A survey of the literature reveals several key strategies for the synthesis of oxetanes, thietanes, and azetidines, some of which are shown in Scheme 2.2. Such strategies include: (i) intramolecular displacement of a leaving group (e.g. halogen or activated alcohol nucleofuge) by the heteroatom; (ii) ring formation via the reaction of a nucleophile with doubly activated substrates bearing LGs in a 1,3 relationship; (iii) [2+2]‐cycloaddition reactions, including Paternò−Büchi or reduction of lactams obtained by Staudinger ketene cycloaddition. Controlling the regioselectivity of cycloadditions is difficult, and often limits the scope and utility of such approaches; (iv) Gold‐catalyzed cyclization of propargylic alcohols and amines toward their 3‐oxo‐substituted products. The wide range of methods available to prepare optically active propargyl alcohols and amines [9] make this fourth approach particularly attractive, greatly expanding the scope of substitution patterns that can be introduced onto the azetidine and oxetane scaffolds.
Scheme 2.2 Summary of common strategies for the synthesis of common four‐membered rings encountered in spirocyclic systems of medicinal chemistry interest.
Oxetanes have been known for 140 years [50]; however, historically they have received little attention in medicinal chemistry [51] compared with epoxides, tetrahydro‐furans, and pyrans. This is mainly due to their perceived reactivity: due to ring strain, they are generally seen as too reactive for their incorporation in drug candidates, despite the harsh conditions needed for their opening. Early reports have suggested that despite displaying similar ring strain to epoxide [52], their opening with nucleophiles is actually much slower [53]. Only in the last decade have oxetanes and their spiro analogues found applications in medicinal chemistry [48, 51, 54]. Due to its geometry and electronic properties, the oxetane ring has been used as a bioisosteric replacement for gem‐dimethyl and carbonyl units (Figure 2.5). Despite useful to increase the affinity of a ligand to a target receptor, the introduction of a bulky hydrophobic gem‐dimethyl motif generally increases the lipophilicity and decreases the solubility of a lead compound, which is often unwanted. The oxetane geometry mimics that of a gem‐dimethyl, while being more hydrophilic due to its oxygen atom.
Carreira’s lab and collaborators from Hoffman‐La Roche examined the physicochemical and biological properties of spiro‐oxetanes 22–30, along with their corresponding carbonyls and gem‐dimethyl analogues (Table 2.2) [55]. The oxetane ring generally provides higher metabolic stability, lower log P/D, and higher aqueous solubility than its gem‐dimethyl counterparts across all series scrutinized. Unsurprisingly, this trend also correlated with a lower pK a (1.2–1.6 units) for the amine, resulting from the inductive effect of the oxygen. While being less water soluble than carbonylated analogues, spiro‐oxetanes displayed much improved metabolic stability toward oxidation in the piperidine and pyrrolidine series. Properties of the azetidin‐3‐one derivative could not be determined due to its low stability.
Figure 2.5 Comparison between carbonyl, gem‐dimethyl, and oxetane groups, highlighting similar spatial arrangement of lone pairs and hydrophobic bulk.
Four‐membered ring containing spirocycles have also been scrutinized for their potential as bioisosteres of morpholines (Y = O), piperidines (Y = CR2), piperazines (Y = NR), and thiomorpholines (Y = S, SO, SO2), for the development of lead molecules with improved physicochemical and pharmacokinetic properties (Figure 2.6a) [48, 56]. While both classes display a similar relative positioning of polar and apolar groups, the comparison has its limitations. For example, the N–Y distance in these six‐membered heterocycles is approximately 3 Å, while it is significantly higher and is slightly above 4 Å in their spirocyclic counterparts (Figure 2.6a). Similarly, the C–O distance in an oxetane is around 2 Å, while the C–O distance in the parent carbonyl group is approximately 1.2 Å. Another important difference is the introduction of exit vectors “out‐of‐plan” in these isosteres compared with the parent groups (Figure 2.6b). Nevertheless, such differences also contribute to the unique properties of the spiro analogues and make them interesting structural features for bioisosteric replacement (Figure 2.6c). For example, incorporation of morpholine rings into drug scaffolds is an established strategy for improving aqueous solubility. However, oxidative metabolism is a known inactivation pathway of morpholines, most often resulting from ring opening. Spiro‐oxetanes 26 and 29 (Table 2.3) were proposed as replacements for morpholine 31, due to the similar relative spatial disposition of their hydrophobic and polar features. Interestingly, 29 displays higher solubility and lower logP than the parent morpholine, while also remaining stable to oxidative metabolism. A number of other potential spirocyclic morpholine mimics based on [3.3], [3.4], and [3.5] motifs have been reported by Carreira, displaying a range of unique dipoles and exit vectors for tunable properties [55].
Another illustrative successful example is the bioisosteric replacement of the metabolically labile morpholine unit of BTK inhibitor 32 (Figure 2.7). This led to the development of 2‐oxa‐6‐azaspiro[3.3]heptane functionalized analogue 33, which displays enhanced stability and solubility while retaining similar potency to its target compared with the parent molecule [57].
Thalidomide 34 (Figure 2.8) was introduced on the market in the late 1950s–early 1960s as an antiemetic and sedative, and banned a few years later due to its shattering side‐effects. Overall, thalidomide is thought to have caused numerous and severe birth defects in more than 10 000 children during that time [25]. While thalidomide was initially administered as a racemic mixture, it has since been shown that the (R) enantiomer has sedative effects and the (S) enantiomer is teratogenic. It is also known that racemization spontaneously takes place in vivo due to the acidity of Ha (Figure 2.8), although as of today the exact mechanism underlying the teratogenicity of thalidomide is still under investigation. In 2013, Carreira and coworkers reported derivative 35 where one of the imide carbonyls is replaced by an oxetane [56], with the goal of limiting in vivo metabolism and racemization, and potentially altering teratogenicity (Figure 2.8). Not unexpectedly, 35 displayed higher pK a, lower lipophilicity, and increased solubility. It displayed similar intrinsic clearance rates in human microsomes compared with 34; however, showed much improved plasma stability after five hours. Pleasingly, 35 was configurationally stable to racemization in human blood plasma