaki-Hiyama reaction were applied. Using six equivalents of iodide 84 and 20.0 equivalents of chromium-(II)-chloride, the product was obtained with an acceptable yield of 34%.

For the subsequent cyclization experiments, both the free alcohol 90 and the TES-protected derivative 91 were employed. The TES group was installed using TES triflate in 88% yield.^[86]

The experiments of the ring-closing metathesis are depicted in Table 13. Surprisingly, treatment of the TES-protected compound 91 with several catalysts (Figure 6) in dichloromethane led to no conversion, even at elevated temperatures (entries 1-3). When toluene was used as the solvent, traces of product could be observed by ESI-MS when the catalyst was used stoichiometrically at 60 °C (entry 4). However, the material decomposed with prolonged reaction time. Employing the third generation Grubbs catalyst led to no conversion at ambient temperature and decomposition at 50 °C (entries 5-6).

Table 13. Attempts for the ring-closing metathesis.

8 Grubbs II (0.60 eq.) PhMe H r.t. 6 h traces decomp.

9 Grubbs I (0.50 eq.) PhMe H r.t. ⟶ 65 °C 3 h traces decomp.

With the unprotected substrate 90, traces of product were observed by ESI-MS with the Grubbs II catalyst at 60 °C (entry 7). Increasing the amount of catalyst led to product formation at room temperature (entry 8) - however, decomposition set in with prolonged reaction time. The same observation could be made with the first-generation Grubbs catalyst (entry 9).

Although product formation was observed via ESI-MS in some of the experiments, no material could be isolated. No significant change on TLC was observed, indicating that the product probably has a very similar retention value as the starting material. Hence, isolation of the desired product was not possible using standard chromatographic methods. With prolonged reaction time, the formation of a baseline spot on TLC was observed for most experiments.

Probably, the terminal double bonds are not reactive enough, or the macrocycle is too strained.

With a longer reaction time, the internal double bonds might begin to react with the catalyst, leading to the decomposition of the material.

Figure 6. Overview of the different Grubbs catalysts.