NASA's recent announcement of the formation of an Astrobiology Institute to study life's origins prompted Lenny Dawidowicz and Mitchell Sogin of the Marine Biological Laboratory, Woods Hole, Massachusetts), to organize a NASA-sponsored workshop in October on "Evolution: A Molecular Point of View." The meeting brought together researchers from diverse fields including geochemistry, paleontology, molecular biology, developmental biology, and polymer chemistry to discuss the origin and diversification of life. Of the many provocative talks, here we briefly review a few that we felt either challenge traditional views, raise particularly contentious issues, or point to areas in critical need of further research efforts.

Sherwood Chang (NASA Ames Research Center, Moffett Field, California) opened the program with the cautious reminder that any canonical scenario for the stepwise progression toward the origin of life is still just a "convenient fiction." That is, we have almost no data to support the historical transitions from chemical evolution to prebiotic monomers, polymers, replicating enzymes, and finally cells. He continued with a reanalysis of the setting for the origin of life and suggested that ubiquitous environments, such as marine hydrothermal systems and the frothy surface of oceans, might have sustained equable conditions for prebiotic evolution as surprisingly early as 4.4 billion years ago. This is just shortly after the formation of the earth, much sooner than we had previously thought, and hence it widens the window for events leading to the origin of life.

New data on the potential catalytic activities of RNA in an "RNA world" came from Peter Unrau and David Bartel (Whitehead Institute, Boston) who used in vitro selection to find a novel ribozyme (RNA enzyme) that synthesizes uridine by forming the glycosidic bond between the nucleoside base and ribose. This represents one of the most difficult steps in prebiotic nucleotide synthesis. Similarly, Andrew Ellington (University of Texas at Austin) is using in vitro selection to ask, What does it take to make a tRNA? Using an assay for binding an aminoacyl-tRNA synthetase (the enzyme that charges tRNA with its cognate amino acid), Ellington and colleagues found that phenylalanine-tRNA-synthetase aptamers (RNA molecules that bind a target) could be readily isolated from random sequence populations. These molecules thus mimic one property of a tRNA, binding to a synthetase (which even DNA can perform [1]), but they were not selected for several other essential properties of tRNA, such as serving as substrates for RNase P and the CCA-adding enzyme.

A surprising feature of the complete archaeal genome sequence of Methanococcus jannaschii [2] was its apparent lack of the genes encoding four tRNA synthetases. Michael Ibba presented research in Dieter Soll's lab (Yale University, New Haven, Connecticut) which has characterized alternative pathways for the charging of two of the relevant tRNAs, glutaminyl- and asparaginyl-tRNA. A two-step transamidation reaction first charges the tRNA with the wrong amino acid (for example, aspartate instead of asparagine) and then a second protein promptly converts it to the correct form [3, 4]. Thus M. jannaschii can live without two "conventional" tRNA synthetases. In addition, they found that several Archaea, including M. jannaschii, actually contain a novel lysyl-tRNA synthetase (LysRS) with no resemblance to known bacterial or eukaryal LysRSs or any other sequences in the database. [5] Strikingly, the archaeal type of LysRS is also found in the Lyme disease spirochete Borrelia burgdorferi. These data can be explained by either redundancy of LysRSs in the last common ancestor or by lateral gene transfer among prokaryotes.

Continuing the discussion of exceptional genes, Laura Landweber (Princeton University) introduced the notion of "cellular computing" to describe how some processes, such as gene scrambling and RNA editing, use simple base pairing to create functional coding genes out of seemingly disordered sequence. The unscrambling of genes in the macronuclei of ciliates such as Oxytricha and now Stylonychia offer a unique opportunity to study the origin of complex genetic systems and even nature's solution to a path assembly problem, rather like Adleman's [6] DNA solution to the traveling salesman problem.

Bill Schopf (University of California at Los Angeles) drew attention to the conflict between molecular data and the paleobiologic record in terms of the origin and diversification of microbial life. Schopf has meticulously described numerous cyanobacterial-like fossils that are 3.0 to 3.5 billion years old. In contrast, Doolittle et al.'s [7] comparison of 57 protein families from 15 major groups of organisms dated the time of coalescence to the last common ancestor to only c. 2 billion years ago with the divergence of cyanobacteria being only 1.5 billion years ago. Schopf argued that reconciling the 1.5-2 billion year discrepancy between these data sets for both the origin of the last common ancestor and diversification of cyanobacteria presents one of the most important challenges to the field.

Russell Doolittle (University of California at San Diego) followed with an analyis of an expanded data set of protein sequences. [8] In their reanalysis, Doolittle and colleagues first consider the topologies of the genealogies generated by different proteins: eight sets support the sister relationship between Archaea and Eukaryotes; 18 unite Bacteria and Eukarya; and 13 indicate paraphyletic relationships among Archaea and Bacteria. Reconciling these differences, Doolittle et al. now move the date of the last common ancestor to between 3-4 billion years ago. While this estimate increases the concordance between the fossil and molecular data, Doolittle's reanalysis still dates the divergence of cyanobacteria to only 2 billion years ago.

Peter Gogarten (University of Connecticut, Storrs) argued that horizontal gene transfer may explain why different genes give conflicting relationships among species. Specifically, recent archaeal genome sequences indicate that most archaeal genes closely resemble their bacterial homologues and that many archaeal genes do not form exclusive groups. This means that a majority consensus tree based on these genes would dissolve representatives of Archaea and Bacteria into paraphyletic groups. Either lateral gene transfer among early diverging lineages or high levels of ancestral polymorphism in the last common ancestor, or both, preclude our ability to resolve a single tree of life.

Both Patrick Forterre (University of Paris, Orsay) and Jeff Palmer (Indiana University) echoed this point as they raised doubts about the power of current phylogenetic methods to determine either the position of the last universal common ancestor of all living organisms or the branching patterns among early lineages. Patrick Forterre also challenged the traditional view that the last common ancestor must have been simple, like a prokaryote. Instead, referring to the "streamlined" and perhaps deceptively "simple" appearance of prokaryotes, he suggested that the last universal common ancestor may have been complex and that simpler systems, such as are found in extant prokaryotes, could represent a derived state.

Gary Olsen (University of Illinois at Urbana) elegantly reminded us that gene trees are hypotheses of species relationships. Olsen, along with Michael Gray (Dalhousie University, Halifax, Nova Scotia, Canada) and Susan Douglas (National Research Council, Halifax, Canada), discussed data emerging from recent genome projects. For example, Olsen argued that the data from M. jannaschii indicate some archaeal genes and biochemical pathways are more similar to eukaryotic ones while others are more similar to bacteria. Similarly, Gray discussed data from the sequencing of entire mitochondrial genomes that are contributing to our understanding of the diversification of early eukaryotes. Douglas updated us on the origin of chloroplast genomes: an analysis of recently accumulated data from gene sequences, gene order and the distribution of plastids or plastid-like sequences among eukaryotes indicates a single origin of plastids.

Perhaps one of the most radical revisions at the meeting was the destruction of the Archezoa [9] (e.g. Giardia, figure 1) as primitive, amitochondrial eukaryotes. Graham Clark (London School of Hygiene and Tropical Medicine) and Masami Hasegawa (Institute of Statistical Mathematics, Tokyo) presented evidence that suggests several members of the Archezoa are secondarily amitochondrial. Gogarten, Palmer and Ford Doolittle (Dalhousie University) all pointed to the removal of Microsporidia from the base of the eukaryotic tree; recent molecular data from several proteins suggests that this taxon may instead be a fungus. Removal of yet another amitochondrial taxon from the Archezoa challenges the stepwise model of the origin of eukaryotes whereby the nucleus (and eukaryotic flagella) were derived first, followed later by acquisition of the endosymbiont that gave rise to mitochondria.

Based on these and other data, Ford Doolittle also led the audience through a period of deconstruction, abandonment and rethinking of the origin of Eukarya. With the rapid accumulation of molecular data, Doolittle suggests we need to rethink the origin of eukaryotes and the endosymbioses that gave rise to them. If instead, the acquisition of mitochondria occurred simultaneously with the formation of a eukaryotic cell, then we are left with the notion of Archezoa as "the kingdom that never was" (figure 1). Indeed a critical issue Ford Doolittle raised is whether endosymbiosis was inevitably "the sine qua non of the eukaryotic condition". The introduction of a foreign entity into the cell as a permanent guest may have provided the selective pressure needed to capture a nuclear membrane, ensuring the separation of these two genomes.

Searching for potentially early branching or at least understudied eukaryotes, David Patterson (University of Sydney, Australia) is leading an investigation into the world of eukaryotes in anoxic environments, low oxygen fresh and brackish water habitats. The broad survey of life in anoxic environments may lead to the discovery and characterization of genuinely primitively amitochondrial protists, if such exist. This will in turn lead to a refined and potentially profound understanding of the origin of eukaryotes. As Gray concluded, "protists are where the action is" for understanding the origin and evolution of eukaryotes - a call for others to join our efforts in probing the nature of these lost worlds: from the pre-RNA world, to the RNA world, to the last common ancestor and Archezoa.