Content uploaded by Ulrich Mayr
Author content
All content in this area was uploaded by Ulrich Mayr
Content may be subject to copyright.
DOI: 10.1126/science.1140738
, 1622 (2007); 316Science
et al.William T. Harbaugh,
Reveal Motives for Charitable Donations
Neural Responses to Taxation and Voluntary Giving
www.sciencemag.org (this information is current as of December 11, 2007 ):
The following resources related to this article are available online at
http://www.sciencemag.org/cgi/content/full/316/5831/1622
version of this article at:
including high-resolution figures, can be found in the onlineUpdated information and services,
http://www.sciencemag.org/cgi/content/full/316/5831/1622/DC1
can be found at: Supporting Online Material
found at:
can berelated to this articleA list of selected additional articles on the Science Web sites
http://www.sciencemag.org/cgi/content/full/316/5831/1622#related-content
http://www.sciencemag.org/cgi/content/full/316/5831/1622#otherarticles
, 8 of which can be accessed for free: cites 26 articlesThis article
4 article(s) on the ISI Web of Science. cited byThis article has been
http://www.sciencemag.org/cgi/content/full/316/5831/1622#otherarticles
1 articles hosted by HighWire Press; see: cited byThis article has been
http://www.sciencemag.org/cgi/collection/psychology
Psychology
: subject collectionsThis article appears in the following
http://www.sciencemag.org/about/permissions.dtl
in whole or in part can be found at: this article
permission to reproduce of this article or about obtaining reprintsInformation about obtaining
registered trademark of AAAS.
is aScience2007 by the American Association for the Advancement of Science; all rights reserved. The title
CopyrightAmerican Association for the Advancement of Science, 1200 New York Avenue NW, Washington, DC 20005.
(print ISSN 0036-8075; online ISSN 1095-9203) is published weekly, except the last week in December, by theScience
on December 11, 2007 www.sciencemag.orgDownloaded from
The scaffold protein Dvl was previously thought
to act downstream of LRP6 because dsh over-
expression activates ß-catenin signaling in Drosoph-
ila LRP6 (arrow) mutants (23) and because the
constitutively active Dfz2-Arrow fusion protein is
inactive in dsh mutants (24). The explanation for this
discrepancy may be that overexpressing Dsh/Dvl
leads to artificial sequestration of Axin or that the
protein has multiple functions in the Wnt pathway.
Taken together, the results suggest that Dvl-
mediated co-aggregation triggers LRP6 phos-
phorylation by CK1g. In this model (Fig. 4D),
upon Wnt signaling Dvl aggregates form at the
plasma membrane, where they co-cluster LRP6
with other pathway components including Fz,
Axin, and GSK3b, in a “LRP6-signalosome.”
The role of Wnt would be to bridge LRP6 and
Fz (25, 5), which copolymerize on a Dvl plat-
form. Clustering of LRP6 then provides a high
local receptor concentration that triggers phos-
phorylation by CK1g and Axin recruitment.
Predictions of this model are as follows: (i)
artificial oligomerization of LRP6 should activate
the receptor and (ii) oligomerized LRP6 should
signal independent of Dvl. Indeed, forced oligo-
merization of LRP6 using a synthetic multimerizer
is sufficient to induce Wnt signaling, and this
oligomerization bypasses the need for Dvl (25).
(iii) Constitutively active LRP6 should signal
independently of Dvl because its self-aggregation
should bypass the need for Dvl polymers. This is
also the case as shown in reporter assays with Dvl
siRNA knockdown (fig. S5, B and C), which
supports previous findings (26, 25). (iv) If LRP6
aggregation is a prerequisite for phosphorylation
by CK1g rather than its consequence, LRP6
aggregates should form even when the kinase is
blocked. This is the case: Nonphosphorylated
LRP6 aggregates were observed in response to
Wnt treatment in cells transfected with dominant-
negative CK1g (Fig. 4C). The model of LRP6-
signalosomes not only provides a mechanism
for Wnt signal transduction but may also be
relevant for the understanding of intracellular
transport of maternal Wnt det erminants in the
fertilized Xenopus egg (27).
References and Notes
1. P. Polakis, Genes Dev. 14, 1837 (2000).
2. R. T. Moon, B. Bowerman, M. Boutros, N. Perrimon,
Science 296, 1644 (2002).
3. C. Y. Logan, R. Nusse, Annu. Rev. Cell Dev. Biol. 20, 781
(2004).
4. R. T. Moon, A. D. Kohn, G. V. De Ferrari, A. Kaykas,
Nat. Rev. Genet. 5, 691 (2004).
5. X. He, M. Semenov, K. Tamai, X. Zeng, Development 131,
1663 (2004).
6. N. S. Tolwinski, E. Wieschaus, Trends Genet. 20, 177 (2004).
7. G. Davidson et al., Nature 438, 867 (2005).
8. X. Zeng et al., Nature 438, 873 (2005).
9. J. Klingensmith, R. Nusse, N. Perrimon, Genes Dev. 8,
118 (1994).
10. K. Itoh, B. K. Brott, G. U. Bae, M. J. Ratcliffe, S. Y. Sokol,
J. Biol. 4, 3 (2005).
11. C. C. Malbon, H. Y. Wang, Curr. Top. Dev. Biol. 72, 153 (2006).
12. J. Mao et al., Mol. Cell 7, 801 (2001).
13. H. Yamamoto, H. Komekado, A. Kikuchi, Dev. Cell 11,
213 (2006).
14. J. D. Axelrod, J. R. Miller, J. M. Shulman, R. T. Moon,
N. Perrimon, Genes Dev. 12, 2610 (1998).
15. U. Rothbächer et al., EMBO J. 19, 1010 (2000).
16. T. Schwarz-Romond et al., Nat. Struct. Mol. Biol. 14, 484
(2007).
17. D. G. Capelluto et al., Nature 419, 726 (2002).
18. J. T. Blitzer, R. Nusse, BMC Cell Biol. 7, 28 (2006).
19. S. Kishida et al., Mol. Cell. Biol. 19, 4414 (1999).
20. T. Schwarz-Romond, C. Merrifield, B. J. Nichols, M. Bienz,
J. Cell Sci. 118, 5269 (2005).
21. M. J. Smalley et al., J. Cell Sci. 118, 5279 (2005).
22. A. Cliffe, F. Hamada, M. Bienz, Curr. Biol. 13, 960 (2003).
23. M. Wehrli et al., Nature 407, 527 (2000).
24. N. S. Tolwi nski et al., Dev. Cell 4, 407 (2003).
25. F. Cong, L. Schweizer, H. Varmus, Development 131,
5103 (2004).
26. L. Li, J. Mao, L. Sun, W. Liu, D. Wu, J. Biol. Chem. 277,
5977 (2002).
27. C. Weaver, D. Kimelman, Development 131, 3491 (2004).
28. We thank R. Pepperkok for support in the EMBL Advanced
Light Microscopy Facility; the Nikon Imaging Center at
the University of Heidelberg and M. Boutros and
D. Ingelfinger for help with siRNA experiments; A. Glinka
for advice; N. Maltry for technical help; and J. Axelrod,
A. Helenius, J. Nathans, R. Nusse, T. Schwarz-Romond,
and M. Semenov for reagents. This work was supported by
the European Union (Endotrack) and the Deutsche
Forschungsgemeinschaft.
Supporting Online Material
www.sciencemag.org/cgi/content/full/316/5831/1619/DC1
Materials and Methods
Figs. S1 to S5
Movie S1
1 November 2006; accepted 11 May 2007
10.1126/science.1137065
Neural Responses to Taxation and
Voluntary Giving Reveal Motives
for Charitable Donations
William T. Harbaugh,
1,2
* Ulrich Mayr,
3
* Daniel R. Burghart
1
Civil societies function because people pay taxes and make charitable contributions to provide
public goods. One possible motive for charitable contributions, called “pure altruism,” is satisfied
by increases in the public good no matter the source or intent. Another possible motive, “warm
glow,” is only fulfilled by an individual's own voluntary donations. Consistent with pure altruism,
we find that even mandatory, tax-like transfers to a charity elicit neural activity in areas linked to
reward processing. Moreover, neural responses to the charity's financial gains predict voluntary
giving. However, consistent with warm glow, neural activity further increases when people make
transfers voluntarily. Both pure altruism and warm-glow motives appear to determine the hedonic
consequences of financial transfers to the public good.
E
very society needs public goods, but the
mechanisms used to fund them vary. For
example, taxation and government spending
are lower in the United States than in most European
countries, but philanthropy is higher (1). To
economists, this charitable giving is a puzzle:
Money is a good, so why are people willing to give
it away? One possible explanation is in terms of a
“pure altruism” motive (2). Individuals with such a
motive receive satisfaction from increases in a
public good, such as the provision of basic services
to the needy. This altruistic concern provides a
motive to give, but there is also an incentive to keep
money for oneself, because the cost of such charity
is entirely paid by the giver, whereas the benefits are
spread out over all those people who care about the
needy. Only those people with a very large pure
altruism motive would give voluntarily, and taxation
is the normal social solution to the resulting free-
riding. Pure altruism implies that people should get
some satisfaction even when public goods are
supplied through mandatory taxation, because, by
this account, people care only about how much of
the public good is provided and not about the
process by which the transfer occurs. A second
possible motive for charitable giving is the sense of
agency associated with the act of voluntary giving.
This reward from giving has been termed “warm
glow” (3, 4). If givers were driven exclusively by
the warm-glow motive, they should derive satisfac-
tion from making a voluntary gift, rather than from
the increase in the level of the public good itself. On
the other hand, taxation should not produce a warm
glow, because paying taxes typically does not
involve a voluntary choice.
The distinction between pure altruism and warm-
glow motives for giving is important for several
reasons. First, if giving is motivated by pure altruism,
tax-funded government expenditures to provide a
public good will reduce private giving, potentially
dollar for dollar, as people cut their voluntary con-
tributions in response to these higher taxes (5). There
should be no similar effect with warm-glow givers,
as their benefit derives from the amount of their gift.
Second, a warm-glow motive for altruism provides
an argument in favor of policies that encourage
voluntary giving, because the warm-glow benefit
provides a reward to the giver that exceeds the ben-
efit from paying an equivalent amount in taxes (6).
Neural evidence may help clarify the relative
importance of pure altruism and warm-glow mo-
tives for charitable giving. Although there is
1
Department of Economics, University of Oregon, Eugene, OR
97403–1285, USA.
2
National Bureau of Economic Research
(NBER), Cambridge, MA 02138–5398, USA.
3
Department of
Psychology, University of Oregon, Eugene, OR 97403–1227, USA.
*To whom correspondence should be addressed. E-mail:
mayr@uoregon.edu (U.M.) or harbaugh@uoregon.edu(W.T.H.)
15 JUNE 2007 VOL 316 SCIENCE www.sciencemag.org1622
REPORTS
on December 11, 2007 www.sciencemag.orgDownloaded from
considerable evidence linking neural activity in
the ventral striatum and the insulae to the pro-
cessing of concrete rewards such as money, food,
and drugs, less is known about how the brain
processes more abstract rewards such as those
often provided by public goods. For money, ac-
tivity in the ventral striatum increases as people
anticipate increases in payoffs and when they
receive unexpected increases in payoffs (7, 8).
Neural responses in the ventral striatum and insu-
lae to information about products and their prices
also predict purchase decisions (9). This work
supports the theory that these areas provide
information on the relative rewards of different
outcomes, which serve as an input to decisions
about consumption and tradeoffs regarding risk
and money (10). Other studies have shown that
activity in the ventral striatum and the insulae is
correlated with more abstract rewards, including
social rewards such as punishing unfair players in
sharing games (11), voluntary contributions to
charities (12, 13), and decisions to trust others
(14, 15). These results motivate our focus on the
ventral striatum and the insulae.
To test for the pure altruism and warm-glow
motives, we used functional magnetic resonance
imaging while subjects played a dictator game.
Subjects received $100 and then made decisions
about whether or not to give money to a local
food bank. They also observed mandatory, tax-
like transfers of their money to the food bank
(Fig. 1, A and B) (16). The behavioral results in
this experiment are consistent with economic
theory and are similar to those reported in earlier
economic experiments (17–19). As shown in Fig.
2A, increases in the amounts going to the charity
and decreases in the cost to the giver both in-
creased the likelihood that a voluntary transfer
was accepted. Self-reported satisfaction with the
transaction followed the same pattern in both the
voluntary and the mandatory conditions (Fig. 2B).
To investigate the neural activity associated with
pure altruism, we used data from the mandatory
treatments, which involved exogenous changes in
subject and charity payoffs. Contrasts of parameter
estimates (Fig. 3) show that activation in very
similar areas of the ventral striatum increased with
the monetary payoff to both the subject and to the
charity. Regression analyses to explain activation
data extracted from anatomical regions of interest
(ROIs) show the same result (table S4) (16). This is
the first evidence we know of demonstrating that
mandatory taxation for a good cause can produce
activation in specific brain areas that have been tied
to concrete, individualistic rewards.
The pure altruism model predicts that people
who highly value increases in the charity’s payoff,
relative to the value they place on getting money
for themselves, will be more likely to give. The
evidence economists have typically used to support
this model has been indirect: Relative values have
been inferred from observed decisions (20). Our
experiment allowed us to observe brain activation,
in areas known to respond to rewards, as we varied
the money the subject received and the money the
charity received. This provides a direct test of the
model: Do across-subject differences in neural
responses to subject and charity payoffs predict
who is more likely to give to the charity?
We are able to address this question out of
treatment by using neural responses in those
“pure” mandatory conditions where only the
subject or only the charity got money (orange and
green cells, respectively, in Fig. 1B). These re-
sponses potentially serve as an indicator of how
much subjects valued money for themselves and
for the charity. In fact, regression coefficients
show that subjects with larger activation re-
sponses to money for themselves were less likely
to give to the charity (black columns in Fig. 4A),
and subjects with larger activation responses to
money for the charity were more likely to give
(gray columns in Fig. 4A). To illustrate this rela-
Fig. 1. (A) Study protocol. We scanned 19 fe-
males using functional magnetic resonance im-
aging (fMRI) while they were presented with
transfers that affected their own account (starting
amount, $100) and the account of a local charity.
Half the transfers were mandatory, to resemble
taxation; the other half were voluntary. We ex-
plained that the experimenters would not know
their choices and that one mandatory and one
voluntary transfer would be randomly chosen
and implemented after the experiment. Events
for each trial occurred as presented in the time
line, details are in the supporting online material
(16). After a 1-s fixation dot, the screen revealed
whether this trial's transfer was mandatory or voluntary, as well as the dollar amount change to the
accounts of the subject and the charity. After 9 s, two vertically aligned labels were added in the lower
portion of the screen, specifying the vertically aligned buttons on a response box. For mandatory
transfers, one of the labels read “acknowledge” and the other “invalid button.” For voluntary transfers,
one of the labels read “accept” and the other “reject.” Label positions varied randomly from trial to trial.
Immediately after the subject’s response, a four-point satisfaction rating scale was shown, to which
subjects responded by pressing one of four laterally oriented keys on the button box. The rating scale
disappeared after 6 s, and there was a blank screen for an intertrial period that was randomly jittered
between 6, 7, and 8 s. (B) Study design. The cells show the dollar transfers. Each design cell was im-
plemented three times as a mandatory transfer and three times as a voluntary decision. Orange cells
indicate pure gains to the subject; green cells indicate pure gains to the charity. These pure-gain design
cells from the mandatory condition were used to predict voluntary giving in the purple cells, where there
was a tradeoff between the subject and the charity (see Fig. 4, A and B).
Fig. 2. (A) Subjects' choices
during voluntary transfers as
a function of payoffs to the
subject and the charity. Many
transfers that were costly to
the subject but benefited the
charity were accepted, and the
rate of acceptance increased
as the cost of making a given
transfer declined. (B) Sub-
jective satisfaction ratings as
a function of payoffs to the
subject and the charity, as
well as the voluntary-mandatory factor. Subjective satisfaction increased as transfers increased and costs
decreased and was higher in the voluntary (solid lines) than in the mandatory conditions (dashed lines).
Fig. 3. Neural response in the ventral
striatum to mandatory payoffs for the sub-
ject (yellow), the charity (blue), and both
(green).
www.sciencemag.org SCIENCE VOL 316 15 JUNE 2007 1623
REPORTS
on December 11, 2007 www.sciencemag.orgDownloaded from
tionship graphically, in Fig. 4B we plotted in-
dividual acceptance rates against the difference
between the neural response to pure charity gains
and pure subject gains, pooled over all the re-
gions in Fig. 4A. We also split the sample into
“altruists” (n = 10) and “egoists” (n = 9) de-
pending on whether they had a larger neural
response to the charity's payoff or to their own
payoff. Altruists gave money nearly twice as
often as egoists (58% versus 31%, P = 0.015).
This supports the existence of a purely altruistic
motive: The larger a person's neural response to
increases in the public good, no matter the
source, the more likely they will give voluntarily.
How then is voluntary giving different from
tax-like transfers? Reported satisfaction ratings
were about 10% higher for voluntary than for the
mandatory transfers (P < 0.01, see Fig. 2B and
table S2) (16). The neural evidence shows a sim-
ilar result; t tests indicate higher activation in
the caudate (left, P = 0.015; right, P = 0.004); the
right nucleus accumbens (P = 0.01); and the
insulae (left, P = 0.063; right, 0.075) in the case
of voluntary transfers (table S4) (16).
Of course, these results might simply reflect
the basic economic principle that adding choices
cannot make the decision-maker worse off. This
follows because a person who likes the payoffs in
a given mandatory transfer can always obtain that
same result in the corresponding voluntary con-
dition by accepting the transfer. However, if the
subject does not like the proposed transfer, only
the voluntary conditions give them the option of
rejecting it and keeping the money. Overall, 55%
of the voluntary transfers that involved a subject’s
giving up money to the charity (purple cells in
Fig. 1B) were rejected. This led to an increase in
the expected payoff to the individual of $13 or
33%, and a decrease in the expected payoff to the
charity of $7 or 10%, relative to the mandatory
condition. So, although the opportunity for free
choice means higher activation in the caudate, the
left nucleus accumbens, and the insulae, as well
as higher payoffs to the individual, it reduced the
level of funding for the public good.
An important question, then, is to what degree
the observed higher activation comes from the
ability to make a choice and to what extent it results
from the differences in payoffs from that choice.
We looked again at the differences in activation
and satisfaction ratings between the mandatory and
voluntary conditions, but this time controlling for
the consequences of rejection by replacing those
payoff changes with $0. The voluntary-mandatory
activation difference remained reliable for the
caudate (left, P = 0.023; right, P = 0.011) and the
right nucleus accumbens (P = 0.042), even after we
controlled for payoffs (table S6) (16). Also,
reported satisfaction was higher for voluntary than
for mandatory transfers after controlling for payoffs
(P < 0.065 for the complete design; P < 0.001
using the cells involving tradeoffs, purple in Fig.
1B). The pure altruism motive for giving, along
with the story about adding choices described
above, would imply that there should be no
mandatory-voluntary differences after controlling
for the payoff effects. Thus, our results suggest that
both the increased payoffs and the ability to choose
lead to increased neural activity and satisfaction.
Previous results have demonstrated that ac-
tivity in the areas we examined is larger when
reward can be linked to one's own actions rather
than to extraneous factors (21–24). Our results
extend these findings about the role of agency in
reward-processing to the important situation in-
volving a choice between the subject’s private
payoff and the public good. What is not clear
from earlier reports is whether agency-linked
modulation of reward activity is actually asso-
ciated with a modulation of hedonic value. Our
study shows that neural activity in the caudate
and right nucleus accumbens, as well as subjec-
tive satisfaction, is larger in the voluntary than in
the mandatory situation. The fact that this effect
persists even after controlling for payoffs sup-
ports the warm-glow theory of giving (3).
In summary, we find that three very different
things—monetary payoffs to oneself, observing a
charity get money, and a warm-glow effect related
to free choice—all activate similar neural sub-
strates. This result supports arguments for a
common “neural currency” of reward (25–29)
and shows that this model can be applied not just
to choice over money, risk, and private con-
sumption goods, but also to more abstract policy
choices involving taxation and charitable giving
(12). Our results are also important for un-
derstanding why people give money to charitable
organizations. First, these transfers are associated
with neural activation similar to that which comes
from receiving money for oneself. The fact that
mandatory transfers to a charity elicit activity in
reward-related areas suggests that even mandatory
taxation can produce satisfaction for taxpayers. A
better understanding of the conditions under
which taxation elicits “neural rewards” could
prove useful for evaluating the desirability of
different tax policies. Second, we show that the
opportunity for free choice is associated with
increased activity in regions implicated in process-
ing rewards, as well as with higher reported
satisfaction. Furthermore, this effect is not entirely
accounted for by increased payoffs. In the context
of charitable giving, this choice-related benefit is
consistent with a warm-glow motive for giving.
In combination, these results suggest that
both pure altruism and warm glow are important
motives for charitable giving. Future work may
reveal whether the free-choice effect found here
extends to other situations, and under which con-
ditions taxation elicits “neural” rewards. A related
question is whether people who vote for a tax to
provide a public good get a warm-glow benefit.
Fig. 4. (A) Predicting giving from ac-
tivations in mandatory “pure-gain” con-
ditions. We created measures of neural
activation in response to “pure subject
gain” and “pure charity gain” by averag-
ing activation from the mandatory con-
ditions where the subject received money
at no cost to the charity and where the
charity received money at no cost to the
subject (orange and green cells, respec-
tively, in Fig. 1B). We used these two sets
of activations as independent predictors of
the average acceptance rate in the nine
design cells involving a tradeoff (purple
cells in Fig. 1B). The figure shows stan-
dardized probit regression coefficients from
models including subject and charity stakes
as control variables and neural response to
pure subject gains and pure charity gains as independent predictors. The
dashed lines indicate P = 0.05 significance. Higher response to pure sub-
ject gain was consistently associated with less giving. Higher response to
pure charity gains was consistently associated with more giving. Co-
efficients for individual predictors were reliable in seven out of 12 cases.
(B) Differences in activation predict giving. As an overall measure, we
averaged the neural activation measures across all six brain areas and
computed the difference between neural responses to the charity's pure
gains and the neural responses to the subject's pure gains (orange and
green cells in Fig. 2B). Giving increased as the neural response to pure
charity gains outweighed the neural response to pure subject gains (R
2
=
27%, P = 0.02).
15 JUNE 2007 VOL 316 SCIENCE www.sciencemag.org1624
REPORTS
on December 11, 2007 www.sciencemag.orgDownloaded from
Last, public goods by their very nature are seldom
traded in markets, and so we cannot observe the
prices people will pay and then use these to
measure value. The finding that neural activity
predicts voluntary donations suggests that such
activity could eventually help measure values and
determine optimal levels of public goods.
References and Notes
1. Center for Civil Society Studies, www.jhu.edu/cnp/
research/compdata.html.
2. S. C. Kolm, in Public Economics, J. Margolis and H. Guitton,
Eds. (Macmillan, London, 1969), pp. 145–200.
3. J. Andreoni, Econ. J. 100, 464 (1990).
4. W. T. Harbaugh, J. Pub. Econ. 67, 269 (1998).
5. T. Bergstrom, L. Blume, H. Varian, J. Pub. Econ. 29, 25 (1986).
6. P. Diamond, J. Pub. Econ. 90, 897 (2006).
7. W. Schultz, P. Dayan, P. R. Montague, Science 275, 1593
(1997).
8. C. M. Kuhnen, B. Knutson, Neuron 47, 763 (2005).
9. B. Knutson, S. Rick, G. E. Wimmer, D. Prelec, G. Loewenstein,
Neuron 53, 147 (2007).
10. P. W. Glimcher, A. Rustichini, Science 306, 447 (2004).
11. D. J. F. de Quervain et al., Science 305, 1254 (2004).
12. J. Moll et al., Proc. Natl. Acad. Sci. U.S.A. 103, 15623 (2006).
13. D. Tankersley, C. J. Stowe, S. A. Huettel, Nat. Neurosci.
10, 150 (2007).
14. J. K. Rilling et al., Neuron 35, 395 (2002).
15. T. Singer, S. J. Kiebel, S. W. Joel, J. D. Dolan, C. Frith,
Neuron 41, 653 (2004).
16. Materials and methods are available as supporting
material in Science Online.
17. R. Forsythe, J. L. Horowitz, N. E. Savin, M. Se fton,
Games Econ. Behav. 6, 347 (1994).
18. C. C. Eckel, P. J. Grossman, J. Pub. Econ. 87, 681 (2003).
19. J. Andreoni, J. H. Miller, Econometrica 70, 737 (2002).
20. T. R. Palfrey, J. E. Prisbrey, Am. Econ. Rev. 87, 829 (1997).
21. R. Elliott, J. L. Neuman, O. A. Longe, J. F. Deakin,
Neuroimage 21, 984 (2004).
22. J. O'Doherty et al., Science 304, 452 (2004).
23. E. M. Tricomi, M. R. Delgado, J. A. Fiez, Neuron 41 , 281
(2004).
24. C. F. Zink, G. Pagnoni, M. E. Martin-Skurski, J. G. Chappelow,
G. S. Berns, Neuron 42, 509 (2004).
25. P. R. Montague, G. S. Berns, Neuron 36, 265 (2002).
26. B. King-Casas et al., Science 308, 78 (2005).
27. M. R. Delgado, L. E. Nystrom, C. Fissell, D. C. Noll,
J. A. Fiez, J. Neurophysiol. 84, 3072 (2000).
28. B. Knutson, G. W. Fong, C. M. Adams, J. L. Varner,
D. Hommer, Neuroreport 12, 3683 (2001).
29. B. Knutson, C. M. Adams, G. W. Fong, D. Hommer,
J. Neurosci. 21, RC159 (2001).
30. Author contributions: Lead authors hip was determined by
a coin flip between the first two authors. Supported by
the National Institute of Aging R01 AG1929601A1 and
NSF SES-0112157. We would like to thank J. Andreoni,
R. Bryck, T. Cameron, J. Chalmers, C. Rode, M. Taylor,
and S. Frey, as well as the staff at the Lewis Center for
Neuroimaging at the University of Oregon.
Supporting Online Material
www.sciencemag.org/cgi/content/full/316/5831/1622/DC1
Materials and Methods
Fig. S1
Tables S1 to S7
References
2 February 2007; accepted 9 May 2007
10.1126/science.1140738
Sequence Finishing and
Mapping of Drosophila
melanogaster Heterochromatin
Roger A. Hoskins,
1
* Joseph W. Carlson,
1
* Cameron Kennedy,
1
David Acevedo,
1
Martha Evans-Holm,
1
Erwin Frise,
1
Kenneth H. Wan,
1
Soo Park,
1
Maria Mendez-Lago,
2
Fabrizio Rossi,
3
Alfredo Villasante,
2
Patrizio Dimitri,
3
Gary H. Karpen,
1,4
Susan E. Celniker
1
†
Genome sequences for most metazoans and plants are incomplete because of the presence of
repeated DNA in the heterochromatin. The heterochromatic regions of Drosophila melanogaster
contain 20 million bases (Mb) of sequence amenable to mapping, sequence assembly, and
finishing. We describe the generation of 15 Mb of finished or improved heterochromatic sequence
with the use of available clone resources and assembly methods. We also constructed a bacterial
artificial chromosome–based physical map that spans 13 Mb of the pericentromeric
heterochromatin and a cytogenetic map that positions 11 Mb in specific chromosomal locations.
We have approached a complete assembly and mapping of the nonsatellite component of
Drosophila heterochromatin. The strategy we describe is also applicable to generating substantially
more information about heterochromatin in other species, including humans.
H
eterochromatin is a major component of
met azoan and plant genomes (e. g.,
~20% of the human genome) that regu-
lates chromosome segregation, nuclear organiza-
tion, and gene expression (1–4). A thorough
description of the sequence and organization of
heterochromatin is necessary for understanding
the essential functions encoded within this region
of the genome. However, difficulties in cloning,
mapping, and assembling regions rich in repeti-
tive elements have hindered the genomic analysis
of heterochromatin (5–7). The fruit fly Drosoph-
ila melanogaster is a model for heterochromatin
studies. About one-third of the genome is con-
sidered heterochromatic and is concentrated in
the pericentromeric and telomeric regions of
the chromosomes (X, 2, 3, 4, and Y) (5, 8). The
heterochromatin contains tandemly repeated sim-
ple sequences (including satellite DNAs) (9),
middle repetitive elements [such as transposable
elements (TEs) and ribosomal DNA], and some
single-copy DNA (10).
The whole-genome shotgun sequence (WGS3)
was the foundation for finishing and mapping
heterochromatic sequences and for elucidating
the organization and composition of the nonsat-
ellite DNA in Drosophila heterochromatin (5, 6).
WGS3 is an excellent assembly of the Dro-
sophila euchromatic sequence, but it has lower
contiguity and quality in the repeat-rich hetero-
chromatin. We undertook a retrospective analysis
of these WGS3 scaffolds (11). Moderately repeti-
tive sequences, such as transposable elements,
are well represented in WGS clones and sequence
reads, but they tend to be assembled into shorter
scaffolds with many gaps and low-quality regions
because of the difficulty of accurately assigning
data to a specific copy of a repeat. The typical
WGS heterochromatic scaffold is smaller [for
scaffolds mapped to an arm, N50 ranged from 4 to
35 kb (11)] than a typical WGS euchromatic
scaffold (N50 = 13.9 Mb) (5). Relative to the
euchromatic scaffolds, the WGS3 heterochromatic
scaffolds have 5.8 times as many sequence gaps
per Mb, as well as lower sequence quality.
To produce the Release 5 sequence, we iden-
tified a set of 10-kb genomic clones from a li-
brary representing 15× clone coverage by paired
end reads (mate pairs) and used this set as tem-
plates to fill small gaps and improve low-quality
regions (11). Higher-level sequence assembly
into Mb-sized linked scaffolds used relationships
determined from bacterial artificial chromosome
(BAC)–based sequence tag site (STS) physical
mapping (see below) and BAC end sequences. In
addition to the WGS data, we incorporated data
from 30 BACs (3.4 Mb; 15 BACs finished since
Release 3) that were originally sequenced as part
of the euchromatin sequencing effort (5, 10).
Sequence finishing resulted in fewer gaps,
longer scaffolds, and higher-quality sequence
relative to WGS3 (fig. S1). About 15 Mb of this
sequence has been finished or improved, and
50% of the sequence is now in scaffolds greater
than 378 kb (N50). Table 1 summarizes the
Release 5 sequence statistics by chromosome
arm. Improved sequence was generated for 145
WGS3 scaffolds, and a set of 90 new scaffolds
were produced by joining or filling 694 gaps of
previously unknown size between WGS3 scaf-
folds. The relationships between the initial WGS
scaffolds and the Release 5 scaffolds can be com-
plex (Fig. 1 and figs. S2 to S7); for example, there
were eight cases in which small scaffolds were
used to fill gaps within larger scaffolds, and two
scaffolds whose gaps interdigitated. As expected,
the sequence consists largely of nests of frag-
1
Department of Genome and Computational Biology,
Lawrence Berkeley National Laboratory, Berkeley, CA
94720, USA.
2
Centro de Biologia Molecular Severo Ochoa,
CSIC-UAM, Cantoblanco 28049, Madrid, Spain.
3
Dipartimento
di Genetica e Biologia Molecolare “Charles Darwin,”
Universita “La Sapienza,” 00185 Roma, Italy.
4
Department
of Molecular and Cell Biology, University of California,
Berkeley, CA 94720, USA.
*These authors contributed equally to this work.
†To whom correspondence should be addressed. E-mail:
celniker@fruitfly.org
www.sciencemag.org SCIENCE VOL 316 15 JUNE 2007 1625
REPORTS
on December 11, 2007 www.sciencemag.orgDownloaded from
