30. Subdecadal correlations are based on annual mean
data after removing a 10-year running mean. The 10-
to 50-year correlations are based on the difference
between the 10- and 50-year running means, and
greater than 10-year correlations are based on 10-
year running means alone. The threshold for signifi-
cance (at the P ? 0.05 level) for the sub–10-year,
10 to 50-year, and greater than 10-year correlations
are 0.15, 0.62, and 0.47, respectively.
31. Potential predictability is given by
var(ens. mean) – var(control)/4
var(ens. mean) – var(control)/4 ? var(control)
where var(ens. mean) is the variance of the decadal
mean temperatures in the ensemble mean and var-
(control) is the variance of the decadal mean tem-
peratures in the control integration. The var(con-
trol)/4 terms account for finite ensemble size, giving
a measure of potential skill in predicting a single
observed series with a hypothetical infinite ensemble,
assuming (as appears to be the case for all diagnos-
tics considered) linear superposition of signals and
noise, and normal distributions.
32. G. J. Boer, Clim. Dyn. 16, 469 (2000).
33. M. Collins, M. R. Allen, Hadley Centre Technical Note
No. 21 (Hadley Centre for Climate Prediction and
Research, Bracknell, UK, 2000).
34. We apply a standard optimal detection methodology,
as used in previous studies (3, 13, 35) by projecting
decadal mean temperature changes onto the main
modes of internal variability. The procedure is opti-
mal in giving more weight to less variable compo-
nents of the patterns. We calculate the probability
that the ensemble mean is consistent with the ob-
servations, where the details of the optimization
procedure are given in Tett et al. (10) and we do not
scale the ALL signal. We analyze the period 1 Decem-
ber 1899 to 30 November 1999 and restrict the
analysis to the highest 40 spatiotemporal modes of
variability, the maximum that is estimated to be
adequately sampled by the control (10).
35. M. R. Allen, S. F. B. Tett, Clim. Dyn. 15, 419 (1999).
36. P. A. Stott, S. F. B. Tett, G. S. Jones, M. R. Allen, J. F. B.
Mitchell, G. J. Jenkins, data not shown.
37. D. T. Shindell, R. L. Miller, G. A. Schmidt, L. Pandolfo,
Nature 399, 452 (1999).
38. T. J. Osborn, K. R. Briffa, S. F. B. Tett, P. D. Jones, R. M.
Trigo, Clim. Dyn. 15, 685 (1999).
39. D. E. Parker, P. D. Jones, C. K. Folland, A. Bevan, J.
Geophys. Res. 99, 14373 (1994).
40. A. Jones, D. L. Roberts, M. J. Woodage, Hadley Centre
Technical Note No. 14 (Hadley Centre for Climate
Prediction and Research, Bracknell, UK, 1999).
41. B. A. Albrecht, Science 245, 1227 (1989).
42. M. R. Allen, P. A. Stott, J. F. B. Mitchell, R. Schnur, T. L.
Delworth, Nature 407, 617 (2000).
43. J. U. Beer et al., in The Sun as a Variable Star: Solar
and Stellar Irradiance Variability, Proceedings of IAU
Colloquium 143, 20 to 25 June 1993, J. M. Pap, C.
Frohlich, H. S. Hudson, S. K. Solanki, Eds. (Cambridge
University Press, Cambridge, 1994), pp. 291–300.
44. M. Lockwood, R. Stamper, M. N. Wild, Nature 399,
45. D. V. Hoyt, K. H. Schatten, J. Geophys. Res. 98, 18895
46. P.A.S., S.F.B.T., G.S.J., and G.J.J. were funded by the UK
Department of the Environment, Transport and the
Regions under contract PECD 7/12/37. M.R.A. was sup-
ported by a Research Fellowship from the UK Natural
Environmental Research Council. J.F.B.M. was supported
by the UK Public Meteorological Service Research and
Development Programme. Supplementary support was
provided by European Commission contract ENV4-
CT97-0501 (QUARCC). We thank the reviewers and N.
Rayner for their comments on the manuscript, and the
many colleagues who developed HadCM3.
11 September 2000; accepted 7 November 2000
Extended Life-Span Conferred
by Cotransporter Gene
Mutations in Drosophila
Blanka Rogina, Robert A. Reenan, Steven P. Nilsen,
Stephen L. Helfand*
Aging is genetically determined and environmentally modulated. In a study of
longevity in the adult fruit fly, Drosophila melanogaster, we found that five
independent P-element insertional mutations in a single gene resulted in a near
doubling of the average adult life-span without a decline in fertility or physical
activity. Sequence analysis revealed that the product of this gene, named Indy
(for I’m not dead yet), is most closely related to a mammalian sodium dicar-
boxylate cotransporter—a membrane protein that transports Krebs cycle in-
termediates. Indy was most abundantly expressed in the fat body, midgut, and
oenocytes: the principal sites of intermediary metabolism in the fly. Excision
of the P element resulted in a reversion to normal life-span. These mutations
may create a metabolic state that mimics caloric restriction, which has been
shown to extend life-span.
Single gene mutations can greatly enhance
our understanding of complex biological pro-
cesses such as aging. Mutations in Caeno-
rhabditis elegans and mice have highlighted
the importance of hormone signal transduc-
tion, mitochondrial function, food intake, and
the growth hormone–prolactin–thyroid-stim-
ulating hormone system in life-span exten-
sion (1–9). To date, only one mutation that
extends life-span in Drosophila has been re-
ported. A partial loss-of-function mutation in
the methuselah (mth) gene extends the aver-
age life-span of Drosophila by 35%, but nei-
ther the function of the methuselah gene
product nor its tissue localization is known
(10). In mammals, the only intervention that
extends life-span is caloric restriction, and it
has been postulated that the mechanism by
which some of the mutations in C. elegans
(for example, daf) extend life-span may be
through a similar alteration in energy use
In studies of Drosophila enhancer-trap
lines (11), we noticed that male and female
flies of two lines, 206 and 302, showed a
doubling of mean life-span (from ?37 to
?70 days) and a 50% increase in maximal
life-span. This occurred when only one
copy of the enhancer-trap chromosome was
present (in heterozygotes) (Fig. 1). Chromo-
somal in situ hybridization revealed that the P
element in both 206 and 302 was inserted at
the same cytological location (12). Genomic
DNA flanking the site of insertion in the two
enhancer-trap lines (206 and 302) was ob-
tained by plasmid rescue (13) and sequenced.
The insertion sites in the 206 and 302 enhanc-
er-trap lines were 5753 base pairs (bp) from
each other and were in the same gene, which
we have named Indy (for I’m not dead yet).
Sequence analysis identified three ex-
pressed sequence tags (ESTs) from the Dro-
sophila genome project (LD13803, LD16220,
and HL01773). Genomic and cDNA sequences
predicted a 572–amino acid protein with 34%
identity and 50% similarity to human and rat
renal sodium dicarboxylate cotransporters
(14–16) (Fig. 2). Mammalian dicarboxylate
cotransporters are membrane proteins respon-
sible for the uptake or reuptake of di- and
tricarboxylic acid Krebs cycle intermediates
such as succinate, citrate, and alpha-keto-
glutarate. They are found in a variety of
tissues, including brush border cells of the
small intestine, colon, and placenta; the baso-
lateral membrane of perivenous cells in the
liver; and epithelial cells of the renal proxi-
mal tubule and the brain (14–16).
Information on the chromosomal location
of Indy was used to identify additional muta-
tions in the Indy gene from other laboratories.
We examined several candidate lines with
P-element insertions in the same cytogenetic
region as Indy and found a third enhancer-
trap line with a P element inserted 734 bp
from the site of the 206 insertion (Fig. 2A).
As a heterozygote, this line, 159, showed the
same extension in life-span (Fig. 1). Two
further P-element insertions in Indy were ob-
tained through site-selected mutagenesis of
the Indy locus. In a polymerase chain reac-
tion–based screen of 10,000 mutagenized third
chromosomes, we identified two new insertions
into the Indy locus (12) (Fig. 2A). Flies het-
also showed a large extension in life-span (12).
Department of Genetics and Developmental Biology,
School of Medicine, University of Connecticut Health
Center, 263 Farmington Avenue, Farmington CT
*To whom correspondence should be addressed. E-
R E P O R T S
www.sciencemag.org SCIENCEVOL 29015 DECEMBER 2000
Fig. 1. Life-span extension in Indy mutants. Survival curves of males
heterozygous for three different Indy mutations, a precise excision of the
P -element from Indy 302 (revertant), and an enhancer-trap control are
shown. All flies were tested as heterozygotes over a wild-type Canton-S
strain. The Indy mutants are Indy302 (open green circles), Indy206 (solid
blue circles), and Indy159 (open orange circles) (see Fig. 2 for mutation
map). The excision line (open red squares) is one of four exact excisions
(sequence confirmed) of the P element obtained by mobilizing the P
element from either the Indy302 or Indy206 line, using delta 2-3 trans-
posase (17). The control (solid black squares) is one of four other enhancer-
trap control lines from the same mutagenesis that generated Indy302 and
Indy206 (11), tested as a heterozygote over Canton-S (28). A similar
control survival curve was found for a control from the mutagenesis that
gave rise to Indy159 (12, 13). The mean 25°C life-spans of controls were
37 days, whereas the mean life-spans for Indy206, Indy302, and Indy159
were 71, 69, and 69 days, respectively. Indy206, Indy302, and Indy159
extended mean life-span by 92, 87, and 87% respectively. Extension of 1%
maximal life-span of these Indy mutants was greater than 45%. At 18°C,
the increase in mean life-span conferred by Indy mutations approached
100%, whereas the increase in 1% maximum life-span approached 50%
(12). Flies were maintained in a humidified, temperature-controlled environmental chamber at 25°C and were transferred to fresh food vials and scored
for survival every 2 to 3 days as in (20). Each survivorship curve represents data from over 300 male flies. A total of 5430 male and female Indy
heterozygote flies were tested.
Fig. 2. (A) Genomic organization of the Indy locus, with insertion sites of
all five P-element alleles. The organization of the Indy transcription unit
is shown. Solid black boxes represent conserved regulatory sequences
(AntC and FasI). The red rectangle represents the conserved Hoppel
transposable element. PlacW insertion sites in the 206, 302, and 159
insertion lines are shown, as well as the orientation of the insertions. The
positions of Birmingham-2 P-element insertions (PBm) in 92 and 265
insertion lines are also shown. The insertions in the 206, 159, and 265
lines are within a Hoppel element in the first intron of the Indy gene
just upstream of the putative translational start site. The Hoppel
element is present in the same position in wild-type animals, includ-
ing P1 clones from the Drosophila Genome Project. The insertion in
the 302 line is within 50 bp of the putative transcriptional start site.
PlacW (10 kb) is not drawn to scale. (B) Sequence comparisons (29).
The proteins most homologous to the Indy protein (GenBank acces-
sion no. AE003519) were identified by Blast. Indy-2 is a highly
homologous Drosophila gene (accession no. AE003728). hNaDC-1
(accession no. U26209) is a human dicarboxylate cotransporter, and
SDCT1 (accession no. AF058714) and SDCT2 (accession no. AF081825) are
rat sodium dicarboxylate cotransporters. Red boxes indicate identity across
all proteins. Blue indicates identity to Indy. Yellow indicates amino acid
similarity to Indy.
R E P O R T S
15 DECEMBER 2000VOL 290SCIENCE www.sciencemag.org
To confirm that the P-element insertion in
Indy caused the life-span extension, we remo-
bilized and excised the P element from the
Indy gene in the 302 and 206 lines (17). Four
independent lines of flies, shown by sequence
analysis to carry exact excisions, reverted to
normal life-span (Fig. 1). A nonexcision con-
trol line isolated at the same time, which passed
through the same genetic background as the
excision lines, remained long-lived (12).
The reporter protein ?-galactosidase (?-
Gal) had an identical pattern of expression in
each of the three enhancer-trap lines (206,
302, and 159), despite the P-element inser-
tions being almost 6.5 kb away from each
other in the three lines. In adult flies, Indy
was expressed in the fat body, midgut, and
oenocytes (Fig. 3): organs that are thought to
be the primary sites of intermediary metabo-
lism, absorption, and metabolic storage in
Drosophila. The fat body is involved in the
metabolism and storage of fat, glycogen, and
protein and is most often compared to the
liver of vertebrates (18, 19). Indy was also
expressed at low levels in the halteres; por-
tions of the alimentary canals, including the
procardia and restricted regions of the esoph-
agus and hindgut; and the base of the legs.
These are regions that have been identified as
storage depots for glycogen (18). Finally,
Indy was expressed in a subset of cells in the
third segment of the antennae (20).
To exclude the possibility that the extend-
ed life-span of the Indy mutants was due to
the rescue of uncharacterized deleterious mu-
tations accumulating in our wild-type Can-
ton-S stock, we crossed the Indy mutation
into several different genetic backgrounds
that were distinct from our Canton-S stock.
These included the Hyperkinetic, Shaker, and
drop dead stocks, each of which was isolated
from other wild-type stocks 25 to 30 years
ago, as well as the long-lived laboratory-
selected lines of Luckinbill (21, 22). In all
cases, there was an extension in life-span. For
all genetic backgrounds, mean life-span was
extended by 40 to 80%, except in the case of
the long-lived lines of Luckinbill, in which
life-span was additionally extended by only
15%. These data indicate that the mechanism
by which Indy mutations extend life-span is a
positive effect of the mutation on life-span
and not simply of the rescue of deleterious
mutations. The smaller increase in life-span
seen with the laboratory-selected long-lived
lines provides additional evidence that the
mechanisms by which Indy acts to increase
life-span may represent physiological sys-
tems already partially optimized by laborato-
A decline in fertility (23, 24) or a reduc-
tion in physical activity (25) can lead to an
extension of life-span in flies. Indy long-lived
heterozygote males and females were com-
pared to controls and found to be normal or
superior in fertility and fecundity (Table 1).
Qualitative observations of flight, courtship,
feeding behavior, and negative geotaxis re-
vealed no significant differences between
Indy long-lived males and females and con-
trol flies during early life. Differences oc-
curred later in life when physical measures of
behavior and locomotor function were main-
tained at high levels in the Indy long-lived
animals but not in normal-lived controls. For
instance, one physiological milestone of ag-
ing in flies is the onset of female infertility.
Indy heterozygous long-lived females contin-
ued to produce viable adult offspring 40%
longer on average than did control flies (23.2
versus 16.5 days). This was a true extension
of the period of fertility and was not associ-
ated with a compensatory delay in fertility
during early life, as is seen in laboratory-
selected long-lived lines (26). Indy long-lived
females showed the same early peak of egg
laying and fertility as control females but
sustained the ability to produce larger num-
bers of offspring for a longer period of time
(Table 1). There was no alteration in the rate
or timing of developmental events in Indy
long-lived mutant animals, as in the C. el-
egans clock mutants (3). The time from egg
to adult at 25°C was the same as that for
normal-lived controls (9 to 10 days).
The expression of Indy in the fat body,
gut, and oenocytes, and the amino acid se-
quence similarity of the Indy protein to dicar-
boxylate cotransporters, suggest that Indy
may play a role in both the absorption of
metabolites and in intermediary metabolism.
Fig. 3. Expression of Indy in adult flies. Whole-mount X-Gal staining showing nuclear localization
of ?-Gal in cells from lines carrying an enhancer-trap insert in the Indy gene: Indy302, Indy206, and
Indy159. Expression is seen in oenocytes (A and B) and the gut (C and D). (A) Low-power view of
oenocytes in the (v) ventral and (d) dorsal abdominal segments. (B) High-power view of dorsal
midline oenocytes. (D) A 5-?m section showing X-Gal staining within the cells of the gut.
Whole-mount staining was performed as in (20). After whole-mount X-gal staining, the tissue in
(D) was postfixed in 6.25% glutaraldehyde, embedded in paraffin, and then sectioned. Scale bar in
(A) through (C), 100 ?m; in (D), 10 ?m.
Table 1. Fertility and fecundity of long-lived Indy mutant animals. CS indicates the Canton-S control,
302/CS indicates heterozygotes of 302 and Canton-S, and 206/CS indicates heterozygotes of 206 and
Canton-S. Single-pair mates were kept in individual vials. Each day throughout their entire life-span, they
were passed to new vials, the number of eggs from the vial of the previous day was counted, and the vial
was saved in order to determine the number of adult offspring.
Genotype of parents
Eggs per pair (n)
Up to day 14
Day 14 to 21
Total over life-span
Offspring per pair (n)
Up to day 14
Day 14 to 21
Total over life-span
53 2626 2325
R E P O R T S
www.sciencemag.orgSCIENCEVOL 290 15 DECEMBER 2000
Thus, insertion of a P element in the Indy Download full-text
gene may alter normal metabolism. In mam-
mals, a moderate caloric restriction increases
life-span, whereas a more severe restriction
leads to starvation and decreased life-span
(27). We postulate that the level of Indy
expression is critical for life-span extension
and that P-element insertions in Indy reduce
Indy expression. When the level of Indy is
mildly reduced, as in flies heterozygous for
Indy insertions, there is a large extension in
life-span. The extension in life-span is less
dramatic with a further reduction in Indy
activity, such as that seen in Indy homozy-
gous flies, in which a 10 to 20% increase in
mean life-span was seen (12). This model
predicts that a further reduction in Indy ac-
tivity would shorten life-span, and indeed this
was observed when Indy gene activity was
further reduced by placing an Indy mutation
over a chromosome deleted for the Indy re-
gion. Flies with only a single copy of a
mutant Indy gene and no normal copy of Indy
(Indy mutant over deletion for the Indy re-
gion) had a 10 to 20% shorter mean and
maximum life-span than did homozygous
Indy mutants (12).
The mechanism by which caloric restric-
tion mediates life-span extension is not un-
derstood, but it is likely to involve alterations
in energy utilization. In contrast to many of
the previously identified genes associated
with life-span extension in metazoans, which
have indirect effects on metabolism (5–9),
Indy appears to be directly involved in inter-
mediary metabolism and thus may represent a
new class of longevity genes. A genetically
induced reduction in the amount or efficiency
of a dicarboxylic acid cotransporter in the
Indy mutants may be creating a metabolic
state similar to caloric restriction. Further
characterization of the Indy mutants may pro-
vide direct genetic insight into the role of
energy balance and aging, and a point of
access for genetic and pharmacological inter-
ventions for extending life-span.
References and Notes
1. T. E. Johnson, Science 249, 908 (1990).
2. C. J. Kenyon, E. Chang, A. Rudner, R. Tabtiang, Nature
366, 461 (1993).
3. B. Lakowski, S. Hekimi, Science 272, 1010 (1996).
4. K. D. Kimura, H. A. Tissenbaum, Y. Liu, G. Ruvkun,
Science 277, 942 (1997).
5. J. Z. Morris, H. A. Tissenbaum, G. Ruvkun, Nature 382,
6. S. Ogg et al., Nature 389, 994 (1997).
7. S. Hekimi, in The Molecular Genetics of Aging, Results
and Problems in Cell Differentiation, S. Hekimi, Ed.
(Springer-Verlag, New York, 2000), vol. 29, pp. 81–
8. H. M. Brown-Borg, K. E. Borg, C. J. Meliska, A. Bartke,
Nature 384, 33 (1996).
9. A. Bartke, in The Molecular Genetics of Aging, Results
and Problems in Cell Differentiation, S. Hekimi, Ed.
(Springer-Verlag, New York, 2000), vol. 29, pp. 181–
10. Y. J. Lin, L. Seroude, S. Benzer, Science 282, 943
11. S. Boynton, T. Tully, Genetics 131, 655 (1992).
12. B. Rogina, R. A. Reenan, S. P. Nilsen, S. L. Helfand, data
13. E. Bier et al., Genes Dev. 3, 1273 (1989).
14. X. Z. Chen, C. Shayakul, U. V. Berger, W. Tian, M. A.
Hediger, J. Biol. Chem. 273, 20972 (1998).
15. A. M. Pajor, Am. J. Phys. 270, F642 (1996).
16. ????, Annu. Rev. Physiol. 61, 663 (1999).
17. M. J. Palladino, L. P. Keegan, M. A. O’Connell, R. A.
Reenan, Cell 102, 437 (2000).
18. V. B. Wigglesworth, The Principles of Insect Physiol-
ogy (Chapman and Hill, London, ed. 7, 1972).
19. L. Sondergaard, Trends Genet. 9, 193 (1993).
20. S. L. Helfand et al., Genetics 140, 549 (1995).
21. The FlyBase database of the Drosophila genome
projects and community literature, Nucleic Acids Res.
27, 85 (1999) (http://flybase.bio.indiana.edu/).
22. L. Luckinbill, R. Arking, M. J. Clare, W. C. Cirocco, S.
Buck, Evolution 38, 996 (1984).
23. L. Partridge, A. Green, K. Fowler, J. Insect Physiol. 33,
24. K. Fowler, L. Partridge, Nature 338, 760 (1989).
25. R. S. Sohal, P. B. Buchan, Exp. Gerontol. 16, 157
26. M. R. Rose, Evolutionary Biology of Aging (Oxford
Univ. Press, New York, 1991).
27. R. S. Sohal, R. Weindruch, Science 273, 59 (1996).
28. Because genetic background has such a powerful
effect on life-span, the genetic backgrounds of the
lines being compared must be as similar as possible.
The Indy302 and Indy206 mutants are both from the
mutagenesis reported in (11), where an effort was
made to use “Cantonized” stocks. In addition, each
enhancer-trap line was further backcrossed to a Can-
ton-S stock with the w1118mutation multiple times
(11). Our controls included four other enhancer-trap
lines from the original mutagenesis (11), compared as
heterozygotes with the same Canton-S stock as the
Indy302 and Indy206 lines. The control for Indy159
was similar; we compared another enhancer-trap line
insertion from the same mutagenesis (13). The con-
trol line had an insert in the same cytological region
as Indy, but not inserted into the Indy locus.
29. Single-letter abbreviations for the amino acid resi-
dues are as follows: A, Ala; C, Cys; D, Asp; E, Glu; F,
Phe; G, Gly; H, His; I, Ile; K, Lys; L, Leu; M, Met; N, Asn;
P, Pro; Q, Gln; R, Arg; S, Ser; T, Thr; V, Val; W, Trp; and
30. We thank J. Jack, B. Graveley, and M. Tanzer for
helpful discussions. Supported by grants from the
NIH-supported Claude Pepper Older Americans Inde-
pendence Center at the University of Connecticut
Center on Aging and the American Federation for
Aging Research (B.R.); by NSF grant 9728737 and a
grant from The Patrick and Catherine Weldon Dona-
ghue Medical Research Foundation (R.A.R); and by
the National Institute on Aging (grants AG14532 and
AG16667), a Burroughs Wellcome Travel Grant, and a
Donaghue Investigator Award from The Patrick and
Catherine Weldon Donaghue Medical Research Foun-
17 October 2000; accepted 14 November 2000
Docosahexaenoic Acid, a Ligand
for the Retinoid X Receptor in
Alexander Mata de Urquiza,1Suya Liu,2*
Maria Sjo ¨berg,3* Rolf H. Zetterstro ¨m,1William Griffiths,2
Jan Sjo ¨vall,2Thomas Perlmann1†
The retinoid X receptor (RXR) is a nuclear receptor that functions as a ligand-
activated transcription factor. Little is known about the ligands that activate
RXR in vivo. Here, we identified a factor in brain tissue from adult mice that
activates RXR in cell-based assays. Purification and analysis of the factor by
Previous work has shown that DHA is essential for brain maturation, and
and other abnormalities. These data suggest that DHA may influence neural
function through activation of an RXR signaling pathway.
The nuclear hormone receptors are ligand-acti-
vated transcription factors [reviewed in (1)].
Included in this family are the so-called “or-
phan receptors,” whose ligands have not been
identified [reviewed in (1, 2)]. By screening
chemical libraries, progress has been made in
identifying ligands for these receptors [for ex-
amples, see (3–8)]; however, in most cases
endogenous ligands have remained elusive.
RXR is an obligatory component of a large
number of nuclear receptor heterodimers and is
activated in vitro by the vitamin A metabolite
9-cis retinoic acid (9-cis RA), which binds with
high affinity to the RXR ligand binding domain
(9, 10); however, 9-cis RA acid has been diffi-
cult to detect in vivo (11). Nonetheless, because
numerous studies have demonstrated striking
synergism in biological responses in vitro and
in vivo when both partners of the RXR-RAR
heterodimer are activated by their respective
ligand [for examples, see (12–14)], it seems
likely that ligand-induced activation of RXR
does occur in vivo. This conclusion has been
corroborated by experiments with transgenic
mice (15, 16).
To search for endogenous ligands that acti-
1Ludwig Institute for Cancer Research, Stockholm
Branch, Box 240, S-171 77 Stockholm, Sweden.2De-
partment of Medical Chemistry and3Department of
Cell and Molecular Biology, Karolinska Institutet,
S-171 77 Stockholm, Sweden.
*These authors contributed equally to this work.
†To whom correspondence should be addressed. E-
R E P O R T S
15 DECEMBER 2000VOL 290SCIENCE www.sciencemag.org