摘要
Nitric oxide (NO) is a soluble gas that acts as a novel type of intercellular messenger molecule in the brain (4Bredt D.S. Snyder S.H. Neuron. 1992; 8: 3-11Abstract Full Text PDF PubMed Scopus (1809) Google Scholar). NO has been suggested to play a role in learning and memory (Table 1) and in types of neuronal plasticity thought to contribute to learning, including both long-term potentiation (LTP) in hippocampus and long-term depression (LTD) in cerebellum (for reviews see7Hawkins R.D. Zhuo M. Arancio O. J. Neurobiol. 1994; 25: 652-665Crossref PubMed Scopus (124) Google Scholar, 17Zhuo M. Hawkins R.D. Rev. Neurosci. 1995; 6: 259-277Crossref PubMed Scopus (70) Google Scholar). However, the involvement of NO in mammalian learning, LTP, and LTD is controversial (1Bannerman D.M. Chapman P.F. Kelly P.A.T. Butcher S.R. Morris R.G.M. J. Neurosci. 1994; 14: 7404-7414Crossref PubMed Google Scholar, 2Bannerman D.M. Chapman P.F. Kelly P.A.T. Morris R.G.M. J. Neurosci. 1994; 14: 7415-7425PubMed Google Scholar, 10Linden D.J. Connor J.A. Eur. J. Neurosci. 1992; 4: 10-15Crossref PubMed Scopus (77) Google Scholar), and the balance of the evidence has shifted back and forth. In this issue of Neuron, U. 14Müller U. Neuron. 1996; 16 (this issue)Abstract Full Text Full Text PDF PubMed Scopus (175) Google Scholar describes experiments in which he took a new approach to this problem by using the honeybee to investigate the function of NO. He reports that there are several different memory systems that are engaged by different numbers of training trials and that cover different (but overlapping) time spans in honeybee. Inhibition of NO synthase (NOS) during training blocks only one of these: long-term (24 hr) memory following multiple training trials. NOS inhibitors have no effect on initial learning or medium-term (3 or 8 hr) memory and no effect on the modest level of long-term memory following a single training trial. Thus, in honeybee, NO seems to be critically involved in only one type of memory: long-term memory following repeated training trials. This conclusion is reminiscent of similar conclusions about the role of the transcription factor CREB in long-term memory in Aplysia, Drosophila, and mice (reviewed by6Frank D.A. Greenberg M.E. Cell. 1994; 79: 5-8Abstract Full Text PDF PubMed Scopus (338) Google Scholar), and Müller speculates that NO might act via CREB in honeybee. Consistent with this idea, NO is thought to act via CREB in mammalian PC12 cells (16Peunova N. Enlikolopov G. Nature. 1993; 364: 450-453Crossref PubMed Scopus (268) Google Scholar).Table 1Effects of NOS Inhibitors on LearningSpeciesTaskTraining (Trails/Day)Test TimeLearning DeficitReferencesRatSpatial Learning”Watermaze4Daily and TransferYes1””5DailyPartial2, 3, 4””6Daily and TransferPartial5””1”No””Eight-arm maze1DailyPartial6”Three-panel runway6DailyYes7”Social memory130 minYes6”Olfactory memory (mice)124 hrNo8”Passive avoidance124 hrNo6””124 hrYes9Chick”130 min to 24 hrYes10, 11, 12RabbitEyeblink conditioning120DailyPartial1GoldfishVOR adaptation3 hrDuringPartial13OctopusTactile discrimination2DailyYes14, 15HoneybeePER conditioning324 hrPartial16””1”No”1. Chapman et al. (1992). Neuroreport 3, 567–570. 2. Estall et al. (1993). Pharmacol. Biochem. Behav. 46, 959–962. 3. Mogensen et al. (1995). Neurobiol. Learning Memory 63, 54–65. 4. Mogensen et al. (1995). Neurobiol. Learning Memory 64, 17–24. 5. Bannerman et al. (1994). J. Neurosci. 14, 7404–7414. 6. Bohme et al. (1993). Proc. Natl. Acad. Sci. USA 90, 9191–9194. 7. Ohno et al. (1993). Brain Res. 632, 36–40. 8. Brennan and Kishimoto (1993). Brain Res. 619, 306–312. 9. Fin et al. (1995). Neurobiol. Learning Memory 63, 113-115. 10. Holscher and Rose (1992). Neurosci. Lett. 145, 165–167. 11. Holscher and Rose (1993). Brain Res. 619, 189–194. 12. 8Holscher C. Learning Memory. 1994; 1: 213-216PubMed Google Scholar. Learning Memory 1, 213–216. 13. Li et al. (1995). J. Neurophys. 74, 489–494. 14. Robertson et al. (1994). Proc. Roy. Soc. (Lond.) B 256, 269–273. 15. Robertson et al. (1995). Proc. Roy. Soc. (Lond.) B 261, 167–172. 16. 14Müller U. Neuron. 1996; 16 (this issue)Abstract Full Text Full Text PDF PubMed Scopus (175) Google Scholar. Neuron 16, this issue. Open table in a new tab 1. Chapman et al. (1992). Neuroreport 3, 567–570. 2. Estall et al. (1993). Pharmacol. Biochem. Behav. 46, 959–962. 3. Mogensen et al. (1995). Neurobiol. Learning Memory 63, 54–65. 4. Mogensen et al. (1995). Neurobiol. Learning Memory 64, 17–24. 5. Bannerman et al. (1994). J. Neurosci. 14, 7404–7414. 6. Bohme et al. (1993). Proc. Natl. Acad. Sci. USA 90, 9191–9194. 7. Ohno et al. (1993). Brain Res. 632, 36–40. 8. Brennan and Kishimoto (1993). Brain Res. 619, 306–312. 9. Fin et al. (1995). Neurobiol. Learning Memory 63, 113-115. 10. Holscher and Rose (1992). Neurosci. Lett. 145, 165–167. 11. Holscher and Rose (1993). Brain Res. 619, 189–194. 12. 8Holscher C. Learning Memory. 1994; 1: 213-216PubMed Google Scholar. Learning Memory 1, 213–216. 13. Li et al. (1995). J. Neurophys. 74, 489–494. 14. Robertson et al. (1994). Proc. Roy. Soc. (Lond.) B 256, 269–273. 15. Robertson et al. (1995). Proc. Roy. Soc. (Lond.) B 261, 167–172. 16. 14Müller U. Neuron. 1996; 16 (this issue)Abstract Full Text Full Text PDF PubMed Scopus (175) Google Scholar. Neuron 16, this issue. NOS inhibitors have previously been shown to interfere with learning and memory in a variety of other species including rat, chick, rabbit, goldfish, and octopus, and in a number of tasks, including spatial learning, social and olfactory memory, passive avoidance, eyeblink conditioning, vestibular–ocular reflex (VOR) adaptation, and tactile discrimination (Table 1). In vertebrates, these tasks are thought to be dependent on several different brain areas including hippocampus and cerebellum. The inhibitor studies have generally included appropriate controls for the specificity of action of the drugs and for possible effects on processes other than learning and memory, such as sensory capabilities, motor performance, and motivation. A consistent finding of these studies has been that NOS inhibitors produce a deficit when given before training but not before subsequent testing; that is, they interfere with acquisition but not with retention of memory. Furthermore, the deficit is often partial: acquisition is delayed, but with continued training the animals eventually learn. In addition, there have been a few reports of failures of NOS inhibitors to interfere with learning of similar tasks (Table 1). This mixed pattern of results has sometimes been taken as evidence that NO is not really involved in learning. This conclusion is based on an underlying assumption that learning is a unitary process and that there should be a 1:1 correspondence between it and NOS-dependent mechanisms. However, the results of Müller in honeybee suggest a different interpretation: that there are multiple, overlapping memory systems, only some of which involve NO. If these systems are engaged to different extents under different circumstances, then inhibiting NOS may have different effects in different experiments. Thus, for example1Bannerman D.M. Chapman P.F. Kelly P.A.T. Butcher S.R. Morris R.G.M. J. Neurosci. 1994; 14: 7404-7414Crossref PubMed Google Scholar found that NOS inhibitors block spatial learning in rats with multiple training trials per day but not with one trial per day and concluded that NO is not really involved in learning. By contrast, based on similar results on conditioning in honeybee, Müller concluded that NO is involved in only one of several types of learning. A variation on this pluralistic view is that redundant memory systems can compensate when the primary system is not functioning. This idea has received support from experiments showing that the delayed learning that occurs in the presence of NOS inhibitors appears to engage learning systems that are not normally strongly engaged (12Mogensen J. Wortwein G. Hasman A. Nielsen P. Wang Q. Neurobiol. Learning Memory. 1995; 63: 54-65Crossref PubMed Scopus (51) Google Scholar, 13Mogensen J. Wortwein G. Gustafson B. Ermens P. Neurobiol. Learning Memory. 1995; 64: 17-24Crossref PubMed Scopus (32) Google Scholar). Unfortunately, the particular pattern of results that Müller found for conditioning in the honeybee does not seem to apply generally to other species and tasks; the effectiveness of NOS inhibitors can not in general be explained by number of training trials or time between training and testing (Table 1). A more general explanatory concept might be that the effect of NO depends on the pattern of activity in the target neurons, which would be different under different behavioral circumstances. This idea has received experimental support in vitro (18Zhuo M. Kandel E.R. Hawkins R.D. Neuroreport. 1994; 5: 1033-1036Crossref PubMed Scopus (103) Google Scholar), but is more difficult to test in vivo. Another difficulty in testing the role of NO in vivo has been that most NOS inhibitors act not only on neuronal NOS but also on endothelial NOS, which causes large changes in blood pressure in vertebrates. This may not be a problem in honeybee, which is not known to have endothelial NOS. A new inhibitor that is relatively specific for neuronal NOS has recently been found to produce learning deficits in chicks similar to those previously obtained with conventional NOS inhibitors, supporting a role of NO in vertebrate learning (8Holscher C. Learning Memory. 1994; 1: 213-216PubMed Google Scholar). As in studies of learning, NOS inhibitors have produced a mixed pattern of results in studies of LTP in hippocampus and LTD in cerebellum: the inhibitors block induction but not maintenance, the block is sometimes partial, and there are reports of failure to produce any effect at all (2Bannerman D.M. Chapman P.F. Kelly P.A.T. Morris R.G.M. J. Neurosci. 1994; 14: 7415-7425PubMed Google Scholar, 10Linden D.J. Connor J.A. Eur. J. Neurosci. 1992; 4: 10-15Crossref PubMed Scopus (77) Google Scholar). Moreover, a number of studies have shown that NOS inhibitors block LTP under some experimental circumstances but not others (reviewed by7Hawkins R.D. Zhuo M. Arancio O. J. Neurobiol. 1994; 25: 652-665Crossref PubMed Scopus (124) Google Scholar). Although there are exceptions, in several cases NOS inhibitors have been more effective in blocking LTP when the tetanic stimulation producing potentiation was weaker. One possible interpretation of these results is that the inhibitors do not completely block NOS with strong tetanic stimulation. Another possible interpretation is that there are multiple mechanisms of LTP (only some of which involve NO) that are engaged to different extents under different circumstances. For example, if LTP involves both pre- and postsynaptic mechanisms but NO is involved in only the presynaptic effect, then inhibiting NOS would produce a greater block of LTP under circumstances that favor the relative importance of presynaptic changes. Thus far, genetic knockouts have failed to support a role of NO in either hippocampal LTP or cerebellar LTD (15O'Dell T.J. Huang P.L. Dawson T.M. Dinerman J. Snyder S.H. Kandel E.R. Fishman M.C. Science. 1994; 265: 542-546Crossref PubMed Scopus (360) Google Scholar, 11Linden D.J. Dawson T.M. Dawson V.L. J. Neurosci. 1995; 15: 5098-5105Crossref PubMed Google Scholar). Mice with a genetic knockout of neuronal NOS have normal LTP, although NOS inhibitors block LTP in the knockout animals. A relatively specific inhibitor of neuronal NOS has also recently been shown to block LTP in normal animals (5Doyle L. Holscher C. Rowan M.J. Anwyl R. J. Neurosci. 1996; 16: 418-424Crossref PubMed Google Scholar). One possible explanation of these results is that endothelial NOS, which is present in hippocampal neurons, can compensate for neuronal NOS in the knockout animals. This possibility should be testable in experiments on animals with a double knockout of both neuronal and endothelial NOS. Finally, attempts to produce either hippocampal LTP or cerebellar LTD by applying NO have also had mixed success (7Hawkins R.D. Zhuo M. Arancio O. J. Neurobiol. 1994; 25: 652-665Crossref PubMed Scopus (124) Google Scholar, 3Boulton C.L. Irving A.J. Southam E. Potier B. Garthwaite J. Collingridge G.L. Eur. J. Pharmacol. 1994; 6: 1528-1535Google Scholar, 10Linden D.J. Connor J.A. Eur. J. Neurosci. 1992; 4: 10-15Crossref PubMed Scopus (77) Google Scholar). This inconsistency may arise in part because NO can produce opposite effects in the same neurons depending on the experimental circumstances (18Zhuo M. Kandel E.R. Hawkins R.D. Neuroreport. 1994; 5: 1033-1036Crossref PubMed Scopus (103) Google Scholar) and in part because many NO donors are themselves toxic. Recent experiments with a new, nontoxic caged NO have shown that releasing NO inside cerebellar Purkinje cells can produce long-lasting depression (9Lev-Ram V. Makings L.R. Keitz P.F. Kao J.P. Tsien R.Y. Neuron. 1995; 15: 407-415Abstract Full Text PDF PubMed Scopus (155) Google Scholar), supporting a role of NO in LTD.