B R U C E B U F F E T T Turbulent flow in Earth’s liquid­iron core generates the planet’s magnetic field through a process known as the geodynamo. This process is sustained by energy drawn from the core as it slowly cools1. Thermal convection is thought to be crucial, but revised estimates2,3 of thermal conducti­ vity in liquid iron at high pressure have called into question the adequacy of the commonly cited energy sources1. On page 387 of this issue, O’Rourke and Stevenson4 propose a solution to this energy crisis. They argue that if magnesium had dissolved in the liquid iron at high temperature when the core formed, then subsequent precipitation of magnesium­ bearing minerals on cooling would be an important source of energy. The authors’ theory warrants a serious reassessment of magnetic­field generation in other rocky (terrestrial) planets. Sustaining a magnetic field is difficult for a terrestrial planet. Creeping flow of the planet’s rocky shell (mantle) restricts heat loss from the underlying core. By comparison, the liquid­ iron core is an efficient thermal conductor. Thermal convection in the core ceases when heat flow into the mantle falls below the core’s capacity to deliver this heat by conduction alone, so high thermal conductivity may push the threshold for thermal convection beyond reach. In this scenario, turbulent flow in the core is driven mainly by buoyancy effects due to variations in the abundance of core constitu­ ents — as the core cools, some of the iron solid­ ifies and accumulates on the solid inner core, leaving lighter elements in the liquid outer core and thus causing convection1. A problem with this conventional view of the geodynamo’s energy sources emerges when we extrapolate back in time. Before the inner core formed (possibly less than 1 billion years ago1), the only energy source was thermal con­ vection. But current estimates of iron’s ther­ mal conductivity suggest that the heat flow required to sustain such convection at that time was extremely high. Even if this heat flow was feasible, an implausibly high core tempera­ ture would be needed to sustain it over geologi­ cal time. Despite these difficulties, Earth has somehow maintained a magnetic field for at least the past 3.4 billion years5. O’Rourke and Stevenson address this quan­ dary by proposing a new energy source. They suggest that magnesium can enter the core to form an iron alloy, even though it is normally considered to be nearly insoluble in liquid iron. Other alloying elements are more com­ monly proposed6 to explain why estimates of the core’s density, based on seismic data, are less than that of pure iron. But theoreti­ cal predictions7 and some experiments (see ref. 8, for example) suggest that magnesium can dissolve in liquid iron at sufficiently high temperatures. The authors argue that, because of its insolubility in iron, magnesium would probably become supersaturated as the core cools. The subsequent precipitation of magnesium­bearing minerals would leave behind a residual liquid enriched in iron, pro­ viding a compositional buoyancy that would drive fluid flow (Fig. 1). Two factors determine whether magne­ sium precipitation is a substantial energy source. First, the amount of magnesium that dissolved in liquid iron during core formation must be sufficient to meet the energy demands of the geodynamo. Second, the temperature dependence of magnesium’s solubility must be strong enough to promote supersaturation of magnesium with only a modest temperature decrease of the core (possibly just several hundred kelvin). Otherwise, a delay in the onset of magnesium precipitation could shut off the energy source in the past or present. O’Rourke and Stevenson tackled the first issue using previously reported experimental data9 that describe the partitioning of elements between liquid iron and silicate melts, a mix­ ture that represents the composition and state of the mantle during core formation. These data allowed them to estimate the concen­ tration of magnesium, oxygen and silicon in liquid iron, as well as the abundance of sidero­ phile elements (those that have an affinity for iron: nickel, cobalt, chromium, vanadium, niobium and tantalum) in the silicate melt, for two models of core formation. They then used a computational technique (a Monte Carlo method) to assess the average temperature and pressure conditions of core formation in the two models, by sampling many possible outcomes. Crucially, the researchers could account for the observed abundances of siderophiles in the mantle9 by using a model that permits a small fraction of the core to equilibrate with silicate melt at high temperatures (greater than 5,000 K). Many of the Monte Carlo outcomes for this model are also compatible with seismological constraints on the abun­ dance of light elements in the core6. These results include outcomes in which the liquid core contains 1–2% magnesium by weight. In other words, enough magnesium to power the geodynamo could have been dissolved in the early liquid core without violating known constraints on the composition of the core and mantle6,9. However, uncertainties prevent a defini­ tive assessment of magnesium precipitation. Some of the realizations sampled by the Monte Carlo method delay magnesium precipita­ tion into the distant future, whereas others permit precipitation much earlier; early and E A R T H S C I E N C E Another energy source for the geodynamo Magnesium is not usually considered to be a constituent of Earth’s core, but its presence there has now been proposed to explain an ongoing enigma — the identity of the energy sources that drive our planet’s magnetic field. See Letter p.387 Core Iron-rich �uid Mantle Buoyant particle Accumulated magnesium minerals Figure 1 | Possible processes at the boundary between Earth’s liquid core and rocky mantle. O’Rourke and Stevenson4 propose that magnesium dissolved in liquid iron at high temperature when the core formed. Subsequent precipitation of magnesium­containing minerals would produce buoyant solid particles that would rise through the liquid iron and accumulate at the top of the core. The residual liquid would be denser than the surrounding fluid because it is enriched in iron, and would therefore sink. The resulting vigorous convection would stir the core and generate Earth’s magnetic field, at modest rates of core cooling. 2 8 8 | N A T U R E | V O L 5 2 9 | 2 1 J A N U A R Y 2 0 1 6 NEWS & VIEWSRESEARCH © 2016 Macmillan Publishers Limited. All rights reserved J E F F S E T T L E M A N Many modern cancer drugs target mutationally activated proteins, but this treatment strategy has limitations. Only a relatively small num­ ber of mutations are seen recurrently across human tumours1, and drug resistance develops rapidly2. Targeting the epigenome3 — the chemically modified form of DNA, and of associated histones and other proteins that facilitate the packaging of DNA as chroma­ tin, all of which influence gene expression — is one of the alternative approaches being explored. Along with two papers4,5 published in Nature last year, a paper6 on page 413 of this issue provides some insight into the potential of epigenome­targeting drugs called BET inhibitors, and outlines the mechanisms by which tumours might become resistant to these drugs. It has long been recognized7 that tumour cells have distinct epigenomic features, which can lead to the overproduction of cancer­ promoting transcription factors such as MYC. Transcription factors are challenging thera­ peutic targets, because they lack structures that can be readily targeted with drugs. But developments in our understanding of the epigenome­regulating factors that influence gene expression, many of which seem to be ‘druggable’, have provided a potential way to sidestep this hurdle. Among these factors is the bromodomain protein family, which includes the BET subfamily8 (BRD2, 3, 4 and T). BET proteins contain two bromodomains, each with small pockets. These pockets bind to histones that have been tagged with acetyl groups, enabling BET proteins to recruit the cell’s transcriptional machinery to specific sites in the genome to regulate gene expression. BET subfamily mem­ bers such as BRD4, which can regulate MYC gene transcription, have been implicated in various tumours (particularly in cancers of the blood) and are therefore candidate targets for therapy8. A few years ago, the first of several small­ molecule BET inhibitors (JQ1) was discovered, and shown8 to effectively disrupt cancer­cell proliferation. This effect seemed to reflect inhibition of BET­mediated regulation of MYC expression. Early clinical trials of BET inhibi­ tors in leukaemia and lymphoma have been encouraging. Investigators are now seeking other disease contexts in which these inhibitors might work, and predicting the acquired resist­ ance mechanisms that will inevitably arise. The two 2015 studies4,5 converge on a potential mechanism of resistance to BET inhibition in acute myelogenous leukaemia (AML). In the first, Rathert et al.4 screened mouse AML cells for chromatin­modifying factors that are required for AML­cell sur­ vival. They confirmed that AML cells need Brd4, and identified several other factors for which inhibition confers AML­cell resistance to JQ1. In AML cells that were JQ1­resistant, the authors observed changes in specific epi­ genome features in DNA enhancer regions, abundant precipitation is required to provide an effective solution to the geodynamo energy crisis. Much of the uncertainty derives from the experimental estimates of element parti­ tioning between iron and silicate, particularly at high temperature. For example, the present work used information from a single set of experiments to derive magnesium’s solubility8. More work is clearly required to address these uncertainties, but the potential contribution of magnesium precipitation to the geodynamo provides plenty of motivation to improve our current knowledge. Magnesium precipitation would produce a buoyant solid that rises to the top of the core10. The dense, iron­rich residual fluid would also contribute to vigorous convection, offering ample energy for the geodynamo at relatively modest cooling rates. Such low cooling rates would allow warm fluid to accumulate at the top of the core, although convection due to magnesium precipitation might mix this warm fluid back to the core’s interior. Further complications are suggested by experimental evidence9 that the core’s liquid is not saturated with oxygen and silicon, indicating that these elements might transfer into the core from the mantle. The potential for two­way transfer across the core–mantle boundary in the light of O’Rourke and Stevenson’s theory is likely to send Earth scientists back to the drawing board. ■ Bruce Buffett is in the Department of Earth and Planetary Science, University of California, Berkeley, Berkeley, C A N C E R Bet on drug resistance Inhibitors of the BET bromodomain proteins are promising cancer therapeutics, but tumour cells are likely to become resistant to these drugs. Anticipated mechanisms of resistance have now been described. See Letter p.413 Resistance in AML Resistance in TNBC Histone MYC DNA BET inhibitor β-Catenin K K MYC K K MYC K K MYC K K MED1 CK2 BRD4 BRD4 BRD4 P Figure 1 | Circumventing BET inhibition. The BET protein BRD4 can bind to acetyl groups (K) on histone proteins around which DNA is packaged as chromatin. BRD4 recruits the cell’s transcriptional machinery, upregulating expression of the cancer­promoting MYC gene. Treatment with BET inhibitors can prevent BRD4–chromatin binding, stilting MYC transcription, but cancer cells rapidly develop drug resistance. Rathert et al.4 and Fong et al.5 report that, in acute myelogenous leukaemia (AML), drug resistance is conferred by activation of the Wnt­signalling pathway, which leads to DNA binding and MYC activation by the protein β­catenin. By contrast, Shu et al.6 find that resistance in triple­negative breast cancer (TNBC) arises owing to activation of the casein kinase 2 (CK2) enzyme. CK2 phosphorylates (P) BRD4, allowing BRD4 to bind to the transcriptional regulator protein MED1 to activate MYC. California 94720-4767, USA. e-mail: bbuffett@berkeley.edu 1. Nimmo, F. in Treatise on Geophysics 2nd edn (ed. Schubert, G.) 31–55 (Elsevier, 2015). 2. Pozzo, M., Davies, C., Gubbins, D. & Alfè, D. Nature 485, 355–358 (2012). 3. Gomi, H. & Hirose, K. Phys. Earth Planet. Inter. 247, 2–10 (2015). 4. O’Rourke, J. G. & Stevenson, D. J. Nature 529, 387–389 (2016). 5. Tarduno, J. A. et al. Science 327, 1238–1240 (2010). 6. Badro, J., Côté, A. S. & Brodholt, J. P. Proc. Natl Acad. Sci. USA 111, 7542–7545 (2014). 7. Wahl, S. M. & Militzer, B. Earth Planet. Sci. Lett. 410, 25–33 (2015). 8. Takafuji, N., Hirose, K., Mitome, M. & Bando, Y. Geophys. Res. Lett. 32, L06313 (2005). 9. Fischer, R. A. et al. Geochim. Cosmochim. Acta 167, 177–194 (2015). 10. Buffett, B. A., Garnero, E. J. & Jeanloz, R. Science 290, 1338–1342 (2000). 2 1 J A N U A R Y 2 0 1 6 | V O L 5 2 9 | N A T U R E | 2 8 9 NEWS & VIEWS RESEARCH © 2016 Macmillan Publishers Limited. All rights reserved Another energy source for the geodynamo Note References