RESEARCH NEWS & VIEWS net magnetization of zero). However, a small magnetization arises from canting (tilting) of the anti-aligned magnetic moments. The weak magnetization of bismuth ferrite can be amplified through strong interfacial magnetic coupling to a ferromagnet5,6. So, if the weak magnetization could be coupled to the material’s ferroelectric polarization, then the direction of magnetization of an adjacent ferromagnetic layer could be altered using an electric field. Unfortunately, direct reversal of ferroelectric polarization in bismuth ferrite was expected to leave the orientation of the canted magnetic moment unchanged7, which means that magnetoelectrically driven mag netization reversal would be impossible. Enter Heron and colleagues, who find that ferroelectric-polarization reversal in a strained bismuth ferrite film follows an indirect path way. When grown as an elastically strained layer on a substrate of DyScO3 (Dy is dysprosium; Sc is scandium) covered with an electrode of strontium ruthenate (SrRuO3), bismuth ferrite forms stripe-like ferroelectric domains con taining stable polarization of electric dipoles, in which the orientations of the dipoles are constrained to two of the four space diagonals of the material’s pseudocubic unit cell. The authors studied local ferroelectric switching in the electric field of the tip of a scanning force microscope at the surface of the bismuth fer rite, obtaining data at microsecond resolution during repeated scans under an applied directcurrent voltage. They observed an overall full reversal of polarization (a 180° switch) nearly everywhere, but this was reached through an intermediate switch by 71° or 109° via the other ‘active’ space diagonal of the unit cell (Fig. 1). The researchers’ computational modelling con firmed that the energy barrier to this two-step switching is smaller than for direct 180° switch ing, and predicted that the indirect switching path reverses the canted magnetic moment. The authors next deposited a spin valve consisting of two layers of a ferromagnet (Co0.9Fe0.1) separated by a copper layer onto an underlayer of bismuth ferrite. They found that the device exhibited about the same val ues of high and low electrical resistance in a field — either in an in-plane magnetic field or in an electric field at the vertical to the bismuth ferrite layer — when the field was cycled between positive and negative values. This result requires the magnetization of the Co0.9Fe0.1 to be aligned with the canted mag netic moment in the adjacent bismuth ferrite film6. The authors used a technique called X-ray magnetic circular dichroism photo emission electron microscopy to show the local magnetization reversal that occurs in Co0.9Fe0.1 after voltage pulses are applied to the bismuth ferrite underlayer, demonstrat ing that magnetic coupling at the interface is strong enough to induce the switching. The crucial message from Heron and colleagues’ work is that the switching path
matters: the final orientation of coupled ferroic order parameters (such as spontaneous magnetization, electric polarization and elas tic strain) during a switching process depends not only on the initial orientation of the para meters and the direction of applied fields, but also on intermediate steps during switching. These steps are governed by kinetic barri ers that affect the dynamics of the switching process, and can be assessed using theory and controlled by elastic strains in films. The authors’ findings open up the prospect of multistep switching processes that access unexpected states of multiferroic materials other than bismuth ferrite. This in turn might enable alternative strategies for engineering magneto electric switching. The study also confirms that the weak magnetism common to many multi ferroics can be amplified by magnetic coupling to a conventional ferromagnetic metal5. Techniques that allow high-quality inter faces between ferromagnetic metals and multiferroics to be prepared have enabled experiments that advance our understand ing of magnetic coupling at such interfaces6. However, these interfaces are unstable in large, cycling electric fields, and this issue must be resolved for practical applications. The same problem afflicts interfaces between ferro electric materials and metal electrodes, but has been partly solved following long-standing research efforts8. Nevertheless, the stability of ferroelectric–metal interfaces is still a major factor in the lifetimes of devices that use such interfaces.
The walls of newly formed ferroelectric domains cross from one side to the other of the stripe domains observed by Heron et al. during switching. So, if the stripe domains can be engineered to be narrow (of the order of 100 nanometres wide), this would help the fast operation of multiferroic spintronic devices, such as spin valves, by reducing the crossing time to a few nanoseconds in a suf ficiently large electric field. A considerable amount of work may be needed to develop fast magnetoelectric switching and to stabilize multiferroic-to-ferromagnet interfaces. Even so, the electrical control of ferromagnetism demonstrated by Heron and co-workers is a decisive step towards the realization of multi ferroic spintronic devices. ■ Kathrin Dörr is at the Institute for Physics, Martin Luther University Halle-Wittenberg, 06099 Halle, Germany. Andreas Herklotz is in the Materials Science and Technology Division, Oak Ridge National Laboratory, Oak Ridge, Tennessee 37831, USA. e-mail: [email protected] 1. Grünberg, P. A. Rev. Mod. Phys. 80, 1531–1540 (2008). 2. Fert, A. Rev. Mod. Phys. 80, 1517–1530 (2008). 3. Heron, J. T. et al. Nature 516, 370–373 (2014). 4. Spaldin, N. A. & Fiebig, M. Science 309, 391–392 (2005). 5. Heron, J. T. et al. Phys. Rev. Lett. 107, 217202 (2011). 6. Trassin, M. et al. Phys. Rev. B 87, 134426 (2013). 7. Ederer, C. & Spaldin, N. A. Phys. Rev. B 71, 060401(R) (2005). 8. Boyn, S. et al. Appl. Phys. Lett. 104, 052909 (2014).
I N FLUEN Z A
An RNA-synthesizing machine Crystal structures of the complete RNA polymerases from influenza A and B viruses provide insight into how these enzymes initiate RNA synthesis, and reveal targets for antiviral drug design. See Articles p.355 & p.361 ROBERT M. KRUG
I
nfluenza A and B viruses cause a highly contagious disease in humans that results in approximately 250,000 to 500,000 deaths worldwide each year1. In addition, influenza A viruses are responsible for periodic human pandemics that can have substantially higher mortality rates; the most severe pandemic, in 1918, caused around 40 million deaths2. The primary defence against influenza infections has been vaccination, but antiviral drugs also play a key part, for example in the elderly, who are not well protected by vaccines. Circulating influenza viruses have developed
3 3 8 | NAT U R E | VO L 5 1 6 | 1 8 / 2 5 D E C E M B E R 2 0 1 4
© 2014 Macmillan Publishers Limited. All rights reserved
resistance to several of the available antivirals3,4, highlighting the need to develop new drugs. The viral RNA polymerase — the enzyme that catalyses the synthesis of virus-specific RNAs in infected cells — is an attractive tar get for antiviral drug development, but pro gress has been hampered by the absence of a three-dimensional structure of the enzyme. Two papers from the Cusack research group published in this issue (Pflug et al.5 on page 355 and Reich et al.6 on page 361) now report the first structures of the viral polymerases of influenza A and B, respectively. Influenza A and B viral polymerases are each composed of three proteins: PB1, PB2
NEWS & VIEWS RESEARCH a
b
Capped RNA primer Cap
PA endonuclease domain
PB2 cap-binding domain
Catalytic site
Catalytic site
RNA polymerase
Figure 1 | RNA transcription by influenza polymerase. The influenza A and B RNA polymerases are made up of three viral proteins: PB1, PB2 and PA. The complete structures presented by Pflug et al.5 and Reich et al.6 show the location of key elements of these protein complexes. a, Synthesis of viral messenger RNA is initiated by a process called cap-snatching, in which a short RNA primer, containing a cap structure, is cleaved from a cellular pre-mRNA (not shown) by the PA endonuclease domain of the viral polymerase. The PB cap-binding domain and the endonuclease domain at the two ends of the U-shaped viral polymerase initially face each other across a channel, and the endonuclease cleaves the pre-mRNA to produce the capped RNA primer. b, The cap-binding domain then moves the capped primer away from the endonuclease domain and directs it down into the RNA-synthesizing catalytic site.
and PA (ref. 7). The tripartite polymerase is activated by RNA sequences found at the two extremities of the viral RNA, which interact with each other to form a partially doublestranded structure referred to as the promoter for viral RNA synthesis. The new three-dimen sional structures contain the viral polymerase in association with this RNA promoter. The viral polymerases catalyse two types of RNA synthesis: transcription to produce mes senger RNA; and RNA replication to produce templates for the production of viral RNA. These two processes are initiated by different mechanisms. The synthesis of viral mRNA is initiated by a process called cap-snatching, which was discovered more than 30 years ago8. The viral polymerase binds to a chemi cal structure, the cap, located at the ends of the precursors of cellular mRNAs (pre-mRNAs), and the polymerase then uses its endonucle ase (nucleic-acid cleaving) activity to cleave the pre-mRNAs at a position 10–15 nucleo tides downstream from the cap. The resulting cap-containing fragment acts as a primer to initiate viral mRNA synthesis. By contrast, the RNA-replication reaction is initiated without a primer7. The viral polymerases analysed in the new papers catalyse the initiation of both transcription and RNA replication. The structures of the influenza A and B RNA polymerases show that the three protein subunits make multiple complex interactions with each other, and that all three subunits participate in most of the key poly merase functions. The polymerase forms a U-shaped structure, with the cap-binding domain (which is part of the PB2 protein) at the top of one of the arms of this struc ture and the endonuclease domain (which is located in the amino-terminal region of the PA protein) at the top of the other arm. All
three polymerase subunits are involved in positioning the endonuclease domain. In the influenza A polymerase structure, the capbinding and endonuclease domains face each other across a channel whose breadth cor responds to the length of the cap-containing fragment that is produced (Fig. 1). Reich et al. determined two influenza B polymerase structures. On the basis of their observation that the cap-binding domain of one of these crystal structures is rotated by 70° compared to the other B polymerase struc ture and Pflug and colleagues’ influenza A polymerase structure, the authors propose that this domain moves the capped primer away from the endonuclease domain and directs it down into the RNA-synthesizing catalytic site of the polymerase, which is in the PB1 protein (Fig. 1). The N-terminal endo nuclease-containing domain of PA is on the opposite side of the polymerase to the larger carboxy-terminal PA domain. The two PA domains are connected by a PA linker region that wraps around the external face of the PB1 protein. All three polymerase protein subunits participate in the binding of the two strands of the RNA promoter that is located close to the PB1 active site. Reich and colleagues deduce a possible mechanism by which viral RNA replication is initiated in the absence of a primer, by com paring the influenza A polymerase structure with the structures of other viral RNA poly merases, particularly those of hepatitis C and dengue viruses. These two viral polymerases contain a ‘priming loop’ in which an aromatic amino-acid residue (tyrosine or trypto phan) stabilizes the positioning of the initial ribonucleotide triphosphates that form base pairs with the template RNA in the absence of an RNA primer9. The PB1 protein in the
influenza A polymerase contains an analogous putative priming loop (amino acids 641–657) in which histidine (an aromatic amino-acid residue) at position 649 could interact with the initial incoming ribonucleoside triphosphates to facilitate unprimed initiation. Pflug and colleagues’ influenza A polymer ase structure includes the region of the PB2 protein that has been implicated in the adap tation of avian influenza A viruses for repli cation in mammalian hosts10, and this should provide insight into the mechanisms by which this region functions. Indeed, the polymerase structures will provide the basis for much future work. Previous structures of fragments of the polymerase identified the cap-binding and endonuclease sites11–13, both of which are potential targets for the development of antiinfluenza drugs. The new, complete structures provide further targets, including the PB1 active site, the binding sites of the two strands of the viral RNA promoter and numerous sites that are required for essential rearrangements during viral RNA synthesis. The structures also give us a much better understanding of the mechanisms of influenza virus RNA synthesis — both for transcrip tion and viral RNA replication. Nonetheless, there are considerable gaps in our knowledge of these processes. For example, we do not know how the 3ʹ end of the viral RNA tem plate is relocated to the PB1 active site, nor how the polymerase progresses from initia tion of viral RNA synthesis to the elongation of RNA chains. Furthermore, because the viral RNA template in infected cells is coated with multiple viral nucleoprotein molecules along almost its entire length, it is not clear how the viral RNA polymerase accesses and copies such a template. ■ Robert M. Krug is in the Department of Molecular Biosciences, Center for Infectious Disease, Institute of Cellular and Molecular Biology, University of Texas at Austin, Austin, Texas 78712, USA. e-mail: [email protected] 1. World Health Organization. www.who.int/ mediacentre/factsheets/fs211/en/ 2. Reid, A. H., Taubenberger, J. K. & Fanning, T. G. Microbes Infect. 3, 81–87 (2001). 3. Deyde, V. M. et al. J. Infect. Dis. 196, 249–257 (2007). 4. Bloom, J. D., Gong, L. I. & Baltimore, D. Science 328, 1272–1275 (2010). 5. Pflug, A., Guilligay, D., Reich, S. & Cusack, S. Nature 516, 355–360 (2014). 6. Reich, S. et al. Nature 516, 361–366 (2014). 7. Boivin, S., Cusack, S., Ruigrok, R. W. & Hart, D. J. J. Biol. Chem. 285, 28411–28417 (2010). 8. Plotch, S. J., Bouloy, M. & Krug, R. M. Proc. Natl Acad. Sci. USA 76, 1618–1622 (1979). 9. Lescar, J. & Canard, B. Curr. Opin. Struct. Biol. 19, 759–767 (2009). 10. Cauldwell, A. V., Long, J. S., Moncorgé, O. & Barclay, W. S. J. Gen. Virol. 95, 1193–1210 (2014). 11. Dias, A. et al. Nature 458, 914–918 (2009). 12. Yuan, P. et al. Nature 458, 909–913 (2009). 13. Guilligay, D. et al. Nature Struct. Mol. Biol. 15, 500–506 (2008).
1 8 / 2 5 D E C E M B E R 2 0 1 4 | VO L 5 1 6 | NAT U R E | 3 3 9
© 2014 Macmillan Publishers Limited. All rights reserved
matters: the final orientation of coupled ferroic order parameters (such as spontaneous magnetization, electric polarization and elas tic strain) during a switching process depends not only on the initial orientation of the para meters and the direction of applied fields, but also on intermediate steps during switching. These steps are governed by kinetic barri ers that affect the dynamics of the switching process, and can be assessed using theory and controlled by elastic strains in films. The authors’ findings open up the prospect of multistep switching processes that access unexpected states of multiferroic materials other than bismuth ferrite. This in turn might enable alternative strategies for engineering magneto electric switching. The study also confirms that the weak magnetism common to many multi ferroics can be amplified by magnetic coupling to a conventional ferromagnetic metal5. Techniques that allow high-quality inter faces between ferromagnetic metals and multiferroics to be prepared have enabled experiments that advance our understand ing of magnetic coupling at such interfaces6. However, these interfaces are unstable in large, cycling electric fields, and this issue must be resolved for practical applications. The same problem afflicts interfaces between ferro electric materials and metal electrodes, but has been partly solved following long-standing research efforts8. Nevertheless, the stability of ferroelectric–metal interfaces is still a major factor in the lifetimes of devices that use such interfaces.
The walls of newly formed ferroelectric domains cross from one side to the other of the stripe domains observed by Heron et al. during switching. So, if the stripe domains can be engineered to be narrow (of the order of 100 nanometres wide), this would help the fast operation of multiferroic spintronic devices, such as spin valves, by reducing the crossing time to a few nanoseconds in a suf ficiently large electric field. A considerable amount of work may be needed to develop fast magnetoelectric switching and to stabilize multiferroic-to-ferromagnet interfaces. Even so, the electrical control of ferromagnetism demonstrated by Heron and co-workers is a decisive step towards the realization of multi ferroic spintronic devices. ■ Kathrin Dörr is at the Institute for Physics, Martin Luther University Halle-Wittenberg, 06099 Halle, Germany. Andreas Herklotz is in the Materials Science and Technology Division, Oak Ridge National Laboratory, Oak Ridge, Tennessee 37831, USA. e-mail: [email protected] 1. Grünberg, P. A. Rev. Mod. Phys. 80, 1531–1540 (2008). 2. Fert, A. Rev. Mod. Phys. 80, 1517–1530 (2008). 3. Heron, J. T. et al. Nature 516, 370–373 (2014). 4. Spaldin, N. A. & Fiebig, M. Science 309, 391–392 (2005). 5. Heron, J. T. et al. Phys. Rev. Lett. 107, 217202 (2011). 6. Trassin, M. et al. Phys. Rev. B 87, 134426 (2013). 7. Ederer, C. & Spaldin, N. A. Phys. Rev. B 71, 060401(R) (2005). 8. Boyn, S. et al. Appl. Phys. Lett. 104, 052909 (2014).
I N FLUEN Z A
An RNA-synthesizing machine Crystal structures of the complete RNA polymerases from influenza A and B viruses provide insight into how these enzymes initiate RNA synthesis, and reveal targets for antiviral drug design. See Articles p.355 & p.361 ROBERT M. KRUG
I
nfluenza A and B viruses cause a highly contagious disease in humans that results in approximately 250,000 to 500,000 deaths worldwide each year1. In addition, influenza A viruses are responsible for periodic human pandemics that can have substantially higher mortality rates; the most severe pandemic, in 1918, caused around 40 million deaths2. The primary defence against influenza infections has been vaccination, but antiviral drugs also play a key part, for example in the elderly, who are not well protected by vaccines. Circulating influenza viruses have developed
3 3 8 | NAT U R E | VO L 5 1 6 | 1 8 / 2 5 D E C E M B E R 2 0 1 4
© 2014 Macmillan Publishers Limited. All rights reserved
resistance to several of the available antivirals3,4, highlighting the need to develop new drugs. The viral RNA polymerase — the enzyme that catalyses the synthesis of virus-specific RNAs in infected cells — is an attractive tar get for antiviral drug development, but pro gress has been hampered by the absence of a three-dimensional structure of the enzyme. Two papers from the Cusack research group published in this issue (Pflug et al.5 on page 355 and Reich et al.6 on page 361) now report the first structures of the viral polymerases of influenza A and B, respectively. Influenza A and B viral polymerases are each composed of three proteins: PB1, PB2
NEWS & VIEWS RESEARCH a
b
Capped RNA primer Cap
PA endonuclease domain
PB2 cap-binding domain
Catalytic site
Catalytic site
RNA polymerase
Figure 1 | RNA transcription by influenza polymerase. The influenza A and B RNA polymerases are made up of three viral proteins: PB1, PB2 and PA. The complete structures presented by Pflug et al.5 and Reich et al.6 show the location of key elements of these protein complexes. a, Synthesis of viral messenger RNA is initiated by a process called cap-snatching, in which a short RNA primer, containing a cap structure, is cleaved from a cellular pre-mRNA (not shown) by the PA endonuclease domain of the viral polymerase. The PB cap-binding domain and the endonuclease domain at the two ends of the U-shaped viral polymerase initially face each other across a channel, and the endonuclease cleaves the pre-mRNA to produce the capped RNA primer. b, The cap-binding domain then moves the capped primer away from the endonuclease domain and directs it down into the RNA-synthesizing catalytic site.
and PA (ref. 7). The tripartite polymerase is activated by RNA sequences found at the two extremities of the viral RNA, which interact with each other to form a partially doublestranded structure referred to as the promoter for viral RNA synthesis. The new three-dimen sional structures contain the viral polymerase in association with this RNA promoter. The viral polymerases catalyse two types of RNA synthesis: transcription to produce mes senger RNA; and RNA replication to produce templates for the production of viral RNA. These two processes are initiated by different mechanisms. The synthesis of viral mRNA is initiated by a process called cap-snatching, which was discovered more than 30 years ago8. The viral polymerase binds to a chemi cal structure, the cap, located at the ends of the precursors of cellular mRNAs (pre-mRNAs), and the polymerase then uses its endonucle ase (nucleic-acid cleaving) activity to cleave the pre-mRNAs at a position 10–15 nucleo tides downstream from the cap. The resulting cap-containing fragment acts as a primer to initiate viral mRNA synthesis. By contrast, the RNA-replication reaction is initiated without a primer7. The viral polymerases analysed in the new papers catalyse the initiation of both transcription and RNA replication. The structures of the influenza A and B RNA polymerases show that the three protein subunits make multiple complex interactions with each other, and that all three subunits participate in most of the key poly merase functions. The polymerase forms a U-shaped structure, with the cap-binding domain (which is part of the PB2 protein) at the top of one of the arms of this struc ture and the endonuclease domain (which is located in the amino-terminal region of the PA protein) at the top of the other arm. All
three polymerase subunits are involved in positioning the endonuclease domain. In the influenza A polymerase structure, the capbinding and endonuclease domains face each other across a channel whose breadth cor responds to the length of the cap-containing fragment that is produced (Fig. 1). Reich et al. determined two influenza B polymerase structures. On the basis of their observation that the cap-binding domain of one of these crystal structures is rotated by 70° compared to the other B polymerase struc ture and Pflug and colleagues’ influenza A polymerase structure, the authors propose that this domain moves the capped primer away from the endonuclease domain and directs it down into the RNA-synthesizing catalytic site of the polymerase, which is in the PB1 protein (Fig. 1). The N-terminal endo nuclease-containing domain of PA is on the opposite side of the polymerase to the larger carboxy-terminal PA domain. The two PA domains are connected by a PA linker region that wraps around the external face of the PB1 protein. All three polymerase protein subunits participate in the binding of the two strands of the RNA promoter that is located close to the PB1 active site. Reich and colleagues deduce a possible mechanism by which viral RNA replication is initiated in the absence of a primer, by com paring the influenza A polymerase structure with the structures of other viral RNA poly merases, particularly those of hepatitis C and dengue viruses. These two viral polymerases contain a ‘priming loop’ in which an aromatic amino-acid residue (tyrosine or trypto phan) stabilizes the positioning of the initial ribonucleotide triphosphates that form base pairs with the template RNA in the absence of an RNA primer9. The PB1 protein in the
influenza A polymerase contains an analogous putative priming loop (amino acids 641–657) in which histidine (an aromatic amino-acid residue) at position 649 could interact with the initial incoming ribonucleoside triphosphates to facilitate unprimed initiation. Pflug and colleagues’ influenza A polymer ase structure includes the region of the PB2 protein that has been implicated in the adap tation of avian influenza A viruses for repli cation in mammalian hosts10, and this should provide insight into the mechanisms by which this region functions. Indeed, the polymerase structures will provide the basis for much future work. Previous structures of fragments of the polymerase identified the cap-binding and endonuclease sites11–13, both of which are potential targets for the development of antiinfluenza drugs. The new, complete structures provide further targets, including the PB1 active site, the binding sites of the two strands of the viral RNA promoter and numerous sites that are required for essential rearrangements during viral RNA synthesis. The structures also give us a much better understanding of the mechanisms of influenza virus RNA synthesis — both for transcrip tion and viral RNA replication. Nonetheless, there are considerable gaps in our knowledge of these processes. For example, we do not know how the 3ʹ end of the viral RNA tem plate is relocated to the PB1 active site, nor how the polymerase progresses from initia tion of viral RNA synthesis to the elongation of RNA chains. Furthermore, because the viral RNA template in infected cells is coated with multiple viral nucleoprotein molecules along almost its entire length, it is not clear how the viral RNA polymerase accesses and copies such a template. ■ Robert M. Krug is in the Department of Molecular Biosciences, Center for Infectious Disease, Institute of Cellular and Molecular Biology, University of Texas at Austin, Austin, Texas 78712, USA. e-mail: [email protected] 1. World Health Organization. www.who.int/ mediacentre/factsheets/fs211/en/ 2. Reid, A. H., Taubenberger, J. K. & Fanning, T. G. Microbes Infect. 3, 81–87 (2001). 3. Deyde, V. M. et al. J. Infect. Dis. 196, 249–257 (2007). 4. Bloom, J. D., Gong, L. I. & Baltimore, D. Science 328, 1272–1275 (2010). 5. Pflug, A., Guilligay, D., Reich, S. & Cusack, S. Nature 516, 355–360 (2014). 6. Reich, S. et al. Nature 516, 361–366 (2014). 7. Boivin, S., Cusack, S., Ruigrok, R. W. & Hart, D. J. J. Biol. Chem. 285, 28411–28417 (2010). 8. Plotch, S. J., Bouloy, M. & Krug, R. M. Proc. Natl Acad. Sci. USA 76, 1618–1622 (1979). 9. Lescar, J. & Canard, B. Curr. Opin. Struct. Biol. 19, 759–767 (2009). 10. Cauldwell, A. V., Long, J. S., Moncorgé, O. & Barclay, W. S. J. Gen. Virol. 95, 1193–1210 (2014). 11. Dias, A. et al. Nature 458, 914–918 (2009). 12. Yuan, P. et al. Nature 458, 909–913 (2009). 13. Guilligay, D. et al. Nature Struct. Mol. Biol. 15, 500–506 (2008).
1 8 / 2 5 D E C E M B E R 2 0 1 4 | VO L 5 1 6 | NAT U R E | 3 3 9
© 2014 Macmillan Publishers Limited. All rights reserved
