scientific correspondence endofullerenes are self-assembled chains of receptor channels, which may account for value or Hill coefficient. This non-competi-

C60. The observation of these structures many of xenon’s attractive pharmacological tive inhibition indicates that xenon should raises the hope that refined processing tech- properties. strongly inhibit neural transmission, niques can be developed to produce them We found that xenon had virtually no despite the high glutamate concentrations in large quantities. effect on GABAA receptors. Currents acti- in synaptic clefts. Brian W. Smith*, Marc Monthioux*†, vated by 3 ȖM GABA, both in voltage- We then tested this in microisland cul- David E. Luzzi* clamped cultured rat hippocampal neurons tures of hippocampal neurons that form *Department of Materials Science and Engineering, and in voltage-clamped PA3 cells4 that synapses with themselves (autapses)9. A University of Pennsylvania, 3231 Walnut Street, stably expressed defined GABA subunits, typical glutamatergic postsynaptic current A 8 Philadelphia, Pennsylvania 19104-6272, USA were not significantly affected even by recorded from a hippocampal neuron is e-mail: [email protected] 100% xenon (to function as a human shown in Fig. 1b. The control records show †CEMES, UPR A-8011 CNRS, BP 4347, anaesthetic, the half-maximal effective con- a characteristic biphasic time course, with a

F-31055 Toulouse cedex 4, France centration (EC50) is 71% v/v; ref. 5). Xenon fast component mediated by non-NMDA 1. Rinzler, A. G. et al. Appl. Phys. A 67, 29–37 (1998). also had little effect on functional GABA- receptors and a much slower component 2. Nikolaev, P., Thess, A., Rinzler, A. G., Colbert, D. T. & Smalley, releasing synapses in hippocampal neurons, mediated by NMDA receptors. This NMDA R. E. Chem. Phys. Lett. 266, 422–426 (1997). with 80% xenon reducing peak inhibitory receptor-mediated component could be 3. Heiney, P. A. J. Phys. Chem. Solids 53, 1333–1352 (1992). Ϯ 4. Yakobson, B. I. & Smalley, R. E. Am. Sci. 85, 324–337 (1997). postsynaptic currents by only 8 2%. This readily identified as it was blocked by the result indicates that the presynaptic effects highly selective competitive antagonist AP5 of xenon must also be very modest. (DL-2-amino-5-phosphonopentanoate)10. Ȗ Apart from the GABAA receptor, the only Addition of 200 M AP5 almost com- How does xenon generally accepted neuronal target of con- pletely blocked the slow component, leav- ventional anaesthetics is the NMDA receptor. ing only a fast component, with a single produce anaesthesia? This subtype of glutamate-activated iono- exponential time course very similar to that tropic channels is implicated in synaptic of the control fast component. The effect of Since the discovery that the gas xenon can mechanisms underlying learning, memory xenon on the glutamatergic postsynaptic produce general anaesthesia1 without caus- and the perception of pain6. The NMDA current resembled that of AP5 (Fig. 1b). ing undesirable side effects, we have receptor is also believed to be a target of the The slow, NMDA-receptor-mediated com- remained surprisingly ignorant of the mol- intravenous general anaesthetic agent keta- ponent was reduced by over 70%, whereas ecular mechanisms underlying this clinical mine7, and possibly nitrous oxide8. the fast component barely changed. So, not activity of an ‘inert’ gas. Although most gen- We therefore looked at the effects of only did xenon inhibit synaptic NMDA eral anaesthetics enhance the activity of xenon on NMDA-activated currents in receptors, it had little apparent effect on ȍ inhibitory GABAA ( -aminobutyric acid cultured hippocampal neurons. We found non-NMDA receptors. type-A) receptors2,3, we find that the effects that 80% xenon, which will maintain surgi- If xenon exerts its effects by inhibiting of xenon on these receptors are negli- cal anaesthesia, reduced NMDA-activated NMDA receptors, then this explains some gible. Instead, xenon potently inhibits the currents by about 60% (Fig. 1a), with no important features of its pharmacological excitatory NMDA (N-methyl-D-aspartate) significant change in the NMDA EC50 profile, particularly as NMDA-receptor a antagonists can relieve pain and cause Figure 1 Xenon inhibits NMDA 100 NMDA NMDA amnesia, which are features of xenon anaes- receptors in cultured rat hip- thesia. Like nitrous oxide (‘laughing gas’), pocampal neurons. a, NMDA 80 Control which may also act, at least partly, on 8 activates an inward current (in NMDA receptors , xenon can induce a state 60 neurons clamped at Ϫ 60 mV) Xenon of euphoria. Other neuronal targets for

Ϯ Ȗ 500 pA with an EC50 of 24 2 M NMDA 40 10 s xenon may emerge, but its powerful inhibi- Ϯ and a Hill coefficient of 1.2 0.1. Control Xenon tion of the NMDA receptor is likely to be Xenon inhibited the current by 20 instrumental in the anaesthetic and anal- approximately 60% but did not gesic effects of this ‘inert’ gas. 0 Percentage of maximum control current of maximum Percentage significantly change either the 0 10 100 1,000 N. P. Franks, R. Dickinson, S. L. M. de Sousa,

EC50 or the Hill coefficient. Each Concentration of NMDA (µM) A. C. Hall, W. R. Lieb data point represents the mean b Biophysics Section, peak current from at least 6 0 AP5 The Blackett Laboratory, cells. Inset, typical current –2 Imperial College of Science, Technology and Ȗ traces (at 100 M NMDA) in the Time (ms) Medicine, Prince Consort Road, –4 presence and absence of 0 50 100 150 200 London SW7 2BZ, UK Xenon 0 xenon. b, Xenon selectively –6 e-mail: [email protected] –1 inhibits the NMDA-receptor- Controls Xenon –8 1. Cullen, S. C. & Gross, E. G. Science 113, 580–582 (1951). mediated component of gluta- current (nA) Postsynaptic 2. Franks, N. P. & Lieb, W. R. Nature 367, 607–614 (1994). –2 Control matergic excitatory postsynaptic –10 3. Mihic, S. J. et al. Nature 389, 385–389 (1997). NMDA component (nA) NMDA 4. Hadingham, K. L. et al. Proc. Natl Acad. Sci. USA 89, 6378–6382 currents (EPSCs). Neurons were –12 (1992). Ϫ voltage-clamped at 60 mV; 0 50 100 150 200 5. Cullen, S. C., Eger, E. I. II, Cullen, B. F. & Gregory, P. synaptic responses were stimu- Time (ms) Anesthesiology 31, 305–309 (1969). lated by a 2-ms depolarizing pulse to ϩ 20 mV. Control glutamatergic EPSCs displayed a characteristic 6. Rang, H. P., Dale, M. M. & Ritter, J. M. Pharmacology 3rd edn Ȗ (Churchill Livingstone, Edinburgh, 1995). biphasic decay. The slow component was completely blocked by 200 M AP5, leaving the fast compo- 7. Anis, N. A., Berry, S. C., Burton, N. R. & Lodge, D. Br. J. nent almost unaffected. Inset, the NMDA-receptor-mediated component (the difference between the Pharmacol. 79, 565–575 (1983). control EPSC and that in the presence of AP5) and its size in the presence of xenon (calculated by tak- 8. Jevtovic-Todorovic, V. et al. Nature Med. 4, 460–464 (1998). 9. Bekkers, J. M. & Stevens, C. F. Proc. Natl Acad. Sci. USA 88, ing the difference between the EPSC in the presence of xenon and that in the presence of AP5). Con- 7834–7838 (1991). trol solutions were equilibrated at room temperature with 80% N2 and 20% O2, and test solutions with 10.Watkins, J. C. & Evans, R. H. Annu. Rev. Pharmacol. Toxicol. 21, 165–204 (1981). 80% Xe and 20% O2. 324 Nature © Macmillan Publishers Ltd 1998 NATURE | VOL 396 | 26 NOVEMBER 1998 | www.nature.com