Selective Inhibition of Cocaine

letters to nature

10. Ca¯isch, A. & Karplus, M. Acid and thermal denaturation of Barnase investigated by molecular 25. Ferguson, N., Capaldi, A. P., James, R., Kleanthous, C. & Radford, S. E. Rapid folding with and dynamics simulations. J. Mol. Biol. 252, 672±708 (1995). without populated intermediates in the homologous four-helix proteins Im7 and Im9. J. Mol. Biol. 11. Boczko, E. M. & Brooks, C. L. III. First principles calculation of the folding free energy of a three-helix 286, 1597±1608 (1999). bundle protein. Science 269, 393±396 (1995). 26. Taketomi, H., Ueda, Y. & GoÅ, N. Studies on protein folding, unfolding and ¯uctuations by computer 12. Guo, Z. Y., Brooks,, C. L. III. & Boczko, E. M. Exploring the folding free energy surface of a three-helix simulations. Int. J. Pept. Protein Res. 7, 445±459 (1975). bundle protein. Proc. Natl Acad. Sci. USA 94, 10161±10166 (1997). 27. Ferrenberg, A. M. & Swendsen, R. H. Optimized Monte Carlo data analysis. Phys. Rev. Lett. 63, 1195± 13. Duan, Y. & Kollman, P. A. Pathways to a protein folding intermediate observed in a 1-microsecond 1197 (1989). simulation in aqueous solution. Science 282, 707±744 (1998). 28. Zhou, Y., Karplus, M., Wichert, J. M. & Hall, C. K. Equilibrium thermodynamics of homopolymers 14. Gouda, H. et al. Three-dimensional solution structure of the B domain of Staphylococcal protein A: and clusters: Molecular dynamics and Monte Carlo simulations of systems with square-well Comparisons of the solution and crystal structures. Biochemistry 31, 9665±9672 (1992). interactions. J. Chem. Phys. 107, 10691±10708 (1997). 15. Zhou, Y., Vitkup, D. & Karplus, M. Native proteins are surface-molten solids: Application of the 29. Ballew, R. M., Sabelko, J. & Gruebele, M. Direct observation of fast protein folding: The initial collapse Lindemann criterion for the solid versus liquid state. J. Mol. Biol. 285, 1371±1375 (1999). of apomyoglobin. Proc. Natl Acad. Sci. USA 93, 5759±5764 (1996). 16. Ptitsyn, O. B. Molten globule and protein folding. Adv. Protein Chem. 47, 83±230 (1995). 30. Munoz, V., Thompson, P. A., Hofrichter, J. & Eaton, W. A. Folding dynamics and mechanism of b- 17. Privalov, P. L. Stability of proteins: Small globular proteins. Adv. Protein Chem. 33, 167±241 (1979). hairpin formation. Nature 390, 196±199 (1997). 18. Zhou, Y. & Karplus, M. Folding thermodynamics of a model three-helix bundle protein. Proc. Natl Acad. Sci. USA 94, 14429±14432 (1997). 19. Schlunegger, M., Bennett, M. & Eisenberg, D. Oligomer formation by 3D domain swapping: A model Acknowledgements for protein assembly and misassembly. Adv. Protein Chem. 50, 61±122 (1997). We thank Y. Duan and P. A. Kollman for sending us the structure of the intermediate 20. Lazaridis, T. & Karplus, M. Multiple unfolding simulations reconcile the ``new view'' of protein observed in their trajectory. This work was supported in part by a grant from the NSF and a folding with the old. Science 278, 1928±1931 (1997). grant from Pittsburgh Supercomputing Center, and from UC Berkeley Network of 21. Roder, H. & Colon, W. Kinetic role of early intermediates in protein folding. Curr. Opin. Struct. Biol. 7, Workstations (NOW) through NAPCI (National Partnership for Advanced Computa- 15±28 (1997). tional Infrastructure). Y.Z. is an NIH postdoctoral fellow. The calculations at Harvard 22. Frauenfelder, H. & McMahon, B. Dynamics and function of proteins: The search for general concepts. were conducted on HP 9000/735, DEC Alpha and SUN UltraSparc workstations. The Proc. Natl Acad. Sci. USA 95, 4795±4797 (1998). drawing of the three-helix bundle were made with QUANTA (Molecular Simulations 23. Burton, R. E., Myers, J. K. & Oas, T. G. Protein folding dynamicsÐquantitative comparison between Inc.). theory and experiment. Biochemistry 37, 5337±5343 (1998). 24. Cavagnero, S., Dyson, H. J. & Wright, P. E. Effect of H helix destabilizing mutations on the kinetic and Correspondence and requests for materials should be addressed to M.K. equilibrium folding of apomyoglobin. J. Mol. Biol. 285, 269±282 (1999). (e-mail: [email protected]. edu).

...... errata

Selective inhibition of cocaine- Binding of double-strand breaks in seeking behaviour by a partial DNA by human Rad52 protein dopamine D3 receptor agonist Eric Van Dyck, Alicja Z. Stasiak, Andrzej Stasiak Maria Pilla, Sylvie Perachon, FrancËois Sautel, & Stephen C. West Fabrice Garrido, AndreÂ Mann, Camille G. Wermuth, Nature 398, 728±731 (1999) Jean-Charles Schwartz, Barry J. Everitt & Pierre Sokoloff ...... In Fig. 4, one of the strands of duplex DNA was deleted. The Nature 400, 371±375 (1999) ...... corrected ®gure is shown below. M Spurious symbols were introduced into Fig. 1d during reproduc- tion. The correct ®gure is shown here.

Rad50/Mre11/Xrs2 DSB Rad50/Mre11/Xrs2 Rad52 Ku

Rad52-dependent Ku-dependent homologous recombination non-homologous end-joining

Rad52 end-binding Single-strand Recruitment DNA ligase IV annealing of Rad51 Xrcc4

Trimming Strand Ligation invasion

NATURE | VOL 401 | 23 SEPTEMBER 1999 | www.nature.com © 1999 Macmillan Magazines Ltd 403 letters to nature the active pool. These resources then relax into the recovered pool through the inactive pool. We assume that the decay from active to inactive pools is rapid so that the effective time constant for relaxation into the recovered Selective inhibition of cocaine- pool is t. For simplicity, we express the total amount of synaptic resources as the maximal EPSP amplitude A, which is the magnitude of the evoked seeking behaviour by a partial EPSP when all the synaptic resources are used. EPSP amplitudes during a stimulus train are given by the product of the synaptic strength and the dopamine D3 receptor agonist synaptic resources in the recovered pool at the time of the action potential. Maria Pilla*†, Sylvie Perachon*‡§, Franç ois Sautel‡, Combining the model with the experimental observations that there is a Fabrice Garridok, Andre´ Mannk, Camille G. Wermuthk, linear relationship between the dendritic excitatory postsynaptic current Jean-Charles Schwartz‡, Barry J. Everitt† & Pierre Sokoloff‡ (EPSC) and somatic EPSP when the somatic EPSP is less than ϳ5mV24, † Department of Experimental Psychology, University of Cambridge, and that somatic EPSPs add linearly29, we derived an equation relating Downing Street, Cambridge CB2 3EB, UK normalized EPSP amplitudes during a stimulus train. After k ‡ Unite´ de Neurobiologie et Pharmacologie Molećulaire, INSERM U 109, stimulus intervals of a regular stimulus train of frequency f, normalized Centre Paul Broca, 2ter rue d’Ale´sia, 75014 Paris, France ⋅ Ϫ ð1=t Ϫ f ⋅lnð1 Ϫ SÞÞt 1=ðf tÞ Ϫ = 1=ðf tÞ Ϫ Ϫ EPSPk ¼ {S e þðe 1Þ} ðe ð1 SÞÞ. Thus, a continuous § Laboratoire Bioprojet, 9 rue Rameau, 75002 Paris, France curve drawn through normalized EPSP amplitudes evoked by the stimulus train k Laboratoire de Pharmacochimie de la Communication Cellulaire CNRS 1=ðf tÞ 1=ðf tÞ has an exponential form with a steady state of ðe Ϫ 1Þ=ðe Ϫ ð1 Ϫ SÞÞ ERS 655, Faculte´ de Pharmacie, 74 route du Rhin BP 24, 67401 Illkirsch, France Ϫ and a time constant of decay of ð1=t Ϫ f ⋅lnð1 Ϫ SÞÞ 1. * These authors contributed equally to this work...... Received 22 March; accepted 3 June 1999. Environmental stimuli that are reliably associated with the effects 1. Buonomano, D. V. & Merzenich, M. M. Cortical plasticity: from synapses to maps. Annu. Rev. Neurosci. 21, 149–186 (1998). of many abused drugs, especially stimulants such as cocaine, can 2. Gilbert, C. D. Adult cortical dynamics. Physiol. Rev. 78, 467–485 (1998). produce craving and relapse in abstinent human substance 3. Merzenich, M. M. et al. Progression of changes following median nerve section in the cortical 1–4 representation of the hand in areas 3b and 1 in adult owl and squirrel monkeys. Neuroscience 10, 639– abusers . In animals, such cues can induce and maintain 665 (1983). drug-seeking behaviour and also reinstate drug-seeking after 4. Robertson, D. & Irvine, D. Plasticity of frequency organization in auditory cortex of guinea pigs with extinction5–7. Reducing the motivational effects of drug-related partial unilateral deafness. J. Comp. Neurol. 282, 456–471 (1989). 3 5. Kaas, J. H. et al. Reorganization of retinotopic cortical maps in adult mammals after lesions of the cues might therefore be useful in the treatment of addiction . retina. Science 248, 229–231 (1990). Converging pharmacological8,9, human post-mortem10 and 6. Glazewski, S. & Fox, K. Time course of experience-dependent synaptic potentiation and depression in 11 12 barrel cortex of adolescent rats. J. Neurophysiol. 75, 1714–1729 (1996). genetic studies implicate the dopamine D3 receptor in drug 7. Gilbert, C. D. & Wiesel, T. N. Receptive field dynamics in adult primary visual cortex. Nature 356, addiction. Here we have designed BP 897, the first D3-receptor- 150–152 (1992). selective agonist, as assessed in vitro with recombinant receptors 8. Fox, K. A critical period for experience-dependent synaptic plasticity in rat barrel cortex. J. Neurosci. 12, 1826–1838 (1992). and in vivo with mice bearing disrupted D3-receptor genes. BP 897 9. Diamond, M. E., Huang, W. & Ebner, F. F. Laminar comparison of somatosensory cortical plasticity. is a partial agonist in vitro and acts in vivo as either an agonist or Science 265, 1885–1888 (1994). 10. Kirkwood, A., Rioult, M. C. & Bear, M. F. Experience-dependent modification of synaptic plasticity in an antagonist. We show that BP 897 inhibits cocaine-seeking visual cortex. Nature 381, 526–528 (1996). behaviour that depends upon the presentation of drug-associated 11. Glazewski, S., Chen, C.-M., Silva, A. & Fox, K. Requirement for ␣-CaMKII in experience-dependent cues, without having any intrinsic, primary rewarding effects. plasticity of the barrel cortex. Science 272, 421–423 (1996). 12. Garraghty, P. E. & Muja, N. NMDA receptors and plasticity in adult primate somatosensory cortex. J. Our data indicate that compounds like BP 897 could be used for Comp. Neurol. 367, 319–326 (1996). reducing the drug craving and vulnerability to relapse that are 13. Carvell, G. E. & Simons, D. J. Biometric analyses of vibrissal tactile discrimination in the rat. J. Neurosci. 10, 2638–2648 (1990). elicited by drug-associated environmental stimuli. 14. Welker, C. & Woolsey, T. A. Structure of layer IV in the somatosensory neocortex of the rat: description Until now, it has proved difficult to identify or design a D3 and comparison with the mouse. J. Comp. Neurol. 158, 437–454 (1974). 15. Markram, H. & Tsodyks, M. Redistribution of synaptic efficacy between neocortical pyramidal cells. Nature 382, 807–810 (1996). 16. Abbot, L. F., Varela, J. A., Sen, K. & Nelson, S. B. Synaptic depression and cortical gain control. Science 275, 220–224 (1997). 17. Tsodyks, M. V. & Markram, H. The neural code between neocortical pyramidal neurons depends on neurotransmitter release probability. Proc. Natl Acad. Sci. USA 94, 719–723 (1997). 18. McCasland, J. S. & Woolsey, T. A. High-resolution 2-deoxyglucose mapping of functional cortical columns in mouse barrel cortex. J. Comp. Neurol. 278, 555–569 (1988). 19. Armstrong-James, M. Diamond, M. E. & Ebner, F. F. An innocuous bias in whisker use in adult rats modifies receptive fields of barrel cortex neurons. J. Neurosci. 14, 6978–6991 (1994). 20. Darian-Smith, C. & Gilbert, C. D. Axonal sprouting accompanies functional reorganization in adult cat striate cortex. Nature 368, 737–740 (1994). 21. Florence, S. L., Taub, H. B. & Kaas, J. H. Large-scale sprouting of cortical connections after peripheral injury in adult macaque monkeys. Science 282, 1117–1121 (1998). 22. Mason, A., Nicoll, A. & Stratford, K. Synaptic transmission between individual pyramidal neurons of the rat visual cortex in vitro. J. Neurosci. 11, 72–84 (1991). 23. Nicolelis, M. A. L., Baccala, L. A., Lin, R. C. S. & Chapin, J. K. Sensorimotor encoding by synchronous neural ensemble activity at multiple levels of the somatosensory system. Science 268, 1353–1358 (1995). 24. Stuart, G. & Sakmann, B. Amplification of EPSPs by axosomatic sodium channels in neocortical pyramidal neurons. Neuron 15, 1065–1076 (1995).

25. Connors, B. W., Malenka, R. C. & Silva, L. R. Two inhibitory postsynaptic potentials, and GABAA receptor-mediated responses in neocortex of rat and cat. J. Physiol. (Lond.) 406, 443–468 (1988). 26. Dobrunz, L. E. & Stevens, C. F. Heterogeneity of release probability, facilitation, and depletion at central synapses. Neuron 18, 995–1008 (1997). 27. Raastad, M., Storm, J. F. & Andersen, P. Putative single quantum and single ﬁbre excitatory postsynaptic currents show similar amplitude range and variability in rat hippocampal slices. Eur. J. Neurosci. 4, 113–117 (1992). Figure 1 BP 897 is a potent and partial D3 receptor agonist and a weak D2 receptor 28. Volgushev, M., Voronin, L. L., Chistiakova, M., Artola, A. & Singer, W. All-or-none excitatory antagonist. a, Chemical structure of BP 897. b, Inhibition by BP 897 of postsynaptic potentials in the rat visual cortex. Eur. J. Neurosci. 7, 1751–1760 (1995). 125 29. Cash, S. & Yuste, R. Input summation by cultured pyramidal neurons is linear and position- [ I]iodosulpride binding to recombinant D2 and D3 receptors. c, In NG 108-15 independent. J. Neurosci. 18, 10–15 (1998). cells expressing the D3 receptor, BP 897 activated mitogenesis and this response

Acknowledgements. This work was supported by a Wellcome Trust Advanced Training Fellowship to was antagonized by the preferential D3 receptor antagonist nafadotride (NAF, G.T.F. and a grant to B.W.C. from NIH. We thank S. Patrick for technical assistance with histology; 1 ␮M). BP 897 also partially antagonized the response induced by quinpirole A. Akima for discussions on the optimal slicing angle; D. Pinto for checking the mathematics; and M. Bear, J. Gibson, M. Beierlein and D. Pinto for helpful comments on a earlier version of the manuscript. G.T.F. (10 nM). d, In CHO cells expressing the D2 receptor, BP 897, while having no effect thanks R. Jones and D. Smith for encouragement with the current project. by itself, reversibly antagonized quinpirole-induced mitogenesis. The Schild plot

Correspondence and requests for materials should be addressed to B.W.C. (e-mail: [email protected]). (inset) allowed us to calculate a pKi value for BP 897 of 7.29 (Ki ¼ 51 nM).

Figure 2 Agonistic effects of BP 897 on rotations in 6-OHDA-lesioned rats. a, Contralateral rotations induced by BP 897 (1 mg kg−1) alone or in combination with SKF 38393 (SKF,10 mg kg−1), before (open bars) and after (ﬁlled bars) repeated twice daily administration of levodopa for 10 days (F½5; 45ÿ¼6:35, P ¼ 0:0002 by analysis of variance; * P Ͻ 0:001 vs. SKF after L-DOPA, # P Ͻ 0:05 vs. SKF þ BP 897

before L-DOPA). Results are means Ϯ s:e:m: of data obtained from 6–13 animals. b, BP 897 dose-dependently increased contralateral rotations induced by SKF 38393 (5 mg kg−1). Squares, saline; triangles, 2.5 mg kg−1 SKF; circles, 5mgkg−1 SKF (F½3; 39ÿ¼3:41, P ¼ 0:027; * P Ͻ 0:05 vs. SKF). c, The effect of BP 897 (1 mg kg−1) on rotations induced by SKF (5 mg kg−1) was antagonized by Figure 3 Antagonistic effects of BP 897 on c-fos gene expression in the islands of nafadotride (1 mg kg−1)(½F½2; 18ÿ¼9:20, P ¼ 0:0018 by analysis of variance, Calleja. a, Effect of BP 897 (0.3–1 mg kg−1), nafadotride (NAF,1 mg kg−1), SKF 39393 − − − * P Ͻ 0:01 vs. SKF and # P Ͻ 0:01 vs. SKF þ BP 897). (SKF,10 mg kg 1) or the combination of SKF (10 mg kg 1) and BP 897 (1 mg kg 1)on the expression of c-fos mRNA in rats. * P Ͻ 0:05, ** P Ͻ 0:01, *** P Ͻ 0:001 vs. receptor agonist that has no agonist activity at the closely related D2 saline; # P Ͻ 0:05 vs. SKF by analysis of variance. b, In situ hybridization signals of 13 receptor . By screening a series of newly designed molecules for c-fos mRNA in wild-type (+/+) or D3 receptor knockout mice (−/−) treated with SKF −1 −1 their differential affinity at recombinant human dopamine D2 and 38393 (10 mg kg ) or the combination of SKF and BP 897 (1 mg kg ). c, Pictures as D3 receptors, we selected BP 897 (Fig. 1a), which displayed a high in b were analysed from 5–10 animals. Two-way analysis of variance indicates a : Ϯ : Ϯ : : : ; : : affinity at the D3 receptor (Ki ¼ 0 92 0 2 nM, mean s e m , significant effect of genotype (F½1 55ÿ¼4 17, P ¼ 0 04) and a significant treat- n ¼ 12) and a 70 times lower affinity at the D2 receptor ment × genotype interaction (F½3; 55ÿ¼3:48, P ¼ 0:02). * P Ͻ 0:001 vs. all other Ϯ : Ϯ : : : (Ki ¼ 61 0 2 nM, mean s e m , n ¼ 8; Fig. 1b). BP 897 also groups. ␮ displayed low affinities at D1 and D4 receptors (Ki ¼ 3 and 0.3 M, ␣ ␣ respectively), as well as at 1 and 2 adrenergic receptors (Ki ¼ 60 We have assessed further the agonist/antagonist potency of BP and 83 nM, respectively), 5HT1A and 5HT7 receptors (Ki ¼ 84 and 897 in vivo on two D3-receptor-mediated responses. In hemipar- Ͼ ␮ 345 nM, respectively), and negligible affinities (Ki 1 M) at kinsonian rats, rotations contralateral to the lesion are elicited by muscarinic, histamine and opiate receptors (not shown). stimulation of hypersensitive postsynaptic dopamine receptors in In NG 108-15 cells expressing the human D3 receptor, BP 897 the dopamine-denervated striatum. Repeated pretreatment with inhibited forskolin-induced cyclic AMP accumulation with an EC50 levodopa (L-DOPA) sensitizes the animals to this drug, an effect of 1:0 Ϯ 0:3 nM, although to a maximal extent ( Ϫ 59% Ϯ 4%) attributable to the induction of D3 receptor gene expression in 16 lower than that elicited by dopamine or the full agonist quinpirole neurons containing D1 receptors . Thus, monitoring rotations in ( Ϫ 82 Ϯ 6%). This response was completely reversed by 1 ␮M these animals provides a reliable behavioural model of D3-receptor haloperidol, a D2/D3 receptor antagonist (data not shown). In stimulation. In such animals, BP 897 potentiated rotations elicited these cells, BP 897 also increased mitogenesis, another D3-recep- by the D1-receptor-selective agonist SKF 38393 in a dose-dependent 14 Ϯ tor-mediated response (EC50 ¼ 3 1 nM), which at maximum manner (Fig. 2a, b). This potentiation did not occur before L-DOPA was only 55% of that elicited by quinpirole and was antagonized by treatment, that is, before induction of D3-receptor expression, and 15 nafadotride, a preferential D3 receptor antagonist (Fig. 1c). BP 897 was abolished by co-treatment with nafadotride used at a dose 16 also antagonized quinpirole-induced mitogenesis, up to the compatible with a selective D3-receptor blockade (Fig. 2a, c). maximal level reached with BP 897 alone (Fig. 1c). In a second in vivo model, we measured c-fos expression in In contrast, in cells expressing the D2 receptor, BP 897 (1 ␮M), granule cells of the islands of Calleja, co-expressing D1 and D3 unlike quinpirole, did not inhibit cyclic AMP accumulation or receptors, which mediate tonic but opposite effects on c-fos expres- trigger mitogenesis; it reversibly antagonized quinpirole-induced sion. For instance, D1 receptor agonists and antagonists enhance mitogenesis, but only at concentrations much larger than those and impair c-fos expression, respectively, whereas corresponding D3 required to stimulate the D3 receptor (Fig. 1d). Hence, BP 897 is a receptor agents (also active at the D2 receptor) exert the opposite 17 selective, potent, but partial (intrinsic activity ϳ0.6) D3 receptor actions . In the islands of Calleja of rats, BP 897 enhanced c-fos −1 agonist and a weak D2 receptor antagonist in vitro. messenger RNA with an ED50 of ϳ0.3 mg kg , an effect similar in Initial in vivo binding experiments in mouse striatum using direction and amplitude to that of nafadotride (Fig. 3a); in addition, 3 17 [ H]N-propylnorapomorphine showed that D2-receptor occu- in contrast to D2/D3 receptor agonists , BP 897 potentiated the −1 pancy is achieved by BP 897 with an ED50 of ϳ15 mg kg and response to a D1 receptor agonist (Fig. 3a). maximally maintained for 2 h. The absence of dopamine agonist We used mice bearing mutations invalidating the D3-receptor 18 activity at the D2 receptor was shown by the inability of BP 897 to gene to assess the specificity of the effects of BP 897 on c-fos gene induce stereotypies (J. Costentin, personal communication). BP 897 expression. In mice (wild-type or mutant), BP 897 had no marked produced catalepsy in rats, a classical response to D2-receptor effects on c-fos mRNA (Fig. 3c). However, BP 897 strongly poten- −1 blockade, with an ED50 of ϳ12 mg kg . Taking into account the tiated the effects of the D1 receptor agonist in wild-type mice respective affinities and potencies of BP 897 at D2 and D3 receptors, (Fig. 3b, c), an effect that was not seen in mice lacking the D3 this indicated that D3-receptor occupancy was to be expected at an receptor. BP 897 shows extreme region-related specificity: it −1 ED50 below 0.5 mg kg . increased c-fos mRNA levels in wild-type mice only in the islands

Figure 4 BP 897 inhibits cue-controlled cocaine-seeking but has no reinforcing (F½3; 27ÿ¼0:21). d, Effects of replacing cocaine with BP 897 (0.5 mg ml−1,100␮l effects. a, Cumulative response records for a representative rat receiving a pre- per infusion: 0.05 mg i.v.) or saline (100 ␮l per infusion) on i.v. cocaine self- injection of saline (top) or BP 897 0.5 mg kg−1 (bottom) on responding during the administration. After 2 days of replacement, responding on the active lever in all first or second interval of a cocaine self-administration session under a subjects had decreased to levels not significantly different from the inactive lever, FI15(FR10:S) second-order schedule of reinforcement (see Methods). The that is, self-administration behaviour had extinguished. There was no difference arrows indicate contingent,1 s presentations of the CS+, following the completion between the effects of replacement with BP 897 or saline. Re-substituting cocaine of each fixed ratio of 10 responses on the active lever. In the second interval, but for BP 897 or saline resulted in the prompt reinstatement of drug self-administra- not the first, responding is under the influence of the self-administered cocaine. tion in both groups. Analysis of variance revealed no effect of treatment (replace- Pretreatment with BP 897 markedly reduces responding in the first, but not the ment with BP 897 or saline, F½1; 10ÿ¼0:37, P Ͼ 0:5); a main effect of lever second, interval. b, Dose-related effects of BP 897 on cocaine-seeking behaviour (F½1; 10ÿ¼219:12, P Ͻ 0:0001) reflecting more responses on the active than the evaluated as in a (n ¼ 11). BP 897 decreased responding for cocaine during the inactive lever; a main effect of day (Fð15; 150Þ¼18:91, P Ͻ 0:0001); and a first, cocaine-free, interval of a second-order schedule (F½3; 30ÿ¼6:45, P ¼ 0:01 lever ϫ day interaction (F½15; 150ÿ¼50:90, P Ͻ 0:001) that reflected the decrease by analysis of variance, * P Ͻ 0:05 and ** P Ͻ 0:01 vs. saline), but not during the in responding on the active lever in both groups following BP 897 or saline second interval (F½3; 30ÿ¼0:93). c, BP 897 has no significant effect on the self- substitution on day 3 and on cocaine re-substitution on day 14. There were no administration of cocaine under a continuous reinforcement schedule significant interactions of treatment with either lever or day. of Calleja. ‘primed’) and depends critically upon the contingent presentation Thus, in vivo, BP 897 increases D1-receptor-mediated responses of the cocaine cue; its omission greatly reduces responding, even by acting as either an agonist or an antagonist, depending upon the though intravenous cocaine is still available7. Administration of BP response considered, consistent with its partial agonist properties in 897 before testing reduced cocaine-seeking behaviour before the vitro. This dual activity may be explained by various models of first infusion of cocaine, in a dose-dependent manner, at doses receptor activation, in which the efficacy of a ligand depends not similar to those at which BP 897 produced its responses on rotations only on its intrinsic property, but also on receptor–G-protein and c-fos expression (Fig. 4a, bottom, and 4b). Importantly, the coupling efficiency19,20, which may vary between tissues. The direc- effect of BP 897 occurred without disrupting the pattern of tion of the response may also depend on the level of the endogenous responding and without increasing the latency to initiate respond- − transmitter. Accordingly, BP 897 acts as a receptor agonist on ing at any dose up to 1 mg kg 1 (data not shown), indicating the rotations elicited in the dopamine-depleted brain and as a receptor absence of any nonspecific, for example motor, effects that might antagonist on c-fos expression maintained by a dopaminergic tone. have disrupted performance. Preliminary data indicate that the Taken in the context of the importance of the D3 receptor in reduction in cocaine-seeking behaviour following administration mediating the reinforcing effects of cocaine8,21, the theoretical use of of BP 897 is prevented by pretreatment with nafadotride (M.P., partial dopamine receptor agonists as treatments for addictive J.-C.S., P.S. and B.J.E., unpublished observations). behaviour22 and the impact of drug-associated cues on drug craving Under the second-order schedule used here, self-administered and relapse in humans2,5,23,24, we have investigated the effects of BP cocaine increases subsequent drug-seeking behaviour7, but BP 897 897 in a model of cocaine-seeking behaviour in which drug cues had no effect on this cocaine-induced increase in responding in the have a demonstrably important controlling role7,24,25. second interval (Fig. 4a, b), at a time when BP 897 is still active Rats were trained to self-administer cocaine under a second- according to our in vivo binding experiments (see above). These 7,24,25 order schedule of reinforcement , responses being reinforced by results indicate a dissociation in the effects of a partial D3 receptor presentation of a light stimulus paired with each cocaine infusion. agonist on drug-cue-controlled cocaine-seeking behaviour and on The light-conditioned stimulus, acting as a conditioned reinforcer, the reinforcing, including response-rate-increasing, effects of maintains the lever-pressing response for a protracted period cocaine, as the effects of BP 897 were only apparent in the first (usually a fixed interval of 15 min; FI15) before cocaine infusion (cocaine-unaffected) interval of a session. (Fig. 4a, top). This drug-seeking behaviour7 occurs in the absence of To investigate this dissociation more specifically, we first tested any pharmacological effects of cocaine (as the animals are never whether BP 897 pretreatment could modify intravenous cocaine

© 1999 Macmillan Magazines Ltd NATURE | VOL 400 | 22 JULY 1999 | www.nature.com 373 letters to nature self-administration under a continuous reinforcement schedule, hydroxydopamine (8 ␮gin4␮l of 0.05% ascorbic acid in saline) in the medial and found that it had no effect (Fig. 4c). Next, BP 897 (at an forebrain bundle16. Four weeks later, we measured rotations for 5 min, starting intravenous dose equivalent to the highest effective dose when given 35 min after administration of drugs after a 5-min period of habituation in intraperitoneally) or saline were substituted for cocaine to deter- vertical Perspex cylinders (30 cm diameter). Rotations were measured again mine whether the drug possessed reinforcing properties sufficient to after animals had received, twice a day for 10 days, levodopa (as L-DOPA- − − maintain drug-taking in animals with a daily history of intravenous methylester, 50 mg kg 1 i.p., in combination with benserazide, 12.5 mg kg 1 cocaine self-administration. Responding declined identically over i.p.). days in rats in which either BP 897 or saline had been substituted for In situ hybridization. Rats were habituated to handling and injections by cocaine (Fig. 4d). Thus, from three sets of independent observa- repeated intraperitoneal administration of saline, during four days before the tions, BP 897 appears to have no intrinsic reinforcing effects, but can experiments. They were killed 30 min after administration of the drugs and nevertheless reduce significantly conditioned cue-controlled their brains frozen in isopentane maintained at −30 ЊC. Slices (10 ␮m) were cocaine-seeking behaviour. hybridized with a 33P-labelled c-fos antisense RNA probe as described17.We The ability of BP 897 to reduce drug-seeking behaviour in the analysed autoradiograms with an image analyser, the islands of Calleja being absence of any intrinsic reinforcing properties, although unprece- visualized after superimposition of images obtained after hemalun counter- dented, is consistent with data indicating that dissociable neural staining. Similar experiments were performed on 8 ␮m slices from wild-type mechanisms underlie responding with conditioned reinforcement and D3 receptor knock-out mice (generations F3 and F4), created by S. Fuchs and responding for cocaine itself 6,25,26. However, these dissociable and D. Accili (Weizmann Institute, Rehovot, Israel)18 and initially bred in mechanisms clearly interact and projections from the basolateral France by W. Rostene and C. Betancur (INSERM U336, Paris). amygdala to the nucleus accumbens may provide the neuroanato- Cocaine self-administration under continuous and second-order sched- mical basis of this interaction, with the accumbens shell region ules of reinforcement. Male Listar hooded rats (Charles River) were 27 7 (where neurons express high D3 receptor levels ) providing a site implanted with indwelling intravenous catheters as described . Each rat was for dopaminergic modulation, for example by the D3-receptor- placed in an operant chamber where two retractable levers were available and selective partial agonist used here. The partial agonistic character trained to self-administer cocaine under a continuous reinforcement schedule and D3-receptor selectivity of BP 897 seem to be essential for its (a fixed ratio 1 of responding: FR1). Depression of the active lever resulted in an − dissociated actions, as the full and non-selective agonist quinpirole, infusion of 100 ␮l of a solution of cocaine hydrochloride (2.5 mg ml 1, in addition to cocaine itself, also reduces cue-controlled cocaine- dissolved in 0.9% saline; 0.25 mg per infusion), extinction of the house light, seeking behaviour (A. Markou, M. Arroyo & B.J.E., unpublished illumination of a stimulus light (the conditioned stimulus; CS+) above the results), whereas, in marked contrast to BP 897, it possesses intrinsic active lever for 20 s, and then retraction of both levers from the chamber. rewarding properties, being both self-administered and also able Depression of the inactive lever had no programmed consequence and was also to enhance the reinforcing effects of cocaine8. Furthermore, followed by retraction of both levers. Active and inactive levers were counter- PNU99194A, a D3 receptor antagonist, has no effect on drug- balanced left and right across subjects. Cocaine self-administration under seeking behaviour (M.P., M. Arroyo & B.J.E., unpublished results). continuous reinforcement was used to evaluate the effects of BP 897 (adminis- The absence of effects of BP 897 on cocaine reinforcement may also tered i.p. 30 min before the session) on cocaine self-administration and when be attributable to its partial agonism, as full agonists and antagonists substituted for cocaine. have opposite effects9. Partial agonism of BP 897 may confer a When rats had acquired a stable pattern of responding (three days of ‘buffering’ capacity, allowing it to oppose, as an antagonist, a chain constant responding Ϯ10%), we introduced a second-order schedule of the of neural events initiated by a conditioned increase in dopamine type FRx(FRy:S), as described7. Progressively, the value of x was increased to 10 release28,29, while maintaining, as an agonist, a moderate degree of (that is, each active lever response produced a CS+, and 10 such responses were D3-receptor stimulation. Finally, as D1 receptor-mediated mechanisms followed by a drug infusion and a 1 s illumination of the light CS+). In a second seem to be important in a model of reinstatement of cocaine self- stage, the value of y was also increased progressively to 10, so that 10 responses administration after extinction30, BP 897 may exert part of its actions were required to produce the CS+, and 10 such units (100 responses; 10 CS+ by potentiating some D1-receptor-mediated effects in the restricted presentations) resulted in a cocaine infusion (and a 20 s presentation of the 17 subset of neurons in which D1 and D3 receptors co-localize . Ⅺ CS+). Ultimately, the form of the second-order schedule in operation was an ...... overall fixed interval (FI) 15 min (i.e. FI15(FR10:S)) in which every FR10 Methods response resulted in presentation of the CS+ as before, but the cocaine infusion Drug screening. The synthesis of BP 897 will be described elsewhere. We was delivered only on completion of the first FR10 responses after the 15 min performed binding studies at D1,D2,D3 and D4 receptors using Chinese fixed interval had timed out. Responding for the first cocaine infusion in a daily hamster ovary (CHO) cells expressing recombinant human receptors and session, evaluated 30 min after intraperitoneal administration of saline or BP 3 125 3 [ H]SCH23390 (for D1), [ I]iodosulpride (for D2 and D3) and [ H]spiperone 897, therefore occurred in the absence of any effects of cocaine and represents 27 (for D4) as described . Additional receptor screening was performed by drug-cue-controlled drug-seeking behaviour. Following the first cocaine Panlabs Taiwan Ltd. infusion, responding in a second 15 min interval is affected by the rewarding Measurement of cAMP and mitogenesis. We used NG 108-15 cells and other (for example, response-rate-enhancing) effects of cocaine. 14 expressing the D3 receptor or CHO cells expressing the D2 receptor . cAMP In substitution experiments, rats first acquired i.v. cocaine self-administra- levels were measured after a 10 min stimulation by forskolin (1 ␮M) and drugs tion under continuous reinforcement. When stable baseline rates of cocaine in the presence of isobutylmethylxanthine (10 ␮M) using the Rianen self-administration were established (3 consecutive days with same number of − [125I]cAMP radioimmunoassay kit (DuPont NEN). Mitogenesis was measured responses Ϯ10%), BP 897 (0.5 mg kg 1) or saline were substituted for cocaine by incorporation of [3H]thymidine during a 2 h pulse following an overnight and self-administration was allowed to continue for a further 10 days. At this incubation with drugs14. point, cocaine at the training dose was re-substituted for BP 897 or saline. Measurement of striatal dopamine receptor occupancy and catalepsy. Received 4 February; accepted 4 May 1999. We measured striatal dopamine receptor occupancy in mice killed 15 min after 3 1. Wikler, A. Dynamics of drug dependence. Arch. Gen. Psychiatry 28, 611–6161 (1973). an intravenous (i.v.) injection of [ H]N-propyl-norapomorphine and 30 min 2. O’Brien, C. P., Childress, A. R., McMellan, A. T. & Ehrman, R. A. A learning model of addiction. Res. −1 after an intraperitoneal (i.p.) injection of saline or BP 897 (1–30 mg kg ). For Publ. Assoc. Res. Nerv. Ment. Dis. 70, 157–177 (1992). assessment of catelepsy, rats were placed, 30 min after an i.p. injection of BP 897 3. O’Brien, C. P.& McMellan, A. T. Myths about the treatment of addiction. Lancet 347, 237–240 (1996). −1 4. Grech, D. M., Spealman, R. D. & Bergman, J. Self-administration of D1 receptor agonists by squirrel (1–30 mg kg ), in a device in which the forepaws were held 10 cm above the monkeys. Psychopharmacology 125, 97–104 (1996). floor with a horizontal bar. Dose–response curves were generated from the 5. de Wit, H. & Stewart, J. Reinstatement of cocaine-reinforced responding in the rat. Psychopharmacology 75, 134–143 (1981). percentage of animals holding the position for at least 5 s. 6. Meil, W. M. & See, R. E. Lesions of the basolateral amygdala abolish the ability of drug associated cues Rotational behaviour. Anaesthetized Wistar male rats were infused with to reinstate responding during withdrawal from self-administered cocaine. Behav. Brain Res. 87, 139–

148 (1997). physiological responses to two classes of olfactory stimuli. These 7. Arroyo, M., Markou, A., Robbins, T. W. & Everitt, B. J. Acquisition, maintenance and reinstatement of intravenous cocaine self-administration under a second-order schedule of reinforcement in rats: rhythms are observed in wild-type flies during light–dark cycles effects of conditioned cues and continuous access to cocaine. Psychopharmacology 140, 331–344 and in constant darkness, but are abolished in per or tim null- (1999). mutant flies (per01 and tim01) which lack rhythms in adult 8. Caine, S. B. & Koob, G. F. Modulation of cocaine self-administration in the rat through D3 dopamine receptors. Science 260, 1814–1816 (1993). emergence and locomotor behaviour. Olfactory rhythms are 9. Caine, S. B. et al.D3 receptor test in vitro predicts decreased cocaine self-administration in rats. also abolished in the per 7.2:2 transgenic line in which per Neuroreport 8, 2373–2377 (1997). 4 10. Staley, J. K. & Mash, D. C. Adaptive increase in D3 dopamine receptors in the brain reward circuits of expression is restricted to the lateral neurons of the optic lobe . human cocaine fatalities. J. Neurosci. 16, 6100–6106 (1996). Because per 7.2:2 flies do not express per in peripheral oscillators, 11. Duaux, E. et al. Homozygosity at the dopamine D3 receptor gene is associated with opioid dependence. Mol. Psychiatry 3, 333–336 (1998). our results provide evidence that peripheral circadian oscillators 12. Sokoloff, P., Giros, B., Martres, M.-P., Bouthenet, M.-L. & Schwartz, J.-C. Molecular cloning and are necessary for circadian rhythms in olfactory responses. As characterization of a novel dopamine receptor (D3) as a target for neuroleptics. Nature 347, 146–151 olfaction is essential for food acquisition, social interactions and (1990).

13. Xu, M., Koeltzow, T. E., Cooper, D. C., Tonegawa, S. & White, F. J. Dopamine D3 receptor mutant and predator avoidance in many animals, circadian regulation of wild-type mice exhibit identical responses to putative D3 receptor-selective agonists and antagonists. olfactory systems could have profound effects on the behaviour Synapse 31, 210–215 (1999). of organisms that rely on this sensory modality. 14. Pilon, C. et al. Functional coupling of the human dopamine D3 receptor in a transfected NG 108-15 neuroblastoma-glioma hybrid cell line. Eur. J. Pharmacol. 268, 129–139 (1994). We determined olfactory rhythms by measuring electroantenno- 5,6 15. Sautel, F. et al. Nafadotride, a potent preferential dopamine D3 receptor antagonist, activates gram (EAG) responses to odorants . Typical EAG responses to the locomotion in rodents. J. Pharmacol. Exp. Ther. 275, 1239–1246 (1995). food odorant ethyl acetate (Fig. 1) were robust, dose-dependent and 16. Bordet, R. et al. Induction of dopamine D3 receptor expression as a mechanism of behavioral sensitization to levodopa. Proc. Natl Acad. Sci. USA 94, 3363–3367 (1997). highly reproducible, both within and across preparations. The EAG 17. Ridray, S. et al. Coexpression of dopamine D1 and D3 receptors in rat ventral striatum: opposite and synergistic functional interactions. Eur. J. Neurosci. 10, 1676–1686 (1998). responses did not saturate over the range of ethyl acetate concen-

18. Accili, D. et al. A targeted mutation of the D3 receptor gene is associated with hyperactivity in mice. trations used in these experiments (Fig. 1b), as described in studies Proc. Natl Acad. Sci. USA 93, 1945–1949 (1996). of Drosophila by others7. Some desensitization was observed during 19. Costa, T., Ogino, Y., Munson, P. J., Oraran, H. O. & Rodbard, D. Drug efficacy at guanine nucleotide- binding regulatory protein linked receptors: thermodynamic interpretation of negative antagonism odorant application (over a period of several seconds; Fig. 1c), and of receptor activity in the absence of ligand. Mol. Pharmacol. 41, 549–560 (1992). but recovery from desensitization was invariably complete when 20. Leff, P. The two-state model of receptor activation. Trends Pharmacol. Sci. 16, 89–97 (1995). 21. Caine, S. B. & Koob, G. F. Pretreatment with the dopamine agonist 7-OH-DPATshifts the cocaine self- odorant pulses were delivered at 30 s intervals (Fig. 1d). Because it administration dose-effect function to the left under different schedules in the rat. Behav. Pharmacol. was not possible to measure EAG amplitudes continuously from a 6, 333–347 (1995). single fly for long periods of time, we obtained mean EAG 22. Pulvirenti, L. & Koob, G. F. Dopamine agonists, partial agonists and psychostimulant addiction. Trends Pharmacol. Sci. 15, 374–379 (1994). amplitudes from groups of animals at different times of day. 23. Grant, S. et al. Activation of memory circuits during cue-elicited cocaine craving. Proc. Natl Acad. Sci. Complete dilution–response curves were obtained for every animal. USA 93, 12040–12045 (1996). 24. Weissenborn,R., Robbins, T. W. & Everitt, B. J. Effects of medial prefrontal or anterior cingulate cortex To determine whether diurnal rhythms were present, recordings lesions on responding for cocaine under fixed-ratio and second-order schedules of reinforcement in were made from flies maintained in light–dark (LD) 12:12 cycles. rats. Psychopharmacology 134, 242–257 (1997). Recordings in the dark were made under infrared optics. Figure 2a 25. Whitelaw, R. B., Markou, A., Robbins, T. W. & Everitt, B. J. Excitotoxic lesions of the basolateral amygdala impair the acquisition of cocaine-seeking behaviour under a second-order schedule of shows that mean EAG responses to ethyl acetate were elevated reinforcement. Psychopharmacology 127, 213–224 (1996). during the middle of the night in wild-type flies during LD cycles, 26. Everitt, B. J. & Robbins, T. W. in The Amygdala: Neurological Aspects of Emotion, Memory and Mental 01 01 Dysfunction (ed. Aggleton, J.) 401–429 (Wiley, New York, 1992). but not in per or tim mutant flies. Data shown in this and

27. Le´vesque, D. et al. Identification, characterization and localization of the dopamine D3 receptor in rat subsequent figures are based on mean responses to the highest brain using 7-[3H]-hydroxy-N,N di-n-propyl-2-aminotetralin. Proc. Natl Acad. Sci. USA 89, 8155– concentration of ethyl acetate tested (a dilution of 1:104). However, 8159 (1992). 28. Fontana, D. J., Post, R. M. & Pert, A. Conditioned increases in mesolimbic dopamine overflow by similar temporal patterns were observed at lower odorant concen- stimuli associated with cocaine. Brain Res. 629, 31–39 (1993). trations (data not shown). Analyses of these data by two-way 29. Di Ciano, P., Blaha, C D. & Phillips, A. G. Conditioned changes in dopamine oxidation currents in the Ͻ : nucleus accumbens of rats by stimuli paired with self-administration or yoked-administration of d- ANOVA indicate statistically significant (P 0 0001) effects of amphetamine. Eur. J. Neurosci. 10, 1121–1127 (1998). time of day, genotype and their interaction. The significant 30. Self, D. W., Barnhart, W. J., Lehman, D. A. & Nestler, E. J. Opposite modulation of cocaine-seeking interaction effect indicates that genotype alters the daily rhythm behavior by D1-like and D2-like dopamine-receptor agonists. Science 271, 1586–1589 (1996). in mean EAG amplitude. Post hoc analyses indicate that mean EAG Acknowledgements. We thank C. Pilon for technical assistance and S. Fuchs, W. Rostene and C. Betancur responses observed at Zeitgeber time (ZT) 13–17 were significantly for providing the D3-receptor-deficient mice. The work was supported by a Biomed II Programme grant Ͻ : to P.S., C.G.W. and B.J.E., a National Institute on Drug Abuse grant to P.S. and C.G.W. and an MRC (P 0 005) greater than those observed at ZT21, ZT1, ZT5 or ZT9 Programme grant to B.J.E. in wild-type flies. In contrast, we observed no statistically significant Correspondence and requests for materials should be addressed to P.S.(e-mail: [email protected]) or changes in mean EAG responses to ethyl acetate at different times of B.J.E. (e-mail: [email protected]). day in per01 or tim01 mutant flies in LD (Fig. 2a). To determine whether these rhythms are endogenously controlled, we made EAG recordings from flies on day 2 of constant darkness (DD) after 3 days’ entrainment in LD. A similar temporal Circadian rhythms in pattern of olfactory sensitivity was observed, indicating circadian control (Fig. 2b). The overall effects of time of day, genotype and olfactory responses of their interactions were statistically significant (P Ͻ 0:0001). Mean EAG responses in wild-type flies increased gradually between Drosophila melanogaster circadian time (CT)9 and CT13, and peaked during the middle of the subjective night (CT17). EAG responses returned to basal levels Balaji Krishnan, Stuart E. Dryer & Paul E. Hardin at CT21 and CT1. Mean EAG responses observed at CT13–17 were Department of Biology and Biochemistry and Biological Clocks Program, significantly (P Ͻ 0:005) greater than those observed at CT21, CT1, University of Houston, Houston, Texas 77204-5513, USA CT5 or CT9. In contrast, there was no significant change in mean ...... EAG responses as a function of the time of day in per01 or tim01 The core mechanism of circadian timekeeping in arthropods and mutant flies in DD (Fig. 2b). A similar pattern was observed in vertebrates consists of feedback loops involving several clock olfactory responses to benzaldehyde (Fig. 2c), an odorant that, in genes, including period (per) and timeless (tim)1,2. In the fruitfly contrast to ethyl acetate, causes behavioural avoidance in Drosophila, circadian oscillations in per expression occur in Drosophila8. The mean amplitude of EAG responses to benzalde- chemosensory cells of the antennae, even when the antennae are hyde (dilution of 1:103) was smaller than those evoked by ethyl excised and maintained in isolated organ culture3. Here we acetate, as described by others7, possibly owing to the lower vapour demonstrate a robust circadian rhythm in Drosophila in electro- pressure of this odorant. However, the phase of the rhythm was

7. Griswold, M. D. Interactions between germ cells and Sertoli cells in the testis. Biol. Reprod. 52, 211± 216 (1995). 8. Lai, C. & Lemke, G. An extended family of protein-tyrosine kinase genes differentially expressed in the vertebrate nervous system. Neuron 6, 691±704 (1991). 9. Mark, M. R. et al. RSE, a novel receptor-type tyrosine kinase with homology to Axl/Ufo, is expressed at high levels in the brain. J. Biol. Chem. 269, 10720±10728 (1994). 10. Taylor, I. C., Roy, S., Yaswen, P., Stampfer, M. R. & Varmus, H. E. Mouse mammary tumors express elevated levels of RNA encoding the murine homology of SKY, a putative receptor tyrosine kinase. J. Biol. Chem. 270, 6872±6880 (1995). 11. Rescigno, J., Mansukhani, A. & Basilico, C. A putative receptor tyrosine kinase with unique structural topology. Oncogene 6, 1909±1913 (1991). 12. Janssen, J. W. et al. A novel putative tyrosine kinase receptor with oncogenic potential. Oncogene 6, 2113±2120 (1991). 13. Jia, R. & Hanafusa, H. The proto-oncogene of v-eyk (v-ryk) is a novel receptor-type protein tyrosine kianse with extracellular Ig/GN-III domains. J. Biol. chem. 269, 1839±1844 (1994). 14. Stitt, T. N. et al. The anticoagulation factor protein S and its relative, Gas6, are ligands for the Tyro 3/ Axl family of receptor tyrosine kinases. Cell 80, 661±670 (1995). 15. Godowski, P.J. et al. Reevaluation of the roles of protein S and Gas6 as ligands for the receptor tyrosine kinase Rse/Tyro 3. Cell 82, 355±358 (1995). 16. Chen, J., Carey, K. & Godowski, P. J. Identi®cation of Gas6 as a ligand for Mer, a neural cell adhesion molecule related receptor tyrosine kinase implicated in cellular transformation. Oncogene 14, 2033± 2039 (1997). 17. Nagata, K. et al. Identi®cation of the product of growth arrest-speci®c gene 6 as a common ligand for Axl, Sky, and Mer receptor tyrosine kinases. J. Biol. Chem. 271, 30022±30027 (1996). 18. Joseph, D. R. Sequence and functional relationships between androgen-binding protein/sex hor- mone-binding globulin and its homologs protein S, Gas6, laminin, and agrin. Steroids 62, 578±588 (1997). 19. Capecchi, M. R. Altering the genome by homologous recombination. Science 244, 1288±1292 (1989). 20. Camenisch, T. D., Koller, B. H., Earp, H. S. & Matsushima, G. K. A novel receptor tyrosine kinase, Mer, inhibits TNF-a production and lipopolysaccaride-induced endotoxic shock. J. Immunol. 162, 3498± 3503 (1999). Figure 1 DNA end-binding and end-to-end interactions promoted by hRad52. 21. Smiley, S. T., Stitt, T. N. & Grusby, M. J. Cross-linking of protein S bound to lymphocytes promotes Complexes formed between hRad52 and linear duplex DNA containing single- aggregation and inhibits proliferation. Cell. Immunol. 181, 120±126 (1997). stranded tails were visualized by electron microscopy. a, b and f,39 single- 22. Leblond, C. P. & Clemont, Y. De®nition of the stages of the cycle of the seminiferous epithelium in the rat. Ann. N. Y. Acad. Sci. 55, 548±573 (1952). stranded (ss) DNA tails. c, d and e,59 ssDNA tails. Inset in c shows a close-up 23. HenrikseÂn, K., Hakovirta, H. & Parvinen, M. Testosterone inhibits and induces apoptosis in rat view of an end-binding complex in which hRad52 rings are evident. Reactions seminiferous tubules in a stage-speci®c manner. Endocrinology 136, 3285±3291 (1995). 24. Sun, Y.-T, Wreford, N. G., Robertson, D. M. & De Kretser, D. M. Quantitative cytological studies of (20 ml) contained 10 mM(a, d) or 4.7 mM(b, c, e, f) tailed duplex DNA in 20 mM spermatogenesis in intact and hypophysectomized rats: Identi®cation of androgen-dependent stages. triethanolamine-HCl (pH 7.5). After 5 min at 37 8C, hRad52 was added to 0.16 mM Endocrinology 127, 1215±1223 (1991). (a, d) or 0.08 mM(b, c, e, f) and incubation was continued for 15 min. 25. Wong, V. & Russell, L. D. Three-dimensional reconstruction of a rat stage V Sertoli cell: I. Methods, basic con®guration, and dimensions. Am. J. Anat. 167, 143±161 (1983). 26. Yomogida,K. et al. Developmental stage- and spermatogenic cycle-speci®c expression of transcription factor GATA-1 in mouse Sertoli cells. Development 120, 1759±1766 (1994). 2 27. Lewis, P. A., Crosier, K. E., Wood, C. R. & Crosier, P. S. Analysis of the murine Dtk gene identi®es primarily by Ku-dependent non-homologous end-joining . The conservation of genomic structure within a new receptor tyrosine kinase subfamily. Genomics 31, 13± contribution of homologous recombination to vertebrate DSB 19 (1996). repair, however, is important3,4. Biochemical studies indicate that 28. Biesecker, L. G., Giannola, D. M. & Emerson, S. G. Identi®cation of alternative exons, including a 5 novel exon, in the tyrosine kinase receptor gene Etk2/tyro3 that explains differences in 59 cDNA Ku binds to DNA ends and facilitates end-joining . Here we show sequences. Oncogene 10, 2239±2242 (1995). that human Rad52, like Ku, binds directly to DSBs, protects them 29. Schulz, A. S., Schleithoff, L., Faust, M., Bartram, C. R. & Janssen, J. W. G. The genomic structure of the human UFO receptor. Oncogene 8, 509±513 (1993). from exonuclease attack and facilitates end-to-end interactions. 30. Bellosta, P., Zhang, Q., Goff, S. P. & Basilico, C. Signaling through the ARK tyrosine kinase receptor A model for repair is proposed in which either Ku or Rad52 binds protects from apoptosis in the absence of growth stimulation. Oncogene 15, 2387±2397 (1997). the DSB. Ku directs DSBs into the non-homologous end-joining Acknowledgements. This work was supported by grants from the NIH (to G.L., S.P.G., H.S.E., G.K.M.), repair pathway, whereas Rad52 initiates repair by homologous and by a Markey fellowship to M.G., who was a student in the Neurosciences graduate program at the recombination. Ku and Rad52, therefore, direct entry into alter- University of California, San Diego. S.P.G. is an Investigator of the Howard Hughes Medical Institute. We thank A. Prieto for advice and for Gas6 antibody, G. Yancopoulos for Gas6, protein S, and Axl cDNA native pathways for the repair of DNA breaks. probes, and D. OrtunÄo, P. Burrola, and D. Baynes for technical assistance. In lower eukaryotes, double-strand breaks in meiotic and mitotic Correspondence and requests for materials should be addressed to G.L. (e-mail: [email protected]). cells are processed to form single-stranded tails, in reactions that involve the Mre11/Rad50/Xrs2 protein complex6,7. Similar processing events are thought to occur in higher cells, where homologues of Mre11, Rad50 and Xrs2 (p95) exist2. In yeast rad52 mutants, Binding of double-strand exonuclease degradation is more extensive than in wild-type cells7, indicating that the Rad52 protein may be involved in the protection breaks in DNA by or processing of the initial DSB. We therefore tested whether Rad52 acts at the early stages of double-strand-break repair by direct human Rad52 protein interaction with the DSB or partially resected DSB. Complexes were formed between human Rad52 protein (hRad52) and linear Eric Van Dyck*, Alicja Z. Stasiak², Andrzej Stasiak² duplex DNA, and were visualized by electron microscopy. Using & Stephen C. West* substrates containing 59 or 39 single-stranded tails (approximately * Imperial Cancer Research Fund, Clare Hall Laboratories, South Mimms, 300 nucleotides in length), we observed that hRad52 bound Herts EN6 3LD, UK preferentially to the ends of the DNA (Fig. 1). Molecules in which ² Laboratoire d'Analyse Ultrastructurale, UniversiteÂ de Lausanne, both ends of the linear DNA were bound by hRad52 are shown in CH-1015 Lausanne-Dorigny, Switzerland Fig. 1a±c...... We found that a large percentage of the linear DNA (,90%) Double-strand breaks (DSBs) in DNA are caused by ionizing present on each grid was held together by hRad52-mediated inter- radiation. These chromosomal breaks can kill the cell unless molecular interactions. Indeed, end-to-end associations between repaired ef®ciently, and inef®cient or inappropriate repair can DNA molecules, mediated by hRad52, resulted in the formation of lead to mutation, gene translocation and cancer1. Two proteins large DNA±protein networks (data not shown). However, away that participate in the repair of DSBs are Rad52 and Ku: in lower from the networks, isolated DNA molecules were observed in which eukaryotes such as yeast, DSBs are repaired by Rad52-dependent Rad52 protein clearly provided an intramolecular bridge between homologous recombination, whereas vertebrates repair DSBs two ends, resulting in their recircularization (Fig. 1d±f). When 46

a b a 50 Cohesive ends Blunt ends –– +hRad52 –– + hRad52 M – + + DNA ligase M – + + DNA ligase 40

P (%) 30 32 Linear multimers Linear multimers Linear dimer Linear dimer 20 no hRad52 Nicked circle Nicked circle 0.24 µM hRad52 Linear monomer Linear monomer µ Acid-soluble 0.48 M hRad52 10 µ ccc ccc 0.96 M hRad52

0 1234 12 34 0 102030 Time (min) Figure 2 Stimulation of DNA ligation by hRad52. a, b, Effect of hRad52 on the ligation of EcoRI- or SmaI-linearized DNA, respectively. M, size markers (1-kb b 100 ladder). See Methods for experimental details.

75 well-spread individual 39-tailed molecules were examined, we P (%) found that 34 were recircularized by hRad52, three contained 32 hRad52 on both ends of a linear molecule, and nine contained 50 hRad52 on one end only. Similarly, with 59-tailed DNA, examina- tion of 30 molecules revealed that 16 were recircularized, seven were bound at both ends and seven were bound at one end only. 25 Although the self-association of hRad52 has been demonstrated Acid-soluble by two-hybrid analysis8, our observations indicate a direct role for these protein±protein interactions in promoting both inter- and 0 intramolecular DNA contacts. hRad52 exhibited a clear preference 0 102030 for binding DNA ends, and only rare molecules were observed in Time (min) which hRad52 bound internally. In these instances, hRad52± hRad52 interactions occasionally led to `loop' formation (Fig. 1c). Figure 3 DNA-end protection by hRad52 protein. a, b, hRad52 was incubated with Electron microscopic studies of yeast9 and human10 Rad52 have uniformly 32P-labelled EcoRI-linearized DNA (30 mM), followed by the addition of revealed the formation of ring-shaped structures, both in solution either exonuclease III (a) or DNAse I (b). Samples were taken at the indicated and on DNA. Electron microscopic analyses of the end-binding times and analysed for the release of acid-soluble 32P counts as described in complexes showed that they were generally amorphous in shape Methods. The ®nal concentrations of hRad52 are indicated. and ranged in size from approximately 15 to 60 nm. Within these complexes, hRad52 rings could occasionally be seen (Fig. 1c, inset). Because hRad52 exhibits preferential binding to single-stranded DNA compared with double-stranded DNA10,11, we next deter- striking with blunt-ended DNA (Fig. 2b: compare lanes 3 and 4). mined whether the single-stranded tails were required for DNA Although our observation that end-ligation occurs in the presence end-binding and end-to-end interactions promoted by hRad52. of hRad52 may indicate that the ends remain freely accessible to Protein±DNA complexes were prepared using blunt-ended linear further enzymatic processing, it is perhaps more likely that ligation duplex DNA. End-binding and end-to-end associations were again results from Rad52-mediated network formation. Protein- observed by electron microscopy (data not shown). With this mediated stimulation of end-ligation has also been observed with substrate, however, the total number of molecules bound by hRad51 protein12. hRad52 was reduced compared to that observed with the tailed Our discovery that hRad52 possesses DNA end-binding activity substrates, and the complexes were generally smaller (,15±30 nm). implies that the proteins should be capable of protecting linear DNA These results indicate that hRad52 may have a higher af®nity for from exonuclease digestion. To test this, uniformly 32P-labelled (or greater stability on) tailed DNA than for blunt-ended linear EcoRI-linearized duplex DNA (7 nM DNA ends) was incubated duplex DNA. with hRad52 (0.24±0.96 mM) before the addition of Escherichia coli Given that hRad52 causes end-to-end interactions, its effect on exonuclease III. Remarkably, hRad52 afforded almost complete the ef®ciency of DNA ligation was investigated. Using linear DNA protection from exonuclease attack (Fig. 3a). Indeed, less than 5% molecules containing cohesive (Fig. 2a) or blunt (Fig. 2b) ends, we of the 32P was released as acid-soluble counts at concentrations of found that preincubation of the DNA with hRad52 stimulated hRad52 $0.48 mM. In contrast, using 0.96 mM hRad52, we failed to ligation catalysed by T4 DNA ligase. With the cohesive-ended observe any protection from digestion by an endonuclease such as DNA, most hRad52-facilitated events were intermolecular and DNase I (Fig. 3b), indicating that the bulk of the DNA remained approximately 65% of the DNA was converted into multimeric accessible. Similar results were obtained using blunt-ended DNA linear forms (Fig. 2a, lane 4). Analysis by digestion with restriction (data not shown). enzymes revealed that ligation involved both head-to-head and We have shown that hRad52 proteins binds double-strand DNA head-to-tail associations (data not shown). In the absence of breaks, protects them from exonuclease attack and promotes end- hRad52, ligation resulted primarily in the formation of circular to-end interactions. These properties are remarkably similar to (covalently closed and nicked circular) species (Fig. 2a, lane 3). The those exhibited by Ku5, leading us to propose a model for the stimulatory effect of hRad52 on DNA ligation was particularly repair of DSBs in human cells in which recognition of the DSB by

property of hRad52 raises the possibility that Rad52 directly controls the extent of exonuclease resection, and offers an explana- tion for the more extensive exonuclease digestion observed in yeast Rad50/Mre11/Xrs2 DSB Rad50/Mre11/Xrs2 7 Rad52 Ku rad52 mutants compared with wild-type cells . Genetic studies in yeast have shown that RAD52 is required for two homologous recombination pathways: (1) single-strand Rad52-dependent Ku-dependent annealing which occurs between repeated DNA sequences, and (2) homologous recombination non-homologous end-joining recombination promoted by the RAD51, RAD54, RAD55 and RAD57 gene products19. Consistent with a direct role in these Rad52 processes, Rad52 anneals complementary single-strands9,20±22 and end-binding can stimulate in vitro recombination reactions promoted by Rad51 Single-strand Recruitment DNA ligase IV annealing of Rad51 Xrcc4 (refs 11, 23±25). Rad52, Rad51 and RP-A all co-localize during meiosis at sites of potential DSBs (ref. 26); also, Rad52 is required for the pre-assembly of recombination complexes containing Rad51 (ref. 26). Rad52 may help to target Rad51 to the site of the resected Trimming Strand Ligation invasion DSB, where it will initiate strand exchange with homologous duplex DNA. Assembly of the recombination complex at the site of bound Rad52 is also likely to involve Rad55, Rad57 and RP-A (refs 23±27). The frequency of spontaneous allelic recombination in higher eukaryotic cells is extremely low28, of the order of 10-8, so gene Figure 4 Model for the initiation of double-strand break repair. In this model, a DSB therapy by in vivo gene targeting will remain impractical unless a can be bound by hRad52, directing repair by homologous recombination (left), or dramatic improvement in targeting ef®ciency can be achieved. The 29 by Ku, which initiates repair by non-homologous end-joining (right). Rad52- frequency of homologous recombination can be stimulated and 30 mediated repair can occur by single-strand annealing or by Rad51-mediated resistance to ionizing radiation can be increased by overexpression strand invasion of a homologous duplex DNA. Rad52 is indicated by shaded of Rad52 in mammalian cells, consistent with the DSB-repair model circles, Rad51 by white elipses, and DNA-PK (that is, Ku/DNA-PKcs) by shaded presented here. Moreover, our model predicts that signi®cant rectangles. increases in the frequency of homologous gene targeting using linear DNA vectors may be achieved by downregulating or inacti- vating Ku in coordination with Rad52 overexpression. M either hRad52 or Ku directs repair by Rad52-dependent homo- ...... logous recombination or Ku-dependent non-homologous end- Methods joining (Fig. 4). Proteins. Recombinant human Rad52 was puri®ed from baculovirus-infected That Rad52 and Ku provide alternative means to process DSBs is Sf9 cells as described previously, except that the Ni±NTA±agarose column was consistent with the phenotypes of rad52/Ku mutants. In higher replaced with Talon10. hRad52 was stored and diluted in 20 mM Tris-HCl (pH organisms, the importance of Ku (in association with the DNA- 8.0), 100 mM KCl, 1 mM EDTA, 0.5 mM dithiothreitol, and 10% glycerol. T4 5 activated protein kinase DNA-PKcs ) is illustrated by the X-ray- DNA ligase (NEB), exonuclease III (BRL), l exonuclease (Pharmacia) and sensitive and V(D)J-recombination-de®cient phenotype exhibited restriction enzymes were used as recommended by the suppliers. -/- by homozygous mouse Ku knockouts13,14. Because Ku is DNA substrates. Uniformly 32P-labelled plasmid pPB4.3 DNA (4,270 bp)11 abundant15, defects in hRad52 alone fail to cause a severe pheno- was prepared from cells grown in 32P-orthophosphate. To produce linear type, as indicated by the apparently normal radiation sensitivity duplex DNA with 59 or 39 single-stranded DNA tails, 32P-labelled form-I found with RAD52-/- knockout mouse16 and chicken17 cells. In yeast plasmid DNA was cut with SmaI and digested with predetermined saturating cells, however, where Ku-directed non-homologous end-joining amounts of either exonuclease III or l exonuclease. Exonuclease treatment was represents only a minor pathway for DSB repair5, rad52 mutants controlled to produce 59 or 39 tails with average lengths of 350 (59 tail) or 280 exhibit a severe recombination defect and are highly sensitive to nucleotides (39 tail). The extent of digestion was determined by measuring the ionizing radiation18. release of acid-soluble 32P counts after precipitation with perchloric acid. All Although vertebrate cells defective in Rad52 do not exhibit an DNA concentrations are expressed in moles of nucleotide residues. extreme recombination/repair-defective phenotype, the contribu- Electron microscopy. Protein±DNA complexes were ®xed by addition of tion of homologous recombination to the repair of DSBs should glutaraldehyde to 0.2% followed by a 15 min incubation at 37 8C. Samples not be underestimated. In hamster cells, endonuclease-generated were diluted and washed in 5 mM magnesium acetate before uranyl acetate DSBs in direct repeat sequences are repaired by homologous staining10. Complexes were visualized at magni®cations of 20,500´ or 25,000´ recombination3. Moreover, studies with the RAD54-/-/Ku70-/- using a Philips C100 electron microscope. double knockout from the chicken B-cell line DT40 show that, Stimulation of DNA ligation. hRad52 (0.32 mM) was incubated for 15 min at whereas non-homologous end-joining (NHEJ) is more important 37 8C with 32P-labelled linearized pPB4.3 DNA (10 mM) in 20 mM triethano- for repairing g-radiation-induced DSBs during G1±early S phase, lamine-HCl (pH 7.5). To this mixture (10 ml) was added 1 mlof10´ buffer -1 recombinational repair is preferentially used in late S±G2, when an (500 mM Tris-HCl (pH 7.5), 100 mM MgCl2, 10 mM ATP, and 250 mgml intact sister chromatid is available4. The lack of a strong phenotype bovine serum albumin) and either 40 (Fig. 2a) or 200 (Fig. 2b) units of T4 in the RAD52-knockout mouse may be due to a redundancy of DNA ligase. Incubation was at room temperature for 2.5 hours. Reactions Rad52 function, with some other function (for example, a Rad59 were stopped and deproteinized by addition of 2 ml `stop' buffer (2% SDS, homologue2) taking over the end-binding role in the absence of 0.6 mg ml-1 proteinase K), followed by 15 min incubation at 37 8C. Products Rad52. were analysed by 0.7% agarose gel electrophoresis using TAE buffer containing In the model shown in Fig. 4, hRad52 binds DSBs induced in 1 mgml-1 ethidium bromide followed by autoradiography. somatic cells by ionizing radiation or chemical damage. DSB- Exonuclease protection. Mixtures (120 ml) containing hRad52 and 32P- binding by hRad52 is equally applicable, however, at the initiation labelled linear duplex DNA in 20 mM triethanolamine-HCl (pH 7.5) were of recombination in germline cells at meiosis. Binding may occur incubated for 15 min at 37 8C. Then, 0.1 vol of 10´ exonuclease III buffer before, during or after resection of the DSB by Mre11/Rad50/Xrs2, (500 mM Tris-HCl (pH 7.5), 50 mM MgCl2) and either 16 units exonuclease III resulting in the formation of duplex molecules containing short or 250 units DNAse I were added, and incubation was continued at 37 8C. At single-stranded DNA tails bound by hRad52. The end-protection various times, aliquots (30 ml) were taken into 10 ml 0.5 M EDTA. The extent of

730 © 1999 Macmillan Magazines Ltd NATURE | VOL 398 | 22 APRIL 1999 | www.nature.com letters to nature digestion was determined following precipitation with perchloric acid by 17. Yamaguchi-Iwai, Y. et al. Homologous recombination, but not DNA repair, is reduced in vertebrate measuring the release of acid-soluble 32P counts. cells de®cient in RAD52. Mol. Cell Biol. 18, 6430±6435 (1998). 18. Game, J. C. DNA double-strand breaks and the RAD50-RAD57 genes in Saccharomyces. Semin. Cancer Biol. 4, 73±83 (1993). Received 14 January; accepted 1 March 1999. 19. Petes, T. D., Malone, R. E. & Symington, L. S. in The Molecular and Cellular Biology of the Yeast 1. Friedberg, E. C., Walker, G. C. & Siede, W. DNA Repair and Mutagenesis (American Society for Saccharomyces: Genome Dynamics, Protein Synthesis and Energetics 407±521 (Cold Spring Harbor Microbiology, Washington, 1995). Laboratory Press, New York, 1991). 2. Kanaar, R., Hoeijmakers, J. H. J. & van Gent, D. C. Molecular mechanisms of DNA double-strand 20. Mortensen, U. H., Bendixen, C., Sunjevaric, I. & Rothstein, R. DNA strand annealing is promoted by break repair. Trends Cell Biol. 8, 483±489 (1998). the yeast Rad52 protein. Proc. Natl Acad. Sci. USA 93, 10729±10734 (1996). 3. Liang, F., Han, M. G., Romanienko, P. J. & Jasin, M. Homology-directed repair is a major double- 21. Reddy, G., Golub, E. I. & Radding, C. M. Human Rad52 protein promotes single-strand DNA strand break repair pathway in mammalian cells. Proc. Natl Acad. Sci. USA 95, 5172±5177 (1998). annealing followed by branch migration. Mutat. Res. 377, 53±59 (1997). 4. Takata, M. et al. Homologous recombination and non-homologous end-joining pathways of DNA 22. Sugiyama, T., New, J. H. & Kowalczykowski, S. C. DNA annealing by Rad52 protein is stimulated by double-strand break repair have overlapping roles in the maintenance of chromosomal integrity in speci®c interaction with the complex of replication protein A and single-stranded DNA. Proc. Natl vertebrate cells. EMBO J. 17, 5497±5508 (1998). Acad. Sci. USA 95, 6049±6054 (1998). 5. Critchlow, S. E. & Jackson, S. P. DNA end-joining: from yeast to man. Trends Biochem. Sci. 23, 394± 23. New, J. H., Sugiyama, T., Zaitseva, E. & Kowalczykowski, S. C. Rad52 protein stimulates DNA strand 398 (1998). exchange by Rad51 and replication protein-A. Nature 391, 407±410 (1998). 6. Sun, H., Treco, D. & Szostak, J. W. Extensive 39-overhanging, single-stranded DNA associated with the 24. Shinohara, A. & Ogawa, T. Stimulation by Rad52 of yeast Rad51-mediated recombination. Nature meiosis-speci®c double-strand breaks at the ARG4 recombination initiation site. Cell 64, 1155±1162 391, 404±407 (1998). (1991). 25. Sung, P. Function of yeast Rad52 protein as a mediator between replication protein-A and the Rad51 7. Sugawara, N. & Haber, J. E. Characterization of double-strand break-induced recombination: recombinase. J. Biol. Chem. 272, 28194±28197 (1997). homology requirements and single-stranded DNA formation. Mol. Cell. Biol. 12, 563±575 (1992). 26. Gasior, S. L., Wong, A. K., Kora, Y., Shinohara, A. & Bishop, D. K. Rad52 associates with RP-A and 8. Shen, Z. et al. Self-association of human Rad52 protein. Mutat. Res. 364, 81±89 (1996). functions with Rad55 and Rad57 to assemble meiotic recombination complexes. Genes Dev. 12, 2208± 9. Shinohara, A., Shinohara, M., Ohta, T., Matsuda, S. & Ogawa, T. Rad52 forms ring structures and 2221 (1998). cooperates with RPA in single-strand DNA annealing. Genes to Cells 3, 145±156 (1998). 27. Sung, P. Yeast Rad55 and Rad57 proteins form a heterodimer that functions with replication protein- 10. Van Dyck, E., Hajibagheri, N. M. A., Stasiak, A. & West, S. C. Visualization of human Rad52 protein A to promote DNA strand exchange by Rad51 recombinase. Genes Dev. 11, 1111±1121 (1997). and its complexes with hRad51 and DNA. J. Mol. Biol. 284, 1027±1038 (1998). 28. Subramani, S. & Seaton, B. L. in Genetic Recombination (eds Kucherlapati, R. & Smith, G. R.) 549±573 11. Benson, F. E., Baumann, P. & West, S. C. Synergistic actions of Rad51 and Rad52 in genetic (American Society for Microbiology, Washington, DC, 1988). recombination and DNA repair. Nature 391, 401±404 (1998). 29. Johnson, B. L., Thyagarajan, B., Krueger, L., Hirsch, B. & Campbell, C. Elevated levels of recombina- 12. Baumann, P., Benson, F. E. & West, S. C. Human Rad51 protein promotes ATP-dependent tional DNA repair in human somatic cells expressing the Saccharomyces cerevisiae RAD52 gene. Mutat. homologous pairing and strand transfer reactions in vitro. Cell 87, 757±766 (1996). Res. DNA Repair 363, 179±189 (1996). 13. Nussenzweig, A. et al. Requirement for Ku80 in growth and immunoglobulin V(D)J recombination. 30. Park, M. S. Expression of human RAD52 confers resistance to ionizing radiation in mammalian cells. Nature 382, 551±555 (1996). J. Biol. Chem. 270, 15467±15470 (1995). 14. Nussenzweig, A., Sokol, K., Burgman, P., Li, L. G. & Li, G. C. Hypersensitivity of Ku80-de®cient cell Acknowledgements. We thank our colleagues for their interest, suggestions and careful reading of the lines and mice to DNA damage: the effects of ionizing radiation on growth, survival, and manuscript. We also thank J. Dubochet for his interest in the project, and N. Hajibagheri for related development. Proc. Natl Acad. Sci. USA 94, 13588±13593 (1997). electron microscopic analyses. This work was supported by the Imperial Cancer Research Fund, the 15. Anderson, C. W. & Carter, T. H. in Molecular Analysis of DNA Rearrangements in the Immune System Human Frontiers Science Program, the Swiss National Foundation and the Swiss-British Council Joint (eds Jessberger, R. & Lieber, M. R.) 91±112 (Springer, Heidelberg, 1996). Research Program. E.V.D. was supported in part by an EC fellowship. 16. Rijkers, T. et al. Targeted inactivation of MmRAD52 reduces homologous recombination but not resistance to ionizing radiation. Mol. Cell. Biol. 18, 6423±6429 (1998). Correspondence and requests for materials should be addressed to S.C.W. (e-mail: [email protected]).