- Email: [email protected]
A Toggle Switch in Plant Nitrate Uptake
2007). Thus, GDE2 could influence neurogenesis by altering the level of Notch ligand-receptor signaling. Alternatively, GDE2 catalytic activity could modify the cytoskeletal network that maintains neural progenitors (Farkas and Huttner, 2008), for example by severing their attachments to the neuroepithelium and causing the cells to undergo differentiation. Finally, although motor neuron differentiation was found by Yan et al. to be delayed in animals lacking Prdx1 or Gde2, motor neuron formation eventually recovers later in development. These findings suggest that either additional glycerophosphodiester phosphodiester ase and peroxiredoxin proteins also con-
tribute to neuronal differentiation or the Prdx1/GDE2 pathway works in parallel with other regulatory systems that balance neural progenitor proliferation and differentiation. Further insights into the function of the Prdx1/GDE2 pathway will require the identification of the signals produced by GDE2, the means by which these signals are perceived, and determining how this pathway interfaces with core neurogenic factors.
Farkas, L.M., and Huttner, W.B. (2008). Curr. Opin. Cell Biol. 20, 707–715.
References
Skwarek, L.C., Garroni, M.K., Commisso, C., and Boulianne, G.L. (2007). Dev. Cell 13, 783–795.
Barber, S.C., Mead, R.J., and Shaw, P.J. (2006). Biochim. Biophys. Acta 1762, 1051–1067.
Ulloa, F., and Briscoe, J. (2007). Cell Cycle 6, 2640–2649.
Briscoe, J., and Novitch, B.G. (2008).Philos. Trans. R. Soc. Lond. B Biol. Sci. 363, 57–70.
Yan, Y., Sabharwal, P., Rao, M., and Sockanathan, S. (2009). Cell, this issue.
Hall, A., Karplus, P.A., and Poole, L.B. (2009). FEBS J. 276, 2469–2477. Holmberg, J., Hansson, E., Malewicz, M., Sandberg, M., Perlmann, T., Lendahl, U., and Muhr, J. (2008). Development 135, 1843–1851. Jang, H.H., Lee, K.O., Chi, Y.H., Jung, B.G., Park, S.K., Park, J.H., Lee, J.R., Lee, S.S., Moon, J.C., Yun, J.W., et al. (2004). Cell 117, 625–635. Rao, M., and Sockanathan, S. (2005). Science 309, 2212–2215.
A Toggle Switch in Plant Nitrate Uptake Grégory Vert1,* and Joanne Chory2
BPMP, CNRS UMR 5004, 34060 Montpellier Cedex 1, France Plant Biology Laboratory and Howard Hughes Medical Institute, The Salk Institute for Biological Studies, La Jolla, CA 92037, USA *Correspondence: [email protected] DOI 10.1016/j.cell.2009.09.005
1 2
In plants, the uptake of nitrate from the soil is a critical process controlled by complex regulatory networks that target nitrate transporters in the roots. In this issue, Ho et al. (2009) show that phosphorylation of the CHL1 nitrate transporter allows the plant root to sense and respond to different nitrate concentrations in the soil. As sessile organisms that are bound to one location, plants have developed sophisticated mechanisms to ensure appropriate adaptation to constantly changing environmental conditions. To cope with unfavorable nutrient availability in the soil, plants harbor on the root surface a battery of specialized transporters to maintain efficient uptake of nutrients. These transporters are controlled by the integration of complex regulatory networks that underlie external and internal cues to modulate nutrient uptake capacity in accordance with the nutrient demand of the plant and the nutrient availability of the soil. Reverse genetics in the model plant Arabidopsis thaliana has provided a wealth of information on the roles of the different transporters. How-
ever, the molecular mechanisms regulating the acquisition of nutrients have remained largely unknown. Given the vital role nitrate plays as a nitrogen source for most plants, the study of nitrate uptake serves as a test case to characterize plant nutrient transport systems and to unravel the signaling pathways governing nutrient acquisition. Nitrate is taken up from the soil by several transporters in the plant root—including the high-affinity NRT2.1 transporter and the high- and low-affinity (dual-affinity) CHL1 transporter (also called NRT1.1) —and directly induces the expression of these transporters as part of the primary nitrate response (Tsay et al., 2007). In this issue of Cell, Ho et al. (2009) shed light on the molecular mechanisms of nitrate signaling by CHL1 in
1064 Cell 138, September 18, 2009 ©2009 Elsevier Inc.
Arabidopsis, highlighting the complexity of nutrient transport, sensing, and signaling in plants. The first hints of a potential role for CHL1 in nitrate signaling came from studies in chl1 loss-of-function mutant Arabidopsis plants, which suggested that CHL1 regulates the expression of NRT2.1 in response to nitrate. These plants lacking CHL1 indeed failed to downregulate the expression of NRT2.1 in the presence of high concentrations of nitrogen (Munos et al., 2004) and were unable to increase the proliferation of lateral roots in nitraterich zones in the soil (Remans et al., 2006), another well-established response to nitrate. In their new work, Ho and coworkers now unambiguously unravel the role of CHL1 in nitrate signaling by
characterizing the chl1-9 mutant allele. They show that the chl1-9 allele encodes a mutant CHL1 protein that is defective in nitrate transport but is still able to trigger responses to nitrate, as visualized by the nitrate-dependent and CHL1-mediated induction of NRT2.1 expression. Genetic and biochemical approaches show that low levels of nitrate are detected by CHL1 and signaled through calcineurin B-like interacting protein kinase CIPK23-dependent phosphorylation of threonine 101 (T101) in CHL1. Phosphorylation of this residue has previously been implicated in the switch from the low- to high-affinity modes of nitrate transport (Liu and Tsay, 2003) and now appears to also mediate the downregulation of primary nitrate responses (Figure 1A). In contrast, a high nitrate concentration converts CHL1 to its unphosphorylated form, leading to the full nitrate response and CHL1-mediated low-affinity nitrate transport (Figure 1A). Whether CHL1 acts as a bona fide nitrate sensor or is simply part of the nitrate signaling pathway remains to be elucidated. Notably, a better understanding of the mechanisms leading to the CIPK23-mediated phosphorylation of CHL1 is required to fully grasp the possible role of CHL1 in direct sensing of external nitrate. CIPKs are activated by calcium ions via the calcineurin B-like (CBL) calcium-binding proteins (Batistic and Kudla, 2009). Although CBL9 is required for CIPK23mediated phosphorylation of CHL1 in frog oocytes and in vitro, evidence of a calcium signal that is triggered by nitrate remains elusive. If CHL1 indeed acts as a nitrate sensor, an early CHL1dependent calcium signal is expected to participate in the establishment of responses to nitrate. The rationale for the strong and rapid induction of CIPK23 expression observed by Ho and coworkers in plants exposed to nitrate is also unclear. This response to nitrate suggests either that nitrate sensing takes place far upstream of CIPK23mediated CHL1 phosphorylation, or that CIPK23 expression is feedback regulated by a CHL1-dependent nitrate signaling pathway. Testing whether the nitrate-dependent early induction of CIPK23 expression requires functional CHL1 should help to position CHL1 in the early steps of nitrate signaling.
Figure 1. Nitrate Signaling and Transport in Plant Roots (A) When plants are exposed to conditions of low external nitrate (NO3−; <1 mM), the dual-affinity nitrate transporter CHL1 recruits or activates the calcineurin B-like CBL9-CBL-interacting protein kinase CIPK23 complex through an unknown mechanism, leading to the phosphorylation of CHL1 on threonine 101 (T101). Once phosphorylated, CHL1 switches to a high-affinity mode of nitrate transport to ensure efficient nitrate uptake under nitrate-poor conditions. CHL1 phosphorylated at T101 also restricts primary nitrate responses to the high-affinity phase of nitrate signaling. When high external nitrate concentrations (>1 mM) are sensed by CHL1, the CBL9-CIPK23-dependent phosphorylation of CHL1 at T101 does not occur. CHL1 not phosphorylated at T101 remains in the low-affinity nitrate transport mode and activates increased primary nitrate responses. Establishment of the primary nitrate responses in the low-affinity phase also requires CIPK8. CIPK8 is activated by an unknown CBL and likely phosphorylates CHL1 at a different residue from T101. (B) In the plant root, CHL1 (red) senses and signals the external nitrate concentration to control the mode of nitrate transport (low- or high-affinity) and to regulate the expression of another nitrate transporter, NRT2.1 (blue), in different root cell layers through an unknown signal. NRT2.1 gene expression is under dual regulation—by the CHL1-dependent external nitrate signaling pathway and by negative feedback regulation driven by the levels of nitrogen-containing metabolites of the shoot—to tightly regulate nitrate uptake according to plant demand and nitrate availability in the soil.
When plants grown in the absence of nitrate are challenged with different concentrations of external nitrate, they rapidly initiate a biphasic primary nitrate response (Hu et al., 2009). This biphasic nitrate response could be controlled by two distinct nitrate-sensing mechanisms that drive either a high-affinity (nitrate concentration <1 mM) or a low-affinity (nitrate concentration >1 mM) nitrate signaling pathway. CIPK8 is a positive regulator of the primary nitrate response in the low-affinity phase of nitrate signaling (Hu et al., 2009). The work of Ho et al. now adds to the complexity of the regulation of nitrate responses with their identification of CIPK23 as a negative regulator of the high-affinity phase. Thus, nitrate
responses in plants appear to be regulated by the interplay of different CIPKs, which drive different phases of nitrate signaling and likely target different CHL1 residues for phosphorylation (Figure 1A). Interestingly, CIPK23 is known to regulate K+ uptake by the AKT1 K+ channel in Arabidopsis (Xu et al., 2006). Although the exact molecular mechanisms of K+ sensing in plants are not known, the lack of crosstalk between the regulation of K+ uptake and the nitrate responses is puzzling, as both processes require CIPK23dependent phosphorylation. A possible explanation may rely on the existence of different CIPK23-protein complexes in the plant cell, with low nitrate only activating the CHL1-containing CIPK23
Cell 138, September 18, 2009 ©2009 Elsevier Inc. 1065
complex. Alternatively, the locations of nitrate and K+ sensing may be very different in the root. Indeed, AKT1 is mostly expressed in peripheral cell layers of the mature root, similar to the pattern of CIPK23 expression (Lagarde et al., 1996; Xu et al., 2006), whereas CHL1 is mostly expressed in nascent root organs such as root tips, emerging lateral roots, and the central cylinder where vascular tissues are located (Figure 1B) (Guo et al., 2001; Remans et al., 2006). Therefore, not only is it crucial to examine the colocalization of CIPK23 and CHL1 expression patterns in the root to validate the direct molecular interaction between the two proteins, it is of key importance to determine where nitrate sensing occurs in the plant root. Is this sensing mechanism restricted to a few cells at the root tip or is it a general feature of all plant cells that express CHL1? Similarly, it is still not clear how
CHL1-dependent nitrate signaling modulates NRT2.1 gene expression in mature epidermal and cortical root cells where CHL1 is not found. The identity of the mobile signals involved in nitrate sensing is currently unknown, leaving a major gap in our understanding of the signaling pathways that drive nitrate uptake in plants. Nonetheless, the new findings of Ho et al. provide much-needed insight into the regulation of nutrient sensing and signaling in plants. It will be of great interest to explore in future work whether CIPKs control other responses to nitrate such as the plasticity of root development that leads to the proliferation of lateral roots.
Guo, F.Q., Wang, R., Chen, M., and Crawford, N.M. (2001). Plant Cell 13, 1761–1777.
References
Tsay, Y.F., Chiu, C.C., Tsai, C.B., Ho, C.H., and Hsu, P.K. (2007). FEBS Lett. 581, 2290–2300.
Batistic, O., and Kudla, J. (2009). Biochim. Biophys. Acta 1793, 985–992.
Xu, J., Li, H.D., Chen, L.Q., Wang, Y., Liu, L.L., He, L., and Wu, W.H. (2006). Cell 125, 1347–1360.
1066 Cell 138, September 18, 2009 ©2009 Elsevier Inc.
Ho, C.-H., Lin, S.-H., Hu, H.-C., and Tsay, Y.-F. (2009). Cell, this issue. Hu, H.C., Wang, Y.Y., and Tsay, Y.F. (2009). Plant J. 57, 264–278. Lagarde, D., Basset, M., Lepetit, M., Conejero, G., Gaymard, F., Astruc, S., and Grignon, C. (1996). Plant J. 9, 195–203. Liu, K.H., and Tsay, Y.F. (2003). EMBO J. 22, 1005– 1013. Munos, S., Cazettes, C., Fizames, C., Gaymard, F., Tillard, P., Lepetit, M., Lejay, L., and Gojon, A. (2004). Plant Cell 16, 2433–2447. Remans, T., Nacry, P., Pervent, M., Filleur, S., Diatloff, E., Mounier, E., Tillard, P., Forde, B.G., and Gojon, A. (2006). Proc. Natl. Acad. Sci. USA 103, 19206–19211.
tribute to neuronal differentiation or the Prdx1/GDE2 pathway works in parallel with other regulatory systems that balance neural progenitor proliferation and differentiation. Further insights into the function of the Prdx1/GDE2 pathway will require the identification of the signals produced by GDE2, the means by which these signals are perceived, and determining how this pathway interfaces with core neurogenic factors.
Farkas, L.M., and Huttner, W.B. (2008). Curr. Opin. Cell Biol. 20, 707–715.
References
Skwarek, L.C., Garroni, M.K., Commisso, C., and Boulianne, G.L. (2007). Dev. Cell 13, 783–795.
Barber, S.C., Mead, R.J., and Shaw, P.J. (2006). Biochim. Biophys. Acta 1762, 1051–1067.
Ulloa, F., and Briscoe, J. (2007). Cell Cycle 6, 2640–2649.
Briscoe, J., and Novitch, B.G. (2008).Philos. Trans. R. Soc. Lond. B Biol. Sci. 363, 57–70.
Yan, Y., Sabharwal, P., Rao, M., and Sockanathan, S. (2009). Cell, this issue.
Hall, A., Karplus, P.A., and Poole, L.B. (2009). FEBS J. 276, 2469–2477. Holmberg, J., Hansson, E., Malewicz, M., Sandberg, M., Perlmann, T., Lendahl, U., and Muhr, J. (2008). Development 135, 1843–1851. Jang, H.H., Lee, K.O., Chi, Y.H., Jung, B.G., Park, S.K., Park, J.H., Lee, J.R., Lee, S.S., Moon, J.C., Yun, J.W., et al. (2004). Cell 117, 625–635. Rao, M., and Sockanathan, S. (2005). Science 309, 2212–2215.
A Toggle Switch in Plant Nitrate Uptake Grégory Vert1,* and Joanne Chory2
BPMP, CNRS UMR 5004, 34060 Montpellier Cedex 1, France Plant Biology Laboratory and Howard Hughes Medical Institute, The Salk Institute for Biological Studies, La Jolla, CA 92037, USA *Correspondence: [email protected] DOI 10.1016/j.cell.2009.09.005
1 2
In plants, the uptake of nitrate from the soil is a critical process controlled by complex regulatory networks that target nitrate transporters in the roots. In this issue, Ho et al. (2009) show that phosphorylation of the CHL1 nitrate transporter allows the plant root to sense and respond to different nitrate concentrations in the soil. As sessile organisms that are bound to one location, plants have developed sophisticated mechanisms to ensure appropriate adaptation to constantly changing environmental conditions. To cope with unfavorable nutrient availability in the soil, plants harbor on the root surface a battery of specialized transporters to maintain efficient uptake of nutrients. These transporters are controlled by the integration of complex regulatory networks that underlie external and internal cues to modulate nutrient uptake capacity in accordance with the nutrient demand of the plant and the nutrient availability of the soil. Reverse genetics in the model plant Arabidopsis thaliana has provided a wealth of information on the roles of the different transporters. How-
ever, the molecular mechanisms regulating the acquisition of nutrients have remained largely unknown. Given the vital role nitrate plays as a nitrogen source for most plants, the study of nitrate uptake serves as a test case to characterize plant nutrient transport systems and to unravel the signaling pathways governing nutrient acquisition. Nitrate is taken up from the soil by several transporters in the plant root—including the high-affinity NRT2.1 transporter and the high- and low-affinity (dual-affinity) CHL1 transporter (also called NRT1.1) —and directly induces the expression of these transporters as part of the primary nitrate response (Tsay et al., 2007). In this issue of Cell, Ho et al. (2009) shed light on the molecular mechanisms of nitrate signaling by CHL1 in
1064 Cell 138, September 18, 2009 ©2009 Elsevier Inc.
Arabidopsis, highlighting the complexity of nutrient transport, sensing, and signaling in plants. The first hints of a potential role for CHL1 in nitrate signaling came from studies in chl1 loss-of-function mutant Arabidopsis plants, which suggested that CHL1 regulates the expression of NRT2.1 in response to nitrate. These plants lacking CHL1 indeed failed to downregulate the expression of NRT2.1 in the presence of high concentrations of nitrogen (Munos et al., 2004) and were unable to increase the proliferation of lateral roots in nitraterich zones in the soil (Remans et al., 2006), another well-established response to nitrate. In their new work, Ho and coworkers now unambiguously unravel the role of CHL1 in nitrate signaling by
characterizing the chl1-9 mutant allele. They show that the chl1-9 allele encodes a mutant CHL1 protein that is defective in nitrate transport but is still able to trigger responses to nitrate, as visualized by the nitrate-dependent and CHL1-mediated induction of NRT2.1 expression. Genetic and biochemical approaches show that low levels of nitrate are detected by CHL1 and signaled through calcineurin B-like interacting protein kinase CIPK23-dependent phosphorylation of threonine 101 (T101) in CHL1. Phosphorylation of this residue has previously been implicated in the switch from the low- to high-affinity modes of nitrate transport (Liu and Tsay, 2003) and now appears to also mediate the downregulation of primary nitrate responses (Figure 1A). In contrast, a high nitrate concentration converts CHL1 to its unphosphorylated form, leading to the full nitrate response and CHL1-mediated low-affinity nitrate transport (Figure 1A). Whether CHL1 acts as a bona fide nitrate sensor or is simply part of the nitrate signaling pathway remains to be elucidated. Notably, a better understanding of the mechanisms leading to the CIPK23-mediated phosphorylation of CHL1 is required to fully grasp the possible role of CHL1 in direct sensing of external nitrate. CIPKs are activated by calcium ions via the calcineurin B-like (CBL) calcium-binding proteins (Batistic and Kudla, 2009). Although CBL9 is required for CIPK23mediated phosphorylation of CHL1 in frog oocytes and in vitro, evidence of a calcium signal that is triggered by nitrate remains elusive. If CHL1 indeed acts as a nitrate sensor, an early CHL1dependent calcium signal is expected to participate in the establishment of responses to nitrate. The rationale for the strong and rapid induction of CIPK23 expression observed by Ho and coworkers in plants exposed to nitrate is also unclear. This response to nitrate suggests either that nitrate sensing takes place far upstream of CIPK23mediated CHL1 phosphorylation, or that CIPK23 expression is feedback regulated by a CHL1-dependent nitrate signaling pathway. Testing whether the nitrate-dependent early induction of CIPK23 expression requires functional CHL1 should help to position CHL1 in the early steps of nitrate signaling.
Figure 1. Nitrate Signaling and Transport in Plant Roots (A) When plants are exposed to conditions of low external nitrate (NO3−; <1 mM), the dual-affinity nitrate transporter CHL1 recruits or activates the calcineurin B-like CBL9-CBL-interacting protein kinase CIPK23 complex through an unknown mechanism, leading to the phosphorylation of CHL1 on threonine 101 (T101). Once phosphorylated, CHL1 switches to a high-affinity mode of nitrate transport to ensure efficient nitrate uptake under nitrate-poor conditions. CHL1 phosphorylated at T101 also restricts primary nitrate responses to the high-affinity phase of nitrate signaling. When high external nitrate concentrations (>1 mM) are sensed by CHL1, the CBL9-CIPK23-dependent phosphorylation of CHL1 at T101 does not occur. CHL1 not phosphorylated at T101 remains in the low-affinity nitrate transport mode and activates increased primary nitrate responses. Establishment of the primary nitrate responses in the low-affinity phase also requires CIPK8. CIPK8 is activated by an unknown CBL and likely phosphorylates CHL1 at a different residue from T101. (B) In the plant root, CHL1 (red) senses and signals the external nitrate concentration to control the mode of nitrate transport (low- or high-affinity) and to regulate the expression of another nitrate transporter, NRT2.1 (blue), in different root cell layers through an unknown signal. NRT2.1 gene expression is under dual regulation—by the CHL1-dependent external nitrate signaling pathway and by negative feedback regulation driven by the levels of nitrogen-containing metabolites of the shoot—to tightly regulate nitrate uptake according to plant demand and nitrate availability in the soil.
When plants grown in the absence of nitrate are challenged with different concentrations of external nitrate, they rapidly initiate a biphasic primary nitrate response (Hu et al., 2009). This biphasic nitrate response could be controlled by two distinct nitrate-sensing mechanisms that drive either a high-affinity (nitrate concentration <1 mM) or a low-affinity (nitrate concentration >1 mM) nitrate signaling pathway. CIPK8 is a positive regulator of the primary nitrate response in the low-affinity phase of nitrate signaling (Hu et al., 2009). The work of Ho et al. now adds to the complexity of the regulation of nitrate responses with their identification of CIPK23 as a negative regulator of the high-affinity phase. Thus, nitrate
responses in plants appear to be regulated by the interplay of different CIPKs, which drive different phases of nitrate signaling and likely target different CHL1 residues for phosphorylation (Figure 1A). Interestingly, CIPK23 is known to regulate K+ uptake by the AKT1 K+ channel in Arabidopsis (Xu et al., 2006). Although the exact molecular mechanisms of K+ sensing in plants are not known, the lack of crosstalk between the regulation of K+ uptake and the nitrate responses is puzzling, as both processes require CIPK23dependent phosphorylation. A possible explanation may rely on the existence of different CIPK23-protein complexes in the plant cell, with low nitrate only activating the CHL1-containing CIPK23
Cell 138, September 18, 2009 ©2009 Elsevier Inc. 1065
complex. Alternatively, the locations of nitrate and K+ sensing may be very different in the root. Indeed, AKT1 is mostly expressed in peripheral cell layers of the mature root, similar to the pattern of CIPK23 expression (Lagarde et al., 1996; Xu et al., 2006), whereas CHL1 is mostly expressed in nascent root organs such as root tips, emerging lateral roots, and the central cylinder where vascular tissues are located (Figure 1B) (Guo et al., 2001; Remans et al., 2006). Therefore, not only is it crucial to examine the colocalization of CIPK23 and CHL1 expression patterns in the root to validate the direct molecular interaction between the two proteins, it is of key importance to determine where nitrate sensing occurs in the plant root. Is this sensing mechanism restricted to a few cells at the root tip or is it a general feature of all plant cells that express CHL1? Similarly, it is still not clear how
CHL1-dependent nitrate signaling modulates NRT2.1 gene expression in mature epidermal and cortical root cells where CHL1 is not found. The identity of the mobile signals involved in nitrate sensing is currently unknown, leaving a major gap in our understanding of the signaling pathways that drive nitrate uptake in plants. Nonetheless, the new findings of Ho et al. provide much-needed insight into the regulation of nutrient sensing and signaling in plants. It will be of great interest to explore in future work whether CIPKs control other responses to nitrate such as the plasticity of root development that leads to the proliferation of lateral roots.
Guo, F.Q., Wang, R., Chen, M., and Crawford, N.M. (2001). Plant Cell 13, 1761–1777.
References
Tsay, Y.F., Chiu, C.C., Tsai, C.B., Ho, C.H., and Hsu, P.K. (2007). FEBS Lett. 581, 2290–2300.
Batistic, O., and Kudla, J. (2009). Biochim. Biophys. Acta 1793, 985–992.
Xu, J., Li, H.D., Chen, L.Q., Wang, Y., Liu, L.L., He, L., and Wu, W.H. (2006). Cell 125, 1347–1360.
1066 Cell 138, September 18, 2009 ©2009 Elsevier Inc.
Ho, C.-H., Lin, S.-H., Hu, H.-C., and Tsay, Y.-F. (2009). Cell, this issue. Hu, H.C., Wang, Y.Y., and Tsay, Y.F. (2009). Plant J. 57, 264–278. Lagarde, D., Basset, M., Lepetit, M., Conejero, G., Gaymard, F., Astruc, S., and Grignon, C. (1996). Plant J. 9, 195–203. Liu, K.H., and Tsay, Y.F. (2003). EMBO J. 22, 1005– 1013. Munos, S., Cazettes, C., Fizames, C., Gaymard, F., Tillard, P., Lepetit, M., Lejay, L., and Gojon, A. (2004). Plant Cell 16, 2433–2447. Remans, T., Nacry, P., Pervent, M., Filleur, S., Diatloff, E., Mounier, E., Tillard, P., Forde, B.G., and Gojon, A. (2006). Proc. Natl. Acad. Sci. USA 103, 19206–19211.