Organoautocatalysis: Challenges for experiment and theory

The idea that molecules could make countless exact copies of themselves offers fascinating prospects in materials science and holds interesting implications for the origin of life on earth. Oparin was the first to realize the importance of self-replication for life processes [1, 2]. Self-replication appeared for a long time to be a sole domain of RNA and DNA molecules replicating via enzymatic pathways [3], until the pioneering studies of von Kiedrowski [4–7], who first demonstrated that oligonucleotides could self-replicate even non-enzymatically via template-directed autocatalysis. Self-replication has been also invoked as an integral part of systems chemistry [8, 9].

That even much smaller and simpler molecules are capable of exhibiting self-replication, was first shown by Rebek and co-workers for artificial synthetic models (Figure 1) [10, 11].

The finding has been much debated. Challenged by Menger et al. [12, 13], who argued that the rate enhancement is due to amide-catalysis and not due to template-autocatalysis, Rebek's interpretation of self-replication has been vindicated by Reinhoudt's group later [14]. Since then, a few other scattered reports about self-replicating molecules have appeared in the literature [15–18].

The potential enantioselectivity of the self-replicating autocatalytic process was implied, but has not drawn particular attention at that time.

Asymmetric autocatalysis, a term first introduced by Wynberg, is the process of automultiplication of a chiral compound in which the chiral product acts as a chiral catalyst for its own formation [19]. Catalyst and product possess of the same absolute configuration and are structurally related. The first example for such a process was reported by Soai in 1990, in the irreversible enantioselective addition of dialkylzinc reagents to pyridine-3-carbaldehyde (Figure 2) [20]. Thus, when product of the reaction was used as catalyst at 20 mol% loading and with 86% ee, the newly generated product was isolated in 67% yield and 35% ee. No autoamplification of product enantiomeric excess was observed.

In 1995, Soai reported the ability of a chiral pyrimidyl alkanol to amplify a tiny initial product enantiomeric excess - in the presence of i- Pr₂Zn - to almost enantiomeric purity in a sequential batch reaction protocol (Figure 3) [21–23]. This process is highly advantageous, because product and catalyst don't need to be separated after completion of the reaction, allowing required product purity easier to be obtained [22, 23].

The Soai reaction is therefore able to generate impressive enantioenrichment from nominally achiral initial conditions, a behaviour unprecedented in stereochemistry, and as an example of true "absolute asymmetric synthesis" in absence of external chiral influences [24]. Soai observed a positive non-linear effect ((+)-NLE) in this reaction [21], indicating the involvement of catalyst aggregation [25]. The reaction is self-accelerating, because the rate-determining step is of the quadratic (or even higher) reaction order in the product concentration, due to formation of a catalytically active homochiral dimeric product Zn-complex (Figure 4) [26–28]. As a result, one of the asymmetric autocatalytic product enantiomers is outrun by its antipode, which forms faster.

The first example of asymmetric autocatalysis for an organocatalytic (metal-free) system was reported by Mauksch and Tsogoeva in 2007 for the reversible Mannich reaction of acetone and N-PMP-protected α-imino ethyl glyoxylate (Figure 5) [29], followed by demonstration of spontaneous asymmetric amplification under nominally achiral starting conditions for the same reactive system and by the same authors [30, 31].

The enantioselectivity observed in the presence of product catalyst is comparable to that obtained with known external catalysts, like proline. The majority of the newly formed product has the same absolute configuration as the initially added product catalyst, which might suggest a template-directed (self-replicating) mechanism. The Mannich product was assumed to bind non-covalently via hydrogen bonds to the reactant, which is attacked by the nucleophile (activated ketone in enol or enamine form) (Figure 6).

This mechanistic proposal has a high appeal, because it is resembling existing mechanistic concepts for classical non-covalent (enantioselective) organocatalysis [32, 33]. Further evidence supporting this idea was found by DFT computations, which allowed to locate the transition state structures for this transformation.

In 2002 Philp reported a non-asymmetric experimental example of a minimal self-replicator in the bimolecular reaction A + B → T (also the initiation step): reactant molecules A and B, both bound by secondary interactions (hydrogen bridges) to a product template T, react to give a dimer [T^.T] in a template directed synthesis (Figure 7) [16], based on the earlier expectations of von Kiedrowski for related systems [5]. The initially formed product template dimers then could facially release the monomeric autocatalysts through dissociation [5, 16].

This mechanism, extended to account for the chirality of the template [31], provides a simple explanation for the observed chiral induction in the organoautocatalytic Mannich reaction: selective transition state structures (where the chiral product template catalyzes formation of new product molecules of the same absolute configuration) may yield homochiral dimers, while antiselective transition state structures (where the product template catalyzes formation of new product with opposite absolute configuration) may yield heterochiral dimers. For the Mannich reaction, the formation of homochiral dimers in the autocatalytic step was indeed found to be kinetically preferred, in accord with the observed enantioselectivity [29].

Furthermore, such organoautocatalytic reactions should involve merely linear autocatalysis (unlike to Soai's example) in the light of lacking coordination sites at a metal allowing to form multiple catalytic aggregates. Linear autocatalysis alone, though, cannot result in the observed asymmetric amplification [34].

Hence, to explain the unprecedented spontaneous mirror symmetry breaking observed in the Mannich reaction [30], Ribó and co-workers proposed the reversible exergonic formation of a heterochiral dimer of the product autocatalyst [35], resulting in mutual inhibition of autocatalyst formation through reduction of the antipode's concentration - in analogy to the seminal theoretical proposal of such spontaneous asymmetric amplification by Frank in 1953 [36]. However, such thermodynamically stable dimers were not yet located computationally or observed experimentally for this reactive system. As an alternative, recycle kinetics, involving endergonic formation of labile heterochiral dimers which take part in closed reaction loops, was invoked recently to explain the observation of spontaneous mirror symmetry breaking in such formally closed reversible (homogenous) reactive systems [30, 37]. Non-equilibrium quasi-steady states might form temporarily in open subsystems of closed systems and with cyclic kinetics [37, 38]. A related theoretical model was also forwarded by Plasson and co-workers, wherein it was proposed that a non-spontaneous reactant recycling step could be driven through coupling to an external source of energy [39, 40]. This situation might apply to several biochemical reaction cycles, driven e.g. by hydrolysis of energy rich compounds.

The reports of the first example for enantioselective organoautocatalysis has drawn considerable attention. Results for a similar reactive system (Mannich reaction of N-PMP-protected α-imino ethyl glyoxylate with cyclohexanone instead of acetone) and in the presence of water were reported in 2008 (Figure 8) [41]. Notably, ¹H NMR studies revealed the acceleration of the rate in course of the reaction and in presence of product catalyst. Such rate acceleration is often seen as a hallmark of autocatalytic reactions.

Most recently, Wang and co-workers further reported the enantioselective organoautocatalytic Mannich reaction of isovaleraldehydes to the same N-PMP-protected α-imino ethyl glyoxylate and employed both product catalysts and their close mimics (Figure 9) [42]. In addition to the often observed near retention of product enantiomeric excess (99% ee), it was also reported that a noteworthy - fairly significant - change of diastereoselectivity in course of the reaction occurs (autocatalyst with syn-configuration provides the formation of anti product). To explain, these authors suggested that the anti product may be formed faster than the syn product under kinetic control.

The generality of asymmetric organoautocatalysis in various organic reactions is conceivable. It might be expected, that this phenomenon may be demonstrated for other reactions than the Mannich reaction in the near future.

Seemingly, presumably well-understood organic reactions appear to have much more complicated mechanisms, than previously expected. This poses a challenge for further mechanistic investigations of organoautocatalytic reactions, both experimentally and theoretically. Classical existing mechanistic concepts may not be sufficient to allow yet a full understanding of all the processes involved. There is no doubt, that the further insights gained will be of great value for the synthetic community both in research laboratories and in industry. A further related enticing prospect might be the deeper understanding of the fundamental question of biological homochirality.