Editors' Choice

Science  11 Apr 2008:
Vol. 320, Issue 5873, pp. 156

    Journey to the Center of the Earth

    The formation of Earth's core was the major differentiation of the early planet, segregating iron and other metal-loving elements to the planet's center, setting the conditions for a magnetic field, and releasing potential energy as heat. How was the iron localized so efficiently and quickly, displacing what was probably an early, semisolid silicate-rich core?

    It is generally thought that heating by impacts (some of which contained iron protocores) and enhanced radiogenic decay allowed blobs of iron metal to form and accumulate in a magma ocean in the outer Earth. Golabek et al. explore the iron's subsequent inward migration. Like others, they propose that enough iron accumulated to allow the dense blobs to move toward the center of Earth, displacing the silicate protocore. They also suggest, and show through numerical simulations scaled to the early Earth, that an extensive network of iron melt channels likely surrounded each sinking diapir. As these drained, they would form additional and larger daughter diapirs to scavenge iron from the upper mantle into the growing core. — BH

    Earth Planet. Sci. Lett. 10.1016/j.epsl.2008.02.033 (2008).


    Of Landmarks and Boundaries

    The striatum and the hippocampus both have important functions in learning and memory. Despite decades of intensive investigation, there are still a number of incompletely resolved questions. What sort of information is processed in each structure? How do the hippocampus and striatum cooperate to influence behavior? What type of learning is performed by each? These issues are tackled in two related papers.

    Using virtual reality and functional magnetic resonance imaging, Doeller et al. show that humans learn and remember the locations of objects relative to both local landmarks and environmental boundaries in parallel. The boundary-related system involves the right posterior hippocampus, whereas the landmark-related system involves the right dorsal striatum. When the hippocampal and striatal systems are in conflict because a landmark has been moved relative to the boundary, they do not compete directly to control behavior; instead, each system independently signals its solution to the task, with the ventromedial prefrontal cortex recruited for arbitration. Doeller and Burgess, using the same virtual reality object-location paradigm in behavioral experiments, show that striatal landmark-related processing of spatial learning obeys associative reinforcement learning, whereas hippocampal boundary-related processing does not. The latter performs purely incidental learning instead. Together, these studies provide evidence for the use of distinct learning rules in the hippocampus and striatum. — PRS

    Proc. Natl. Acad. Sci. U.S.A. 105, 10.1073/pnas.0801489105; 10.1073/pnas.0711433105 (2008).


    Feeling the Unseen

    In uncertain situations, rapid responses may be called for, and one line of thought suggests that assessments of stimulus valence (positive or negative) occur more quickly than the evolution of specific emotions, such as the discrimination between the two negative-valence states fear and disgust. Previous work has demonstrated that valence detection is indeed fast and can be applied to stimuli that are presented so briefly (a 40-ms exposure to a picture) as to lie outside of conscious awareness. Ruys and Stapel have looked at the question of whether specific emotions can similarly be elicited unconsciously by transiently displayed (120-ms duration) pictures. Measures of cognition (a stem-word completion task), feeling (self-reports of specific emotions), and behavior (choosing to avoid fearful or disgusting movies) all indicated that the super-quick exposure evoked only a valence-based response, whereas the merely quick, yet still subliminal, stimulus was capable of evoking specific emotions. — GJC

    Psychol. Sci. 19, 385 (2008).


    A Big World After All?

    From the spread of viral epidemics to the flow of information in a social network, a key question is the nature of the pathway taken. Epidemic-style models assume that propagation should occur from one individual to many others, following small-world principles. If the viral chain does not die out, it can be mapped as a tree-like structure with many branches that have short paths from the root (or original source). Liben-Nowell and Kleinberg looked at the spread of data on the Internet by examining the lists of names in two email petitions and found that, rather than expanding to many individuals in a few steps, the tree structures were very narrow and several hundreds of layers deep, with large variations in the time needed to reach the edges of the network. By running simulations on other networks, the authors confirmed that the distribution could not be modeled by a random epidemic process. However, they successfully reproduced the long, narrow trees by altering the model to allow for variations in the response time of the recipients (asynchronous response) and to allow recipients the option of either forwarding the message or group replying (back-transmission) to those already in the list. The results highlight how the transmission of data can vary from the small-world network in which the dissemination occurs. — MSL

    Proc. Natl. Acad. Sci. U.S.A. 105, 4633 (2008).


    Ticking Down to Metastasis

    The idea that the accumulation of mutations is a type of clock that can be used to chart the developmental or evolutionary history of cells or organisms is well established. Adapting this mode of analysis, Jones et al. have catalogued mutations that occur during the stages of colorectal cancer, focusing on the poorly understood transition from advanced carcinoma to metastasis. They model the rates of appearance of mutations found in a large-scale survey of colorectal cancer (see Wood et al., Research Articles, 16 Nov. 2007, p. 1108). Mutation rates were determined by combining the numbers of mutations with estimates of cell division time, under the assumption that mutation rate and cell division time were constant in each patient. They conclude that it takes roughly 17 years for a large adenoma, which is benign, to evolve into an advanced carcinoma, which is malignant, but less than 2 years for that carcinoma to begin to metastasize. Evolution occurs through cycles of clonal expansion and mutation, and their model provides estimates of the “birth dates” of the founder cells. Although some metastasis-specific alterations were identified, the authors consider that all or most of the mutations needed for metastasis may already be present in the carcinomas. — BRJ

    Proc. Natl. Acad. Sci. U.S.A. 105, 4283 (2008).


    Mix-and-Match Warfare

    Many secondary metabolites are produced by plants for defensive use against insect predators. Insects, in turn, develop their own countermeasures and may end up using the plant's metabolites as a signal to target, rather than to avoid, that plant. Added complications are that the suite of predators changes seasonally, and groups of plants have different vulnerabilities than do isolated plants—with the result that a plant is best served by a nimble tactical approach for the flexible deployment of its defensive weaponry.

    Glucosinolates, one category of secondary metabolites, show considerable biochemical diversity owing to the enzymes that generate their derivatives. Wentzell and Kliebenstein have analyzed the diversity of glucosinolate production in Arabidopsis. Their results show that different pieces of the plant make different product profiles that are fine-tuned by the age of the plant and the local population density. Variability in the profile of glucosinolate derivatives among different accessions of Arabidopsis highlights the influence of additional modifier factors, enabling the plant to balance the risks of detection against the effectiveness of defense in its production of glucosinolate derivatives. — PJH

    Plant Physiol. 146, 10.1104/pp.107.115279 (2008).


    A Locus of Resistance to Activation

    Chagas' disease is caused by a protozoan parasite of the Trypanosomatidae, others of which cause a range of high-morbidity and disfiguring human diseases, usually within impoverished populations. Drug treatment takes several months, is expensive, and can have side effects. Moreover, the nitroheterocyclic compounds nifurtimox and benznidazole have been the frontline drugs for more than 40 years; nitroheterocycles do not act directly upon their targets, but must first be converted by endogenous enzymes to the active compounds.

    Wilkinson et al. have worked out how these drugs are activated. Eight months of selection in vitro generated nitroheterocyclicresistant parasites that had lost a gene encoding a mitochondrially targeted type I nitroreductase, a flavoenzyme that reduces the nitro group and yields agents that damage DNA. Deleting a single copy of this gene enhanced resistance to nifurtimox and other nitroheterocycles severalfold. Intriguingly, knocking out both copies slowed replication and blocked differentiation into the infectious form of the parasite, which suggested that this enzyme is essential for parasite development in its mammalian host. — CA

    Proc. Natl. Acad. Sci. U.S.A. 105, 5022 (2008).