Fig. 2. Comparison of two possible views for the path leading from a ‘primordial soup’ to a rudimentary protocellular structure (bottom). (A) The ‘biopolymer first’ scenario, according to which the emergence of self-replicating informational strings such as RNA and proteins are assumed to have had an independent origin from that of lipid encapsulation. (B) The ‘Lipid World’ scenario, which maintains that the roots of life could have been aggregates of spontaneously assembling lipid-like molecules endowed with capabilities for dynamic self-organization and compositional inheritance. More elaborate structures, including informational and catalytic biopolymers, might then have evolved gradually.

Fig. 1. An assembly of amphiphilic molecules could spontaneously form according to two regimens governed by the following two parameters: N, the number of molecules within an assembly (i.e. assembly size), and N_G, the fixed, global number of different ‘monomer’ compounds attainable under the constraints of chemistry (i.e. repertoire size). For randomly formed large assemblies, fulfilling N>>N_G, (top), every type of molecule is, on average, represented in multiple copies within each assembly, and therefore all assemblies are practically identical. By splitting, they form nearly identical replicas (+) but, because the assemblies manifest no diversity, evolution and selection are impossible. For small assemblies, where N<<N_G, (bottom), diversity is high, each molecular species is typically present in only one copy within an assembly and each assembly a different assortment of molecules. (For example, for an assembly containing N = 100 molecules and having a repertoire size of N_G = 1000, there are in excess of 10¹⁰⁰ different assemblies possible, entering the realm of unlimited heredity (Szathmary and Maynard Smith, 1997). However, as previously pointed out (Morowitz, 1967; Szathmary and Maynard Smith, 1997), a split under the conditions N<<N_G will invariably lead to two very different progeny, resulting in ineffective replication (bottom left, –). In order to reach an intermediate regimen in which both diversity and replication fidelity are achieved, it is necessary for the small assemblies to undergo a process of molecular repertoire restriction. The GARD model (see text) demonstrates that this is possible through the formation of organized, mutually catalytic networks (inset, A, thicker arrows). Importantly, repertoire restriction will not occur under conditions of equilibrium (inset, B) but a limited repertoire will be selected out of equilibrium (inset, C) (Segré et al., 1998a,b). This process might happen only for a minute fraction within the huge diversity of assemblies formed under the N<<N_G conditions. Such assemblies will gradually reach a critical point, at which each molecule type will be present at two or more copies. Above this threshold, termed the ‘Morowitz boundary’ (Morowitz, 1967; Segré and Lancet, 1999), effective replication by splitting would take place (bottom right, +), potentially leading to natural selection.

Doron Lancet and Daniel Segré