Efficient Synthesis and Resolution of Pyrrolizidines

Link: http://www3.interscience.wiley.com/cgi-bin/abstract/114172048/ABSTRACT

From Prof. Mathias Christmann's group at Institut fur Organische Chemie in Aachen, Germany

A recent paper in ACIEE EarlyView highlights a new and interesting way to make an indolizidine carboxylic acid analog 1. This compound is a component embedded in a natural product, telomerase inhibitor UCS1025A. Compound 1 has been made before by Prof. Danishefsky from compounds 4 and 5 in nine steps. But the new approach would require only two steps starting from 4-aminobutyric acid 2 and maleic anhydride 3, followed by a kinetic resolution to form an enantio-pure material.

Reaction of 2 and 3 would involve an in situ formation of maleimide, followed by an enolate (from the carboxylic acid) addition to the maleimide ring carbonyl group. Some concerns were raised here:
- Corboxylic acid is known to be unsuitable for enolate formation
- In the subsequent lactone formation of 1, carboxylic acid is also known to be unsuitable to react in a 1,4-addition

The first concern was resolved with the use of Hoye’s soft-enolization strategy. It was found that this strategy worked well with compounds 2 and 3 to surprisingly give the cis-7a in high enantioselectivity. However, when the reaction started with the maleimide 8, trans-isomer was obtained. These results suggest that the reaction between 2 and 3 went through a different mechanism. And with different workups, 7b or lactone 9 can be obtained selectively.
With the condition worked out, the scope of the reaction was then explored and summarized below (isolated yield in parentheses).
Next, kinetic resolution was explored in converting racemic 7b to lactone 9 in enantio-enriched form. An organocatalytic method, using cinchona alkaloids, was chosen to effect the intramolecular 1,4-addition (oxa-Michael lactonization) as required for the resolution. This is particularly noteworthy as hard nucleophile (the carboxylic group) is not a very good nucleophile for this type of addition. Several reaction conditions were explored and monitored by NMR and are summarized in the table below.
It was found that at rt, dichloromethane worked better than chloroform or benzene. The base of choice was found to be quinine (entry 4). When (-)-7b is the desired enantiomer, quinidine can be used (entry 6) or lactone (-)-9 can also be converted back to (-)-7b using DBU (through beta-elimination).

At this point, trituration can be employed to separate lactone (-)-9 from other impurities. Interestingly, it was found that "when a weakly enriched scalemic mixture of 7b is triturated in hot n-pentane, (-)-7b is readily dissolved, while racemic 7b remains as a solid residue". The difference in spatial arrangements and solubilities of homochiral and heterochiral molecules are believed to play a role in this behavior, especially in this case, a strong intermolecular hydrogen bond between carboxylic acid and the imide carbonyl group seems to contribute to a different crystal structures between recemic 7b and (-)-7b.
Subsequent transformations using the Danishefsky’s protocols, indolizidine (-)-7b could be converted to UCS1025A or the cyclohexyl analogue 11 in only four steps.
This completes a nice total synthesis of UCS1025A and its analogue 11 in only six steps from maleimide 8.