INSIGHTS
REFERENCES
PHOTO: CORA BERGAÑITOS
1. 2. 3. 4. 5. 6. 7. 8.
9. 10. 11. 12.
E. Rignot et al., Geophys. Res. Lett. 41, 3502 (2014). I. Joughin, B. E. Smith, B. Medley, Science 344, 735 (2014). S. Schmidtko et al., Science 346, 1227 (2014). J. A. Griggs, J. L. Bamber, J. Glaciol. 57, 485 (2011). S. S. Jacobs et al., Nat. Geosci. 4, 519 (2011). H. D. Pritchard et al., Nature 484, 502 (2012). M. A. Depoorter et al., Nature 502, 89 (2013). A. P. S. Wong, N. L. Bindoff, A. Forbes, Ocean, Ice, and Atmosphere: Interactions at the Antarctic Continental Margin Antarctic Research Series 75, 173 (1998). Y.-H. Park et al., J. Mar. Syst. 17, 233 (1998). E. Rignot, S. S. Jacobs, Science 296, 2020 (2002). A. K. Wåhlin et al., J. Phys. Oceanogr. 43, 2054 (2013). S. Lee, S. B. Feldstein, Science 339, 563 (2013). 10.1126/science.aaa0886
DEVELOPMENTAL BIOLOGY
Death to the losers Cell competition is linked to innate immunity mechanisms to eliminate unwanted cells and maintain healthy tissue By Ginés Morata and Luna Ballesteros-Arias
I
nsects use the innate immunity signaling pathways Toll and IMD to fight infections by foreign pathogens, bacteria, or fungi (1). These pathways induce the expression of powerful antimicrobial- antifungal peptides (AMPs) that are released from the fat body cells. On page 1199 of this issue, Meyer et al. (2) report that fruit flies make use of components of both pathways to build another pathway that removes its
Imaginal disc. An image of a wing imaginal disc (Drosophilia melanogaster) is shown, containing marked pairs of sibling cell clones generated by mitotic recombination. Cells expressing extra Myc protein (green) are competing against neighboring cells, including those in the sibling clones (red). Nuclei (blue) are stained with the DNA dye Hoechst.
own cells that are abnormal and may impair tissue fitness or compromise the survival of the organism. In contrast to the antimicrobial operation, the killing signal does not come from outside the tissue; it originates from interactions of the doomed cell with neighboring cells through the process of cell competition. Cell competition is an interactive process originally reported in the imaginal discs
SCIENCE sciencemag.org
(the precursors of the adult cuticle) (see the first figure) of Drosophila melanogaster (3, 4), but more recent studies with mouse cells (5–7) indicate that it is a conserved mechanism in animal tissues. Cell competition is a general surveillance process that identifies and eliminates cells that are weaker than their neighbors, or are abnormal or not well adapted to their developmental context. Its function ensures genetic uniformity of cell populations and also contributes to the general fitness of developing tissues (8). As part of its surveillance function, cell competition exerts a therapeutic role by destroying oncogenic cells that may arise during development (9–11). Cell competition is a context-dependent phenomenon. It describes interactions between two types of viable cells that lead to the death, by apoptosis (12), of one of the cell types. As reported in the classical experiments (3, 4), slow-dividing cells (“losers”) are perfectly viable on their own; they are eliminated only if residing in the same population as faster-dividing cells—the “winners.” This is the characteristic feature of loser cells: They are viable but die under the influence of their neighbors. Another feature of interest is that it is a short-range phenomenon—the winners and losers lay very close to each other, possibly in physical contact. A simple scenario for cell competition is a crosstalk between winner and loser cells that triggers apoptosis (12) in the losers and possibly activates engulfment genes in the winners; winners appear to consume (phagocytose) the losers (13). But what mechanisms underlie this interaction? What are the events leading to apoptosis in the loser cells? The study by Meyer et al. opens a new avenue of research into the molecular mechanism linking cell competition to innate immunity. They make the unexpected discovery that in the wing imaginal disc, loser cells are treated as foreign pathogens in their resident populations. The destruction of the loser cells is effected by the innate immune system, and presumably this activation is carried out by neighboring winner cells. Centro de Biología Molecular CSIC-UAM, Universidad Autonoma de Madrid, 28049 Madrid, Spain. E-mail: gmorata@ cbm.csic.es 5 DECEMBER 2014 • VOL 346 ISSUE 6214
Published by AAAS
1181
Downloaded from www.sciencemag.org on December 4, 2014
drive basal melting (2, 5–7, 10). The details of circulation in the basal cavity remain elusive, however, because there is no easy way to drill through hundreds of meters of ice to deploy oceanographic instruments in the cavity. Schmitdko et al. avoid the challenges of in situ measurements in the basal cavity by instead evaluating the properties of the water that can be transported into the cavity. Their analysis of hydrographic ocean data collected since 1975 shows that over the past 40 years, Circumpolar Deep Water has warmed at all longitudes around Antarctica. The authors also report that in some regions of the Antarctic continental shelf (for example, near the Amundsen and Bellingshausen Seas, adjacent to West Antarctica), changes in the Antarctic Continental Shelf Bottom Water appear to mirror changes in Circumpolar Deep Water. Other regions appear protected from the warming Circumpolar Deep Water. The critical difference is that in warming regions, Circumpolar Deep Water slopes upward to the shelf break, possibly as a response to strong winds from the west that are expected to induce upwelling (see the figure). In contrast, in regions without shelf warming, including the Ross and Weddell Seas, Circumpolar Deep Water slopes downward to the shelf break, consistent with winds from the east that limit onshore flow. An analysis of mooring data in the Amundsen Sea (11) also concluded that local winds play a key role in bringing water onto the Antarctic Shelf. Schmidtko et al.’s results highlight the critical role of wind forcing. Winds north of the shelf break determine the properties of water that in turn influence basal melting of ice shelves. Both greenhouse warming and ozone depletion can intensify the Southern Hemisphere westerly winds and displace them southward (12). Although current wind conditions appear to shield the Ross and Weddell Seas from the melting effects associated with Circumpolar Deep Water, future changes in the winds could modify that and thus alter the ice-shelf buttresses, with further effects on global sea level rise. ■
INSIGHTS | P E R S P E C T I V E S
Pro-Spz
Spz TRRs
?
Tube
PGRP-LCs MyD88
IMD dFADD DREDD
Pelle
Loser cell is engulfed by winner cell
dTAK1-dTAB2 P P P
Cactus
KennyIRD5
Dif/Dl P
Loser cell dies through apoptosis
P
Relish
rpr Apoptosis
hid
Aberrant cell arises in normal tissue Elimination. A proposal for how components of the innate immunity Toll and IMD signaling pathways are used in a different pathway (shown in yellow) to eliminate aberrant cells (purple) that have arisen within a normal tissue (in Drosophila). By a mechanism that is still unknown, aberrant cells are detected as different by their neighbors and are determined to be “losers.” The end result is the activation of the proapoptotic genes hid
Meyer et al. assayed cell competition by looking at the cellular response to different amounts of Myc protein (a transcription factor that promotes cell proliferation) and/or the ribosomal protein RpL14. The rule is that viable cells with low amounts of either Myc or RpL14 become losers when surrounded by cells with higher amounts. The key finding of Meyer et al. is that loser cells are rescued from elimination when the wing imaginal disc is mutant for one of various components of the Toll and IMD signaling pathways. It is intriguing that some components of the Toll and IMD pathways are required but not others; this finding suggests that cell competition functions through a cascade of events different from the canonical Toll and IMD pathways (see the second figure). A key factor in the immune response is the cytokine Spätzle (Spz), a ligand of the Toll receptor; upon its absence, cell competition is suppressed. However, the loss of Toll still allows cell competition, which suggests that Spz functions through other Toll-related receptors (TRRs), of which there are several in Drosophila. The loss of some of them is very effective in suppressing cell competition. At the bottom of the cascade, the transcription factors Relish, Dorsal (Dl), and Dif are activated differentially, depending on the assay. These transcription factors activate the expression of AMPs during the antimicrobial response, and Meyer et al. show that 1182
and rpr, and death of the loser cells. They are subsequently engulfed by the neighboring “winner” cells, thereby restoring a healthy tissue. PGRP-LC, peptidoglycan recognition protein LC; IMD, immune deficiency; TAK1, transforming growth factor beta–activated kinase 1; DREDD, death-related ced-3/Nedd2-like caspase; FADD, FAS-associated death domain; TAB2, TAKI-associated binding protein 2; IRD5, immune-response deficient 5.
they also activate apoptosis in the loser cells. The proapoptotic genes called head involution defective (hid) and reaper (rpr) become activated by Relish and either Dl or Dif, although the requirements for hid and rpr appear to be different according to the cell competition assay used. The triggering of apoptosis by Toll and IMD factors is itself a new finding that may have general implications in the understanding of innate immunity and also of the regulation of programmed cell death. The study of Meyer et al. poses some challenging questions about the mechanisms of cell competition. One concerns the activation of Spz in the loser cells in the wing disc. Normally Spz is expressed in several larval tissues, including hemocytes, the fat body (similar to the mammalian liver), and salivary glands but not in the imaginal discs. Because Spz is a secreted molecule, it could originate elsewhere and become accessible to wing cells, or it may be specifically produced and secreted by the winner cells. Moreover, its activation involves several proteolytic steps. How this process occurs autonomously in the loser cells is an intriguing problem. Another question concerns the generality of the phenomenon. The two assays reported, the amounts of Myc or of the ribosomal protein RpL14, are based on the comparison of the metabolic activity of
the interacting cells. It will be of interest to determine whether the innate immunity pathway also functions in other distinct cases of cell competition—for example, the elimination of cells with wrong identity (14) or the removal of oncogenic cells (9, 10). The latter is especially appealing in view of recent evidence (11) that cell competition has an antitumor role also in the mouse thymus. Given the conservation of the innate immunity systems in Drosophila and mammals (1), it seems possible that innate immunity might have a similar antitumor role in mammals. ■ REFERENCES
1. L. A. O’Neill, D. Golenbock, A. G. Bowie, Nat. Rev. Immunol. 13, 453 (2013). 2. S. N. Meyer et al., Science 346, 1258236 (2014). 3. G. Morata, P. Ripoll, Dev. Biol. 42, 211 (1975). 4. P. Simpson, Dev. Biol. 69, 182 (1979). 5. E. R. Oliver, T. L. Saunders, S. A. Tarlé, T. Glaser, Development 131, 3907 (2004). 6. C. Clavería, G. Giovinazzo, R. Sierra, M. Torres, Nature 500, 39 (2013). 7. M. Sancho et al., Dev. Cell 26, 19 (2013). 8. M. Amoyel, E. A. Bach, Development 141, 988 (2014). 9. C. L. Chen, M. C. Schroeder, M. Kango-Singh, C. Tao, G. Halder, Proc. Natl. Acad. Sci. U.S.A. 109, 484 (2012). 10. L. Ballesteros-Arias, V. Saavedra, G. Morata, Oncogene 33, 4377 (2014). 11. V. C. Martins et al., Nature 509, 465 (2014). 12. E. Moreno, K. Basler, G. Morata, Nature 416, 755 (2002). 13. W. Li, N. E. Baker, Cell 129, 1215 (2007). 14. T. Adachi-Yamada, M. B. O’Connor, Dev. Biol. 251, 74 (2002). 10.1126/science.aaa2345
sciencemag.org SCIENCE
5 DECEMBER 2014 • VOL 346 ISSUE 6214
Published by AAAS
Death to the losers Ginés Morata and Luna Ballesteros-Arias Science 346, 1181 (2014); DOI: 10.1126/science.aaa2345
If you wish to distribute this article to others, you can order high-quality copies for your colleagues, clients, or customers by clicking here. Permission to republish or repurpose articles or portions of articles can be obtained by following the guidelines here. The following resources related to this article are available online at www.sciencemag.org (this information is current as of December 4, 2014 ): Updated information and services, including high-resolution figures, can be found in the online version of this article at: http://www.sciencemag.org/content/346/6214/1181.full.html A list of selected additional articles on the Science Web sites related to this article can be found at: http://www.sciencemag.org/content/346/6214/1181.full.html#related This article cites 14 articles, 4 of which can be accessed free: http://www.sciencemag.org/content/346/6214/1181.full.html#ref-list-1
Science (print ISSN 0036-8075; online ISSN 1095-9203) is published weekly, except the last week in December, by the American Association for the Advancement of Science, 1200 New York Avenue NW, Washington, DC 20005. Copyright 2014 by the American Association for the Advancement of Science; all rights reserved. The title Science is a registered trademark of AAAS.
Downloaded from www.sciencemag.org on December 4, 2014
This copy is for your personal, non-commercial use only.
REFERENCES
PHOTO: CORA BERGAÑITOS
1. 2. 3. 4. 5. 6. 7. 8.
9. 10. 11. 12.
E. Rignot et al., Geophys. Res. Lett. 41, 3502 (2014). I. Joughin, B. E. Smith, B. Medley, Science 344, 735 (2014). S. Schmidtko et al., Science 346, 1227 (2014). J. A. Griggs, J. L. Bamber, J. Glaciol. 57, 485 (2011). S. S. Jacobs et al., Nat. Geosci. 4, 519 (2011). H. D. Pritchard et al., Nature 484, 502 (2012). M. A. Depoorter et al., Nature 502, 89 (2013). A. P. S. Wong, N. L. Bindoff, A. Forbes, Ocean, Ice, and Atmosphere: Interactions at the Antarctic Continental Margin Antarctic Research Series 75, 173 (1998). Y.-H. Park et al., J. Mar. Syst. 17, 233 (1998). E. Rignot, S. S. Jacobs, Science 296, 2020 (2002). A. K. Wåhlin et al., J. Phys. Oceanogr. 43, 2054 (2013). S. Lee, S. B. Feldstein, Science 339, 563 (2013). 10.1126/science.aaa0886
DEVELOPMENTAL BIOLOGY
Death to the losers Cell competition is linked to innate immunity mechanisms to eliminate unwanted cells and maintain healthy tissue By Ginés Morata and Luna Ballesteros-Arias
I
nsects use the innate immunity signaling pathways Toll and IMD to fight infections by foreign pathogens, bacteria, or fungi (1). These pathways induce the expression of powerful antimicrobial- antifungal peptides (AMPs) that are released from the fat body cells. On page 1199 of this issue, Meyer et al. (2) report that fruit flies make use of components of both pathways to build another pathway that removes its
Imaginal disc. An image of a wing imaginal disc (Drosophilia melanogaster) is shown, containing marked pairs of sibling cell clones generated by mitotic recombination. Cells expressing extra Myc protein (green) are competing against neighboring cells, including those in the sibling clones (red). Nuclei (blue) are stained with the DNA dye Hoechst.
own cells that are abnormal and may impair tissue fitness or compromise the survival of the organism. In contrast to the antimicrobial operation, the killing signal does not come from outside the tissue; it originates from interactions of the doomed cell with neighboring cells through the process of cell competition. Cell competition is an interactive process originally reported in the imaginal discs
SCIENCE sciencemag.org
(the precursors of the adult cuticle) (see the first figure) of Drosophila melanogaster (3, 4), but more recent studies with mouse cells (5–7) indicate that it is a conserved mechanism in animal tissues. Cell competition is a general surveillance process that identifies and eliminates cells that are weaker than their neighbors, or are abnormal or not well adapted to their developmental context. Its function ensures genetic uniformity of cell populations and also contributes to the general fitness of developing tissues (8). As part of its surveillance function, cell competition exerts a therapeutic role by destroying oncogenic cells that may arise during development (9–11). Cell competition is a context-dependent phenomenon. It describes interactions between two types of viable cells that lead to the death, by apoptosis (12), of one of the cell types. As reported in the classical experiments (3, 4), slow-dividing cells (“losers”) are perfectly viable on their own; they are eliminated only if residing in the same population as faster-dividing cells—the “winners.” This is the characteristic feature of loser cells: They are viable but die under the influence of their neighbors. Another feature of interest is that it is a short-range phenomenon—the winners and losers lay very close to each other, possibly in physical contact. A simple scenario for cell competition is a crosstalk between winner and loser cells that triggers apoptosis (12) in the losers and possibly activates engulfment genes in the winners; winners appear to consume (phagocytose) the losers (13). But what mechanisms underlie this interaction? What are the events leading to apoptosis in the loser cells? The study by Meyer et al. opens a new avenue of research into the molecular mechanism linking cell competition to innate immunity. They make the unexpected discovery that in the wing imaginal disc, loser cells are treated as foreign pathogens in their resident populations. The destruction of the loser cells is effected by the innate immune system, and presumably this activation is carried out by neighboring winner cells. Centro de Biología Molecular CSIC-UAM, Universidad Autonoma de Madrid, 28049 Madrid, Spain. E-mail: gmorata@ cbm.csic.es 5 DECEMBER 2014 • VOL 346 ISSUE 6214
Published by AAAS
1181
Downloaded from www.sciencemag.org on December 4, 2014
drive basal melting (2, 5–7, 10). The details of circulation in the basal cavity remain elusive, however, because there is no easy way to drill through hundreds of meters of ice to deploy oceanographic instruments in the cavity. Schmitdko et al. avoid the challenges of in situ measurements in the basal cavity by instead evaluating the properties of the water that can be transported into the cavity. Their analysis of hydrographic ocean data collected since 1975 shows that over the past 40 years, Circumpolar Deep Water has warmed at all longitudes around Antarctica. The authors also report that in some regions of the Antarctic continental shelf (for example, near the Amundsen and Bellingshausen Seas, adjacent to West Antarctica), changes in the Antarctic Continental Shelf Bottom Water appear to mirror changes in Circumpolar Deep Water. Other regions appear protected from the warming Circumpolar Deep Water. The critical difference is that in warming regions, Circumpolar Deep Water slopes upward to the shelf break, possibly as a response to strong winds from the west that are expected to induce upwelling (see the figure). In contrast, in regions without shelf warming, including the Ross and Weddell Seas, Circumpolar Deep Water slopes downward to the shelf break, consistent with winds from the east that limit onshore flow. An analysis of mooring data in the Amundsen Sea (11) also concluded that local winds play a key role in bringing water onto the Antarctic Shelf. Schmidtko et al.’s results highlight the critical role of wind forcing. Winds north of the shelf break determine the properties of water that in turn influence basal melting of ice shelves. Both greenhouse warming and ozone depletion can intensify the Southern Hemisphere westerly winds and displace them southward (12). Although current wind conditions appear to shield the Ross and Weddell Seas from the melting effects associated with Circumpolar Deep Water, future changes in the winds could modify that and thus alter the ice-shelf buttresses, with further effects on global sea level rise. ■
INSIGHTS | P E R S P E C T I V E S
Pro-Spz
Spz TRRs
?
Tube
PGRP-LCs MyD88
IMD dFADD DREDD
Pelle
Loser cell is engulfed by winner cell
dTAK1-dTAB2 P P P
Cactus
KennyIRD5
Dif/Dl P
Loser cell dies through apoptosis
P
Relish
rpr Apoptosis
hid
Aberrant cell arises in normal tissue Elimination. A proposal for how components of the innate immunity Toll and IMD signaling pathways are used in a different pathway (shown in yellow) to eliminate aberrant cells (purple) that have arisen within a normal tissue (in Drosophila). By a mechanism that is still unknown, aberrant cells are detected as different by their neighbors and are determined to be “losers.” The end result is the activation of the proapoptotic genes hid
Meyer et al. assayed cell competition by looking at the cellular response to different amounts of Myc protein (a transcription factor that promotes cell proliferation) and/or the ribosomal protein RpL14. The rule is that viable cells with low amounts of either Myc or RpL14 become losers when surrounded by cells with higher amounts. The key finding of Meyer et al. is that loser cells are rescued from elimination when the wing imaginal disc is mutant for one of various components of the Toll and IMD signaling pathways. It is intriguing that some components of the Toll and IMD pathways are required but not others; this finding suggests that cell competition functions through a cascade of events different from the canonical Toll and IMD pathways (see the second figure). A key factor in the immune response is the cytokine Spätzle (Spz), a ligand of the Toll receptor; upon its absence, cell competition is suppressed. However, the loss of Toll still allows cell competition, which suggests that Spz functions through other Toll-related receptors (TRRs), of which there are several in Drosophila. The loss of some of them is very effective in suppressing cell competition. At the bottom of the cascade, the transcription factors Relish, Dorsal (Dl), and Dif are activated differentially, depending on the assay. These transcription factors activate the expression of AMPs during the antimicrobial response, and Meyer et al. show that 1182
and rpr, and death of the loser cells. They are subsequently engulfed by the neighboring “winner” cells, thereby restoring a healthy tissue. PGRP-LC, peptidoglycan recognition protein LC; IMD, immune deficiency; TAK1, transforming growth factor beta–activated kinase 1; DREDD, death-related ced-3/Nedd2-like caspase; FADD, FAS-associated death domain; TAB2, TAKI-associated binding protein 2; IRD5, immune-response deficient 5.
they also activate apoptosis in the loser cells. The proapoptotic genes called head involution defective (hid) and reaper (rpr) become activated by Relish and either Dl or Dif, although the requirements for hid and rpr appear to be different according to the cell competition assay used. The triggering of apoptosis by Toll and IMD factors is itself a new finding that may have general implications in the understanding of innate immunity and also of the regulation of programmed cell death. The study of Meyer et al. poses some challenging questions about the mechanisms of cell competition. One concerns the activation of Spz in the loser cells in the wing disc. Normally Spz is expressed in several larval tissues, including hemocytes, the fat body (similar to the mammalian liver), and salivary glands but not in the imaginal discs. Because Spz is a secreted molecule, it could originate elsewhere and become accessible to wing cells, or it may be specifically produced and secreted by the winner cells. Moreover, its activation involves several proteolytic steps. How this process occurs autonomously in the loser cells is an intriguing problem. Another question concerns the generality of the phenomenon. The two assays reported, the amounts of Myc or of the ribosomal protein RpL14, are based on the comparison of the metabolic activity of
the interacting cells. It will be of interest to determine whether the innate immunity pathway also functions in other distinct cases of cell competition—for example, the elimination of cells with wrong identity (14) or the removal of oncogenic cells (9, 10). The latter is especially appealing in view of recent evidence (11) that cell competition has an antitumor role also in the mouse thymus. Given the conservation of the innate immunity systems in Drosophila and mammals (1), it seems possible that innate immunity might have a similar antitumor role in mammals. ■ REFERENCES
1. L. A. O’Neill, D. Golenbock, A. G. Bowie, Nat. Rev. Immunol. 13, 453 (2013). 2. S. N. Meyer et al., Science 346, 1258236 (2014). 3. G. Morata, P. Ripoll, Dev. Biol. 42, 211 (1975). 4. P. Simpson, Dev. Biol. 69, 182 (1979). 5. E. R. Oliver, T. L. Saunders, S. A. Tarlé, T. Glaser, Development 131, 3907 (2004). 6. C. Clavería, G. Giovinazzo, R. Sierra, M. Torres, Nature 500, 39 (2013). 7. M. Sancho et al., Dev. Cell 26, 19 (2013). 8. M. Amoyel, E. A. Bach, Development 141, 988 (2014). 9. C. L. Chen, M. C. Schroeder, M. Kango-Singh, C. Tao, G. Halder, Proc. Natl. Acad. Sci. U.S.A. 109, 484 (2012). 10. L. Ballesteros-Arias, V. Saavedra, G. Morata, Oncogene 33, 4377 (2014). 11. V. C. Martins et al., Nature 509, 465 (2014). 12. E. Moreno, K. Basler, G. Morata, Nature 416, 755 (2002). 13. W. Li, N. E. Baker, Cell 129, 1215 (2007). 14. T. Adachi-Yamada, M. B. O’Connor, Dev. Biol. 251, 74 (2002). 10.1126/science.aaa2345
sciencemag.org SCIENCE
5 DECEMBER 2014 • VOL 346 ISSUE 6214
Published by AAAS
Death to the losers Ginés Morata and Luna Ballesteros-Arias Science 346, 1181 (2014); DOI: 10.1126/science.aaa2345
If you wish to distribute this article to others, you can order high-quality copies for your colleagues, clients, or customers by clicking here. Permission to republish or repurpose articles or portions of articles can be obtained by following the guidelines here. The following resources related to this article are available online at www.sciencemag.org (this information is current as of December 4, 2014 ): Updated information and services, including high-resolution figures, can be found in the online version of this article at: http://www.sciencemag.org/content/346/6214/1181.full.html A list of selected additional articles on the Science Web sites related to this article can be found at: http://www.sciencemag.org/content/346/6214/1181.full.html#related This article cites 14 articles, 4 of which can be accessed free: http://www.sciencemag.org/content/346/6214/1181.full.html#ref-list-1
Science (print ISSN 0036-8075; online ISSN 1095-9203) is published weekly, except the last week in December, by the American Association for the Advancement of Science, 1200 New York Avenue NW, Washington, DC 20005. Copyright 2014 by the American Association for the Advancement of Science; all rights reserved. The title Science is a registered trademark of AAAS.
Downloaded from www.sciencemag.org on December 4, 2014
This copy is for your personal, non-commercial use only.
