News & views - Nature

© 1999 Nature America Inc. • http://neurosci.nature.com

news and views

Experience-dependent development of NMDA receptor transmission Kevin Fox, Jeremy Henley and John Isaac


Light exposure changes the subunit composition and kinetics of NMDA receptors in the developing visual cortex. Quinlan and colleagues suggest that this may be due to rapid synaptic insertion of receptors containing newly synthesized NR2A subunits.

down to adult levels within a few The maturation of cortical circuitdays of such exposure 6 . These ry depends critically on experience, as sensory deprivation changes, which seem to mimic prevents many of the changes in normal development, could be cortical function that normally explained by increases in AMPA occur with age. During the develreceptor currents, changes in opment of glutamatergic transNMDA receptor currents or both. mission in sensory cortex, the This raises the question of proportion of synapses with whether maturational changes in detectable AMPA receptor currents glutamate receptors during normal Visual experience increases, and the decay kinetics of development are experience NMDAR activation NMDA receptors become faster1,2. dependent. Although the proportion of synapses with AMPA recepBoth these changes serve to reduce NR2B NR1 NR2A AMPAR subunits Amy Center tor currents increases during the relative contribution of NMDA receptors to synaptic currents and Fig. 1. Changes in postsynaptic glutamate receptors during develop- development in visual cortex7 and are thought to depend on synaptic ment. Left, immature synapses contain NMDA receptors (NMDAR) in other mammalian sensory sysactivity. The increase in AMPA and no detectable AMPA receptors (AMPAR). NMDA receptors at tems such as barrel cortex1,3, the receptor currents is thought to be these so-called ‘silent synapses’ have slow kinetics and are likely to con- effects of experience on this due to rapid insertion of AMPA sist of NR1 and NR2B subunits. Right, the mature synapse contains process in vivo are not yet known. AMPA receptors as well as NMDA receptors containing NR2A subreceptors into the synaptic memOn the other hand, dark rearing or units, which have fast kinetics. The transition between the two states brane under the control of NMDA depends on visual experience. Note that NMDA receptors are shown activity block by direct application receptors 3 , and the change in as pentamers but may be tetramers. The exact stochiometry of the of tetrodotoxin prevents NMDA NMDA receptor kinetics is subunits is not known, and the particular configurations shown here receptor decay kinetics from reaching their mature form in rat visual believed to result from a develop- are only meant to illustrate the possible state of affairs. cortex 2 . The mature kinetics mental switch in the subunit composition of NMDA receptors (Fig. (decay time constant, 100 ms) are 1). The NR1 subunit combines characteristic of NMDA receptors with various NR2A–D subunits to procomposed of NR1 and NR2A subunits, the adult. In the immature brain, synapduce receptor subtypes with different whereas the immature form (350 ms) is tic transmission is weak, extremely plaskinetics, of which receptors containing characteristic of receptors with NR1 and tic and mediated in large part by NMDA NR2A have the fastest decay times. On NR2B subunits8. The concept that the receptors. Some glutamatergic synapses page 352 of this issue, Quinlan and colin young animals have no AMPA recepdevelopmental change in receptor kinetics leagues demonstrate that rats deprived of tor currents, making them functionally depends on subunit composition in vivo visual experience by dark rearing have low ‘silent’ at resting membrane potentials3. gains support from the correlation levels of NR2A-containing receptors in between the timing of NR2A expression In the adult, transmission is stronger, less visual cortex, that these receptors are and the switch from slow to fast kinetics9. plastic and mainly mediated by AMPA rapidly increased by light exposure, and receptors. These developmental changes Moreover, this correlation extends to that the process is controlled by NMDA require visual experience, because synapindividual cells, as cortical neurons that receptors. tic transmission in the visual cortex can express higher levels of NR2A mRNA Excitatory transmission is very differbe preserved in an immature state by have faster NMDA receptor kinetics10. By 4 ent in the immature brain compared with delaying the onset of light exposure . If showing that NR2A expression is suppressed in the visual cortex in the absence animals are reared in the dark, visual of visual experience, Quinlan and colresponses are weaker and less orientation Kevin Fox is at the Cardiff School of Biosciences, leagues now provide an explanation for selective and include larger NMDA recepCardiff University, Museum Avenue, Cardiff why dark rearing delays the change in tor currents than in light-reared animals CF1 3US, UK. Jeremy Henley and John Isaac NMDA receptor kinetics. of the same age. However, visual responses are at the MRC Centre for Synaptic Plasticity, So what are the possible mechanisms strengthen within hours of the first expoDepartment of Anatomy, Bristol University, for regulation of NMDA receptor subsure to light5, and the NMDA receptor University Walk, Bristol BS8 1TT, UK. units? Receptor proteins are traditionally component of the visual response shrinks e-mail: [email protected]. nature neuroscience • volume 2 no 4 • april 1999

297



news and views

under tight and perhaps inducible regulation. On the Inte rna other hand, PCR was used liza tion to amplify the mRNA in this 2+ Ca study, and very low levels of dendritic mRNA may simply Local synthesis result from a non-specific of NR2A Dissassembly, ‘leak’ of somatic mRNA. As recycling? yet, there is little informaInsertion Assembly tion on the mechanisms by which mRNA may be specifically targeted to dendrites. Local Second, the machinery synthesis NR1 of NR NR2B for the translation and postNR2A subunits translational modification Amy Center of receptor complexes and Fig. 2. Mechanisms for rapid synthesis, assembly and insertion for their assembly and packof NMDA receptors. Translation of NR2A mRNA is proposed aging into vesicles should be to be calcium dependent. By activating NMDA receptors, visual present in dendrites close to experience leads to translation of new NR2A, which is com- spines. There is evidence bined with other NMDA receptor subunits in a Golgi-like endo- that dendritic shafts and some and inserted in the membrane. The source of the other spines contain polyribosubunits may be recycled NMDA receptors from the postsynapsomes associated with memtic membrane or subunits transported from the cell body. branous cisterns. These are the so-called synapse-associated polyribosome complexes or SPRCs for short; (for review, thought to be translated in the nucleus see ref. 12). Although there are no puband delivered to the synapse via axon lished data to show that translation actutransport. However, several factors sugally occurs at these sites, there is gest that another mechanism must evidence for translation at isolated account for these results. First, NR2A progrowth cones, where these structures are tein levels increase extremely rapidly also found12. SPRCs are thought to be (within 1 hour) in the synaptoneurosome fraction (a synapse-enriched biochemical capable of translation and post-translapreparation), too fast to rely on axon tional modifications such as glycosylatransport. Second, the increase is blocked tion and phosphorylation. In this case, by the protein synthesis inhibitor cyclothey would also have to assemble the heximide. Taken together, these results various subunits and package them in imply that NR2A subunits are synthesized vesicles for membrane insertion. In this locally in the dendrites. Third, a funcsense, SPRCs may act like a mini endotional change occurs within the same time plasmic reticulum and Golgi apparatus span, which suggests that synthesis is folfor dendrites. lowed by rapid local assembly of subunits Third, synaptic activity should be able and insertion of NR2A-containing recepto induce local translation. Unfortunately, tors into the postsynaptic membrane (Fig. there is no direct evidence for local trans2). Finally, because all this is set in motion lation of NMDA receptor subunits resultby visual experience and blocked by ing from synaptic activity at present. NMDA receptor antagonists, some synapHowever, there is some evidence for how tically activated second messenger system local translation of another mRNA might such as calcium must control the process. be regulated by synaptic activity. An iniSeveral conditions would be required tial step in translation involves mRNA for such a mechanism to occur in neupolyadenylation, which in turn depends rons. First, mRNA for NR2A should be on a regulatory protein known as cytopresent in dendrites. Although there is plasmic polyadenylation element binding no direct evidence in visual cortical neuprotein (CPEB). It has been found that rons, mRNA for glutamate receptors is CPEB-dependent polyadenylation of the present in dendrites of cultured hipα subunit of calcium/calmodulin-depenpocampal neurons11. However, in this dent protein kinase II (αCaMKII) mRNA occurs within 30 minutes of exposing study, NR2A mRNA (together with dark-reared animals to the light14. This NR2C) was the scarcest of all the glutamate receptors. This may indicate that visual activity probably triggers NR2A mRNA very rarely reaches the polyadenylation via a second messenger dendrites or, more interestingly, that it is system such as calcium influx through 298

NMDA receptors (Fig. 2) and hence leads to local translation of the αCaMKII mRNA. Importantly, in this study, CPEB was found to be enriched in postsynaptic densities, and αCaMKII mRNA was present in dendrites, so the required machinery for translation was present near the synapse. Therefore, these results provide a model for how local translation of NR2A mRNA might depend on activity. Glycine receptors may also be inserted via a mechanism similar to the one proposed for NMDA receptors by Quinlan and colleagues. There is good evidence that the α subunits of the glycine receptor are synthesized dendritically and then inserted in complexes together with β subunits, which are produced by conventional somatic synthesis (see ref. 14). Interestingly, the mRNAs encoding both glycine receptor and NMDA receptor subunits are rather unusual in having very long 3´ untranslated regions (3´UTRs), which is where CPEB binds on the αCaMKII mRNA. Although the role of these long 3´ UTRs is not fully understood, they could be involved in the regulation of translation14. Whatever the exact mechanisms of expression, one can ask what purpose such changes in NMDA receptors could serve during development. One possibility is that they form part of a negative feedback system, whereby the plasticity of the synapse is downregulated as it becomes more potentiated. The incidence of silent synapses decreases with development, probably as a result of NMDA receptor activation, which causes insertion of AMPA receptors. If the same intracellular pathway were to cause synthesis and insertion of the NR2Acontaining NMDA receptor, then subsequent activation of the same synapse would cause less calcium influx per depolarization because of the faster decay kinetics of this NMDA receptor subtype. This may either make it more difficult to accumulate sufficient postsynaptic calcium to produce further potentiation, or actually favor a process like long-term depression, which has been postulated to result from synaptic activity producing low levels of postsynaptic calcium. Either way, the initial emphasis would be on connecting weak synapses by a potentiation mechanism (perhaps AMPA receptor insertion triggered by NRB-type NMDA receptors). Subsequently, in older animals, the strength of the newly functional synapses could be reduced or erroneously formed connections eliminated by nature neuroscience • volume 2 no 4 • april 1999


news and views

synaptic depression (perhaps dephosphoryation of AMPA channels triggered by NR2A-type NMDA receptors).

4. Fox, K., Daw, N., Sato, H. & Czepita, D. Nature 350, 342–344 (1991). 5. Buisseret, P., Garey-Bobo, E. & Imbert, M. Nature 272, 816–817 (1978).

10. Flint, A. C., Maisch, U. S., Weishaupt, J. H., Kriegstein, A. R. & Monyer, H. J. Neurosci. 17, 2469–2478 (1997).

1. Crair, M. C. & Malenka, R. C. Nature 375, 325–328 (1995).

6. Fox, K. Daw, N., Sato, H. & Czepita, D. J. Neurosci. 12, 2672–2684 (1992).

11. Miyashiro, K., Dichter, M. & Eberwine, J. Proc. Natl. Acad. Sci. USA 91, 10800–10804 (1994).

2. Carmignoto, G. & Vicini, S. Science 258, 1007–1011 (1992).

7. Rumpel, S., Hatt, H. & Gottmann, K. J. Neurosci. 18, 8863–8874 (1998).

13. Wu, L. et al. Neuron 21, 1129–1139 (1998).

3. Isaac, J. T. R., Crair, M. C., Nicoll, R. A. & Malenka, R. C. Neuron 18, 269–280 (1997).

8. Williams, K., Russell, S. L., Shen, Y. M. & Molinoff, P. B. Neuron 10, 267–278 (1993).

14. Kirsh, J., Meyer, G. & Betz, H. Mol. Cell. Neurosci. 8, 93–98 (1996).

A tale of two spikes © 1999 Nature America Inc. • http://neurosci.nature.com

Yves Frégnac Backpropagating action potentials amplify the response to weak dendritic inputs. A new study suggests that this may serve to link simultaneous inputs to different dendritic compartments. Since the discovery that action potentials initiated at the soma can propagate back into dendrites, the role of these spikes in shaping the cell’s response to its inputs has been of intense interest. In a forthcoming issue of Nature, Larkum, Zhu and Sakmann 1 show that backpropagated action potentials in cortical pyramidal neurons can interact with weak synaptic inputs in the apical dendrites, triggering a dendritic calcium spike. The calcium wave is in turn propagated to the soma, causing the neuron to fire a burst of action potentials. This mechanism allows for the all-or-none amplification of weak inputs, and may have important implications for how information impinging on distal dendrites is processed at the cellular level within the cortex. In the classic view of synaptic integration, excitatory postsynaptic potentials (EPSPs) are transmitted passively through the dendritic tree to the cell body and axon hillock, where sodium action potentials are generated. The EPSPs are summed at this ‘decision point’, and if the total depolarization reaches threshold, the neuron fires an action potential, which is then propagated along the axon to the rest of the network. In this model, which has often been applied to pyramidal cells of the hippocampus and neocortex, the dendrites are treated as passive cables, and the effect of a given synapse depends on its location within the dendritic tree. Yves Frégnac is at the Equipe Cognisciences, Institut Alfred Fessard, CNRS, 91 198, Gif sur Yvette, France. e-mail: [email protected] nature neuroscience • volume 2 no 4 • april 1999

9. Fox, K. Neuron 15, 485–488 (1995).

12. Steward, O. Neuron 18, 9–12 (1997).

but also backward into the dendritic tree5. Third, there exists at least one other site for action potential generation, which is distinct from the soma–axon region. This second site, described in layer-V cortical neurons, is in the tuft of the apical dendrite. This region, which is rich in voltage-dependent calcium and sodium channels, gives rise to calcium spikes6,7. Calcium spikes are typically of much longer duration than sodium action potentials, but these regenerative events involving voltage-gated channels are normally attenuated as they spread to the soma. The new study in Nature from Larkum et al.1 is a logical continuation

Inputs from more distant synapses take longer to reach the cell body and are more attenuated than inputs originating at synapses closer to the cell body. Such a neuron is predicted to behave as a spatiotemporal correlator; it is more likely to fire an action potential if all the EPSPs reach the axon hillock simultaneously. For this to occur, the inputs must arrive in an appropriate temporal relationship to compensate 1 and 2 for their different locations and the resulting delays in reaching the site of action potential generation. In reality, however, dendrites are not passive cables. They express a variety of voltage-dependent ion channels that modify their biophysical properties, allowing them not only to transmit synaptic inputs to the cell body, but also to perform 2 considerable local processing. Recently, refined techniques, such as the use of multiple simultaneous 1 patch electrodes at different points on a single neuron, have revealed a complex picture of neuronal function that differs in several important respects from the classical model. First, ‘hot spots’ of excitability within the apical dendrites are 1 2 thought to modify the propagation of EPSPs from synaptic sites to the cell body, so as to reduce the differLayer V Layers I-II ences of timing and amplitude that Fig. 1. The association of a subthreshold distal denresult from their different locations dritic input (1) with a backward propagating somatic on the dendritic tree 2–4 . Second, action potential (2) generates a wide dendritic calcium sodium action potentials originat- spike (1+2) sufficient to reactivate the soma and cause ing at the soma–hillock region can a burst of firing. Courtesy of Dr. Thierry Bal (IAF, propagate not only along the axons, CNRS, France). 299