APPROACHES TO THE ORIGIN OF LIFE ON EARTH

Theories of the origin of life have gone through several stages, with accompanying experimental work. The start of the modern era in this field lies with Stanley Miller's stunning early work showing abiotic synthesis of several amino acids, followed by an era of prebiotic chemistry, including sugars, nucleotides and other abiotically synthesized biomolecules.

With the discovery of the structure of double stranded DNA and RNA, focus shifted, largely in the work of Leslie Orgel, to the search for conditions in which a single stranded RNA molecule might line up free nucleotides according to Watson Crick base paring, and in the absence of an enzyme, ligate these free nucleotides into proper 3', 5' phosphodiester bonds, make a template complementary strand, the two strands would melt apart and cycle. These experiments, after some 40 years, have never worked.

On the theoretical level, Eigen and Schusters' Hypercycle model was build upon template replicating RNA strands which formed a hypercycle in which replicating strand pair I, helped replicating strand pair I + 1 replicate around a cycle of such mutualistic strand pairs. This era was followed by the RNA world hypothesis. The discovery of ribozymes led to a sustained effort in which two RNA world views predominated. In the first, a population of RNA ribozyme sequences would mutually catalyze one another's formation, creating what I will call a collectively autocatalyltic set. In the second, a single ribozyme would act as a polymerase able to copy any RNA sequence, including itself - the first self replicating molecule it was hoped. At present, David Bartel has made a candidate ribozyme that can template copy 14 nucleotides.

We may be entering a third era. In 1971 I published a first paper in which polymers, such as proteins, could form a collectively autocatalytic set. This set arises as a phase transition analagous the phase transition in Erdos Renyi random graphs, but "one level up", in a world of chemical reaction graphs and a hypergraph structure in which some molecules in the reaction graph can catalyze some reactions in that very reaction graph. I will discuss how, at a sufficient diversity of polymers, the formation of such a collectively autocatalytic set arises with probability near one.

On the experimental front, G. von Kiedrowski made a single autocatalytic DNA sequence, then the first collectively autocatalytic set of two DNA sequences. In 1995 Reza Ghadiri at Scripps, made the first self reporducing 31 amino acid sequences, and shortly thereafter a collectively autocatalytic set of peptides. Recently, his former post doc, Gonen Ashkanazi has made a collectively autocatalytic set of 9 peptides. In addition, Ashkanazi has engineered these peptides to realize all Boolean functions of 2 inputs, so autocatalytic sets with complex dynamics can now be studied.

Still more recently Mike Steel and Wim Hordijk have improved on my own initial model, showing that the density of catalysis can be much smaller and still yield a phase transition to collectively autocatalytic sets.

At present we are at the stage of generating random "never before born" peptides. I made the first libraries in the late 80s and 90s. Luigi Luisi and independently, Tetsuya Yomo have made such libraries. In 1996 Thomas LaBean in my lab showed that such random peptides fold to molten globules or better. This work has been repeated by both Luisi and Yomo. It is known from phage display that the probability a small peptide binds an epitope is about one in a million. The probability of catalysis of a specific reaction is still unknown. I personally hope collectively autocatalytic sets will be generated soon.