Formalizing planning and information search in naturalistic decision-making.

Decisions made by mammals and birds are often temporally extended. They require planning and sampling of decision-relevant information. Our understanding of such decision-making remains in its infancy compared with simpler, forced-choice paradigms. However, recent advances in algorithms supporting planning and information search provide a lens through which we can explain neural and behavioral data in these tasks. We review these advances to obtain a clearer understanding for why planning and curiosity originated in certain species but not others; how activity in the medial temporal lobe, prefrontal and cingulate cortices may support these behaviors; and how planning and information search may complement each other as means to improve future action selection.

Nucleic acid sequence bank.

In the report by T. Kakunaga and J. D. Crow (25 July, p. 505), Fig. 1 on page 506 should have been printed as follows: [See figure in the PDF file]

Relationship of promagainin to three other prohormones from the skin of Xenopus laevis: a different perspective.

We observed a striking sequence similarity between precursors for promagainin and procaerulein type I (excluding the caerulein peptide region). Additional comparisons of the promagainin precursor with those of other procaeruleins, proxenopsin, and peptide-Gly-Leu-amide revealed that all possess one or more copies of a structurally similar spacer module, from which an amphiphilic spacer peptide is cleaved. Promagainin yields the magainins, spacer peptides with antimicrobial activity; we suggest other spacer peptides may have similar activity. We propose that the genes for the four kinds of hormones were derived from a common ancestral gene through gene and exon duplications and that the procaerulein and proxenopsin genes are mosaic genes in which the original 3'-ends were replaced by exon shuffling.

Cystatin domains in alpha-2-HS-glycoprotein and fetuin.

We have found that chain A of alpha-2-HS-glycoprotein contains two cystatin domains that show closest similarity to those of kininogen. Most likely, the two proteins diverged after the primary duplication of a single cystatin domain as the two cystatin domains of alpha-2-HS-glycoprotein are more similar, especially in disulfide bonding, to the corresponding domains of kininogen than to each other. We also propose that the carboxyl-terminal (non-cystatin) parts of kininogen and alpha-2-HS-glycoprotein contain homologous segments. We suggest that alpha-2-HS-glycoprotein may act as an inhibitor of the cysteine proteinases responsible for bone resorption. We have also found that fetuin is closely related to alpha-2-HS-glycoprotein.

The prokaryote-eukaryote interface.

Over the past 30 years the study of the sequences of proteins and nucleic acids has produced almost incredible amounts of information, new concepts, and new avenues of research. The beginning was slow: the first peptide hormones sequenced in the early 1950's, the first cytochrome c (horse) in 1961, the first bacterial ferredoxin in 1964, and the first transfer RNA (yeast alanine tRNA) in 1965. In the past 6 years, the rate of data accumulation has accelerated tremendously, primarily due to technological advances in nucleic acid sequencing techniques. For investigators of biological evolution, the sequence data and the new information on genetic mechanisms would prove to be the best evidence for elucidating relationships among the genomes of living organisms and for deducing phylogenetic history. In particular, they needed evidence to decide between the two hypotheses for the origin of eukaryotic cells. Now, less than 20 years since Margulis renewed the investigation of this problem, comparisons of protein and nucleic acid sequences, especially of the small subunit ribosomal RNAs, have answered this question in favor of the endosymbiotic origin of eukaryotic cells. After briefly discussing some of the concepts that helped resolve this controversy and the problems involved in using sequence data for evolutionary studies, we describe a few examples of useful evolutionary trees.

Evolution of protein complexity: the blue copper-containing oxidases and related proteins.

The blue copper proteins and their relatives have been compared by sequence alignments, by comparison of three-dimensional structures, and by construction of phylogenetic trees. The group contains proteins varying in size from 100 residues to over 2,300 residues in a single chain, containing from zero to nine copper atoms, and with a broad variation in function ranging from electron carrier proteins and oxidases to the blood coagulation factors V and VIII. Difference matrices show the sequence difference to be over 90% for many pairs in the group, yet alignment scores and other evidence suggest that they all evolved from a common ancestor. We have attempted to delineate how this evolution took place and in particular to define the mechanisms by which these proteins acquired an ever-increasing complexity in structure and function. We find evidence for six such mechanisms in this group of proteins: domain enlargement, in which a single domain increases in size from about 100 residues up to 210; domain duplication, which allows for a size increase from about 170 to about 1,000 residues; segment elongation, in which a small segment undergoes multiple successive duplications that can increase the chain size 50-fold; domain recruitment, in which a domain coded elsewhere in the genome is added on to the peptide chain; subunit formation, to form multisubunit proteins; and glycosylation, which in some cases doubles the size of the protein molecule. Size increase allows for the evolution of new catalytic properties, in particular the oxidase function, and for the formation of coagulation factors with multiple interaction sites and regulatory properties. The blood coagulation system is examined as an example in which a system of interacting proteins evolved by successive duplications of larger parts of the genome. The evolution of size, functionality, and diversity is compared with the general question of increase in size and complexity in biology.

von Willebrand factor shares a distinctive cysteine-rich domain with thrombospondin and procollagen.

The identification of common domains among different proteins is of great interest at present. We have found that a cysteine-rich domain in thrombospondin, also present in types I and III procollagen alpha 1 chains, is related to two internally homologous domains in von Willebrand factor. In the four proteins these domains are similar in length (64-74 residues) and have nine invariant cysteines, some of which form intramolecular disulfide bonds. The structural and functional similarities of this domain in the four kinds of proteins, and its correspondence in procollagen to an exon, support our hypothesis of a common origin for the domain.

Analysis and organization of protein sequence data: a retrospective spanning four decades.

Protein sequence data are as useful and valuable today as was envisioned by pioneering sequencers and by the organizers of the first sequence database. Sequence analysis was first the province of specialists who developed search, comparison, and tree-building methods. Microcomputers, communication satellites, and the Internet have made these methods accessible to any scientist. The rapid increase in the data has driven a succession of changes in how databases are compiled, distributed, and accessed. Large public databases have become international collaborations. Although they need to develop still more efficient ways to accumulate, organize, annotate, and standardize huge amounts of data, inadequate support is available for such efforts. Thus there will be greater reliance on direct input from the scientific community. The World Wide Web is essential but not sufficient for integrated access to related databases.

Avidin-like domain in an epidermal growth factor homolog from a sea urchin.

We have found that a protein from the purple sea urchin has a carboxyl-terminal domain with striking sequence similarity to chicken avidin and bacterial streptavidin. All our evidence supports the homology of these sequences. Tetramers of avidin and streptavidin bind biotin strongly; the biotin binding site involves two to four tryptophans and probably an adjacent lysine in each chain. The presence of four tryptophans at equivalent positions in the sea urchin protein domain suggests that it may also be able to bind biotin and inhibit cell growth, as do the two other proteins. Alternatively, this domain may have acquired a new role as part of a multidomain protein.

Identifying domains in protein sequences.

Amino acid sequences are often inferred without information about the structure or function of the protein. Just as unknown proteins may be identified through comparison with other known sequences, so can structural and functional domains be inferred through analogy with other known domain sequences. A database of domain sequences, along with their known properties, would be very helpful in the characterization of a protein sequence in terms of its possible conformation and function. Here we describe the type of information to be included in such a database and discuss the techniques that may be used to determine the characteristic features of domains. Recently identified domains in thyroid peroxidase, chondroitin-sulfate proteoglycan core proteins, an epidermal growth factor precursor homolog, and a bacterial beta-amylase are presented as examples.