REPORTS

involve the brain dopamine (DA) systems, in

Nucleus Accumbens D2/3 Receptors particular the mesolimbic and mesocortical DA
pathways innervating the nucleus accumbens and
Predict Trait Impulsivity and prefrontal cortex (19–22). Positron emission
tomography (PET) studies in nonhuman primates

Cocaine Reinforcement have indicated a role for DA D2 receptors in

determining individual differences in intravenous
cocaine self-administration (19, 20). Specifically,
Jeffrey W. Dalley,1,2* Tim D. Fryer,1,3 Laurent Brichard,1† Emma S. J. Robinson,1‡ low D2 receptor availability in the striatum
David E. H. Theobald,1,2 Kristjan Lääne,1,2 Yolanda Peña,1§ Emily R. Murphy,1,2 Yasmene Shah,4 inversely predicts subsequent levels of intra-
Katrin Probst,1,3 Irina Abakumova,1,3 Franklin I. Aigbirhio,1,3 Hugh K. Richards,1,5 Young Hong,1,3 venous cocaine self-administration in rhesus
Jean-Claude Baron,1,6 Barry J. Everitt,1,2 Trevor W. Robbins1,2 monkeys (20), a result apparently similar to that
seen in studies of human cocaine abusers (23).
Stimulant addiction is often linked to excessive risk taking, sensation seeking, and impulsivity, However, it is not clear how individual
but in ways that are poorly understood. We report here that a form of impulsivity in rats predicts differences in D2 receptor availability relate to a
high rates of intravenous cocaine self-administration and is associated with changes in dopamine specific behavioral endophenotype or behavioral
(DA) function before drug exposure. Using positron emission tomography, we demonstrated that process that confers vulnerability to drug addic-
D2/3 receptor availability is significantly reduced in the nucleus accumbens of impulsive rats that tion. In addition, there have been few, if any,
were never exposed to cocaine and that such effects are independent of DA release. These data studies where DA release in vivo has been
demonstrate that trait impulsivity predicts cocaine reinforcement and that D2 receptor dysfunction combined with PET estimates of D2 receptor
in abstinent cocaine addicts may, in part, be determined by premorbid influences. availability. This is important because D2
receptor availability is influenced by both recep-
ccumulating evidence suggests that cer- intravenous amphetamine and cocaine self- tor density and competing DA release (24, 25).

A tain personality traits, including sensa-

tion (or novelty) seeking, impulsivity,
and antisocial conduct disorder, may predispose
administration at lower doses than do rats that
show reduced levels of activity (12, 13). In addition,
rats that are impulsive on a delay-of-reward task,
Thus, there is a need to conduct analogous PET
studies in animals to investigate the predictive
relationship between D2 receptor availability
humans to drug abuse and addiction (1–4). choosing a small immediate reward over a large and trait behavioral markers of drug-abuse
However, from studies of human drug addicts but delayed reward, show an increased propensity vulnerability.
alone, it is difficult to determine whether comor- to self-administer cocaine, as compared to low- We investigated the relevance of a spontane-
bid impulsivity and cognitive dysfunction (5, 6) impulsive rats (18). Finally, the existence of trait ously occurring form of impulsivity in outbred
pre-date the onset of drug use or emerge as a con- variables related to drug-abuse vulnerability is Lister hooded (LH) rats to intravenous cocaine
sequence of chronic drug use. Current hypotheses encouraged by studies in nonhuman primates in self-administration and to underlying changes
suggest that long-term drug use impairs inhibitory which cocaine is more readily self-administered by in striatal DA function, as measured by micro-
control functions mediated by the prefrontal subordinate, rather than dominant, monkeys (19). PET and in vivo microdialysis (26). We de-
cortex and the associated limbic brain circuitry, A key neural substrate underlying individual fined impulsivity as high levels of anticipatory
leading to a loss of inhibition or to impulsivity differences in drug vulnerability is thought to responses made before the presentation of a
(7, 8). However, there is little evidence to date that
chronic exposure to cocaine and other psycho- Fig. 1. Behavioral
stimulant drugs leads to long-term increases in attributes of trait impul-
impulsive behavior in animals (9–11). sivity on the 5-CSRT task.
The view that individual differences in drug (A) Impulsive rats exhib-
abuse reflect distinct behavioral and physiologi- it high levels of prema-
cal traits is richly supported by studies in animals ture responding on days
(12–17). Rats that are selected for high novelty- when visual targets are
induced locomotor activity more readily acquire presented either 5 s after
trial initiation (days 1, 2,
1
Behavioral and Clinical Neuroscience Institute, University 4, and 5) or 7 s after trial
of Cambridge, Downing Street, Cambridge CB2 3EB, UK. initiation (day 3), as com-
2
Department of Experimental Psychology, University of pared to non-impulsive
Cambridge, Downing Street, Cambridge CB2 3EB, UK.
3
Wolfson Brain Imaging Centre, Department of Clinical
rats. Two-way analysis of
Neurosciences, University of Cambridge, Addenbrooke’s variance (ANOVA) of pre-
Hospital, Cambridge CB2 2QQ, UK. 4Sir Peter Mansfield mature responses revealed
Magnetic Resonance Centre, School of Physics and a significant main effect
Astronomy, University of Nottingham, Nottingham NG7 of day [F(4,40) = 144.9,
2RD, UK. 5Neurosurgery Unit, Department of Clinical P < 0.01] and a signifi-
Neurosciences, University of Cambridge, Addenbrooke’s Hos-
pital, Cambridge CB2 2QQ, UK. 6Neurology Unit, Depart- cant main effect of group
ment of Clinical Neurosciences, University of Cambridge, [F(1,10) = 26.1, P <
Addenbrooke’s Hospital, Cambridge CB2 2QQ, UK. 0.01]. However, there
*To whom correspondence should be addressed. E-mail: were no significant effects
jwd20@cam.ac.uk on other measures of task
†Present address: Laboratoire d’Imagerie Moléculaire et performance, including
Fonctionnelle: de la Physiologie à la Thérapie, ERT CNRS/ (B) attentional accuracy
Université Victor Segalen Bordeaux 2, Bordeaux, France.
[F(1,10) = 1.17, P = 0.306], (C) latency to collect food reward [F < 1, not significant (ns)], (D) omissions
‡Present address: Department of Pharmacology, University
of Bristol, Bristol BS8 1TD, UK. (F < 1, ns), (E) latency to respond correctly [F(1,10) = 3.0, P = 0.113], and (F) the time required to
§Present address: Departament de Psiquiatria i de Medicina complete both standard and challenge (long-ITI) sessions (F < 1, ns). Black circles, high-impulsive rats;
Legal, Universita Autònoma de Barcelona, Barcelona, Spain. white circles, non-impulsive rats.

www.sciencemag.org SCIENCE VOL 315 2 MARCH 2007 1267

 REPORTS
food-predictive, brief light stimulus in a five- explanation for the significant reduction in D2/3 We next investigated the consequences of
choice serial reaction time (5-CSRT) task of sus- receptor availability in trait-impulsive rats was a trait impulsivity for the acquisition and mainte-
tained visual attention (27). Typically, 5-CSRT decrease in the number of D2/3 receptors. nance of intravenous cocaine self-administration.
task impulsivity occurs in ~7% of male LH rats
and, once identified, is a stable trait that persists
throughout adulthood.
We first investigated the behavioral specific-
ity of trait impulsivity on the 5-CSRT task and
the correspondence of this trait to the high-
responder (HR) drug-vulnerable phenotype
described previously in rats by Piazza and
colleagues (12). Trait impulsivity on the 5-CSRT
task was characterized by a near-twofold increase
in premature responding under both standard test
conditions and when the intertrial interval (ITI)
was increased to 7 s (Fig. 1A). In contrast to the
HR rat, in our study trait impulsivity on the
5-CSRT task was inversely related to the level of
locomotor activity measured in a novel environ-
ment [correlation coefficient (r) = 0.49, P = 0.02]
(fig. S1).
Twelve cocaine-naïve adult male rats were
then used to investigate DA D2/3 receptor
availability in the ventral and dorsolateral stria-
tum by means of micro-PET and the selective
high-affinity DA D2/3 receptor antagonist [18F]
fallypride (28). Six non-impulsive rats and six
high-impulsive rats were prepared previously
with an intravenous catheter. The ventral striatum
was included as a region of interest (ROI),
because the primary reinforcing and psycho-
stimulant effects of drugs of abuse depend on the
functional integrity of DA afferents to this region
(29–31). In contrast, DA neurotransmission in
the dorsolateral striatum mediates the persistent Fig. 2. Reduced binding potential (BP) of the selective D2/3 receptor antagonist [18F]fallypride in the
or compulsive forms of drug-seeking behavior ventral striatum of drug-naïve trait-impulsive rats (n = 6 rats) as compared to drug-naïve non-
(32), which is consistent with a role for this impulsive rats (n = 6). (Top) ROIs are shown in the schematic coronal sections of the rat forebrain
region in stimulus-response habit-based learning [adapted from (46)]. Dorsal and ventral striatal ROIs are depicted by the shaded and striped circles,
mechanisms (33, 34). respectively. Anterior-posterior coordinates (in millimeters) relative to bregma are shown on each
We found significantly reduced D2/3 receptor coronal section. The ROIs have diameters of 2 mm in the transverse plane. BP values are averages of left
availability in the ventral striatum (P = 0.037) but and right striata. (A to D) Horizontal MR coregistered PET images of [18F]fallypride binding in the dorsal
not the dorsolateral striatum of high-impulsive (upper panels) and ventral (lower panels) striatum of a non-impulsive rat [(A) and (C)] and a high-
rats as compared to non-impulsive rats (Fig. 2), impulsive rat [(B) and (D)]. The images are 4.5 mm [(A) and (B)] and 7.0 mm [(C) and (D)] below the
and we also found a significant inverse correla- dorsal brain surface (BP threshold = 9).
tion between D2/3 receptor availability in the
ventral striatum and impulsivity (r = 0.58, P =
0.048) (fig. S2). An example of the quality of the Fig. 3. Differential escalation of intra-
data fit obtained with the simplified reference venous cocaine self-administration in
tissue model (35) is shown in fig. S3. high-impulsive rats (n = 8) as compared
We also investigated the possibility that to non-impulsive rats (n = 8). On the first
reduced [18F]fallypride binding potential in the 5 days, access to cocaine was restricted to 5
ventral striatum was due to excessive release of hours and a maximum of 50 infusions.
DA in the nucleus accumbens. We postulated After a withdrawal period of 9 days, access
that enhanced DA release may have displaced to cocaine was increased on each of the
following 5 days to 8 hours and a max-
[18F]fallypride from D2/3 receptors in this region,
imum of 150 infusions. This pattern of
leading to an apparent reduction in D2/3 receptor intermittent cocaine self-administration
availability. DA release and metabolism [inferred was repeated on two further occasions. Im-
from levels of 3,4-dihydroxyphenylacetic acid and pulsive rats showed a differential increase
homovanillic acid (4-hydroxy-3-methoxyphenyl- in their rate of cocaine self-administration
acetic acid)] in the nucleus accumbens were not after extended access to cocaine [session:
significantly different between impulsive rats and F(19,133) = 2.04, P = 0.01; group: F(1,14) = 32.82, P < 0.001; session × group: F(19,133) = 1.92, P =
non-impulsive rats. There was also no significant 0.017]. Subsequent pairwise comparisons revealed significant differences (P < 0.05) between the first
difference in levels of the 5-hydroxytryptamine session and sessions 10, 13, 14, and 15 for the non-impulsive group and between the first session and
metabolite 5-hydroxyindoleacetic acid in this all sessions but nos. 2, 3, 4, 10, and 16 for the high-impulsive group. Black circles, high-impulsive rats;
brain region (table S1). Thus, the most likely white circles, non-impulsive rats.

1268 2 MARCH 2007 VOL 315 SCIENCE www.sciencemag.org

 REPORTS
Intravenously catheterized rats that were screened found that prolonged exposure of trait-impulsive greater tendency for escalation of intravenous
previously for impulsivity on the 5-CSRT task, as rats to cocaine decreased levels of premature cocaine self-administration than did their non-
well as their non-impulsive controls, were trained responding on the 5-CSRT task when the ani- impulsive counterparts. We also found un-
to acquire intravenous cocaine self-administration mals were tested in withdrawal (Fig. 4), but that equivocal evidence for a significant inverse
under a continuous-reinforcement schedule (fig. it had no effect on other measures of perform- relationship between D2/3 receptor availability
S4). Impulsive rats exhibited a clear increase in ance, including attentional accuracy, omissions, in the ventral striatum and trait impulsivity that
their rate of intravenous cocaine self-administration and response latencies. Thus, one clear conse- could not be explained by abnormalities in pre-
as compared to non-impulsive rats (Fig. 3), as well quence of long-access cocaine exposure and sub- synaptic DA release and function in this region.
as a vertical shift in the cocaine dose-response sequent cocaine withdrawal in trait-impulsive These results expand on previous findings in
curve (fig. S5). rats was a selective normalization of premature abstinent human cocaine addicts (23, 36) by
In a final experiment, we investigated the ef- responding on the 5-CSRT task, relative to non- demonstrating that decreased D2 receptor avail-
fects of repeated intermittent withdrawal of co- impulsive rats. ability in the striatum may be a predisposing
caine on pretrained performance of the 5-CSRT We have shown that individual differences in neurobiological trait and not only a consequence
task (Fig. 4). Our objective was to determine impulsive behavior on a five-choice task of sus- of chronic cocaine exposure. They also add to the
whether high rates of cocaine self-administration tained visual attention strongly predict individual finding that D2 receptor availability in the
exhibited by trait-impulsive rats resulted in dif- variation in the rate of intravenous cocaine self- striatum inversely predicts rates of intravenous
ferential impairments in sustained visual atten- administration. Specifically, rats exhibiting trait cocaine self-administration in previously cocaine-
tion, as compared to non-impulsive rats. We impulsivity on the 5-CSRT task (27) showed a naïve rhesus monkeys (20), by defining the
characteristics of a drug-vulnerable phenotype in
terms of a specific behavioral construct, as well as
by underlying abnormalities in D2 receptor
function in the striatum. Thus, the present data
demonstrate that trait impulsiveness is a drug-
vulnerable phenotype and highlight a potential
overlapping involvement of D2-like DA recep-
tors, which have been previously implicated in
genetic studies of both impulsivity and vulnera-
bility to drug-abuse disorders (37, 38).
Trait impulsivity on the 5-CSRT task is
negatively correlated with novelty-induced loco-
motor activity and so is unlikely to be equivalent
to the HR drug-vulnerable phenotype (12). In
contrast to HR rats (12, 13), the high-impulsive
rats in the present study did not acquire cocaine
self-administration faster than the low-impulsive
rats. Although impulsive female rats screened by
means of a food-motivated delayed-discounting
task acquired cocaine self-administration more
rapidly than did low-impulsive rats (18), the high-
impulsive rats acquired the delayed-discounting
task more readily, suggesting that a more funda-
mental difference in instrumental learning may
underpin the difference in the initiation of drug-
taking behavior. HR rats also self-administer high
doses of stimulants (12, 13), and so it is con-
ceivable that this phenotype overlaps with the
present high-impulsive animals to some extent.
However, our findings specifically link trait im-
pulsivity, rather than hyperactivity, to the mainte-
nance and escalation of cocaine self-administration
Fig. 4. Effects of intermittent intravenous cocaine self-administration in high-impulsive rats (n = and to reduced D2/3 receptor binding in the nu-
8) and non-impulsive rats (n = 8) on sustained visual attention on the 5-CSRT task. (A) Rats were cleus accumbens.
withdrawn from intravenous cocaine self-administration on four occasions and tested 24 hours Mutant mice lacking D2-like receptors
later for 7 consecutive days on the 5-CSRT task. ShA, tested after short access to cocaine over 5 show high rates of intravenous cocaine self-
days, when rats received 50 cocaine infusions over 5 hours; LgA-1, tested after long access to
administration when high doses of cocaine are
cocaine, when the daily number of cocaine infusions increased to 150 and the session duration to 8
made available (39). Thus, D2 receptors may
hours; LgA-2, tested after a second long-access exposure to cocaine; LgA-3, tested after a third
long-access exposure to cocaine. ANOVA of premature responses revealed significant group × cycle play a role in regulatory processes that nor-
[F(3,42) = 7.86, P = 0.023] and group × cycle × session [F(18,252) = 4.29, P < 0.001] mally act to limit excessive rates of cocaine
interactions. Subsequent analyses revealed that premature responding was higher in high- self-administration. The decreased D2/3 receptor
impulsive rats as compared to non-impulsive rats during the ShA cycle [group: F(1,14) = 20.58, P < availability in the nucleus accumbens of high-
0.001] but not during subsequent LgA-1 [F(1,14) = 0.62, P = 0.44], LgA-2 [F(1,14) = 3.82, P = impulsive rats may therefore contribute to in-
0.071], and LgA-3 [F(1,14) = 0.51, P = 0.49] cycles. There were no significant differences between creased cocaine self-administration. This finding
non-impulsive and high-impulsive rats with respect to (B) attentional accuracy, (C) omissions, and (D) may have broad relevance to a number of abused
correct response latency (F < 1.44, P > 0.24 for all comparisons). Black circles, high-impulsive rats; drugs in humans, including nicotine and opiates,
white circles, non-impulsive rats. where high consumption rates have also been

www.sciencemag.org SCIENCE VOL 315 2 MARCH 2007 1269

 REPORTS
linked to reduced D2 receptor function, specifi- References and Notes 31. H. O. Pettit, A. Ettenberg, F. E. Bloom, G. F. Koob,
cally because of the presence of the D2 Taq I A1 1. N. Chakroun, J. Doron, J. Swendsen, Encephale 30, 564 Psychopharmacology 84, 167 (1984).
(2004). 32. L. J. Vanderschuren, P. Di Ciano, B. J. Everitt, J. Neurosci.
allele (40, 41), which has previously been shown 2. S. Dawe, N. J. Loxton, Neurosci. Biobehav. Rev. 28, 343 25, 8665 (2005).
by PET to be associated with reduced D2 (2004). 33. M. G. Packard, B. J. Knowlton, Annu. Rev. Neurosci. 25,
receptor density in the striatum (42). 3. K. J. Sher, B. D. Bartholow, M. D. Wood, J. Consult. Clin. 563 (2002).
Our results are consistent with the hypothe- Psychol. 68, 818 (2000). 34. H. H. Yin, B. J. Knowlton, B. W. Balleine, Eur. J. Neurosci.
4. J. B. Adams et al., Am. J. Drug Alcohol Abuse 29, 691 (2003). 19, 181 (2004).
sis that dysfunctional DA neurotransmission 5. K. I. Bolla, R. Rothman, J. L. Cadet, J. Neuropsychiatry 35. R. N. Gunn, A. A. Lammertsma, S. P. Hume, V. J.
at D2-like receptors in the nucleus accumbens Clin. Neurosci. 11, 361 (1999). Cunningham, Neuroimage 6, 279 (1997).
confers susceptibility to increased cocaine self- 6. R. Hester, H. Garavan, J. Neurosci. 24, 11017 (2004). 36. D. Martinez et al., Neuropsychopharmacology 29, 1190
administration in high-impulsive rats. Recent 7. J. D. Jentsch, J. R. Taylor, Psychopharmacology 146, 373 (2004).
(1999). 37. G. R. Uhl, Ann. N. Y. Acad. Sci. 1025, 1 (2004).
findings in rhesus monkeys trained to self- 8. M. Lyvers, Exp. Clin. Psychopharmacol. 8, 225 (2000). 38. A. M. Persico, G. Bird, F. H. Gabbay, G. R. Uhl, Biol.
administer cocaine support this view (19, 20), 9. T. A. Paine, H. C. Dringenberg, M. C. Olmstead, Behav. Psychiatry 40, 776 (1996).
but there has been little evidence to date that Brain Res. 147, 135 (2003). 39. S. B. Caine et al., J. Neurosci. 22, 2977 (2002).
10. J. W. Dalley et al., Psychopharmacology 182, 579 40. D. E. Comings, K. Blum, Prog. Brain Res. 126, 325
D2 receptor dysregulation that is present either
(2005). (2000).
as a trait marker or induced by chronic cocaine 11. J. W. Dalley et al., Neuropsychopharmacology 30, 525 41. M. Shahmoradgoli Najafabadi et al., Am. J. Med. Genet.
exposure is restricted to the nucleus accum- (2005). Neuropsychiatr. Genet. 134, 39 (2005).
bens. Indeed, analogous PET studies in monkeys 12. P. V. Piazza, J. M. Deminiere, M. Le Moal, H. Simon, 42. T. Pohjalainen et al., Mol. Psychiatry 3, 256 (1998).
Science 245, 1511 (1989).
(19, 20, 43) and abstinent human cocaine addicts 13. J. R. Mantsch, A. Ho, S. D. Schlussman, M. J. Kreek,
43. M. A. Nader et al., Neuropsychopharmacology 27, 35
(23) report generalized reductions in D2 receptor (2002).
Psychopharmacology 157, 31 (2001). 44. M. S. Hooks et al., J. Neurosci. 14, 6144 (1994).
availability throughout the striatum. In contrast, 14. E. B. Larson, M. E. Carroll, Pharmacol. Biochem. Behav. 45. B. J. Everitt, T. W. Robbins, Nat. Neurosci. 8, 1481
as in the present study, D2 receptor density and 82, 590 (2005). (2005).
15. A. D. Morgan, N. K. Dess, M. E. Carroll,
D2 mRNA content were found to be significant- Psychopharmacology 178, 41 (2005).
46. G. Paxinos, C. Watson, in The Rat Brain in Stereotaxic
ly reduced in the nucleus accumbens but not in Coordinates (Academic Press, San Diego, CA, ed. 4,
16. B. A. Gosnell, Psychopharmacology 149, 286 (2000).
1998), p. xxvi.
the dorsal striatum or medial prefrontal cortex of 17. E. L. van der Kam, B. A. Ellenbroek, A. R. Cools,
47. This research was supported by a joint award from the
HR rats (44). Neuropharmacology 48, 685 (2005).
U.K. Medical Research Council (MRC) and the Wellcome
18. J. L. Perry, E. B. Larson, J. P. German, G. J. Madden,
The contrast between specific reductions in M. E. Carroll, Psychopharmacology 178, 193 (2005).
Trust and by an MRC Pathfinder grant (G0401068) to
D2 receptor availability in the nucleus accumbens J.W.D., B.J.E., and T.W.R. D.E.H.T. was supported by the
19. D. Morgan et al., Nat. Neurosci. 5, 169 (2002).
Wellcome Trust. Y.P. was supported by a predoctoral
in rats before cocaine self-administration, as 20. M. A. Nader et al., Nat. Neurosci. 9, 1050 (2006).
Formación Investigadora scholarship from Generalitat de
compared to the more divergent effects in the 21. P. V. Piazza et al., Brain Res. 567, 169 (1991).
Catalunya. E.S.J.R. holds a Research Councils UK
22. M. Tonissaar, L. Herm, A. Rinken, J. Harro, Neurosci. Lett.
striatum after chronic exposure to cocaine self- Academic Fellowship supported by the British
403, 119 (2006).
administration, suggests the hypothesis that the Pharmacological Society Integrative Pharmacology Fund.
23. N. D. Volkow et al., Synapse 14, 169 (1993).
E.R.M. was supported by a Gates Cambridge scholarship.
ventral striatal changes confer susceptibility to 24. R. H. Mach et al., Pharmacol. Biochem. Behav. 57, 477
We thank Merck, Sharp, & Dohme for originally donating
cocaine taking, which subsequently affects DA (1997).
the micro-PET scanner to Cambridge University.
neurotransmission in the dorsal striatum, with 25. M. Laruelle et al., Neuropsychopharmacology 17, 162
(1997).
corresponding effects on the number of D2 26. Materials and methods are available as supporting
Supporting Online Material
receptors in that region (20, 23, 43). Thus, the material on Science Online. www.sciencemag.org/cgi/content/full/315/5816/1267/DC1
Materials and Methods
development of psychostimulant drug addiction 27. T. W. Robbins, Psychopharmacology 163, 362 (2002).
Figs. S1 to S5
may represent a progression from initial impul- 28. J. Mukherjee et al., Nucl. Med. Biol. 26, 519 (1999).
29. D. C. Roberts, M. E. Corcoran, H. C. Fibiger, Pharmacol. Table S1
sivity mediated by the nucleus accumbens to the Biochem. Behav. 6, 615 (1977).
References
development of compulsive habitual responding 30. S. B. Caine, G. F. Koob, J. Exp. Anal. Behav. 61, 213 1 November 2006; accepted 1 February 2007
mediated by the dorsal striatum (32, 45). (1994). 10.1126/science.1137073

ments have evolved their own polymer-based DNA

Reconstitution of DNA Segregation segregation machinery (4, 5). Plasmid R1 is a
large (100 kb), low–copy number plasmid with
segregation machinery encoded by the par oper-
Driven by Assembly of a on (6, 7). This operon is composed of three
elements: parC, parR, and parM (6). parM
Prokaryotic Actin Homolog encodes an actin-like protein that assembles into
dynamically unstable filaments (8). parR encodes
Ethan C. Garner,1,2,3 Christopher S. Campbell,1,2,3 Douglas B. Weibel,3,4 R. Dyche Mullins1,2,3* a protein that binds cooperatively to sequence
repeats within parC, forming a complex that
Multiple unrelated polymer systems have evolved to partition DNA molecules between daughter binds ParM filaments (9). It has been proposed
cells at division. To better understand polymer-driven DNA segregation, we reconstituted the three- that products of the par operon assemble into a
component segregation system of the R1 plasmid from purified components. We found that the bipolar spindle-like structure whose elongation
ParR/parC complex can construct a simple bipolar spindle by binding the ends of ParM filaments, drives plasmid segregation (9), but it is not yet
inhibiting dynamic instability, and acting as a ratchet permitting incorporation of new monomers
1
and riding on the elongating filament ends. Under steady-state conditions, the dynamic instability Cellular and Molecular Pharmacology, University of Cali-
of unattached ParM filaments provides the energy required to drive DNA segregation. fornia, San Francisco, CA 94158, USA. 2UCSF/UCB Nanomed-
icine Development Center, San Francisco, CA 94158, USA.
3
Physiology Course, Marine Biological Laboratory, Woods
o ensure fidelity of gene transmission, to drive alignment and segregation of chromo-

T DNA molecules must be evenly distrib-

uted among daughter cells before division.
Eukaryotes harness the polymerization of tubulin
somes (1); chromosome segregation in some eu-
bacteria requires assembly of actin-like filaments
(2, 3); and some extrachromosomal DNA ele-
Hole, MA 02543, USA. 4Department of Biochemistry, University
of Wisconsin, Madison, WI 53706, USA.
*To whom correspondence should be addressed. E-mail:
dyche@mullinslab.ucsf.edu

1270 2 MARCH 2007 VOL 315 SCIENCE www.sciencemag.org