Initial structure-activity studies from our laboratory focused on altering the substitution patterns of both the northern and southern aromatic rings attached to the 2,3,8,8a-tetrahydroindolizin-5(1H)-one core. ³ These studies revealed that the lead compound, racemic 2, was the most active in the library and indicated that the southern aromatic ring at C6 did not accommodate substitution well. In subsequent studies, we prepared both enantiomers of 2, and found that only the (R)-2 enantiomer is cytotoxic. ⁵

In an effort to expand the structure-activity relationship information for these compounds, we sought to develop an analog library to probe the ramifications of replacing the southern phenyl ring with various heteroaromatic rings. As the C6 aromatic ring of related analogs had proved not to accommodate substitution well, we could not explore the electronic ramifications of the southern ring by substitution alone. With this limitation in mind, a 14-membered heteroaromatic 6,7-diaryl-2,3,8,8a-tetrahydroindolizin-5(1H)-one analog library was prepared. Ultimately, this library would serve to vary the electronic characteristics of the southern ring without increasing the steric bulkiness by adding substituents to the aromatic ring system.

The retrosynthetic analysis for the analog library is patterned after the stereospecific synthesis developed recently in our laboratories to generate both enantiomers of 2 (Scheme 1). ⁵ Although in this case, without any stereocontrolling element elsewhere in the molecule, the stereocenter is destroyed due to a retro-Michael-Michael racemization. ⁵ Regardless, the use of N-BOC L-homoproline (5) as the starting substrate was convenient and the synthesis was otherwise sound for our purposes of probing the heteroaromatics effects on biological activity. Thus, title compounds 3a–3n can be envisioned as coming from amide 4a–4n following an intramolecular aldol condensation of the amide with the aryl ketone. In turn, amides 4a–4n may be derived from 5 through a Grignard reaction followed by a BOC-deprotection/peptide bond formation sequence.

The synthesis toward 3a–3n was initiated with the Weinreb amidation of (S)-2-(1-(tert-butoxycarbonyl)-pyrrolidin-2-yl)acetic acid (5, Scheme 2). The subsequent Grignard addition into the Weinreb amide supplied aryl ketones 6a–6b. An acidic BOC-deprotection of ketones 6a–6b formed stable hydrochloride amine salts, which were subjected to an EDCI coupling protocol with the appropriate heteroaryl acetic acid to provide the aldol precursors 4a–4n. The final step called for a basic intramolecular aldol condensation to construct the heteroaromatic library of analogs 3a–3n. As previously illustrated, the acidic BOC-deprotection and basic intramolecular aldol conditions led to the retro-Michael-Michael racemization. ⁵ The yields and substitution patterns for the heteroaryl analog library developed in Scheme 2 are shown in Table 1. In addition, the phenyl parent compounds 4a and 4i were carried through the synthesis as well so as to have a standard for cytotoxicity comparison purposes.

The heteroaromatic indolizidine analog library was tested for cytotoxicity against the HCT-116 colon cancer cell line. The results are summarized in Table 2. It is clear that the pyridyl analogs 3b–3d of the parent compound 3a are the least active of the compounds tested, with the 2- and 3-pyridyl derivatives having much reduced activity. The latter two compounds were not pursued further as the thiomethyl derivatives. The thiomethyl complement of the 4-pyridyl derivative 3j, also had the lowest activity of the thiomethy subgroup of compounds. In general, there was little difference in activity amongst the furan and thiophene derivatives of the parent or thiomethyl compounds, although the thiomethyl thiophene derivatives 3k and 3l were the most active of all the derivatives tested and were equal in potency to the parent compound 3a.