PACE REPORT

Evolution of Protocell-embedded Molecular Computation

Lead partner: ALife Group, Dublin City University

---

Introduction

A core objective of the PACE project is to evaluate the potential of synthetic protocells for application in novel information and communications technologies (ICTs). In this section of the final project report we present the results of a focussed investigation into how this might be achieved. As practical synthesis of functional protocells is still some way in the future, this investigation necessarily relies on simulation models. However, to allow investigation of evolutionary phenomena, at the protocell level, these molecular level simulations must still be computationally tractable even when used to realise sizable populations of complete protocells, simulated over extended numbers of protocell generations. Accordingly, the approach adopted here is to deliberately abstract away much physico-chemical detail, while attempting to retain just those dynamic and organisational features that are judged to be critical for demonstrating and investigating the evolution of at least minimal protocell computation capability. To this end, we have developed an “artificial chemistry” based modelling framework (Dittrich et al., 2001), with an accompanying simulator system. In this particular investigation, the computational capability is realised by assemblies or networks of interacting molecular species, embedded within individual protocells (as opposed to being implemented at the higher level of assemblies of interacting protocells, which is also a conceivable approach). The computation function accordingly affects, and is constrained by, the viability, growth and ongoing reproduction of the “host” protocell lineage—which is precisely the vector for its potential evolution. One of the key demarcations between a proto-cell and a cell is that the latter has a translational subsystem, which acts to decouple a family of “purely” informational molecules from a family of active, functional, molecules; or, more precisely, to decouple a space of genotypic variation from a space of phenotypic variation. In modern cells, this is achieved, at the molecular level, though the demarcation between DNA and proteins, and the machinery of transcription, ribosomes, adaptors etc., which translate from the former to the latter. In the abstract theory of evolutionary automata it is represented by the “general constructive automaton” of von Neumann (1966). By contrast, protocells are conceived as functioning, and evolving, without such a translational system. Instead, by analogy with the family of RNA-enzymes or ribozymes, a single family of molecules is envisaged whose members can act both as information carriers and as catalysts of specific functional transformations. In particular, it is supposed that there can be a family of such molecules that can be “replicated” (i.e., generically copied), where the same family also contains enzymes capable of catalysing this replication. This is then analogous to the hypothetical ribozymal RNA-replicases of the RNA-world and related origin-of-life models.¹ Replication of informational molecules is, of course, essential to realise informational inheritance at the protocell level, i.e., across recurrent protocell divisions, when molecules must be distributed between offspring. The current investigation is therefore focussed on developing at least an initial understanding of the distinctive evolutionary dynamics that may arise at the protocell level specifically because of the reliance on such a family of “informational enzymes” as the essential mechanism for hereditary variation. The simulation framework presented here consists of the following components:

• An enzymatically closed family of template-replicator molecules, termed “informational enzymes” or informazymes. These are intended as an abstract generalisation of RNA, specifically including the RNA-enzymes or ribozymes. Informazymes are modelled as polymers generated from a fixed set of monomers, with a variety of enzymatic functions systematically determined by the primary sequence of monomers.² The core functional capability of an informazyme is to act as a replicase for some “recognised” subset of this same family. In this way it is possible to create populations of self-catalysing (collectively or otherwise), modular, molecular replicators. Under suitable conditions, such molecular populations can be self-sustaining, with “unlimited” heredity (Szathmáry and Maynard Smith, 1997). Assuming a resource constraint, this will then give rise to various potential forms of selectional and evolutionary dynamics at the molecular level. These will occur on a relatively fast timescale compared to any separate (but coupled) dynamics at a protocell level. However, note carefully that the model of molecular (self-)replication studied here is based on “trans-” rather than “cis-” acting autocatalysis (Altmeyer et al., 2004). That is, a replication event requires two discrete molecules, one acting as template and the other acting as replicase—even if both are instances of the same molecular species. As we shall see, this strongly affects the forms of “molecular evolution” that arise.
• A “container” subsystem which packages a population of informazyme molecules into a discrete protocell. Growth of a protocell is coupled to growth of the contained molecular population (primarily driven, in the current case, by informazyme-mediated replication activity). Under specific conditions, a protocell becomes unstable and will divide or fission, with independent assortment of the contained molecules between the daughter protocells. This realises reproduction at the protocell level, with statistical inheritance of the molecular-composition structure of the parent; and therefore of any protocell level traits that are determined by this molecular composition. Imposing a protocell population size limit can then generally result in a selectional dynamics at the protocell population level. If there are heritable protocell traits (arising from the underlying, somatic time, informazyme dynamics, and determined by the—heritable—molecular composition), then these can be made the target of an externally imposed fitness function. In this way, artificial (as opposed to natural) selection can be used to direct protocell evolution to desired outcomes (such as achievement of computational function).

In accordance with the generic “complementation” principle applied across the PACE project, this particular investigation assumes that external support systems can ensure a ready availability of free energy, and recycling of inactivated polymers to activated monomers for ongoing processing. That is, in this particular framework there is no explicit modelling of any metabolic subsystem. The detailed report on this investigation is structured in logically successive parts as follows:

• Theoretical analysis and generic characterisation of the potential dynamics of isolated (non-cellular) informazyme systems.
• Implementation and test of an informazyme-based protocell model with minimal enzymatic repertoire (MCS-0): essentially a minimal recognition rule, with replicase as the only enzymatic function. Demonstration of informazyme-based inheritance mechanism at the protocell level, and protocell level evolution (stalemate between conflicting selection dynamics at molecular and protocell levels: evolutionary stasis).
• Extension of enzymatic repertoire to allow more variable recognition, with consequently more varied molecular level dynamics (MCS-1).
• Extension of enzymatic repertoire to incorporate minimal “information processing” or “computational” capability. This is implemented as an extension of the core replicase function, so that as an informazyme processes (“replicates”) a template sequence, the generated sequence can differ from the template in a systematic, programmatically determined, way. The molecular computation is coupled to the protocell level via rate control of “membrane” molecule production. Demonstration of successful directed evolution of desired molecular computation function, based exclusively on selection of protocell level traits.
• Concluding discussion: critical evaluation of the potential and limitations of the approach.