© 2006 Nature Publishing Group
Degradation of Id2 by the anaphase-promoting
complex couples cell cycle exit and axonal growth
Anna Lasorella1,2,3, Judith Stegmu ¨ller5, Daniele Guardavaccaro6, Guangchao Liu1, Maria S. Carro1,
Gerson Rothschild1, Luis de la Torre-Ubieta5, Michele Pagano6, Azad Bonni5& Antonio Iavarone1,2,4
In the developing nervous system, Id2 (inhibitor of DNA binding
2, also known as inhibitor of differentiation 2) enhances cell
proliferation, promotes tumour progression and inhibits the
activity of neurogenic basic helix–loop–helix (bHLH) transcrip-
tion factors1,2. The anaphase promoting complex/cyclosome and
its activator Cdh1 (APC/CCdh1) restrains axonal growth but the
targets of APC/CCdh1in neurons are unknown3–5. Id2 and other
members of the Id family are very unstable proteins that are
eliminated as cells enter the quiescent state, but how they are
targeted for degradation has remained elusive6,7. Here we show
that Id2 interacts with the core subunits of APC/C and Cdh1 in
primary neurons. APC/CCdh1targets Id2 for degradation through
a destruction box motif (D box) that is conserved in Id1 and Id4.
Depletion of Cdh1 stabilizes Id proteins in neurons, whereas Id2
D-box mutants are impaired for Cdh1 binding and remain stable
in cells that exit from the cell cycle and contain active APC/CCdh1.
Mutants of the Id2 D box enhance axonal growth in cerebellar
granule neurons in vitro and in the context of the cerebellar
cortex, and overcome the myelin inhibitory signals for growth.
Conversely, activation of bHLH transcription factors induces a
cluster of genes with potent axonal inhibitory functions including
the gene coding for the Nogo receptor, a key transducer of myelin
inhibition. Degradation of Id2 in neurons permits the accumu-
lation of the Nogo receptor, thereby linking APC/CCdh1activity
findings indicate that deregulated Id activity might be useful to
reprogramme quiescent neurons into the axonal growth mode.
To identify protein complexes of Id2 in human neuroblastoma
cells, we used immunoaffinity chromatography followed by tandem
mass spectrometry. New Id2 partners are the APC/C subunits Apc1,
Apc5 and Apc8/Cdc23 (Supplementary Fig. 1a). The identification
in vivo of complexes of Id2 and subunits of APC/C indicates that Id2
might be targeted for degradation by APC/C. The enzymatic E3
ubiquitin ligase activity of APC/C requires either the Cdc20 or Cdh1
co-activator4. Expression of Cdh1 but not that of Cdc20 led to a
inhibition (Fig. 1a). Moreover, Id2 was eliminated at a faster rate in
the presence of Cdh1 (Fig. 1b). Expression of Cdh1 decreased
endogenous Id2 in asynchronously growing U2OS cells without
changing the cell cycle distribution, and also in cells synchronized
in the S phase of the cell cycle with aphidicolin (Fig. 1c, d, and
Supplementary Fig. 1b, c). Activation of APC/CCdh1also eliminated
terminal 15 amino-acid residues, which have been shown to con-
tribute to Id2 ubiquitination7(Supplementary Fig. 1d, e). Id2
associated with Cdh1 in the absence of proteasomal inhibition, but
the interaction was more efficient in cells treated with MG-132
(Supplementary Fig. 1f, g). To examine whether Id2 becomes
in response to serum deprivation. Id2 was downregulated and
underwent accelerated degradation in NIH3T3 fibroblasts deprived
of serum mitogens (Supplementary Fig. 2a, b). We observed similar
results in serum-starved U2OS-Id2 and SK-N-SH neuroblastoma
cells undergoing arrest in G0/G1 after treatment with retinoic acid
(Supplementary Fig. 3a–c)8. To examine whether APC/CCdh1regu-
APC/C activity through the depletion of Cdh1. Knockdown of Cdh1
by RNA interference led to elevation of the steady-state state levelsof
Figure 1 | Id2 is a substrate of APC/CCdh1. a, HeLa cells were transfected
with Id2, haemagglutinin (HA)-conjugated Cdh1 or HA–Cdc20 and treated
with N-Acetyl-L-leucyl-L-leucyl-L-norleucinal (LLnL). b, Transfected HeLa
cells were treated with cycloheximide. c, Endogenous Id2 in U2OS cells
transfected with HA–Cdh1. d, U2OS-tet-Cdh1 cells were treated with
aphidicoline before the removal of tetracycline. e, f, LAN-1 neuroblastoma
cells (e) and U2OS-Id2 cells (f) transfected with siRNA oligonucleotides.
g, LAN-1 cells were transfected with siRNA; quiescence was induced by
retinoic acid and cells were treated with cycloheximide (CHX).
1Institute for Cancer Genetics, and Departments of2Pathology,3Pediatrics and4Neurology, College of Physicians and Surgeons of Columbia University, New York, New York
10032, USA.5Department of Pathology, Harvard Medical School, 77 Avenue Louis Pasteur, Boston, Massachusetts 02115, USA.6Department of Pathology, NYU Cancer Institute,
New York University School of Medicine, 550 First Avenue MSB 599, New York, New York 10016, USA.
Vol 442|27 July 2006|doi:10.1038/nature04895
© 2006 Nature Publishing Group
Id2 in U2OS cells (Fig. 1e, and Supplementary Fig. 3d). It also
prevented the downregulation of Id2 in quiescent U2OS cells and led
to a tenfold increase in the Id2 half-life in LAN1 cells undergoing cell
cycle arrest after treatment with retinoic acid (Fig. 1f, g, and
Supplementary Fig. 3e, f). However, Id3 was unchanged in Cdh1-
depleted cells (Fig. 1e). In support of a role of APC/CCdh1in Id2
stability, expression of Emi1, a specific inhibitor of APC/C (ref. 9),
led to the accumulation of Id2 (Supplementary Fig. 3g).
atresidues 100–107,indicating that Id2 mightbeadirectsubstrateof
APC/CCdh1. The D-box motif is conserved in Id1 and Id4 but not in
Id3 (Fig. 2a). We generated mutant Id2 proteins in which the key
arginine and leucine residues were changed into glycine and valine,
respectively (Id2-DBM) and the D box was deleted (Id2dDB). Id2-
DBM and Id2dDB mutants were expressed at levels higher than
wild-type Id2 and were resistant to Cdh1-mediated destabilization
(Fig. 2b, and Supplementary Fig. 4a). Conversely, APC/CCdh1elimi-
nated the Id2 mutant that lacks the HLH region (Id2dHLH) and also
wild-type Id1 and Id4 but not Id3 (Supplementary Fig. 4b–e). Next
we investigated whether the D box is essential for the recognition of
Id2 by APC/CCdh1. Glutathione S-transferase (GST)–Id2 but not
GST–Id2-DBM or GST–Id2dDB captured in vitro translated Cdh1
and Cdh1 from HeLa cell extract, whereas all GST–Id2 fusion
proteins captured Apc1 and Cdc27 (Fig. 2c, and Supplementary
Fig. 5a). Similarly, both Flag–Id2and Flag–Id2-DBM associated with
APC/C core subunits, but Flag–Id2-DBM precipitated much less
Figure 2 | D-box-dependent degradation of Id proteins by APC/CCdh1.
a, Alignment of D boxes in Id proteins. b, HeLa cells were transfected with
Id2 or Id2-DBM and HA–Cdh1 and analysed by western blotting. c, Pull-
down assay with GST fusion proteins and Cdh1 and E47 translated in vitro.
d, HeLa cells were transfected with Id2 or Id2-DBM and treated with
cycloheximide (CHX). e, Quantification of Id2 (circles) and Id2-DBM
(squares) from d. f, In vitro ubiquitination assay of Id2 and Id2-DBM by
immunopurified APC/C in the presence or absence of Cdh1 translated
in vitro. Arrowheads, unmodified Id2/Id2-DBM.
Figure 3 | An APC/CCdh1–Id–bHLH pathway controls axonal growth.
a, b, Immunoprecipitation of Id2 (a) and Cdc27 (b) from CGNs. c, CGNs
transfected with siRNA oligonucleotides and analysed by western blotting.
Notch1 is the Val1744-cleaved intracellular fragment. d, GFP (green) and
Tau (red) staining in CGNs transfected withvector and Id2-DBM. e, Axonal
length measured in CGNs transfected with vector or Id2dDB together with
GFP (n ¼ 119; asterisk, P # 0.0001). Results are means ^ s.e.m. f, CGNs
transfected with E47 and U6-CTR or U6-shCdh1 together with GFP.
g, Quantification of axonal length from f (n ¼ 300; asterisk, P # 0.001).
Results are means ^ s.e.m. h, Axonal length measured from CGNs
transfected with Id2-DBM in the absence or presence of E47 (n ¼ 76;
asterisk, P # 0.01). Results are means ^ s.e.m. i, Expression of E47 target
genes in SK-N-SH neuroblastoma cells infected with adeno-vector,
adeno-E47 and adeno-E47 plus adeno-Id2. NogoR, Nogo receptor. Results
are means ^ s.d. (n ¼ 3). j, RhoA activation assay in adenovirus-infected
SK-N-SH cells treated with Nogo-A peptide. k, SF210-E47-ER cells treated
with the indicated concentrations of 4-OHT.
NATURE|Vol 442|27 July 2006
© 2006 Nature Publishing Group
Cdh1 than Flag–Id2 (Supplementary Fig. 5b). To determine the
the rate of degradation of Id2 and Id2-DBM. In transiently trans-
fected HeLa cells the half-life of Id2-DBM was extended more than
tenfold compared with wild-type Id2 (Fig. 2d, e). In addition,
treatment of U2OS cells with the antimitogenic cytokine transform-
ing growth factor-b efficiently depleted Id2 but did not decrease Id2-
DBM (Supplementary Fig. 5c). Accordingly, wild-type Id2 but not
Id2-DBM was polyubiquitinated in vitro by APC/C in the presence,
but not in the absence, of Cdh1 (Fig. 2f, and Supplementary Fig. 5d).
Together, these data indicate that APC/CCdh1can polyubiquitinate
Id2in a D-box-dependent manner. In contrast, Id2 did not affect the
degrade other substrates (Supplementary Fig. 6).
Apart from the regulation of the cell cycle, the expression of
APC/CCdh1in postmitotic neurons has been puzzling11. Previous
studies indicated that APC/CCdh1directs cell-cycle-independent
functions in postmitotic neurons, including the negative control of
axonal growth. At embryonic day 16 (E16), Id2 was physically
associated with brain APC/C (Supplementary Fig. 7a). We therefore
differentiated neurons, and whether APC/CCdh1directs the degra-
dation of Id2 to control axonal growth. We found that lysates from
primary cerebellar granule neurons (CGNs) and differentiated neuro-
blastoma cells contained Id2–core APC/C and Id2–Cdh1 complexes
(Fig. 3a, b, and Supplementary Fig. 7d, e). Furthermore, postmitotic
neurons transfected with Cdh1 short interfering RNA (siRNA)
accumulated Id2, Id1 and Id4 in the absence of cell cycle re-entry
as indicated by the lack of bromodeoxyuridine incorporation,
retinoblastoma gene product phosphorylation or decrease in the
cyclin-dependent kinase inhibitor p27 (Fig. 3c, and Supplementary
Fig. 7f, g)3. Next we determined whether expression of the Id2
mutants that are resistant to APC/CCdh1-mediated degradation
stimulated axonal elongation. CGN axons are identified on the
basis of morphology, positive staining for Tau, and negative staining
for MAP-2 (refs 3, 12). In the presence of green fluorescent protein
(GFP) expression to mark transfected cells, expression of Id2-DBM
and Id2dDB but not wild-type Id2 significantly increased axonal
length in CGNs, cortical neurons and differentiated SK-N-SH cells
(Fig. 3d, e, Supplementary Figs 7h and 8a, b, and data not shown).
with Id2-DBM and cultured on top of organotypic cerebellar slices
from P9 rat pups3(Supplementary Fig. 8c, d).
Having shown that, resembling Cdh1 silencing, the expression of
undegradable Id2 leads to unrestrained axonal growth, we investi-
gated whether restoring bHLH activity to neurons depleted of Cdh1
or expressing degradation-resistant Id2 rescues the deregulated axon
growth. Ectopic expression of the bHLH transcription factor E47
abolished the stimulation ofaxonal growth byCdh1 knockdown and
Id2-DBM (Fig. 3f–h). These results indicate that downstream targets
of bHLH transcription factors mediate the axonal inhibitory func-
tions and activation of these targets requires the controlof Idprotein
accumulation by APC/CCdh1. To identify the relevant targets
recruited by the Cdh1–Id–bHLH pathway for the control of axono-
genesis, we infected SK-N-SH cells with an adenovirus expressing
E47 (Ad-E47) and determined the gene expression profile. E47
induced the expression of genes coding for inhibitors of axonal
growth including secreted molecules (Sema3F), ligands (Jagged-2)
and receptors (Nogo receptor, Unc5A, Notch1) of multiple inhibi-
tory and repellant signals for axons13–15(Supplementary Fig. 9a).
Prominent among them, the Nogo receptor is one of the best-known
inhibitors of axonal growth on which all myelin-inhibitory signals
(Nogo-A, myelin associated glycoprotein (MAG) and oligodendro-
results with quantitative real time polymerase chain reaction from
cells transduced with Ad-E47 and found that coinfection with Id2
adenovirus (Ad-Id2) prevented the E47-mediated induction of anti-
axonal genes (Fig. 3i, and Supplementary Fig. 9b). The main
intracellular transducer of axonal inhibitory signals driven by Nogo
receptor is the small GTP-binding protein RhoA17. Ad-Id2 abolished
the activation of RhoA by Nogo-A inhibitory peptide (Fig. 3j). To
non-neuronal cells, we generated SF210 glioma cells expressing
conditionally active E47 fused to the ligand-binding domain of
oestrogen receptor (E47-ER). Activation of bHLH transcription by
4-hydroxytamoxifen (4-OHT) increased Nogo receptor protein and
theVal1744-cleavedactivefragment of Notch1 (Fig. 3k).Conversely,
knockdown of Cdh1 decreased E47-mediated transcriptional acti-
vation of an E-box-luciferase plasmid and inhibited the basal and
of the Cdh1–Id–bHLH pathway for axonal growth.
To evaluate the significance of deregulated Id2 in neurons whose
axonal growth had been arrested by the inhibitory myelin com-
ponents that require Nogo receptor for signalling, we transfected
CGNs plated on MAG-Fc inhibitory substrate with Id2-DBM or
the myelin inhibitor. Conversely, MAG-Fc markedly inhibited axonal
growth in neurons transfected with the empty plasmid (Fig. 4a, b).
are actively growing axons and on reaching synaptogenesis when
growth is terminated. Cortical neurons undergo axonal elongation
in vitro that culminates with extensive synaptogenesis at DIV (days
Figure 4 | Significance of APC/CCdh1-mediated degradation of Id2 for
axonal growth. a, Axonal outgrowth of Id2-DBM and vector-transfected
CGNs in the presence and absence of MAG-Fc. b, Quantification of axonal
length from a. Open bars, no inhibitor; filled bars, MAG-Fc. Results are
means ^ s.e.m. (n ¼ 150; asterisk, P # 0.0001). c, Co-staining for
presynaptic synapsin-I (green) and postsynaptic PSD95 (red) in DIV 2 and
DIV 12 cortical neurons. d, Id2 expression in cortical neurons undergoing
synaptogenesis. NogoR, Nogo receptor. e, DIV 2 and DIV 12 cortical
neurons treated with cycloheximide (CHX) for the indicated durations. f,
Id2 (red) and nuclei (4,6-diamidino-2-phenylindole (DAPI), blue) staining
NATURE|Vol 442|27 July 2006
© 2006 Nature Publishing Group
in vitro) 12 as shown by the accumulation of synaptophysin and
maturation of the postsynaptic marker PSD-95 into large clusters
juxtaposed to presynaptic synapsin-I puncta (Fig. 4c, d)18. The
steady-state levels of Id2 decreased progressively after DIV 2 and
became almost undetectable at DIV 12, when the Nogo receptor
steady state between DIV 2 and DIV 12 coincided with an increased
rate of Id2 protein degradation (Fig. 4e). APC/CCdh1inhibition by
silencing of Cdh1 led to an increase in Id1 and Id2 levels also in
cortical neurons (Supplementary Fig. 10). APC/CCdh1activity is
localized in the nucleus19. Consistent with this notion is the obser-
neurons led to the accumulation of Id2 in the nucleus at levels
comparable to those detected at DIV 2 (Fig. 4f).
We show that APC/CCdh1primes Id2 for degradation through a D
box as a recognition site for the Cdh1 coactivator. The activity of
Id proteins, which are invariably depleted in cells undergoing
quiescence, corresponds to the consequences of inactivation of
APC/CCdh1in postmitotic cells. Indeed, in a manner consistent
with reactivation of the cell cycle in cells carrying inactive APC/C
(ref. 20), deregulated expression of Id2 prevents cell cycle arrest by a
wide range of antiproliferative signals21–23. Under certain experimen-
tal conditions, ectopic Id2 is able to override the quiescent state and
drive terminally differentiated cells back into the cell cycle24. The
observation that the most aggressive tumours frequently contain the
largest amounts of Id proteins raises the possibility that inactivation
of the control of Id protein stability by APC/CCdh1might also
contribute to Id accumulation in cancer.
By setting the timing of Id protein degradation in the nervous
activation of downstream targets of bHLH transcription factors. The
dual activity of the brain APC/CCdh1as an inducerof the postmitotic
process that restrains Id proteins during distinct stages of neural
development and, as a consequence, sets the timing of bHLH-
dependent transcription. The general implication of our findings is
that neurons in active axonal growth should be viewed as relatively
‘undifferentiated’ compared with neurons that have reached their
targets and are unable to resume neurite outgrowth. We suggest that,
by disabling the APC/CCdh1–Id–bHLH axis, degradation-resistant
forms of Id proteins might provide beneficial effects for axonal
regeneration of damaged neurons.
Transfections and siRNA. Cell transfection was performed with Lipofectamine
2000 (Invitrogen). CGNs, cortical neurons and SK-N-SH cells differentiated
times with 100nM siRNA oligonucleotides and the Gene Silencer siRNA
Transfection Reagent (Gene Therapy Systems)26. siRNAs were Cdh1 Smart Pool
from Dharmacon (5
RhoA assay. SK-N-SH cells were infected with Ad-vector or Ad-Id2 at a
multiplicity of infection of 50. After 24h, cells were stimulated by the addition
was precipitated by using beads with GST–Rho-binding domain (RBD) of
rhotekin, and detected with anti-RhoA antibodies in accordance with the
manufacturer’s guidelines (Cytoskeleton). Additional methods are described
in Supplementary Information.
0) and U6shCdh1 (5
Received 27 March; accepted 12 May 2006.
Published online 28 June 2006.
1.Iavarone, A. & Lasorella, A. Id proteins in neural cancer. Cancer Lett. 204,
189– -196 (2004).
Perk, J., Iavarone, A. & Benezra, R. Id family of helix– -loop– -helix proteins in
cancer. Nature Rev. Cancer 5, 603– -614 (2005).
Konishi, Y., Stegmuller, J., Matsuda, T., Bonni, S. & Bonni, A. Cdh1-APC controls
axonal growth and patterning in the mammalian brain. Science 303, 1026– -1030
Peters, J. M. The anaphase-promoting complex: proteolysis in mitosis and
beyond. Mol. Cell 9, 931– -943 (2002).
Stegmuller, J. & Bonni, A. Moving past proliferation: new roles for Cdh1-APC in
postmitotic neurons. Trends Neurosci. 28, 596– -601 (2005).
Bounpheng, M. A., Dimas, J. J., Dodds, S. G. & Christy, B. A. Degradation of Id
proteins by the ubiquitin– -proteasome pathway. FASEB J. 13, 2257– -2264 (1999).
Fajerman, I., Schwartz, A. L. & Ciechanover, A. Degradation of the Id2
developmental regulator: targeting via N-terminal ubiquitination. Biochem.
Biophys. Res. Commun. 314, 505– -512 (2004).
Wainwright, L. J., Lasorella, A. & Iavarone, A. Distinct mechanisms of cell cycle
arrest control the decision between differentiation and senescence in human
neuroblastoma cells. Proc. Natl Acad. Sci. USA 98, 9396– -9400 (2001).
Reimann, J. D. et al. Emi1 is a mitotic regulator that interacts with Cdc20 and
inhibits the anaphase promoting complex. Cell 105, 645– -655 (2001).
10. Harper, J. W., Burton, J. L. & Solomon, M. J. The anaphase-promoting complex:
it’s not just for mitosis any more. Genes Dev. 16, 2179– -2206 (2002).
11.Gieffers, C., Peters, B. H., Kramer, E. R., Dotti, C. G. & Peters, J. M. Expression
of the CDH1-associated form of the anaphase-promoting complex in
postmitotic neurons. Proc. Natl Acad. Sci. USA 96, 11317– -11322 (1999).
12. Powell, S. K., Rivas, R. J., Rodriguez-Boulan, E. & Hatten, M. E. Development of
polarity in cerebellar granule neurons. J. Neurobiol. 32, 223– -236 (1997).
13. Barallobre, M. J., Pascual, M., Del Rio, J. A. & Soriano, E. The Netrin family of
guidance factors: emphasis on Netrin-1 signalling. Brain Res. Brain Res. Rev. 49,
22– -47 (2005).
14. Fiore, R. & Puschel, A. W. The function of semaphorins during nervous system
development. Front. Biosci. 8, s484– -s499 (2003).
15. Sestan, N., Artavanis-Tsakonas, S. & Rakic, P. Contact-dependent inhibition of
cortical neurite growth mediated by notch signaling. Science 286, 741– -746
16. Schwab, M. E. Nogo and axon regeneration. Curr. Opin. Neurobiol. 14, 118– -124
17. Spencer, T., Domeniconi, M., Cao, Z. & Filbin, M. T. New roles for old proteins
in adult CNS axonal regeneration. Curr. Opin. Neurobiol. 13, 133– -139 (2003).
18. Lesuisse, C. & Martin, L. J. Long-term culture of mouse cortical neurons as a
model for neuronal development, aging, and death. J. Neurobiol. 51, 9– -23
19. Zhou, Y., Ching, Y. P., Chun, A. C. & Jin, D. Y. Nuclear localization of the cell
cycle regulator CDH1 and its regulation by phosphorylation. J. Biol. Chem. 278,
12530– -12536 (2003).
20. Wirth, K. G. et al. Loss of the anaphase-promoting complex in quiescent cells
causes unscheduled hepatocyte proliferation. Genes Dev. 18, 88– -98 (2004).
21. Baghdoyan, S. et al. Id2 reverses cell cycle arrest induced by g-irradiation in
human HaCaT keratinocytes. J. Biol. Chem. 280, 15836– -15841 (2005).
22. Kowanetz, M., Valcourt, U., Bergstrom, R., Heldin, C. H. & Moustakas, A. Id2
and Id3 define the potency of cell proliferation and differentiation responses to
transforming growth factor b and bone morphogenetic protein. Mol. Cell. Biol.
24, 4241– -4254 (2004).
23. Lasorella, A. et al. Id2 is critical for cellular proliferation and is the oncogenic
effector of N-myc in human neuroblastoma. Cancer Res. 62, 301– -306 (2002).
24. Chaudhary, J., Sadler-Riggleman, I., Ague, J. M. & Skinner, M. K. The helix– -
loop– -helix inhibitor of differentiation (ID) proteins induce post-mitotic
terminally differentiated sertoli cells to re-enter the cell cycle and proliferate.
Biol. Reprod. 72, 1205– -1217 (2005).
25. Encinas, M. et al. Sequential treatment of SH-SY5Y cells with retinoic acid and
brain-derived neurotrophic factor gives rise to fully differentiated, neurotrophic
factor-dependent, human neuron-like cells. J. Neurochem. 75, 991– -1003
26. Aarts, M. et al. A key role for TRPM7 channels in anoxic neuronal death. Cell
115, 863– -877 (2003).
Supplementary Information is linked to the online version of the paper at
Acknowledgements We thank S. Gygi and N. Sherman for the identification of
Id2-associated proteins by mass spectrometry; P. K. Jackson for providing us
with purified MBP-Emi1; and W. G. Kaelin and J. Lukas for the U2OS cells
conditionally expressing Myc-Cdh1. This work was supported by grants from the
National Institutes of Health to A.L., A.I., A.B. and M.P., from the Charlotte
Geyer Foundation to A.I., and from the Christopher Reeve Paralysis Foundation
to A.B. J.S. and D.G. are supported by Deutsche Forschungsgemeinschaft and
Emerald Foundation grants, respectively.
Author Information The microarray data have been deposited in the
ArrayExpress database (http://www.ebi.ac.uk/arrayexpress/query/entry)
under accession number E-MEXP-413. Reprints and permissions information is
available at npg.nature.com/reprintsandpermissions. The authors declare no
competing financial interests. Correspondence and requests for materials should
be addressed to A.I. (firstname.lastname@example.org).
NATURE|Vol 442|27 July 2006