146 VOLUME 13 | NUMBER 2 | FEBRUARY 2010 NATURE NEUROSCIENCE NEWS AND VIEWS partners 13 , should all clarify whether these neurons functionally integrate into cortical networks. Until then, we can only speculate whether these new neurons have a positive effect on mild ischemic damage and contribute to the generally favorable clinical outcomes of patients enduring minor vascular accidents. 1. Sanai, N. et al. Nature 427, 740–744 (2004). 2. Arvidsson, A., Collin, T., Kirik, D., Kokaia, Z. & Lindvall, O. Nat. Med. 8, 963–970 (2002). 3. Parent, J.M. et al. J. Neurosci. 17, 3727–3738 (1997). 4. Chen, J., Magavi, S.S. & Macklis, J.D. Proc. Natl. Acad. Sci. USA 101, 16357–16362 (2004). 5. Ohira, K. et al. Nat Neurosci. 13, 173–179 (2010). 6. Cobos, I., Long, J.E., Thwin, M.T. & Rubenstein, J.L. Cereb. Cortex 16, i82–i88 (2006). 7. Batista-Brito, R. & Fishell, G. Curr. Top. Dev. Biol. 87, 81–118 (2009). 8. Rakic, S. & Zecevic, N. Cereb. Cortex 13, 1072–1083 (2003). 9. Wichterle, H., Garcia-Verdugo, J.M., Herrera, D.G. & Alvarez-Buylla, A. Nat. Neurosci. 2, 461–466 (1999). 10. Zhao, J.W., Raha-Chowdhury, R., Fawcett, J.W. & Watts, C. Eur. J. Neurosci. 29, 1853–1869 (2009). 11. Nishiyama, A. & Komitova, M. Nat. Rev. Neurosci. 10, 9–22 (2009). 12. Yan, Y.P. et al. J. Cereb. Blood Flow Metab. 27, 1213–1224 (2007). 13. Burkhalter, A. et al. Nat. Rev. Neurosci. 9, 557–568 (2008). projection neurons seem to be the population most in need of replacement, neurogenesis seemingly produces more interneurons. New interneurons might serve several purposes. First, they may protect projection neurons from further damage by hypoxic/ischemic events, either by secreting substances such as NPY and somatostatin that promote neuronal survival or by dampening excessive excitatory input that may cause neuronal death or seizures. Second, they may act to reroute information from projection neurons in damaged areas to neighboring areas that may have sustained less damage. That is, they might be able to reconfigure cortical circuits in beneficial ways. Whereas the authors provide some evidence that the ischemia-induced interneuron populations integrate into cortical circuits by demonstrating their expression of the immediate early gene product c-Fos, future studies should attempt more direct physiological measures of their function and should provide a more precise picture of their contributions to cortical networks. Assessment of their synaptic inputs and intrinsic physiological properties, as well as paired recordings with efferent synaptic increased and widespread proliferation of glia, including microglia and cells that express the NG2 proteoglycan and the Olig2 transcription factor, markers of glial precursor cells that can generate oligodendrocytes and some astrocytes after injury 10,11 . Identifying the signal(s) for subpial cell proliferation, migration and differentiation may ultimately be a rewarding project, and could inform potential therapeutic strategies designed to stimulate neurogenesis after hypoxia/ischemia. The activated microglial and astrocyte populations are likely sources of such factors, including monocyte chemoattractant protein-1, which is expressed by these cells after an ischemic insult, and which promotes neuroblast migration, interacting with its receptor, CCR2, on immature neurons 12 . All of these observations beg the question of whether the observed neurogenesis represents pathology itself or a reparative process initiated to compensate for ischemic damage. Indeed, is the generation of new interneurons beneficial, and if so, how? Of the various cortical neuronal populations, the large projection neurons are more sensitive to hypoxic/ischemic damage than are the interneurons. And yet, though the large A ‘sustain pedal’ in the hippocampus? Matthew C Walker, Ivan Pavlov & Dimitri M Kullmann A study reveals that a largely ignored cell type in the dentate gyrus, semilunar granule cells, are persistently depolarized after a transient input and recruit interneurons to regulate the gating of information into the hippocampus. In our day-to-day lives, we rely on numerous temporary memory stores, often lasting seconds, for a wide variety of functions, such as performing mental calculations, remembering to turn the gas off and keeping track of which article in this journal we are reading. These memories need to be transient in comparison with longer-term memories such as memorizing the essential message of this article. This large capacity, but transient, store of accessible information has been likened to the “blackboard of the brain” and is termed working memory 1 . That working memory has a neural substrate has been established in experiments in which primates are trained to delay the execution of a movement until some time after a sensory cue 1 . A typical task consists of the animal observing food being placed into one of several wells The authors are in the Institute of Neurology, University College London, London, UK. e-mail: mwalker@ion.ucl.ac.uk (stimulus) that it subsequently has to retrieve. These delayed response trials are associated with sustained neuronal firing during the interval period and this has been proposed to be the correlate of working memory 2 . The ability of a network to maintain short-term increases in activity following an input is a common feature of sensory processing not only in the cortex, but also in the cerebellum 3 . The hippocampus and related structures of the medial temporal lobe have long been recognized as being critical for encoding long- term memory; more recent evidence, however, indicates that the medial temporal lobe is also necessary for the maintenance of working memory for novel items and associations 4 . But how do the medial temporal lobe and hippocampus sustain neuronal activity? The article by Larimer and Strowbridge 5 in this issue describes the behavior of a set of neurons in the dentate gyrus that can be likened to a ‘sample and hold’ circuit, providing an unexpected substrate for working memory. Initial attempts to explain persistent activity following a stimulus centered on recurrent, reinforcing excitatory connections and both computational and experimental support for this exists in some systems 6 . However, there is growing evidence that persistent firing can also be maintained by the intrinsic properties of neurons themselves. In particular, certain neurons are intrinsically bistable and have, in effect, two resting membrane potentials (a relatively negative potential at around –70 mV and a potential around –50 mV close to firing threshold). Notably, a change of state in a few neurons (from down- to up-state) can drive other neurons in the network into an up-state; thus, bistable neurons can result in bistable networks 7 . Some neurons even have multiple stable states; layer V entorhinal cortex pyramidal cells can respond to consecutive transient stimuli with stable, graded changes in firing rates 8 . Bistable neurons are found in many cortical and subcortical structures, but are relatively rare overall in the mammalian CNS. © 2010 Nature America, Inc. All rights reserved.