Fragment deprotection approaches

With the macrocycle 126 in hand, the only remaining step was the global deprotection. Thus, we turned our attention towards finding suitable deprotection conditions. We first wanted to explore the deprotection on smaller fragments before trying them on the real Pulvomycin skeleton.

Our experiments commenced with the ketone fragment 11. The use of ten equivalents of TBAF in THF at –10 °C led to a very slow conversion of the starting material (Scheme 90).

Scheme 90. Global deprotection of ketone fragment 11 at low temperature.

After 22 hours, there still was no full conversion, and only 25% of the fully deprotected product 167 could be isolated. Unfortunately, applying the same conditions to the macrocycle 92 led to rapid decomposition (Scheme 91). The reaction turned red directly after the TBAF addition, and no material could be isolated after workup.

Scheme 91. Treatment of macrocycle 92 with TBAF leads to decomposition, even at low temperature.

We reasoned that the basicity of the TBAF reagent is the main problem. If alkyne 35 was subjected to TBAF, the deprotection to diol 39 proceeded smoothly even at elevated temperatures (Scheme 92).

Scheme 92. Deprotection of alkyne 35.

If vinyl iodide 36 was used under identical conditions, the desired product could not be isolated.

Instead, NMR analysis suggested the formation of the allene compound 168 (Scheme 93). The double allylic position of the C5 alcohol probably makes it very prone to elimination under basic conditions. The eliminated product 169 can then be attacked by the primary alcohol to form the allene by the elimination of hydrogen iodide.

Scheme 93. Formation of the unusual allene 168.

Therefore, we decided to buffer the TBAF using acetic acid. Using five equivalents of TBAF in a 10:1 mixture of THF and acetic acid completely stopped the reaction, while a 12:1 mixture of TBAF and acetic acid led to complete decomposition. Finally, a one-to-one mixture of TBAF and acetic acid successfully removed both TBDPS groups from the macrocycle 92 (Scheme 94).^[104] Although the reaction was carried out at ambient temperature, the deprotection took seven hours. To our surprise, the ¹H-NMR spectrum of the isolated product was missing the C13 CH signal. Instead, the ¹³C spectrum showed an additional signal next to the C12 ketone

signal. After careful analysis of the analytical data, the NMR spectra unambiguously proved the formation of the 1,2-diketone 170.

Scheme 94. Formation of 1,2-diketone 170 under TBAF deprotection conditions.

We first suspected aerial oxidation of the C13 alcohol to the ketone. However, the ketone formation was also observed when both the reaction and the workup were conducted under a protective atmosphere. It seems that the TBAF reagent itself acts as the oxidant, as pointed out in the literature.^[105]

Around the same time of these findings, the group of Moon and coworkers reported the isolation of three new compounds of the Pulvomycin family, as pointed out in chapter 2.^[21] These compounds include Pulvomycin D (4), which exhibits the same 1,2-diketone structure between C12 and C13. We reasoned that Pulvomycin D (4) would be an interesting target for our total synthesis and did not focus on strategies to avoid the oxidation at C13 further.

Applying the above-mentioned buffered TBAF conditions to ketone 11 successfully triggered the Peterson olefination and removed the TBDPS group (Scheme 95). However, the TBS group at C37 remained untouched under these conditions. No full conversion was achieved after 22 hours, resulting in a low yield of only 26% for the TBS-protected fragment 171.

Scheme 95. Application of the buffered TBAF conditions to ketone fragment 11.

A variety of other known deprotection conditions were applied to the ketone fragment to facilitate the removal of the C37 TBS group (Table 21). However, most conditions only triggered the Peterson elimination while leaving the TBDPS and TBS group untouched.

The combination of potassium fluoride and TBACl is known for the in-situ formation of TBAF.

However, only 51% of the eliminated product could be isolated. Combinations of potassium fluoride and crown ether are known to generate highly reactive fluoride species.^[106] In our case, only eliminated product was isolated as well under these conditions. The use of ammonium fluoride in combination with HF led to no conversion.^[107] The TASF-reagent was not able to remove neither the TBDPS nor the TBS group.^[108]

Table 21. Application of other known silyl deprotection conditions to ketone 11.

Conditions solvent t T Result

Finally, we came across a literature report of Paterson and coworkers from 2006, where they successfully removed a TBS group from a similar sugar unit.^[109] Employing a large excess of HF ‧ pyridine, they were able to remove the TBS group in 79% yield. Applying these conditions

to the ketone fragment indeed led to the deprotection of the desired TBS group, although only in a moderate yield of 52% for alcohol 172 (Scheme 96). Interestingly, the TBS group on the sugar was cleaved selectively in the presence of the C26 TBS group, as proven by exclusive NOE contacts between the methyl protons of the remaining TBS group and the ethyl group of the ketone.

Scheme 96. Successful removal of the C37 TBS group by using an excess of HF ‧ pyridine.

Due to the orthogonality of the TBS group at C37 we reasoned, that a two-step deprotection sequence would be the best choice for the global deprotection. Indeed, treatment of the partially deprotected fragment 172 with buffered TBAF for 24 hours finally furnished the fully deprotected triene fragment 167 in 42% yield (Scheme 97).

Scheme 97. Complete deprotection of ketone 172 via two-step deprotection sequence.