UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Modulation of long-term potentiation in the hippocampus Xie, Zheng 1994

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata


831-ubc_1994-954132.pdf [ 6.36MB ]
JSON: 831-1.0088108.json
JSON-LD: 831-1.0088108-ld.json
RDF/XML (Pretty): 831-1.0088108-rdf.xml
RDF/JSON: 831-1.0088108-rdf.json
Turtle: 831-1.0088108-turtle.txt
N-Triples: 831-1.0088108-rdf-ntriples.txt
Original Record: 831-1.0088108-source.json
Full Text

Full Text

MODULATION OF LONG-TERM POTENTIATION IN THE HIPPOCAMPUSByZHENG XIEM.D., Jinan University, Guangzhou, China, 1986A THESIS SUBMITTED IN PARTIAL FULFILLMENT OFTHE REQUIREMENTS FOR THE DEGREE OFDOCTOR OF PHILOSOPHYinTHE FACULTY OF GRADUATE STUDIESDEPARTMENT OF PHARMACOLOGY & THERAPEUTICSFACULTY OF MEDICINEWe accept this thesis as conformingto the required standardTHE UNIVERSITY OF BRITISH COLUMBIAMarch, 1994©Zheng Xie 1994In presenting this thesis in partial fulfilment of the requirements for an advanceddegree at the University of British Columbia, I agree that the Library shall make ftfreely available for reference and study. I further agree that permission for extensivecopying of this thesis for scholarly purposes may be granted by the head of mydepartment or by his or her representatives. it is understood that copying orpublication of this thesis for financial gain shall not be allowed without my writtenpermission.(Signature)Department of iThe University of British ColumbiaVancouver, CanadaDate / , / 1jDE-6 (2188)ABSTRACTBrief trains of high-frequency stimulation of monosynaptic excitatorypathways in the hippocampus can cause a long-term synaptic potentiation (LTP)that can last for hours in vitro or weeks in vivo. This activity-mediated LTP isthought to be involved in certain forms of learning and memory.Although mechanisms underlying the induction and maintenance of LTPprocess have been extensively investigated, factors that modulate LTP remainunresolved. Suppression of y-aminobutyric acid (GABA) -ergic inhibition byGABA antagonists can facilitate LTP of the excitatory postsynaptic potential(EPSP) in the hippocampus. It is, however, unclear whether long-term changesin GABAergic inhibition occur after a tetanic stimulation, and if so, how such achange affects LTP of the EPSP. In this project, experiments were conducted todetermine if the CAl neuronal inhibitory postsynaptic potentials (IPSPs) wereaffected following a high frequency stimulation. Somatostatin co-exists withGABA in some interneurons in the CAl area. Since the peptide might be coreleased with GABA, experiments were carried out to determine if somatostatinmodifies GABAergic IPSPs. Whether the peptide could modulate the inductionof LTP of the EPSP was also tested. Previous studies from this laboratoryshowed that substances collected from the hippocampus or the neocortex duringa tetanic stimulation could induce LTP if applied on hippocampal slices. It isgenerally believed that the induction of LTP requires the activation ofpostsynaptic NMDA receptors and that the maintenance of LTP is at least in partdue to a presynaptic mechanism. It is possible that a release of substances fromthe postsynaptic cells or the nearby glia during a tetanic stimulation of thehippocampal afferents could result in a retrograde interaction of thesesubstances with the presynaptic terminals leading to changes that sustain LTP.IIIIn the present study, these substances were further characterized, and themechanisms of their release as well as LTP-inducing action elucidated. Reportsin the literature suggest that a deficiency in cL-tocopherol (vitamin E) leads toenhancement in lipid peroxide content in the hippocampus as well as impairmentof spatial learning. Oxygen free radicals have also been shown to acceleratethe decay of established LTP. Studies were, therefore, initiated to examine if thevitamin causes LTP and if its deficiency leads to impairment of LTP-induction.EPSPs and IPSPs were recorded from CAl neurons in guinea pig hippocampalslices in response to stratum radiatum stimulation.In control CAl neurons, tetanic stimulation of the stratum radiatumcaused LTP of the EPSP and the fast IPSP without changing the slow IPSP. IfBAPTA (a Ca2 chelator) or K-252b (a PKC inhibitor) was injected into the CAlneurons, LTP of the EPSP did not occur. However, in these drug-injectedneurons LTP of the fast IPSP was enhanced, and LTP of the slow IPSP occurredafter tetanus. With the potentiation of the lPSPs in the drug-injected neurons,the shape of the EPSP was distorted in most neurons. These findings indicatethat LTP occurs not only at excitatory synapses but also at inhibitory synapses.The tetanus-induced increases in intracellular free Ca2 and PKC activitypotentiate the EPSP, while decreasing LTP of the IPSPs so that the distortion ofthe EPSP by the IPSPs is minimized for a better expression of LTP of the EPSP.Somatostatin hyperpolarized the CAl neurons and decreased the inputresistance. The peptide depressed both the fast IPSP and the slow IPSP,without changing the EPSP. These actions of somatostatin were not due tointeractions of the peptide with GABAA or GABAB receptors. However, theactivation of GABAB receptors by baclofen, reduced the somatostatin-inducedhyperpolarization of the CAl neurons. The suppression of the GABAA receptormediated IPSP by somatostatin appears to be through a postsynaptic action ofivthe agent. It is concluded that interactions between somatostatin andGABAergic responses may be related to the peptide- and GABA-receptors beingcoupled to the same channels or to the sharing of the same second messengersystems for effects. By modulating the IPSP5, somatostatin could logicallyinterfere with the induction of LTP of the EPSP. However, application ofsomatostatin failed to facilitate or block LTP.Samples were collected from the rabbit neocortex during a tetanic (50 Hz,5 s) stimulation (tetanized neocortical sample, TNS). Application of TNS causedLTP of the EPSP and population spike without changing the membrane potentialor the input resistance of the CAl neurons. The TNS-induced LTP requiredactivation of afferents. TNS also induced a short-term potentiation (STP) of thefast IPSP without changing the slow IPSP. Since the TNS- and tetanus-inducedLTPs occluded each other, these two LTPs may share some commonmechanisms. Different fractions of TNS (<3 kDa, 3-10 kDa and >50 kDa) wereable to cause LTP. APV did not block the TNS-induced LTP. However, PKCinhibitors such as sphingosine and K-252b blocked the TNS-induced LTP. TNSfailed to induce LTP inCa2-chelated CAl neurons. If TNS was collected fromthe rabbits pretreated with MK-801 (i.p.), a non-competitive NMDA antagonist,this TNS failed to induce LTP. Gel electrophoresis of the substances in TNSrevealed the presence of an acidic protein with a molecular weight of about 69kDa. It, therefore, appears that NMDA receptor activation is required for therelease but not for the LTP-inducing action of substances in TNS.x-Tocopherol phosphate (referred to as x-tocopherol) induced a slowlydeveloping LTP of the EPSP without changing the fast and slow IPSPs, and theelectrical properties of the CAl neurons. The agent failed to induce a furtherpotentiation of the EPSP during a pre-established tetanus-induced LTP. The ctocopherol-induced LTP was decreased by AP3 (an ACPD antagonist) but not byVAPV (a NMDA antagonist). Furthermore, chelation of postsynaptic free Ca2with BAPTA or inhibition of PKC by sphingosine and K-252b prevented the ctocopherol-induced LTP. Sodium ascorbate (a water soluble antioxidant) failedto induce LTP. DMSO (a lipid soluble antioxidant) was able to potentiate theEPSP as long as the application of the agent continued, but the EPSP quicklyreturned to the pre-application level once the application of the agent wasstopped. In hippocampal slices obtained from vitamin E deficient rats, bothtetanic stimulation and c*-tocopheroI failed to induce LTP of the EPSP in the CAlneurons. It is possible that the structure and the function of the membrane orcertain receptors on neurons are affected by the lack of c-tocopherol in thevitamin deficient rats and, therefore, an acute application of x-tocopherol maynot able to correct the changes induced by the long-term vitamin E deficiency.In conclusion, various mechanisms that modulate LTP of the EPSP wereexamined in guinea pig hippocampal slices. It appears that the elevation ofpostsynaptic [Ca2]and the activation of PKC are needed not only to cause LTPof the EPSP but also to diminish LTP of the IPSP5 so that the expression of LTPof the EPSP is not distorted. Somatostatin suppresses GABAergic IPSPsthrough mechanisms other than to interfere with the amino acid receptors andthe peptide appears not to affect LTP of the EPSP. Tetanic stimulation of theneocortical surface causes a release of LTP-inducing substances whose releasebut not action, depends on NMDA receptor activation. While vitamin E appearsto induce LTP, animals with the vitamin deficiency appear to ha a diminishedability to induce LTP.Date: 194valaR.Sast,Ph..viTABLE OF CONTENTSCHAPTER PAGEABSTRACT .11TABLE OF CONTENTS viLIST OF FIGURES xivLIST OF TABLES xviiiLIST OF ABBREVIATIONS xixACKNOWLEDGEMENTS xxDEDICATION xxi1. INTRODUCTION I1.1. LTP and endogenous substances 21.2. LTP and a-tocopherol 41.3. LTP and GABAergic inhibition 51.3.1. LTP and IPSPs 51.3.2. LTP and somatostatin 62. THE HIPPOCAMPAL FORMATION 72.1. General anatomy of the hippocampal formation 72.2. Topography and cytoarchitecture of the hippocampus 82.2.1. The dentate gyrus 82.2.2. The cornu ammonis field 112.2.3. The subicular complex 122.2.4. The entorhinal cortex 132.3. Cell morphology of the hippocampus 132.3.1. Dentate gyrus granule cells 132.3.2. Cornu ammonis pyramidal cells 142.3.3. Hilar region mossy cells 162.3.4. Interneurons 162.4. Intrinsic hippocampal circuitry 18VII2.4.1. Transverse circuit 192.4.2. Longitudinal pathway 202.4.3. Local circuits 212.4.4. Commissural afferents 222.5. Extrinsic afferents to the hippocampus 232.5.1. Entorhinal afferents 232.5.2. Septal afferents 242.5.3. Other afferents 252.6. Extrinsic efferents from the hippocampus 252.6.1. Fornix Fimbria system 252.6.2. Entorhinal projection 263. ELECTROPHYSIOLOGY OF HIPPOCAMPAL NEURONS 273.1. Characteristic of hippocampal neuron 273.1.1. Bursting activity of hippocampal neurons 283.2. Hippocampalinterneurons 293.3. Membrane ionic currents of hippocampal pyramidal neurons 313.3.1. Nacurrents 313.3.2. Ca2 currents 323.3.3. K currents 353.3.4. Cl currents 403.3.5. “Leak” currents 413.4. Evoked field potentials in the hippocampus 413.5. Ephaptic interactions 423.6. Electrotonic coupling 433.7. Excitatory postsynaptic potentials (EPSP5) 434. SYNAPTIC TRANSMISSION IN THE HIPPOCAMPUS 444.1. Excitatory neurotransmitters in the hippocampus 44VIII4.1.1. NMDA receptors and their role in synaptictransmission 454.1.2. Non-NMDA glutamate receptors 514.1.3. AP4 receptors 544.1.4. ACPD receptors 554.2. Inhibitory neurotransmitters in the hippocampus 584.2.1. GABAA receptors 584.2.2. GABAB receptors 594.2.3. GABAA and GABA8receptor-mediated responses 624.2.4. GABAA and GABAB receptor-mediated IPSPs 634.2.5. Presynaptic GABAB receptor-mediated actions 644.2.6. Spontaneous IPSPs 664.2.7. Inhibitory Circuitry 674.3. Other neurotransmitters in the hippocampus 684.3.1. Acetylcholine 694.3.2. Noradrenaline 704.3.3. Serotonin 704.3.4. Dopamine 714.3.5. Histamine 714.3.6. Adenosine 714.4. Neuropeptides in the hippocampus 724.4.1. Oploid peptides 724.4.2. Somatostatin 744.4.3. Cholecystokinin 744.4.4. Neuropeptide Y 754.4.5. Other neuropeptides 765. LONG-TERM POTENTIATION IN THE HIPPOCAMPUS 76ix5.1. Characteristics and properties of L TP 765.1.1. Definition and classification of LTP 765.1.2. Distribution of LTP in the hippocampus 785.1.3. Homosynaptic and heterosynaptic LTP 795.1.4. Cooperativity of LTP 795.1.5. Associativity of LTP 805.2. The induction of LTP 815.2.1. Involvement of glutamate receptors in LTP 825.2.2. The role ofCa2 in the induction of LTP 855.2.3. The role of protein kinases in the induction of LTP 875.2.4. The role of GABA receptors in the induction of LTP 905.3. The maintenance of LTP 915.3.1. Postsynaptic versus presynaptic mechanisms 925.3.2. Signal transduction mechanisms 995.3.2.1. Protein kinases 995.3.2.2. Protein synthesis 1015.3.2.3. Immediate early genes 1025.3.2.4. Release of proteins 1035.3.3. Possible retrograde messengers 1045.3.3.1. Arachidonic acid 1045.3.3.2. Nitric oxide and carbon monoxide 1055.3.3.3. Neurotrophic factors 1075.3.3.4. Released K 1085.4. Modulatory factors on LTP 1095.5. LTP in hippocampal interneurons 1105.6. The physiological significance of L TP 1115.7. Summary 112x5.8. Rationale and specific aims 1145.8.1. The role of released substances in LTP 1145.8.2. The influence of a -tocopherol in LTP 1166. PHARMACOLOGICAL TOOLS FOR EXPERIMENTS 1176.1. Somatostatin 1176.1.1. Distribution of somatostatin in hippocampus 1176.1.2. Actions of somatostatin on hippocampal neurons 1186.1.3. The functional roles of somatostatin 1196.2. a-Tocopherol 1216.2.1. Physicochemical properties 1216.2.2. Antioxidant properties 1226.2.3. The functional role of a-tocopherol in the CNS 1237. METHODS AND MATERIALS 1247.1. Animals 1247.2. Perfusion media 1247.3. Slice Chamber 1267.4. Preparation of slices 1267.5. Recording and stimulating systems 1297.5.1. Amplifiers 1297.5.2. Recording electrodes 1297.5.3. Stimulating units 1307.5.4. Recording units 1307.6. Application of drugs 1307.7. Collection of endogenous samples 1317.8. Separation of endogenous samples 1327.9. Gel electrophoresis 1327.10. Extracellular recordings 134xi7.11. Intracellular recordings .1347.12. Induction of LTP 1357.13. Data analysis 1368. RESULTS 1368.1. LTP and IPSPs 1368.1.1. Tetanic stimulation and IPSPs 1408.1.2. Effects of Ca2 chelator on IPSPs 1418.1.3. Effects of protein kinase C inhibitor on IPSP5 1478.1.4. Effects of IPSPs on EPSP 1548.1.5. Effects ofAPV on IPSPs 1548.1.6. IPSP5 in the absence of glutamatergic transmission 1578.2. Effects of somatostatin on GABAergic inhibition and LTP 1618.2.1. Effects of somatostatin on CAl neurons 1618.2.2. Effects of somatostatin on GABA receptors 1658.2.3. Effects of somatostatin during the activation of GABABreceptors 1668.2.4. The role of Ca2 and PKC in the actions ofsomatostatin 1738.2.5. Effects of QX-314 on the actions of somatostatin 1748.2.6. Effects of somatostatin on the induction of LTP 1758.3. LTP and endogenous substances 1788.3.1. LTP-inducing action of the endogenous substances 1828.3.2. Different fractions of the endogenous substances andLTP 1868.3.3. Involvement of NMDA receptors in the release ofendogenous substances 193XII8.3.4. Possible mechanisms of the action of LTP-inducingsubstances 1948.3.4.1. NMDA receptors and the action of LTPinducing substances 1948.3.4.2. ACPD receptors and the action of LTPinducing substances 1948.3.4.3. Protein kinase C and the action of LTPinducing substances 1988.3.4.4. Ca2 and the action of LTP-inducingsubstances 1988.3.5. Gel electrophoresis of the endogenous substances 1988.4. LTP and a-Tocopherol 1998.4.1. Effects of a-tocopherol on LTP 2038.4.2. Effects of a-tocopherol on IPSPs and electricalproperties of CA I neurons 2048.4.3. Possible mechanisms of LTP-inducing action of atocopherol 2088.4.3.1. Glutamate receptors and LTP-inducing actionof a-tocopherol 2088.4.3.2. Protein kinase C and LTP-inducing action ofa-tocopherol 2098.4.3.3. Ca2 and LTP-inducing action of atocopherol 2098.4.3.4. Antioxidant and LTP-inducing action of atocopherol 2158.4.4. LTP in vitamin E deficient rats 2169. DISCUSSION 222XIII9.1. LTP and endogenous substances .2229.1.1. Characterization of the LTP-inducing substances 2239.1.2. Mechanisms of the release of the endogenoussubstances 2259.1.3. Mechanisms of the LTP-inducing action of theendogenous substances 2299.1.4. The possible feedback mechanisms 2329.2. Involvement of a-tocopherolin LTP 2379.2.1. Possible mechanisms for the actions of a-tocopherol 2399.2.2. The significance of the LTP-inducing action of atocopherol 2439.3. Modulation of LTP by GABAergic inhibition 2449.3.1. LTP and IPSPs 2449.3.2. Possible mechanisms for L TP of the IPSPs 2469.3.3. Somatostatin and GABAergic inhibition 2509.3.4. LTP and somatostatin 25310. SUMMARY AND CONCLUSIONS 25511. REFERENCES 259xivLIST OF FIGURESFIGURE PAGEFigure 1 Overview of the hippocampal formation .9Figure 2 Schematic representation of a transverse section of thehippocampus 10Figure 3 Slice chamber and perfusion system 127Figure 4 Preparation for collection of samples 133Figure 5 Schematic illustration for positioning of stimulating andrecording electrodes in hippocampal slices 137Figure 6 Representative field potentials and LTP in hippocampal CAlarea 138Figure 7 Characteristics of intracellular evoked synaptic potentials inhippocampal CAl cells 139Figure 8 Pharmacologically isolated components of intracellularsynaptic responses and effects of tetanic stimulation on thesecomponents 142Figure 9 Effects of tetanic stimulation on EPSP and slow IPSP in thepresence of picrotoxinin 143Figure 10 Effects of tetanic stimulation on evoked synaptic responses inBAPTA injected neurons 145Figure 11 Effects of tetanic stimulation on EPSP and slow IPSP inBAPTA-injected neurons in the presence of picrotoxinin 146Figure 12 Actions of K-252b on the induction of LTP of the EPSP in CAlneurons 149Figure 13 Actions of K-252b on changes of EPSP, fast and slow IPSPsinduced by tetanic stimulation 150Figure 14 Effects of K-252b on post-tetanic changes of EPSP and slowIPSP in the presence of picrotoxinin 151Figure 15 Equilibium potentials for fast and slow IPSP5 in control,BAPTA- and K-252b-injected neurons 152xvFigure 16 Effects of phaclofen on EPSP in K-252b-injected neuronsbefore and after tetanic stimulation 155Figure 17 Effects of phaclofen on EPSP in BAPTA-injected neuronsbefore and after tetanic stimulation 156Figure 18 Effects of APV on post-tetanic changes of IPSPs 158Figure 19 Effects of APV on post-tetanic changes of slow IPSPs 160Figure 20 Post-tetanic changes of IPSPs in the presence of APV andCNQX 162Figure 21 Post-tetanic changes of monosynaptic slow IPSP 163Figure 22 Effects of somatostatin on the membrane potential, inputresistance and evoked synaptic responses in CAl neurons ofhippocampal slices 167Figure 23 Actions of somatostatin in the presence of picrotoxinin 168Figure 24 The effects of phaclofen on the hyperpolarizing actions ofbaclofen and somatostatin in picrotoxinin-treated slices 169Figure 25 Effects of somatostatin on IPSPs in the presence of 2-OH-saclofen 170Figure 26 The interactions between baclofen and somatostatin 171Figure 27 Possible role of protein kinase C and postsynaptic Ca2 in thehyperpolarizing action of somatostatin 172Figure 28 Effects of QX-31 4 on the actions of somatostatin 176Figure 29 Effects of somatostatin on LTP of extracellular EPSPs 179Figure 30 Induction of LTP of intracellular EPSP in the presence ofsomatostatin 180Figure 31 Effects of weak tetanic stimulation in the presence ofsomatostatin 181Figure 32 Effects of substances collected from rabbit neocortical surfaceon guinea pig hippocampal CAl population spike 184Figure 33 Representative LTP of population spike and intracellularEPSP caused by TNS in hippocampal slices 187xviFigure 34 Changes in EPSP, fast and slow IPSP5 produced by TNS 188Figure 35 Effects of TNS on the membrane potential, input resistanceand the slope of EPSP in the CAl neurons 189Figure 36 The ability of different fractions of TNS to induce LTP 190Figure 37 The effects of different fractions of TNS on the populationsspike 191Figure 38 Effect of TNS collected from rabbits pretreated with MK-801 195Figure 39 Effects of APV on the induction of LTP by different molecularweight fractions of TNS 196Figure 40 Effects of AP3 on the induction of LTP produced by TNS 197Figure 41 Involvement of protein kinase C in LTP induced by TNS 200Figure 42 The role of postsynaptic Ca2 in LTP produced by TNS 201Figure 43 Electrophoretic separation of peptides from TNS and UNS 202Figure 44 Effects of x-tocopherol phosphate on the EPSP of guinea pighippocampal CAl neurons 205Figure 45 Actions of x-tocopherol phosphate on the evoked synapticresponses, input resistance and action potential generation inCAl neurons 206Figure 46 Effects of a-tocopherol on the input-stimulus/output-response(I/O) relationship 207Figure 47 Effects of APV on the induction of cc-tocopherol phosphateinducdLTP 210Figure 48 Effects of L-AP3 on the ct-tocopherol-induced LTP 211Figure 49 Effects of cw-tocopherol phosphate on the NMDA and nonNMDA responses 212Figure 50 Effects of protein kinase C inhibitors on the LTP-inducingaction of c-tocopherol phosphate 213Figure 51 Effects of chelation of postsynaptic Ca2 on the LTP-inducingaction of ct-tocopherol phosphate 214xviiFigure 52 The inability of sodium ascorbate to induce LTP in the CAlneurons 217Figure 53 Effects of DMSO on the EPSP of CAl neurons 218Figure 54 The inability of tetanic stimulation to produce LTP in vitamin Edeficient rat hippocampal CAl neurons 219Figure 55 Tetanic stimulation of the stratum radiatum in vitamin Edeficient and control rat hippocampal slices 220Figure 56 The failure of x-tocopherol phosphate to induce LTP in thehippocampal CAl neurons of vitamin E deficient rats 221xviiiLIST OF TABLESTABLE PAGETable I Effects of intracellular injection of BAPTA and K-252b on posttetanic EPSP, fast and slow IPSPs 154Table 2 Effects of intracellular injection of BAPTA and K-252b on posttetanic EPSP and slow IPSP in picrotoxinin-treated slices 154Table 3 Effects of BAPTA and K-252b on post-tetanic EPSP duration 154Table 4 Effects of intracellular injection of BAPTA and K-252b on posttetanic fast and slow IPSPs in the presence of APV and CNQX 164Table 5 Effects of intracellular injection of BAPTA and K-252b on posttetanic slow IPSP in the presence of APV, CNQX andpicrotoxinin 164Table 6 Effects of various fractions of samples collected from the rabbitneocortical surface on the CAl population spike in guinea pighippocampal slices 192xixLIST OF ABBREVIATIONSABBREVIATION WORDACPD (±)-1 -aminocyclopentane-trans-1 ,3-dicarboxylic acidAMPA (RS)-x-amino-3-hydroxy-5-methyI-4-soxazolepropionic acidAP3 L(+)-2-amino-3-phosphonopropionic acidAPV DL-2-amino-phosphonovalerateAHP AfterhyperpolarizationATP Adenosine triphosphateBAPTA I ,2-bis(2-aminophenoxy)etheane-N’,N’, N’,N’tetraacetic acidcAMP Adenosine 3’:5”-cyclic phosphateCA Cornu ammonisCaMKII a2/caImodulin-dependent protein kinase IICNQX 6-Cyano-7-nitroquinoxaline-2,3-dioneDG Dentate gyrusEPSP Excitatory postsynaptic potentialGABA y-aminobutyric acidGAD Glutamic acid decarboxylaseH-7 I -(5-isoquinolinylsulfonyl)-2-methyl-piperazineIPSP Inhibitory postsynaptic potentialK-252b 9-carboxylic acid derivative of [8R*,9S*,1 I S*][]9hydroxy-9-methoxycarbonyl-8-methyl-2, 3,9,1 0-tetra-hydro-8,1 1-epoxy-I H,8H,I I H-2,7b,I Ia-triazadibenzo[a,g] cycloocta [cde] trinden-I -oneLTP Long-term potentiationMK-801 [+1-5-methyl-I 0,11 -dihydro-5H-dibenzo (a, b)cyclohepten-5, I 0-imine maleateNMDA N-methyl-D-aspartatePC-12 cells Rat adrenal pheochromocytoma cellsPKA Protein kinase APKC Protein kinase CPP Perforant pathwayQX-31 4 N-(2,6-dimethyl-phenylcarbamoylmethyl)-triethylammonium bromideSch Schaffer collateralsSS SomatostatinSTP Short-term potentiationTTX TetrodotoxinTNS Tetanized neocortical sampleUNS Untetanized neocortical sampleTUNS UNS collected after a previous tetanic stimulationxxACKNOWLEDGEMENTSI would like to first thank my supervisor Dr. Bhagavatula R. Sastry for hisencouragement, assistance and academic guidance throughout this study. I amgrateful to Mr. Wade Morishita for his friendship, valuable discussion andcollaboration on some experiments in this study. Also, I thank Dr. Timothy Kam,Mr. Joseph Ip, Mr. Trevor Shew, Mr. Samuel Yip, Ms. Gitanjali Adlakha and Dr.Hermina Maretic for their help and friendship. I wish to thank all the members ofthe Department of Pharmacology & Therapeutics, particularly Dr. MichaelWalker, Dr. David Godin, Dr. Ernest Puil, Ms. Janelle Swetnam and Ms.Margaret Wong for their help and guidance. I particularly thank Ms. May Chanfor her emotional support and help. Lastly, I would like to thank my parents,sister and brother for their selfless love and support.Financial support from the studentship of Medical Research Council ofCanada and the University of British Columbia Graduate Fellowship is greatlyappreciated.DEDICATIONDedicated to my mother and father.xxiXIE I1. INTRODUCTIONBrief trains of high-frequency stimulation of monosynaptic excitatorypathways in the hippocampus can cause an abrupt and sustained increase insynaptic efficiency that can last for hours in vitro or weeks in vivo. This activity-dependent enhancement of synaptic transmission is referred to as long-termpotentiation (LTP) (Bliss and Lomo, 1973). This LTP is characterized as aninput specific long-term enhancement in the excitatory postsynaptic potentials(EPSP) of a single neuron or in the field potentials of a population of neurons.LTP is thought to be a strong candidate for the cellular mechanisms underlyinglearning and memory because it is consistent with the “Hebbian rule” that asynaptic modification for learning and memory occurs as a consequence ofcoincidence between pre- and post-synaptic activity (Hebb, 1949). There isgrowing evidence that LTP is correlated with some forms of learning andmemory (Silva et al., 1992a,b; Grant et al., 1992; Bolhuis and Reid, 1992; Daviset al., 1992; Watanabe et al., 1992). In the last decade, efforts have beenfocused on elucidating the mechanisms underlying the induction andmaintenance of LTP. It is generally believed that the induction of LTP isprimarily mediated through postsynaptic mechanisms (McNaughton, 1982;Malenka et al., 1988). In the CAl regions of the hipppocampus the induction ofthe LTP appears to involve the activation of the N-methyl-D-aspartame (NMDA)receptors (Collingridge et al., 1983). Whether the maintenance of LTP is due topresynaptic (increase in neurotransmitter release) and/or postsynaptic (changein receptor/channel complex or increase in receptor number) mechanisms isdebatable (Bliss and Collingridge, 1993; Otani and Ben-An, 1993). The LTPprocess, including the induction and maintenance, is known to be influenced bya variety of factors such as neurotransmitters, neuromodulators andXIE 2neurohormones (Otani and Ben-An, 1993). The major objective of the studiesreported in this thesis was to examine factors which are involved in themodulation of the LTP process. A knowledge of how these factors modulate theLTP process may lead to a better understanding of the mechanisms of LTP.1.1. LTP and endogenous substancesThe current view on LTP in hippocampal CAl region is that tetanicstimulation activates postsynaptic NMDA receptors which allow Ca2 influxthrough NMDA channels, (Collingridge et al., 1983). The rise of intracellular freeCa2 concentration triggers the activation of a number of protein kinases whichprobably leads to postsynaptic and/or presynaptic modifications andsubsequently the induction of LTP (Otani and Ben-Ari,1993). If presynapticmechanisms such as increases in neurotransmitter release are responsible forthe maintenance of LTP, interactions between the postsynaptic cell and thepresynaptic terminals must occur. In addition to tetanic stimulation of afferents,direct depolarization of the CAl neurons with intracellular current injection, whenpaired with stimulation of afferents, can also induce LTP of the EPSP (Sastry etal., 1986; Gustaffson and Wigstrom, 1986; Kelso et al., 1986). Depolarization ofthe postsynaptic neurons not only facilitates the opening of NMDA channels byremoving Mg2 block (Mayer et al, 1984, Gustafsson et al., 1987), but alsocauses release of some substances, such as proteins into extracellular fluid(Duffy et al., 1981; Nystrom et al., 1986; Sastry et al., 1988a, Chirwa and Sastry,1986). It has been speculated that these released substances can interact withthe presynaptic terminals leading to the induction of LTP. Therefore, retrogrademessengers have been proposed to describe the substances which are releasedfrom the postsynaptic cells during tetanic stimulation, or injection of depolarizingcurrent pulse and act on the presynaptic terminals. (Sastry et al., 1986; Bliss etal., 1986). Retrograde messengers presumably play a key role in mediating theXIE 3interactions between the postsynaptic cell and the presynaptic terminals duringthe development of LTP. Direct evidence for existence of retrogrademessengers is lacking at the present time. Ideally, retrograde messengersshould be released from postsynaptic cells and act quickly on the presynapticterminals leading to the induction of LTP. Arachidonic acid and nitric oxide havebeen proposed to serve as retrograde messengers for LTP (Williams et al.,1989; Bohme et al., 1991; Schuman and Madison, 1991). Both agents arehighly diffusible and are able to induce LTP (Williams et al., 1989; Bohme et al.,1991; Zhuo et al., 1993). However, unlike tetanic stimulation which induces LTPvery rapidly, arachidonic acid (30 mm application) induces a slowly developingLTP (Williams et al., 1989). Nitric oxide can only induce LTP in hippocampalCAl neurons when paired with weak tetanus of afferents (Zhuo et al., 1993).Furthermore, nitric oxide synthesis has not yet been found in the CAl neurons(Bredlt et al., 1991). It appears that the notion that arachidonic acid and nitricoxide act as retrograde messengers for LTP is questionable. Previous studies inour laboratory first demonstrated that endogenous substances collected fromhippocampus (Chirwa and Sastry, 1986) and neocortex (Sastry et al., 1988a)during tetanic stimulation in vivo, when applied to the hippocampal slices, induceLTP in these slices. Like tetanic stimulation, the endogenous substances caninduce a rapidly developing LTP in the CAl neurons of the hippocampus (Sastryet al., 1988a). These substances can also act as trophic factors to enhanceneurite growth in cultured PC-12 cells (Sastry et al., 1988a). If theseendogenous substances are released from the postsynaptic cells during tetanicstimulation, they may serve as retrograde messengers in LTP. In the presentstudy, experiments were conducted to determine the possible mechanismsunderlying the release and LTP-inducing action of the endogenous substancesXIE 4collected from rabbit neocortex during tetanic stimulation in guinea pighippocampal slices.1.2. LTP and a-tocopherolAlthough LTP has been widely studied for the last two decades, thephysiological significance of LTP is unclear. LTP is thought to underlie thecellular mechanisms for learning and memory although clear evidence for this islacking at present. Investigations on effects of factors which are involved in thememory process and diseases associated with loss of memory such asAlzheimer’s disease on LTP will help us to better understand the physiologicalsignificance of LTP. x-Tocopherol (vitamin E) is a major antioxidant in thebiological system (Tappel, 1962). x-Tocopherol, which can scavenge freeradicals attacking from outside of the membrane and within the membrane, isessential for the maintenance of normal structure and function of the humannervous system (Tappel, 1962; Sokol, 1989). The vitamin has been suggestedto be involved in spatial learning and memory in animals (Moriyama et al., 1990).In vitamin E deficient rats, impairment of spatial learning ability is associatedwith an increase in lipid peroxide contents of the hippocampus (Moriyama et al.,1990). lntracerebral administration of liposomes containing cL-tocopherol hasbeen demonstrated to facilitate the recovery of learning ability in rats with braininjury (Stein et al., 1991). In cultured central neurons, cL-tocopherol has beenshown to act as a growth-inducing factor which supports the survival of theneurons and enhances neurite growth (Nakajima et al., 1991; Sato et al., 1993).Changes in cL-tocopherol and free radical levels have been found in patients withAlzheimer’s disease (Jeandel et al., 1989). It thus appears that cc-tocopheroland free radicals have a role in certain types of learning and memory. SinceLTP is thought to underlie the cellular mechanisms for learning and memory, it islogical to speculate that a-tocopherol and free radicals are probably involved inXIE 5LTP. Recently, free radicals have been found to facilitate the decay of LTP(Pellmar et al., 1991). Therefore, it is of interest to determine whether -tocopherol plays a role in LTP. In the present studies, experiments wereconducted to examine the role of c-tocopherol in LTP using hippocampal slicesobtained from both contro.l animals and vitamin E deficient animals.1.3. LTP and GABAergic inhibitionGABAergic inhibition is known to play an important role in LTP. Blockadeof GABAA receptors facilitates the induction of LTP because suppression of theGABAA receptor-mediated fast inhibitory postsynaptic potentials (IPSP) allowsthe NMDA channels to open more readily (Wigstrom and Gustafsson, 1983).The role of GABAB receptors in LTP was not clear until recently (Davies et al.,1991; Mott and Lewis, 1991). Presynaptic GABAB receptors have been reportedto regulate the release of GABA from presynaptic terminals through a negative-feedback mechanism (Davies et al., 1990). Activation of presynaptic GABABreceptors has been suggested to facilitate the induction of LTP by suppressingGABA release and reducing the IPSPs (Davies et al., 1991; Mott and Lewis,1991). The GABAB receptor-mediated slow IPSP occurs in the apical dendritesclose to the site where the EPSP occurs. Phaclofen, a postsynaptic GABABantagonist, has been shown to facilitate the induction of LTP (Olpe andKarlsson, 1990). It is thus apparent that changes in GABAergic inhibition havemodulatory effects on the induction of LTP of the EPSP. Under normalconditions, factors that alter the GABAergic inhibition should also affect theinduction of LTP.1.3.1. LTP and IPSPsStimulation of the stratum radiatum not only evokes EPSPs in the CAlneurons, but also induces fast and slow IPSPs in these neurons. While longterm changes in the EPSP after tetanic stimulation have been widely studied, theXIE 6long-term changes in the IPSPs have been somewhat ignored and are lessclear. Tetanic stimulation has been reported to cause an increase, a decrease,or no change in the IPSPs (Abraham et al., 1987). The inconsistent results arepartly due to the IPSPs in the CAl neurons consisting of monosynaptic andpolysynaptic components. Protein kinase C (PKC) activation and Ca2 influx areknown to occur during the induction of LTP in the CAl neurons (Malinow et al.,1989; Malenka et al., 1989). PKC activation and the rise of intracellular Ca2concentration have been reported to have significant effects on IPSPs in theCAl neurons (Baraban et al., 1985; Dutar and Nicoll, 1988b; Chen et al., 1990).Since changes in IPSPs have been shown to have significant modulatory effectson LTP, the present studies were conducted to examine the long-term effects oftetanic stimulation on IPSPs and the role of the IPSPs changes in LTP.1.3.2. LTP and somatostatinSomatostatin (SS), a peptide that co-exists with GABA in the inhibitoryterminals, has been shown to depress the GABA receptor-mediated IPSPs in theCAl neurons (Scharfman and Schwartzkroin, 1989). The mechanismsunderlying the action of SS on the IPSPs are not clear. SS has also beenreported to hyperpolarize the CAl neurons (Pittman and Siggins, 1981).Depression of the IPSPs and hyperpolarization of the CAl neurons canmodulate the induction of LTP of the EPSP as previously discussed (Section1.3). If SS and GABA are co-released from the same presynaptic terminalsduring tetanic stimulation, the actions of SS on the IPSPs and the membranepotential can affect the induction of LTP of the EPSP in the CAl neurons.Furthermore, depletion of SS in the CNS has been shown to impair animalspatial learning ability, which is thought to be related to hippocampal LTP(Haroutunian et al., 1987). Therefore, it would be of interest to determine therole of this endogenous peptide in LTP and the mechanisms of the interactionsXIE 7between SS and GABA in the CAl neurons. Experiments were conducted toexamine these issues.2. THE HIPPOCAMPAL FORMATIONA comprehensive literature review on the hippocampal formation is usefulfor better interpretation of the electrophysiological signals of the hippocampus.Various nomenclatures have been used to describe the subdivisions of thehippocampus. The terminology introduced by Cajal (1911), Lorente de No(1934) and Blackstad (1956) is widely used in the literature and has beenconsistently employed in this review. The updated information in anatomicalstudies has been included in the discussion. (Teyler and DiScenna, 1984;Schwerdtfeger, 1984; Swanson et al. 1987; Amaral and Witter, 1989; Lopes DaSilva et al., 1990; Bronen, 1992).2.1. Genera! anatomy of the hippocampa! formationThe cortex is divided into two basic types; the allocortex and theneocortex, during ontogenic development (Chronister and White, 1975). Theneocortex, also called the isocortex, is a homogenous unit which completelyseparates from the mantle layer. The allocortex is a heterogeneous unit whichfails to cleave completely from the mantle layer during development (Filimonoff,1947). The regions which lie between the allocortex and the neocortex arecalled periallocortex. These are the peripalaeocortical claustral region, theentorhinal region, the presubicular region, the retrosplenial region and theperiarchicortical cingulate region (Lorente de No, 1934; Stephan, 1975;Chronister and White, 1975). The allocortex is separated into palaeocortex andarchicortex. The palaeocortex is composed of the olfactory bulb and accessoryolfactory bulb, the retrobulbar region, the periamygdalar region, the olfactorytubercie, the septum, the diagonal region and the prepiriform region(Schwerdtfeger, 1984). The archicortex consists of the subiculum, cornuXIE 8ammonis, dentate gyrus, precommissural hippocampus and supracommissuralhippocampus (Chronister and White, 1975; Schwerdtfeger, 1984; Teyler andDiSenna, 1984). Note the terms “archicortex”, “hippocampus” and “hippocampalformation” are synonymous (Schwerdtfeger, 1984; Teyler and DiScenna, 1984;Lopes Da Silva et al., 1990). In this review, the hippocampal formation or thehippocampus refers to the unit consisting of the cornu ammonis, the dentategyrus, the subicular complex and the entorhinal cortex (Swanson et al., 1987;Amaral and Witter, 1989).2.2. Topography and cytoarchitecture of the hippocampusThe hippocampus is a convoluted symmetrical structure, forming themedial margin of the cortical hemisphere and is located on the medial wall of thelateral ventricle. The appearance of the curved and intertwined simple corticallayers account for the name hippocampus, meaning seahorse (Knowles, 1992).The three-dimensional shape of the hippocampus is relatively complex. Thelongitudinal axis, generally referred to as the septotemporal axis, bends in a C-shaped manner from the septal nuclei rostrodorsally to the incipient temporallobe caudoventrally. The transverse axis is oriented perpendicular to theseptotemporal axis (Fig. 1). A transverse section of the hippocampus revealstwo C-shaped interdigitating archicortical fields: the cornu ammonis and thedentate gyrus (Fig. 2) (Teyler and DiScenna, 1984; Amaral and Witter,1989;Witter, 1989).2.2.1. The dentate gyrusThe dentate gyrus consists of a densely packed cell layer, with granulecells as the major cell type. The C-shaped cell layer is named the stratumgranulosum. The apical dendrites of granule cells orient away from the centre ofXIE 9DorsalFigure 1 Overview of the hippocampal formation.The drawing shows the position of the hippocampal formation in guinea pig brainwith part of the neocortex overlying the hippocampus removed. Note thehippocampus is a C-shaped structure with its longitudinal axis running from theseptal (S) nuclei rostrodorsally to the temporal (T) lobe caudoventrally. Thetransverse axis of the hippocampus is oriented perpendicular to the longitudinalaxis. (Modified from Andersen et al., 1971a)XIE 10Figure 2 Schematic representation of a transverse section of the hippocampus.The major structures and afferents of the hippocampal formation are illustratedin a transverse section of the hippocampus. Note the dentate gyrus (DG) andthe cornu ammonis (CA) are two major structures in the hippocampal formation.The hilus (hil) is a transition zone bewteen the dentate gyrus and the cornuammonis. The cornu ammonis contains a densely packed pyramidal cell layer;and the dentate gyrus consists of a densely packed granule cell layer. Theperforant path (pp) is a major extrinsic input to the hippocampus innervating thegranule cells. The axons of the granule cells, the mossy fibres (mf), makesynapse with the pyramidal cells in CA3 field. The CA3 pyramidal cells give offthe Schaffer collaterals (Sch) to innervate CAl pyramidal cells. Thecommissural (comm) input originates from the contralateral hippocampus. Theaxons of CAl pyramidal cells enter the alveus (alv) and then project into thesubiculum (SUB) and the fimbria (ElM). The shadow in the transverse section isenlarged in the inset to show the arrangement of the layers of the hippocampus.basilar dendrizessomavew.S. pinida1eSch&radiatum -apical dendrnesS Iacwosum/motecularefissureS. moireulareS. granulosumnsf s. poIymopheXIE 11the “C’ and project into the stratum moleculare layer that lies directly adjacent tothe hippocampal fissure. The axons of granule cells, termed mossy fibres,project toward the centre of the “C” (Cajal, 1911; Lorente de No, 1934).Because of the curvature of the dentate regions around part of the cornuammonis, the dentate granular layer is divided into a suprapyramidal layer(upper blade) and an infrapyramidal layer (lower blade) (Chronister and White,1975; Teyler and DiScenna, 1984). The field between the two blades is calledhilar region, which contains several layers of polymorphic cells (Cajal, 1911;Lorente de No,1934). Opinions are split on whether the hilar region belongs tothe dentate gyrus or the cornu ammonis (Cajal,1911; Lorente de No, 1934;Blackstad, 1956). Blackstad (1956) considered it as the third layer of thedentate gyrus. Amaral (1978) described at least 21 different cell types in thehilus of rat dentate gyrus, including basket cells and mossy cells. His studiesconfirmed that the cells in the area are most closely related to the dentate gyrus.The hilar region may serve as a transition zone between the dentate gyrus andthe cornu ammonis.2.2.2. The cornu ammonis fieldThe cornu ammonis, also called the hippocampus proper or Ammon’shorn, was divided into four subfields: CAl, CA2, CA3 and CA4 by Lorente de No(1934). The CA4 field corresponding to the poymorphic zone of the dentategyrus as described by Cajal (1911) and Blackstad (1956), has been discussed inthe previous section (section 2.2.1).The cornu ammonis, composed predominantly of pyramidal cells, formsthe other C-shaped cell layer of the hippocampus (Lorente de No, 1934; Teylerand DiScenna, 1984). The pyramidal cell layer is termed the stratumpyramidale. The pyramidal cells contain both apical and basal dendrites. Thesedendrites, together with the axons and their collaterals of the pyramidal cells,XIE 12form another five layers in the cornu ammonis field. There are two layers on thebasal dendrite side of the pyramidal cell layer. The stratum oriens lies next to thestratum pyramidale. It mainly contains the tufts of the basal dendrites of thepyramidal cells, the collaterals of the axons of CA3 pyramidal cells and cellbodies of some interneurons. The alveus is situated next to the stratum oriensand marks the outer boundary of the cornu ammonis. This layer consists of theaxons of pyramidal cells and the incoming fibres. Three layers lie on the apicaldendrite side of the pyramidal cell layer. The stratum radiatum is situated nextto the stratum pyramidale and mainly contains several fibre systems such as theSchaffer collaterals, and some sparse cell bodies. The stratum moleculare liesdirectly adjacent to the hippocampal fissure. It contains predominantly fibresand dendritic terminals. The stratum lacunosum is situated between the stratumradiatum and the stratum moleculare. This layer consists mainly of bundles ofparallel fibres (Cajal, 1911; Lorente de No, 1934; Lopes Da Silva, 1990) (Fig. 2).Some authors combine the stratum moleculare and the stratum lacunosum intothe stratum lacunosum-moleculare (Hjorth-Simonsen, 1977). In the CA3 field, anadditional layer, which mainly contains the mossy fibres from the granule cells,lies between the stratum radiatum and the stratum pyramidale. This layer iscalled the stratum lucidum.2.2.3. The subicular complexThe subicular complex can be divided into three subdivisions: thesubiculum, presubiculum and parasubiculum (Amaral and Witter, 1989). Thesubiculum is considered to constitute the major output structure of thehippocampus. The subiculum consists of three major layers: a deep, thick layerof pyramidal cells with similar structure of the CAl pyramidal cells; anintermediate cell-sparse layer that is more or less continuous with the stratumradiatum of the cornu ammonis; a molecular layer, which is continuous with theXIE 13presubiculum on one side and the CAl field on the other. Fibres from theentorhinal area (the perforant path) run transversely through the subiculum toend in the molecular layer of the cornu ammonis and the dentate gyrus (Lorentede No, 1934; Swanson et al. 1987).2.2.4. The entorhinal cortexThe entorhinal cortex is usually divided into two parts: a lateral and amedial part (Lorente de No, 1934; Hjorth-Simonsen, 1972b; Hjorth-Simmonsenand Jeune, 1972). While the lateral entorhinal cortex sends its fibres mainly tothe septal hippocampus, the medial entorhinal cortex projects to the temporalhippocampus (Witter et al. 1989a,b; Germroth et al., 1991). In the entorhinalcortex, two spinous types of neurons with “long-axon cylinders projecting to thewhite matter”, the spiny stellate cells (the “star” cells) and the pyramidal cells,were described by Cajal (1911) and Lorente de No (1934).2.3. Cell morphology of the hippocampus2.3.1. Dentate gyrus granule cellsGranule cells are the major neurons in the dentate gyrus. The granulecells were first described by Golgi (1886). Further details were given by Sala(1892), Schaffer (1892) and Cajal (1893). These cells form a compact layer,called the stratum granulosum. The body of these neurons is oval shaped andabout 20 by 15 .tm in size. The granule cells are highly polarized in rodents.The dendrites of these neurons extend into the stratum moleculare where theyreceive the input of the entorhinal cortex through the perforant path fibres. Ingeneral, spines are confined to segments beyond the first branch of the stemdendrites (Williams and Matthysse, 1983). In humans and monkeys, asignificant population of granule cells has basal dendrites in the hilus. Somebasal dendrites curve up into the molecular layer where they have similarmorphology as the apical dendrites. Other basal dendrites in the hilus areXIE 14shorter, thinner and only have a few side branches (Seress and Mrzljak, 1987).The granule cells give rise to distinctive axons, the mossy fibres, whichcollateralize in the polymorphic layer before entering the CA3 field where theyform en passant synapses on the proximal dendrites of the pyramidal cells. Themossy fibres are mainly organized in a lamellar fashion (Amaral and Witter,1989).2.3.2. Cornu ammonis pyramidal cellsPyramidal cells are densely packed in the stratum pyramidale. The pear-shaped pyramidal cell body is on an average 40 by 20 tm in size. However, thesize of pyramidal cells varies in different subfields of the cornu ammonis. TheCAl field has the smallest pyramidal cells whereas the CA3 field has the largestcells (Lorente de No, 1934). All these pyramidal cells possess both apical andbasal dendrites.There are morphological differences among the subfields of cornuammonis. In the CAl field, the basal dendrites of pyramidal cells extend into thestratum oriens in a bush-like fashion. The apical dendrites travel for somedistance in the stratum radiatum before they ramify. The distal apical dendritesextend into the stratum lacunosum-moleculare. A common characteristic of CAlpyramidal cells is that their shafts have many fine side branches in the stratumradiatum which do not appear in the CA2 and CA3 fields (Lorente de No, 1934).The axons of CAl pyramidal cells are thin. They arise from the basal side andreach out to the stratum orien and the alveus. Some axons also give off severalcollaterals, which ramify in the stratum radiatum. The CAl field is further dividedinto three subfields: (1) CAIa lies next to the subiculum and contains some cellsbelonging to the subiculum; (2) CA1b contains the smallest pyramidal cells ofcornu ammonis; (3) CAIc pyramidal cells have smooth dendrites with many sideXIE 15branches and are bigger than the CAl b cells (Lorente de No, 1934). However,the borders of these CAl subfields are not very clear.The CA2 field makes up a small portion of the pyramidal cell layer. TheCA2 pyramidal cells are bigger than the CAl cells. They are characterized bypossessing dendrites similar to those of CA3, but without the thick thorns. Thestratum radiatum in the CA2 field is the thinnest among the subfields of cornuammonis. The axons there do not have a Schaffer collateral, but have severalcollaterals for the longitudinal association path. These axons have longhorizontal collaterals in the stratum oriens (Lorente de No, 1934).In the CA3 field, the pyramidal cells are the biggest in cornu ammonis.They are characterized as described by Cajal (1911). The ascending shaftsfrom apical dendrites do not have side branches in the stratum radiatum, butgenerally divide into two or more vertical branches which ascend to the stratummoleculare. The initial part of the shaft has numerous thick thorns, which makecontact with the mossy fibre. The basal dendrites ramify in the stratum orien(Lorente de No, 1934). The axons of CA3 pyramidal cells are thick and highlycollateralized. The axons arise from the basal pole of the CA3 pyramidal cellsand cross the stratum oriens into the fimbria giving off collaterals. Somecollaterals terminate within the CA3 field. Others termed Schaffer collateralscross the stratum pyramidale and radiatum and enter the stratum lacunosummoleculare where they constitute horizontal fibres. Schaffer collaterals are verythick and myelinated. They make contact with the CAIa and CAIb apicaldendrites (Lorente de No,1934). Lorente de No (1934) described that onlysome CA3 pyramidal cells have Schaffer collaterals. He further divided the CA3field into three subfields (CA3a,b,c). The CA3a lies next to the CA2 field whilethe CA3c lies next to the hilus. The CA3b is situated between the CA3a andCA3c. The dendrites of CA3c pyramidal cells make contact with the mossyXIE 16fibres from both infrapyramidal and superpyramidal granule cells, whereas thedendrites of CA3a and CA3b pyramidal cells only make contact with the mossyfibres from superpyramidal granule cells. All CA3c pyramidal cells and half ofCA3b pyramidal cells have Schaffer collaterals. The other half of CA3b and allCA3a pyramidal cells do not have Schaffer collaterals. They have one or twocollaterals which extend to the stratum radiatum and constitute an associationalpathway running in the CA3a and the CA2 fields. However, Ishizuka et al.(1 990)have recently described collaterals from all portions of CA3 field can reachwidespread regions of CA3, CA2 and CAl transversely and longitudinally, a fewfibres enter the subicular complex, but none enter the entorhinal cortex.2.3.3. Hilar region mossy cellsMossy cells are one of the most distinctive and common cell types in thehilus of the hippocampal dentate gyrus (Amaral,1978; Ribak et al. 1985). Thesecells are distinguishable from other neurons of the hilar region primarily by thecluster of complex spines (‘thorny excrescences’) located at several locations onthe soma and proximal dendrites, on the distal dendrites which have moretypical spines (Ribak et al. 1985). Mossy cells have a triangular or multipolarshaped soma with about 20 to 25 p.m in size. Three to four primary dendritesarise from the soma and bifurcate once or more to produce an extensivedendritic arborization restricted, for the most part, to the hilus where they makecontact with the mossy fibres (Ribak et al., 1985). The axons of the mossy cellsbifurcate and project towards the fimbria and the molecular layer of the dentategyrus.2.3.4. InterneuronsThe two principal cell types, the pyramidal cells of the cornu ammonis andthe granule cells of the dentate gyrus, make up 96-98% of the neurophil of thehippocampus (Seress and Pokorny, 1981; Buzsaki, 1984). Interneurons areXIE 17found both in the principal cell layers and other strata of the structure (Cajal,1911; Lorente de No, 1934). There are various types of interneurons based ontheir morphological characteristics, such as pyramidal, horizontal, fusiform,inverted fusiform and multipolar. The somata of these interneurons vary in size(10 to 50 jim). These cells have aspinous dendrites and locally arborizing axons(Ribak and Anderson, 1980; Buzsaki, 1984). The most well-studied interneuronsare basket cells characterized by their axonal terminals forming basket-likestructures around the somata of the target cells (Cajal, 1911; Lorente de No,1934). These basket cells are distributed in both the stratum pyramidale and thestratum granulosum. These cells lie very close to the granule cells and thepyramidal cells. The cell bodies vary in shape from spherical to triangular andare on an average 30-50 j.tm in size. Several dendrites with few branches andspines extend from the soma. They show frequent swelling like a string pearls(Andersen et al., 1969). In the cornu ammonis field, most of the basket celldendrites distribute in the stratum oriens while some extend into the stratumradiatum (Andersen et al. 1969). In the dentate gyrus, the dendrites of basketcells are found in all layers (Ribak and Seress, 1983). The axons of these cellsare very thin. They form an extensive plexus in the pyramidal cells or thegranule cells. Each basket cell may synapse with as many as 200 to 500pyramidal cells or granule cells (Anderson et al. 1969; Buzsaki, 1984). Thesynaptic sites of axon terminals are on the somata and proximal dendrites of thetarget cells.Another group of interneurons is localized at the stratum oriens/alveus(0/A) border. The somata of these cells are oblong-shaped and 20 to 30 jim insize. The longitudinal axis of the soma is parallel to the alvear surface. Thedendrites are long and coarse, and almost parallel to the alvear fibres (Cajal,1911; Lorente de No, 1934; Andersen, 1963). These axons extensively ramify inXIE 18the stratum oriens, pyramidale, and radiatum (Cajal 1911; Lorente de No, 1934)where they make contact with the basal dendrites, the somata and the proximaldendrites of pyramidal cells (Lacaille et al. 1987).A distinct group of interneurons, termed the stratum lacunosummoleculare (L-M) interneurons, are located in the stratum lacunosum-moleculareand the stratum radiatum (Kawaguchi and Hama, 1987; Lacaille andSchwartzkroin, 1988a,b). L-M somata are fusiform in shape and 15 to 25 p.m insize. Their dendrites are aspinous and slightly varicose. They branch profuselyin different strata of the hippocampus (Kunkel et al. 1988). Their axons projectthroughout the stratum lacunosum-moleculare and into the stratum radiatum.Some dendrites and axon collaterals can extend towards the hippocampalfissure and cross into the stratum moleculare of the dentate gyrus (Kunkel et al,1988). The L-M interneuron axons make synaptic contact with the pyramidal celldendrites in the stratum lacunosum-moleculare and the stratum radiatum, andthe granule cell dendrites in the stratum moleculare of the dentate gyrus. The LM interneuron axons also make synaptic contact with the dendrites of non-pyramidal neurons in the stratum radiatum (Kunkel et al, 1988).2.4. Intrinsic hippocampal circuitryThe “lamellar hypothesis” was proposed by Andersen and his colleagues(1971a). They summarized the results of their physiological studies in the rabbitand concluded that “The hippocampal cortex seems to be organized in parallellamellae ... By means of this lamellar organization, small strips of thehippocampal cortex may operate as independent functional units, althoughexcitatory and inhibitory transverse connections may modify the behaviour of thefibre lamellae.” However, subsequent and recent neuroanatomicalinvestigations have demonstrated that the major intrinsic hippocampalprojections are as extensive and highly organized in the longitudinal orXIE 19septotemporal axis of the hippocampus as in the transverse axis (I-IjorthSimonsen, 1973; Laurberg, 1979; Swanson et al., 1978; Amaral and Witter,1989; Ishizuka, 1990). These studies have clearly shown that the hippocampalformation is a three dimensional cortical structure with important informationprocessing taking place in both the transverse and longitudinal axes. In thefollowing discussion, the major intrinsic pathways within the hippocampalformation are roughly divided into transverse (lamellar circuits), longitudinal(septotemporal pathways) and local circuits, It is obvious that these circuits arehighly interdependent.2.4.1. Transverse circuitThe major excitatory pathway within the hippocampus enters from theentorhinal cortex via the perforant pathway across the hippocampal fissure to thegranule cells of the dentate gyrus. The dentate granule cells distribute theirmossy fibres to CA3 pyramidal cells (Cajal 1911; Lorente de No, 1934). Theproximal portions of CA3 pyramidal cells (CA3c) are reached by the mossy fibreswhich originated in both the infrapyramidal and suprapyramidal blades. Only themossy fibres which arise from the suprapyramidal blade, make contact with moredistal portions of the CA3 field (CA3a and CA3b) (Lorente de No, 1934;Blackstad et al. 1970; Chronister and White, 1975; Witter, 1989). All along theircourse, the mossy fibres make numerous synapses en passage with thedendrites they pass in the hilus and the CA3 field (Andersen et al., 1966;OKeefe and Nadel, 1978). The mossy fibres which terminate on the basaldendrites of the proximal CA3 pyramidal cells arise mainly from theinfrapyramidal blade of the dentate gyrus (Witter, 1989). The mossy fibres aremainly organized in a lamellar fashion. Bands of mossy fibres are principallyoriented transverse to the longitudinal axis of the hippocampal formation and theXIE 20bands arising from each septotemporal level of the dentate gyrus only minimallyoverlap those arising from other septotemporal levels (Amaral and Witter, 1989).The CA3 pyramidal cells receive inputs not only from the mossy fibres,but also directly from the perforant path fibres. In turn, the axons of the CA3pyramidal cells divide, with one branch entering the fimbria and going to theseptum, while other branches remain within the hippocampus. Among the latter,Schaffer collaterals are the most prominent projection that arise from the CA3pyramidal cells. The Schaffer collaterals run into the stratum radiatum and thestratum oriens of the CAl field (Lorente de No,1934; Hjorth-Simonsen, 1973).The Schaffer collaterals synapse with the apical dendrites in the stratumradiatum and the basal dendrites in the stratum oriens in the CAl field. TheCAl pyramidal cell axons enter the alveus and project caudally into thesubiculum and rostrally into the lateral septal nuclei and the prefrontal cortex.The connections between the CA3 and the CAl neurons, and the connectionsfrom the CAl to the subiculum show a columnar organization perpendicular tothe cell layer (Witter, 1989). Proximal parts of CA3 and CAl distribute fibres todistal parts of CAl and the subiculum, respectively, while the more distal parts ofCA3 and CAl interact with more proximal parts of CAl and the subiculum,respectively (Witter, 1989; Ishizuka et al., 1990).2.4.2. Longitudinal pathwayThe intrinsic connections between the various subfields of thehippocampus show the major hippocampal projections are as extensive andhighly organized in the longitudinal or septotemporal axis of the hippocampus asin the transverse axis (Amaral and Witter, 1989; Ishizuka et al.; 1990). Theperforant path from the entorhinal cortex to the granule cells spreads for at leastmillimeters along the longitudinal axis of the hippocampus. Certain neurons inthe polymorphic layer of the dentate gyrus preferentially send fibres over lengthyXIE 21longitudinal distances to the granule cells (Hjorth-Simonsen and Laurberg, 1977;Knowles, 1992). The mossy fibres which arise from the granule cells, are theonly projection organized in a lamellar fashion. On the other hand, the CA3projections within the CA3 field and to the CAl field are equally extensive in thetransverse and longitudinal axes (Swanson et al., 1978; Amaral and Witter,1989; Ishizuka, et al., 1990; Knowles, 1992). The Schaffer collateral systemserves not only as a serial link from the CA3 field to the CA2 and CAl fields, butalso as a longitudinal association pathway within the CA3 field and along theCA2 and CAl fields (Hjorth -Simonsen and Laurberg,1977; Knowles, 1992).Lorente de No (1934) described that the associational connections link differentseptotemporal levels of the hippocampus whereas the CA3 and CAl projectionlink the two fields at one septotemporal level. His description seems to beoversimplified.2.4.3. Local circuitsWithin the subfields of the cornu ammonis and the dentate gyrus,interneurons and principal cells form two local synaptic circuits, the feed-forwardand the recurrent inhibitory circuits (Andersen et al., 1969; Buzsaki, 1984).These inhibitory circuits play a crucial role in the hippocampal function and arediscussed further in section 4.2.7 (inhibitory circuitry).Granule cells send extensive collaterals not only to many interneurons,but also to the mossy cells in the hilus. The axon collaterals of the mossy cellsproject into the inner molecular layer of the dentate where they make contactwith the dendrites of the granule cells. These axon collaterals also make contactwith interneurons in the hilus of the dentate gyrus which in turn contact thegranule cells (Schwartzkroin et al. 1990).Both CA3 and CAl pyramidal cells send excitatory axon collaterals toseveral classes of interneurons that inhibit the pyramidal cells (Knowles andXIE 22Schwartzkroin, 1981a, b). The CA3 pyramidal cells also have extensive localaxon collaterals, the CA3 associational collaterals, which project along thelongitudinal axis and excite neighbouring CA3 pyramidal cells (Ishizuka et al.,1990).2.4.4. Commissural afferentsExtensive neural fibres connect the two hippocampi, crossing the midlinein the ventral and dorsal hippocampal commissures (psalteria). The majority ofthe crossed hippocampal fibres appear to course through the ventralcommissure, with only a few traveling through the dorsal commissure (OKeefeand Nadel, 1978). Two major components of the projections have beendescribed by Blackstad (1956) and Swanson et al. (1978). One projection arisesfrom the dentate gyrus and ends on the inner one-third of the dendrites of thegranule cells of the contralateral dentate gyrus, homotopic to the ipsilateralprojection. The second projection arises from the CA3 field and ends in thestratum radiatum and the stratum oriens of the contraJateral CAl, also homotopicto the ipsilateral projection. The origin of the cells of the contralateral projectionto the dentate gyrus molecular layer has been controversial. Recent studieshave shown that the mossy cells of the hilus are the major neurons which projectto the inner molecular layer of the contralateral dentate gyrus, homotopic to theipsilateral projection (Lauberg and Sorensen, 1981; Schwartzkroin et al., 1990;Ribak et al., 1985). The commissural connections of the hippocampal formationin the monkey is less than in the rat (Gottlieb and Cowan, 1973; Rosene andVan Hoesen, 1987; Knowles, 1992). The significance of this difference is notvery clear, It may reflect a growing functional linkage between the hippocampalformation and increasingly lateralized cerebral cortex of primates.XIE 232.5. Extrinsic afferents to the hippocampusThe hippocampus receives extrinsic afferents from a variety of otherstructures; including primarily the entorhinal cortex, the medial septal area andseveral brain-stem sites.2.5.1. Entorhinal afferentsThe perforant path is the major avenue of entorhinal afferents to thehippocampus. As previously described, the entorhinal cortex is usually dividedinto at least two parts; a lateral and a medial part (Lorente de No, 1934; HjorthSimmonsen and Jeune, 1972; Amaral and Witter, 1989). While the lateralentorhinal cortical neurons send their fibres mainly to the septal hippocampus,the medial entorhinal neurons project into the temporal hippocampus (Witter etal., 1989). Whereas fibres from the lateral entorhinal cortex mainly terminate inthe outer one-third of the molecular layer of the dentate gyrus, fibres from themedial entorhinal cortex preferentially distribute to the middle one-third of themolecular layer (Wyss, 1981; Witter, 1989; Steward, 1976). This differentiationbetween the lateral and the medial components of the perforant pathway is moreprominent in rats and cats than in monkeys (Hjorth-Simonsen, 1972; Steward,1976; Wyss, 1981; Witter et al., 1989). In rats, the fibres coming from the lateralentorhinal cortex prefer the suprapyramidal blade of the dentate gyrus, whereasthe medial parts project to both the suprapyramidal and infrapyramidal blades(Witter, 1989). The entorhinal cortex also directly projects into the CAl, CA3and the subiculum fields (Gottlieb and Cowan, 1972; Steward, 1976; Witter etal., 1988; Witter et al., 1989). The perforant path originates from layer I-Ill of theentorhinal cortex (Steward and Scoville, 1976). The projection into the dentategyrus arises mainly from layer II of the entorhinal cortex whereas the projectionto the CAl field arises from the layer Ill of the entorhinal cortex (Steward andXIE 24Scoville, 1976). A few neurons in the layer V also send their axons to thehippocampus (Kohler, 1985).2.5.2. Septal afferentsDaitz and Powell (1954) first demonstrated that neurons in the medialseptal nucleus and the nucleus of the diagonal band (together known as themedial septal complex), project through the fimbria to the hippocampus in rats,rabbits and monkeys. The septo-hippocampal pathway appears to be crucial forthe initiation or maintenance of 9 rhythm activity in the hippocampus (Petsche etal., 1962; Andersen et al. 1979; Mitchel and Ranck, 1982). This projection alsoprovides the major source of extrinsic cholingergic fibres to the hippocampusalthough some septo-hippocampal projection neurons are not cholinergic. (Lewiset al., 1967; Swanson et al., 1987). Some septo-hippocampal projectionneurons in the medial septal nucleus and diagonal band contain glutamic aciddecarboxylase (GAD), a synthesizing enzyme for the neurotransmitter GABA.Therefore, these neurons are presumably GABAergic (Kohler et al, 1984;Swanson et al., 1987). The neurons in the medial septal complex innervate allfields of the hippocampus (Swanson and Cowan, 1976, 1978; Swanson et al.,1987). Septal fibres to the dentate gyrus course primarily through the fimbriaand reach the hilar region and a thin zone of the molecular layer, just superficialto the granule layer (Rose et al., 1976; Swanson et al., 1987). The septal fibresproject into the cornu ammonis ending predominantly in the stratum radiatumand the stratum oriens of the CA3 field, and in the stratum oriens of the CAl field(Nyakas, et al., 1987; Swanson et al.,1987). The input to the CAl field is lessdense than the input to the CA3 field, and appears to course predominantlythrough the dorsal fornix. The septal fibres also project to the subiculum. Thetarget cells of the septo-hippocampal projection are pyramidal and granule cellsas well as interneurons (O’Keefe and Nadel, 1978).XIE 252.5.3. Other afferentsThere are several other projections from different parts of the brain to thehippocampus. The major inputs from the thalamus, including one from theanterior thalamic nuclei and one from the midline nuclei, terminate in the CAlfield and the subiculum. The projections from hypothalamus end in the dentategyrus, the subiculum, as well as the CA3 field (Swanson et al., 1987;Schwerdtfeger, 1984). The raphe nuclei give rise to serotonergic input to thehippocampus from the midbrain. These serotonergic fibres from the raphenuclei innervate all fields of the hippocampus (The densest innervation occurs inthe dentate hilar area) (Kohler, 1982; Swanson et al., 1987). The hippocampusalso receive dopaminergic inputs from the ventral tegmental area and thesubstantia nigra, the pars compacta and the central linear nucleus (Swanson,1982; Swanson et al., 1987). These dopaminergic fibres terminate in thesubiculum entorhinal area, the dentate gyrus and the cornu ammonis (Swanson,1982, Scatton et al., 1980; Swanson et al, 1987). All fields of the hippocampusreceive noradrenergic input from the locus ceruleus of the pons (Blackstad et al.1967; Swanson et al. 1987). The densest noradrenergic innervation is the hilarregion of the dentate gyrus (Koda et al., 1978; Swanson et al., 1987).2.6. Extrinsic efferents from the hippocampusThe hippocampus projects to a series of parahippocampal and isocorticalstructures, including the entorhinal cortex, the septum, the nucleus accumbens,the amygdaloid complex, and the hypothalamus. The fornix-fimbria system andthe entorhinal area are the two main output pathways of the hippocampus.2.6.1. Fornix Fimbria systemThe rostrally directed efferent fibres of the hippocampus and its adjacentareas gather together in the fimbria and the dorsal fornix in which the fibres joinat the columns of the fornix (O’Keefe and Nadel, 1978). The main fibres in theXIE 26fornix continue rostrally to penetrate the septal-fimbrial nucleus at the caudalparts of the septal region where the fibres split into two components to form thepost-commissural and pre-commissural fornices. The latter originates primarilyin the hippocampus, and the former arises mainly from adjacent allocorticalareas (O’Keefe and Nadel, 1978).The pre-commissural fornix arises from neurons throughout thelongitudinal axis of the cornu ammonis and the subiculum and distributes to mostof the nuclei of the septal area, including the lateral septum, the diagonal bandof broca, and the bed nucleus of the anterior commissure. The fibres alsoterminate in the lateral pre-optic region and the lateral hypothalamus (Swansonand Cowan, 1977; O’Keefe and Nadel, 1978; Swanson et al., 1987).The post-commisural fornix divides into two components. One componentarises from the pre- and parasubiculum and project to the thalamus (Chronisterand White, 1975; Swanson and Cowan, 1975; O’Keefe and Nadel, 1978). Theother derives from the subiculum and projects to the mammillary bodies androstral brain stem (Meibach and Siegel; 1975; O’Keefe and Nadel, 1978).2.6.2. Entorhinal projectionA direct projection that originates in the CA3 field of primarily ventralhippocampus and terminates in layer IV of the entorhinal area, has beendescribed by Hjorth-Simonsen (1971). Later, Swanson et al. (1978) describedthat another projection into the entorhinal cortex originating from the CAl field.Several other efferent projections have also been reported by Swanson et al.(1978). They are as follows: the CA3 fibres directly project to the cingulatecortex; the subiculum fibres project to the perirhinal cortex and the medial frontalcortex; and the CAl fibres project to the perirhinal cortex.XIE 273. ELECTROPHYSIOLOGY OF HIPPOCAMPAL NEURONS3.1. Characteristic of hippocampal neuronThe use of slice preparations has provided a great deal of informationabout the physiology of hippocampal neurons. The experimental data obtainedfrom hippocampal slices are gratifyingly similar to results obtained from theintact animal (Kandel and Spencer, 1961; Kandel et al., 1961; Spencer andKandel et al., 1961a, b; Schwartzkroin, 1975). Careful studies of theelectrophysiological properties of the CAl, CA3 pyramidal cells and the dentategranule cells have been performed in hippocampal slices (Brown et al., 1981;Johnston, 1981; Brown and Johnston, 1983; Durand et al., 1983; Turner andSchwartzkroin, 1983).The resting membrane potentials of hippocampal pyramidal and granulecells are on an average -50 to -70 mV and -55 to -80 mV, respectively. Theinput resistances calculated from the slope resistance within the linear range ofthe current-voltage (l-V) curves are 20-45 M2 for pyramidal cells and 40-60 M2for granule cells (Brown et al., 1981; Turner and Schwartzkoin, 1983; Lambertand Jones, 1990). The membrane constants, which are the time for membranepotential to reach 1-l/e of their peak voltage, are on an average 25-35 msec forCA3 pyramidal neurons, 12-16 msec for CAl pyramidal neurons, and 9-li msecfor granule cells (Brown et al., 1981; Durand et al., 1983; Turner andSchwartzkroin, 1983; Schwartzkroin and Mueller, 1987). Assuming a constantspecific membrane capacitance, the differences in the membrane time constantssuggest that there are proportionate differences in the specific membraneresistivity in the different types of hippocampal neurons. Based on theequivalent cylinder modeling theory (RaIl, 1969; RaIl, 1974; RaIl, 1977), theelectrotonic length of the equivalent dendritic cylinder and the conductance ratioof the dendrites to the soma have been estimated. The average estimated valueXIE 28of the electrotonic length and the dendrite to soma conductance ratio ofhippocampal neurons are 0.8-1 and 1.1-1.5, respectively (Brown et al, 1981;Johnston, 1981; Burand et al., 1983; Schwartzkroin and Mueller, 1987).However, using the multipolar cylinder model, Glenn (1988) suggested that theelectrotonic length of neuron dendrites was overestimated. In fact, he found thatthe value of the electrotonic length was only 1/3 of the previous value.Recently, the patch clamp technique has been applied to the slicepreparation (Edward et al., 1989; Blanton et al., 1989; Randall et al., 1990).High input resistances (200 M2 to 5 G) of hippocampal neurons have beenobserved using whole cell patch clamp recording (Edward et al., 1989; Randallet al., 1990). Similar mean resting membrane potentials have been observedbetween conventional intracellular recording and whole cell recording. Themembrane time constants are usually longer when using the whole cellrecording compared to the intracellular recording (Edward et al., 1989). Thediscrepancy between the values obtained from two methods has not beensatisfactorily explained (Edward et a)., 1989; Strom, 1990a).3.1.1. Bursting activity of hippocampal neuronsHippocampal pyramidal cells can fire spikes spontaneously, or duringintracellular depolarizing current pulses. There are significant differences in themode of spike discharge between CAl and CA3 pyramidal cells (Masukawa etal., 1982; Wong et al., 1979). Spontaneous burst generation occurs morecommonly in CA3 neurons than in CAl neurons. CAl cells produceaccommodating trains of spikes when depolarizing current pulses are injectedintracellularly. In contrast, CA3 pyramidal cells can readily generate burstdischarges in response to intracellular depolarizing current pulses (Wong et al.,1979; Schwartzkroin et al., 1990). The spike frequency accommodation in CAlneurons is due to the activation of several K currents, primarily the Ca2-XIE 29dependent K currents. The event in CA3 cells is due to the activation of atleast one type of Ca2 current (Wong and Prince, 1978; Traub, 1982). The burstdischarge property of CA3 cells contributes to the role of the CA3 regions as a“pacemaker” for internal discharge in the hippocampus and as “boosters’ foramplifying incoming excitatory signals from afferent pathways. Granule cells donot fire spontaneous bursts of spikes. This is probably due to the high thresholdrequired for spike generation in these cells.3.2. Hippocampal interneuronsHippocampal interneurons can modulate the hippocampal input and theprincipal cells. These interneurons play a significant role in the hippocampalfunction. However, the electrophysiological characteristics of the hippocampalinterneurons are not as well known as those of hippocampal pyramidal cells andgranule cells. Three types of interneurons in the CAl region, basket cells,oriens/alveus (0/A) interneurons, and lacunosum-moleculare (L-M) interneuronsare discussed here. The resting membrane potentials and the input resistancesof basket cells are on an average -50 to -70 mV and 30 to 50 respectively.Basket cells display properties that clearly distinguish them from pyramidal andgranule cells. Basket cells have a very brief action potential (average 0.8 ms), alarge spike after hyperpolarization (AHP) (5 to 10 mV, 10-30 ms) and a shortmembrane time constant (approximately 3 ms). At resting membrane potential,basket cells generate spontaneous spikes at a high rate (5-25 Hz) and display aconstant barrage of EPSPs. Basket cells also fire non-decrementing bursts ofspikes in response to intracellular depolarizing pulses (Kawaguchi and Hama,1987; Lacaille et al. 1989; Schwartzkrion and Kunkel, 1985, Schwartzkroin andMathers, 1978).0/A interneurons located at the stratum oriens-alveus borders haveelectrophysiological properties quite similar to those of basket cells (Lacaille etXIE 30al., 1987). The resting membrane potentials and the input resistances of 0/Ainterneurons are on an average -50 to -70 mV and 30 to 65 M2, respectively(Lacaille and Williams, 1990). The action potentials are brief (approximately Ims) and the time constants are fast (4-6 ms). 0/A interneurons have a largeAHP (7 to 13 mV, 10-30 ms) following a spike. These cells generatespontaneous spike firing at high frequency (10-27 Hz) and receive a constantbarrage of EPSPs; and they fire in non-accommodating bursts of actionpotentials in response to intracellular depolarizing current pulses.Both basket cells and 0/A interneurons show GABA-likeimmunoreactivity. These interneurons receive excitatory inputs from Schaffercollaterals, commissural fibres as well as the alvear fibres (Schwartzkroin andKunkel, 1985; Lacaille et al., 1989). In turn, they provide feed-forward andfeedback inhibition to pyramidal neurons (Lacaille et al. 1989).The third type of interneurons in the CAl field are called L-M interneuronswhich are located at the border between the stratum lacunosum-moleculare andthe stratum radiatum. L-M interneurons display some distinctive membraneproperties (Kawaguchi and Hama, 1987) and share some similarelectrophysiological properties with other interneurons. The membranepotentials and the input resistances of L-M interneurons are -50 to -70 mV and40 and 70 M2, respectively. L-M interneurons have a prominent AHP (7 to 14mV, 30-34 ms) following an action potential and no spike frequencyaccommodation during depolarizing current pulses. The action potential of L-Minterneurons is longer (approximately 2 ms) than that of basket cells or 0/Ainterneurons. In addition, the membrane time constant of L-M interneurons isslower (8.6 ms) than that of basket cells or 0/A interneurons. Unlike basketcells and 0/A interneurons, the majority of L-M interneurons do not fire actionpotentials spontaneously; some L-M interneurons fire action potentialsXIE 31spontaneously at low frequency (below 10 Hz). L-M also generate littlespontaneous synaptic activity (Lacaille et al. 1989; Lacaille and Schwartzkroin,1988).L-M interneurons also display GABA-like immunoreactivity. Theseinterneurons receive excitatory inputs from Schaffer collaterals and commissuralfibres. Unlike basket cells and 0/A interneurons which provide both feed-forward and feedback inhibition to pyramidal cells, L-M interneurons only givefeed-forward inhibition to pyramidal neurons (Lacaille et al., 1989).3.3. Membrane ionic currents of hippocampal pyramidal neuronsMultiple ionic channels co-exist in the membrane of the hippocampalneurons. Different ionic currents, mediated through the opening of voltagegated and/or non-voltage-gated channels, play different roles in the functioningof the hippocampal neurons.3.3.1. Na currentsTwo types of Na currents have been found in hippocampal neurons: afast current ‘Na (fast) and a slowly inactivating current lNa(slow) (Brown et al.,1990). The fast Na current behaves as an orthodox Hodgkin-Huxley currentwith an activation threshold of -60 mV and a time to peak of about 0.9 ms at 0mV (Sah et al., 1988a; Brown et al., 1990). It is responsible for the depolarizingphase of the cell action potential. Inactivation is complete throughout theactivation range. This current has been observed in both the soma and thedendrites of the hippocampal neurons (Benardo et al., 1982; Sah et al., 1988a;Hugenard et al., 1989). The current is blocked by extracellular application oftetrodotoxin (TTX) and intracellular injection of QX-222 or QX-314(Schwartzkroin and Slawsky, 1977; Connors and Prince, 1982; Schwartzkroinand Mueller, 1987).XIE 32The slowly inactivating Na current has an activation threshold of about 5to 10 mV positive to the resting potential and is triggered and sustained bydepolarizing prepotentials (Lanthorn et al., 1984b; French and Gage, 1985).This current may be confined to the soma of the hippocampal neurons (Benardoet al., 1982). It is responsible for the repetitive firing of action potentials andacts as a “pacemaker’ current for the hippocampal activity (Brown et al., 1990).The current is also TTX-sensitive and blocked by QX-314 (Benardo et al., 1982).3.3.2. Ca2 currentsThere are at least four types of voltage-dependent calcium channelspresently recognized in the hippocampal neurons. They are the high-threshold,sustained current (L-type), the low-threshold transient current (T-type), the highthreshold inactivating current (N-type) (Ozawa et al., 1989; Takahashi et al.,1989; Fisher et al., 1990; Brown et al., 1990; Mogul and Fox, 1991) and thehigh-threshold w-agatoxin IVA-sensitive (o-Aga-IVA) current (P-type) (Hillman etal., 1991; Mintzetal., 1992).The L-type Ca2 current is a high-threshold and long-lasting inwardcurrent, which is activated at potentials between -20 mV and -10 mV andinactivated at potentials between -20 mV and -60 mV (Ozawa et al., 1989; Moguland Fox, 1991). The inactivation rate of the L-type Ca2 current is very slow(time constant> 500 ms at 0 mV) (Ozawa et al., 1989; Tsien et al., 1988). At thesingle channel level, the L-type conductance is 25-27 pS. The L-type currentcan be enhanced by Ba2 and blocked by Co2,Cd2, Ni2 (Fisher et al., 1990;Mogul and Fox, 1991). The current is blocked by verapamil and bydihydropyridine Ca2 antagonists, such as nimodipine, nifedipine and PYIO8-068, and is enhanced by dihydropyridine agonist, BAY K8644 (Brown andGriffith 1983a,b, Brown et al., 1984; Tsien et al., 1988; Brown et al., 1990). -XIE 33Conotoxin suppresses the L-type more strongly than the N-type Ca2 currents(Takahashi et al., 1989).Another high threshold Ca2 current, called N-type current, inactivatesslowly, but much faster than the L-type current. The N-type current is activatedat -20 mV or more positive and inactivated at -30 mV to -110 mV, decaying froma maintained depolarization with a time constant of 30-36 ms at 0 mV (Ozawa etal., 1989). The single channel conductance of the N-type channel is 13-15 pS(Brown et al., 1990; Tsien et al., 1988). The N-type current is enhanced by Ba2and blocked by Cd2 and Ni2. This current is also blocked by co-conotoxin.Dihydropyridine Ca2 antagonists, nicardipine suppresses the N-type current inhippocampal neurons, but not in the peripheral neurons (Takahashi et al., 1989).The third type of Ca2 current, called the T-type current, is a low-threshold transient inward current. The T-type current is activated bydepolarization beyond about -60 mV from holding potential down to -100 mV,and is completely inactivated at -40 to -50 mV (Brown et al., 1990, Takahashi etal., 1989). The time constant of inactivation of the T-type current isapproximately 20 ms at -40 mV. The single channel conductance is 7-8 pS.Unlike the L-type and the N-type current, the T-type current is reduced or notchanged on substituting Ba2 for Ca2 (Brown et al., 1990). The T-type currentis blocked by Ni2 and Cd2. It is sensitive to dihydropyridine Ca2 antagonistssuch as nicardipine. However, the current is resistant to o-conotoxin (Takahashiet al., 1989). The T-type current may contribute to spontaneous depolarizationwaves and acts as a generators of pacemaker activity (Tsien et al., 1988).The fourth type of Ca2 current, the P-type current, was first described incerebellar Purkinje cells (Llinas et al., 1989). The P-current was later found in avariety of central and peripheral neurons, including hippocampal pyramidalneurons (Hillman et al., 1991; Mintz and Adams, 1992). The density of P-typeXIE 34Ca2 channels is lower in hippocampal pyramidal cells than cerebellar Purkinjecells (Hiliman et al., 1991). The P-type current is characterized by beingsensitive to w-Aga-IVA but not to o-conotoxin or dihydropyridine types of blocker(Llinas et al. 1989; Mintz et al., 1992). The P-current has a high activationvoltages (-20 mV or more positive) with little inactivation (Llinas et al., 1992).The P-type current is enhanced by Ba2 and blocked by Cd2 and Co2. The Ptype channels have been suggested to be involved in the release ofneurotransmitters, such as glutamate (Turner et al., 1992).Whether there are more than four types of Ca2 channels in thehippocampal neurons is not very clear. Takahashi (1989) has reported a TTXsensitive Ca2 current, which differs from the L, N, P, T- type currents, inisolated hippocampal CAl pyramidal neurons. This current is activated at morepositive depolarizing pulses than around -65 mV from a holding potential of -100mV, and reaches a peak value at about -35 mV. The inactivation rate is 7-10 msat -35 mV. This current is suppressed by phenytoin and nicardipine, adihydropyridine, and is resistant to o-conotoxin. However, the identity of thiscurrent remains unknown at present. Recently, a “Q-type” Ca2 current hasbeen reported in cerebellar granule cells (Randall et al., 1993). The “Q-type”current is sensitive to o-conotoxin-MVIIC, but is insensitive to nimodipine, oconotoxin-GVIA or low concentrations (<30 nM) of o-Aga-IVA. It is not clearwhether “Q-type” channels also exist in the hippocampal pyramidal neurons.Intradendritic recording has demonstrated that dendrites are probably theprimary sites ofCa2-dependent bursts and Ca2 spikes, particularly in CAlneurons (Wong and Prince, 1979; Wong et al., 1979; Benardo et al., 1982). Thedendritic Ca2 spike is appreciatively larger than the somatic Ca2 spike. Onthe other hand, theCa2-dependent subthreshold depolarization appears to berestricted to the soma. Yaari et al. (1987) have shown that the low-thresholdXIE 35transient current, which probably underlies the subthreshold depolarizations, islargely present in the soma of cultured rat embryonic neurons. The channelscarrying the high-threshold currents are distributed in both soma and processesof hippocampal neurons.3.3.3. K currentsMultiple potassium (Kj channels co-exist and function in hippocampalneurons. At least six types of voltage-dependent K currents and two types ofCa2-dependent currents have presently been identified in hippocampalneurons. The voltage-dependent K currents are as follows: delayed rectifiercurrent (lK); fast transient current (IA); slowly inactivating delay current (ID); Mcurrent (IM); fast inward rectifier current (IK(IR)); and Q-current (la). The formerfour types of K currents are activated by depolarization whereas the latter twotypes of K currents are activated by hyperpolarization. TheCa2-dependent Kcurrents include the fastCa2-dependent K current (Ic) and the slow Ca2-dependent K current (IAHP). They are activated by an influx of Ca2 ionsthrough voltage-gated Ca2 channels (Brown et al., 1990; Storm, 1990b).‘K corresponding broadly to the classical delayed rectifier current firstdescribed by Hodgkin and Huxley (1952) has been recorded in hippocampalneurons (Segal and Barker, 1984a; Numann et al., 1988; Madison et al., 1987a;Storm, 1990b). This current is activated by depolarization beyond -40 mV. Itactivates only during action potential. The activation is rather slow (time topeak, about 180 ms in acutely dissociated cells) (Numann et al., 1987). ‘K isblocked by mM concentrations of external tetraethylammonium (TEA). It isresistant to 4-aminopyridine (4-AP) and external cesium (Csj. The single Kchannels (15-20 pS) may underlie ‘K in hippocampal neurons (Rogawsky, 1986;Brown et al., 1990). tK has been found in both soma and dendrites ofXIE 36hippocampal neurons (Masukawa and Hansen, 1987). The current maycontribute to the repolarization of the action potential.‘A is a prominent current which can be recorded from hippocampalneurons (Gustafsson et al., 1982; Zbicz and Weight; 1985). ‘A is activated bydepolarization beyond about -60 mV and it inactivates between -60 and -40 mV.The current activates very rapidly (time to peak within 5-10 ms) and inactivateswith a time constant of 20-30 ms. ‘A is blocked by a low mM concentration of 4-AP (Segal and Barker, 1984; Numann et al., 1987) and nM of dendrotoxin (DTX)(Harvey and Karlsson, 1980, 1982), but it is resistant to TEA (Segal and Barker,1984a; Gustafsson et al., 1982; Storm, 1990b). Noradrenaline and acetylcholinehave been reported to inhibit ‘A (Sah et al., 1985; Nakajima et al., 1986; Storm,1990). A single transient K channel with the conductance of 15 pS has beenfound primarily in the soma of hippocampal neurons. ‘A plays a major role inregulating the repetitive firing of neurons. It can delay the onset of firing up tolOOms (Gustafsson et al., 1982; Storm, 1988c). The current is also involved inrepolarization of the action potential (Storm, I 988c, I 990a).‘D has been found to co-exist with ‘A in CAl pyramidal cells (Storm,1988a,b). ‘D differs from ‘A in that: (a) ‘D has slow kinetics, particularlyinactivation and recovery from inactivation; (b) it has more negative threshold foractivation and inactivation; (c) it is more sensitive to 4-AP (Storm, 1990). ‘D isactivated by depolarization beyond -70 mV, within about 20 ms, and inactivatesover several seconds. Inactivation starts at about -120 mV and is complete at -60 mV. ‘D is blocked by mM concentration of 4-AP and nM DTX. It is insensitiveto TEA, Ba2 and Cs (Storm, 1990b). The current can cause a long delay inthe onset of firing in response to long-lasting depolarizing stimuli and keep thecell from depolarizing further. ‘D contributes very little to the normal membranepotential and is only effective in cells with a highly negative resting potential.XIE 37The rapid activation of this current suggests that it may participate in spikerepolarization (Storm, I 987a, 1 988c, I 990b). Because of its slow recovery ofinactivation (up to 20s), it enables the cell to “integrate” separate depolarizinginputs over several seconds. Therefore, the response of the cell not onlyreflects the immediate synaptic input, but also takes into account what happenedin the preceding seconds (Storm, 1988a).Hippocampal ‘M was first reported in CAl cells by Halliwell and Adams(1982) and later recorded in CA3 cells (Brown and Griffith, 1 983b; Madison etal., 1987). The hippocampal ‘M resembles closely the one in sympatheticganglia (Brown and Adams, 1980; Adams et al., 1982a). It can be blocked byacetylcholine and other muscarinic agonists (Brown and Adams, 1980). ‘M isactivated by depolarization beyond about -60 mV. It activates and deactivatesslowly (time constant: 50 ms) and does not inactivate (Adams et al., 1982). 1M isblocked by bath-application of Ba2 and mM concentration of TEA, not by Csand 4-AP (Halliwell and Adam, 1982; Storm, 1989). This current is also blockedby serotonin (Colino and Halliwell, 1987, also see Andrade et al., 1986). On theother hand, somatostatin (SS-14 and SS-28) can enhance ‘M (Moore et al.,1988; Watson and Pittman, 1988). ‘M participates in an early phase of spikefrequency accommodation (Madison and Nicoll, 1984). It is also involved in themedium AHP following a single spike or a burst of spikes (Storm, 1989).However, ‘M contributes little to the resting membrane potential of hippocampalneurons unless the resting membrane potentials of the cells are more positivethan -60 mV. ‘M will form a component of outward rectification when these cellsare depolarized, particularly for long duration.‘K(IR) is a fast inward rectifier current, which is activated byhyperpolarization beyond the resting potential and reaches the peak with 5 ms(Owen, 1987). It inactivates at potentials more negative than -100 mV (timeXIE 38constant about 35 ms at -1 l5mV). It is blocked by Cs and Ba2 (Brown,1990b). A variety of neurotransmitters, including GABA (via GABA8 receptors),serotonin and adenosine can induce an inward rectifying K current inhippocampal neurons (Newberry and Nicoll, 1985; Colino and Halliwell, 1987;Andrade and Nicoll, 1987).l is a slow mixed Na/K inward rectifier current, which is activated byhyperpolarization beyond -80 mV (Halliwell and Adams, 1982). Activation isrelatively slow (time constant about 100 ms at -82 mV), but accelerates withincreasing hyperpolarization (time constant 37 ms at -130 mV) (Halliwell andAdams, 1982). l is blocked by external Cs but unaffected by Ba2. l doesnot contribute to the normal resting potentials of hippocampal cells, but itsactivation serves to resist hyperpolarizing deviations from the resting potential,including a characteristic rebound depolarization during a hyperpolarizingcurrent pulse, and hence stabilizes the membrane potential. Deactivation of lcontributes to the rebound depolarization and excitation following ahyperpolarizing pulse. l contributes to spike after-hyperpolarization only whenthe spike is initiated from membrane potentials negative to the normal restingpotential (Storm, 1989). As mentioned above, these six types of voltage-gatedchannels persist whenCa2-influx has been eliminated byCa2-free medium, orby the Ca2 channel blockers, Mn2 or Cd2.There are at least two types of K currents which areCa2-dependent. ICis a ‘fast” and largeCa2-dependent K-current, which is activated rapidly(within 1-2 ms) when Ca2 ions flow through voltage-gated Ca2 channelsfollowing their activation by a depolarization beyond -40 mV or during an actionpotential (Brown and Griffith, I 983b; Lancaster and Adams, 1986; Storm, I 987a;Lancaster and Nicoll, 1987). When the neuron is repolarized, IC deactivateswithin 50-1 50 ms, depending on the voltage (Brown and Griffith, 1983b). TheXIE 39activation of the current requires a relatively high concentration of internal Ca2(threshold > I pM) (Franciolin, 1988). Ic can be eliminated by Ca2-freemedium and bath-application ofCa2-channeI blockers Mn2,Cd2 or Co2, andby injection of fast Ca2 chelator BAPTA, but not by slow chelator EGTA(Lancaster and NicoIl, 1987; Storm 1987b; Madison et al., 1987a; Storm, 1989).IC is also blocked by low concentration external TEA (1-10 mM) below thatrequired to block the ‘K’ and by nM concentration charybdotoxin (CTX)(Lancaster and Nicoll, 1987; Storm, I 987c). Since IC is strongly activated duringa single action potential, its activation contributes to spike repolarization andgenerates the early phase of the spike after-hyperpolarization (Lancaster andNicoIl, 1987; Storm 1987a, b). IC also has an influence on the initial part of thespike accommodation. At single channel level, Ic may be mediated throughlarge-conductance (140-270 pS)Ca2-dependent K channels, BK” channels(Brett and Lancaster, 1985; Brett et al., 1986; Francilin, 1988; Ikemoto et al.,1989).‘AHP rises slowly following Ca2 entry, and peaks in 400-700 ms. Itdeclines more slowly (time constant 1-1.6 s) on repolarization. Its decay islargely voltage-independent between -30 to -90 mV (Lancaster and Adams,1986). ‘AHP is activated by much lower Ca2 concentrations (30 to 60 nM) thanthose required to activate IC (Knopfel et al., 1989). It is not clear whether theslowness of ‘AHP is partly intrinsic to the channels or whether it just reflects thetime-course of the intracellular Ca2 concentration or subsequent biochemicalsteps (Storm,1990b). At single channel level, a class of small conductance(approximately 19 pS) Ca2-activ ted K channels may mediate the 1AHP(Lancaster et al., 1987). ‘AHP can be eliminated byCa2-free medium and Ca2-channel blockers, Mn2, Cd2 and Co2. Intracellular injection of BAPTA orEGTA also blocks the ‘AHP (Lancaster and Nicoll, 1987). ‘AHP differs from ICXIE 40because ‘AHP is resistant to TEA and CTX (Lancaster and Adams, 1986; Storm,1987a; Lancaster and NicoIl, 1987). It is also insensitive to 4-AP in CAl cells(Storm, 1990). In contrast to Ic’ ‘AHP can be regulated by a variety ofneurotransmitters, acetylcholine and muscarinic agonists (via M1 receptors)inhibit ‘AHP (Cole and Nicoll, 1983; Madison et al., 1987a). 1AHP is about 10 foldmore sensitive to muscarinic agonists than ‘M (Madison et al., 1987a).Noradrenaline (via 13i receptors) (Madison and NicoIl, 1982; Haas and Konnerth,1983), histamine (via H2 receptors) (Haas and Konnerth, 1983), serotonin(Andrade and Nicoll, 1987; Colino and Halliwell, 1987) and ACPD (viametabotropic glutamate receptors) (Charpak et al., 1990) can reduce ‘AHP•These substances suppress ‘AHP by blocking the K current itself rather thanaffecting the Ca2 transient (Knoptel et al., 1989). Adenosine (Haas andGreene, 1984) and dopamine (Benardo and Prince, 1982; Haas and Konnerth,1983; also see Malenka and Nicoll, 1986) have been reported to enhance ‘AHPActivation of protein kinase C or cyclic AMP-dependent protein kinase alsoinhibit ‘AHP (Baraban et al., 1985; Malenka et al., 1986a; Storm, 1990). ‘AHPgenerates the long after-hyperpolarization following hippocampal actionpotentials. ‘AHP provides a strong negative feedback control of the activity of thecells. It is activated by Ca2 influx during the action potential and increases asthe number of spikes increase (Lancaster and Adams, 1986). It tends tosuppress further discharge by hyperpolarizing the cell and shunting depolarizinginputs. More importantly, ‘AHP plays a major role in the spike frequencyaccommodation in hippocampal pyramidal cells (Madison and Nicoll, 1982,1984).3.3.4. Ctcurrentslcl(v) is a slow and voltage-dependent C1 current, which is activated byhyperpolarizing steps between -20 and -100 mV in the hippocampal pyramidalXIE 41cells (Madison et al., 1986). This current is notCa2-dependent. It is blocked byCd2 and phorbol dibutyrate (Madison et al., 1986).Another type of C1 current, the icI(cA)’ isCa2-dependent and voltage-insensitive (Owen et al., 1988). Icl(cA) is activated by > 0.5 mM Ca2. It may bemediated through Ca2 activated C[ channels of 20 pS conductance. Thiscurrent may contribute to the longCa2-dependent” tail currents” following Cl--loading coupled with suppression ofCa-activ ted K-currents (Brown andGriffith, I 983b).3.3.5. “Leak’1currents“Leak” currents are the passive currents which remain around restingpotential when voltage- andCa2-gated currents are blocked. One of the “leak”currents is a voltage-insensitive K current, which can be reduced by muscarinicagonists (Madison et al., 1987b; Benson et al., 1988). This effect may bemediated by a pertussis toxin insensitive GTP-binding protein (Brown et al.,1988). Some “leak” currents are mediated through Cl channels (Franciolini andNonner, 1987; Franciolini and Petris, 1988; Owen et al., 1988). Both K and Cl-“leak” currents play an important role in stabilizing the membrane potentialaround -60 to -70 mV.3.4. Evoked field potentials in the hippocampusDue to the close packing and a parallel arrangement to each other of cellsin the hippocampus, stimulation of excitatory afferents leads to a synchronousactivation of a large population of neurons and generates large extracellular fieldpotentials corresponding to EPSP activity (field EPSP5) and action potentialdischarge (population spike) (Andersen, 1975, Schwartzkroin and Mueller,1987). In turn, the axon collaterals from the principal neurons activate a giveninterneuron population and generate field IPSPs. Moreover, direct stimulation ofinhibitory interneurons affects a large number of principal neurons because theirXIE 42axons ramify for hundreds of micrometers (Kunkel and Schwartzkroin, 1982).The discussion here will be focused on the field potentials in the CAl regions.The field synaptic potentials can be generated by the stimulation of Schaffercollaterals or commissural afferents. The field EPSP recorded at the apicaldendrites of the CAl pyramidal cells is characterized as a large negative wavewhen the stratum radiatum is stimulated. The large negative wave actuallyrepresents a mixture of fields, generated largely by EPSPs from the apicaldendrites and by some components of the IPSPs from the soma. This fieldEPSP can be interrupted by a positive peak when the stimulation strength isstrong enough to cause the cells to discharge spikes. The positive peak mayreflect a population spike generated in the somatic region. The negativelydirected field EPSP is reversed in polarity to the positive wave, and thepopulation spike manifests as a sharp negative wave when the recordingelectrode moves close to the CAl cell body layer. The size of the populationspike represents the number of synchronous firing cells. A negatively directedcomponent recorded prior to the field EPSP and the population spike is thepresynaptic volley. The presynatic volley results from the extracellular currentssurrounding the synchronously activated unmyelinated fibres running in thedendritic layer (Andersen et al., 1978) and is usually taken as an index of thenumber of fibres activated. Recently, field IPSPs have been analyzed in thepresence of excitatory amino acid antagonists (Lambert et al., 1991 b).3.5. Ephaptic interactionsThe large field potentials in the hippocampus are capable of influencingindividual neuronal activity (Jefferys and Haas, 1982; Taylor and Dudek, I 982b;Turner et al., 1983). This phenomenon known as ephaptic interaction, is due tothe effects of extracellular currents. The current flow associated with fieldpotentials may discharge (or inhibit) neurons at the subliminal fringe of an activeXIE 43population (Schwartzkroin and Mueller, 1987). Ephaptic interactions can act torecruit additional neurons into an already active population.3.6. Electrotonic couplingElectrophysiological and morphological evidence shows that electrotoniccoupling exists among pyramidal cells and granule cells (MacVicar and Dudek,1981, 1982; Taylor and Dudek, 1982a). Intracellular injection of Lucifer yellow,which does not cross chemical synaptic junctions, reveals gap junctions and thepattern of dye coupling in the hippocampus (MacVicar and Dudek, 1981).Anatomical ultrastructural studies suggest that gap junctions exist betweeninterneurons (Kosaka, 1983; Schlander and Frotscher, 1986). Dudek et al.(1983) have demonstrated electrotonic coupling in a small percentage of cells byusing antidromic stimuli to trigger “short-latency depolarization”. Theelectrotonic coupling may facilitate or disrupt synchronous discharge producedby chemical synapses, depending on the relative strength and timing of chemicalversus electrotonic connections (Traub et aL, 1981; 1982; Traub et al., 1989;Dermietzel and Spray, 1993).3.7. Excitatory postsynaptic potentials (EPSPs)All major hippocampal afferents produce EPSPs in the hippocampalprincipal neurons (Andersen et al., 1966; 1975). Moreover, the afferents tohippocampal interneurons also generate EPSPs in these interneurons (Lacailleet al., 1987). Activation of the perforant path fibres produces EPSPs on thedendrites of granule cells (Blackstad, 1958; Hjorth-Simonsen and Jeune, 1972;Steward, 1976; McNaughton and Barnes, 1977). The unitary EPSP of granulecells is approximately 0.1 to 0.3 mV in amplitude (McNaughton et al., 1981).Activation of the mossy fibres generates EPSPs in the dendrites of CA3pyramidal cells (Andersen and Lomo, 1966; Yamamoto, 1972). In CAl field,excitatory inputs from Schaffer collaterals or commissural produce EPSPs in theXIE 44pyramidal dendrites (Andersen, 1960). The unitary EPSP is on average 0.15 mVin amplitude, probably produced by a single quantum of transmitter (Sayer et al.,1989; Andersen, 1990). EPSP has a reversal potential of approximately 0 mV.The EPSP5 in response to distally located synapses are relatively slower thanthose elicited by proximally located synapses on the same cell. This is inagreement with the cable theory (Turner, 1988). EPSPs produced byneighbouring synaspes can sum linearly or non-linearly with each other and withhyperpolarizing inhibitory potentials, depending on the space between thesynapses and the timing of the synaptic input (Langeoen and Andersen, 1983).The summation effect will be greater for synapses contacting the samesecondary dendrites than for more distributed dendritic contacts. L-Glutamate isthe major excitatory neurotransmitter in the hippocampus. The CAl pyramidalcell EPSPs are mediated by NMDA and non-NMDA receptors.4. SYNAPTIC TRANSMISSION IN THE HIPPOCAMPUS4.1. Excitatory neurotransmitters in the hippocampusGlutamate and/or aspartate are thought to be the major excitatoryneurotransmitters in the vertebrate central nervous system. In the hippocampus,all major fibre systems, including the perforant paths, the mossy fibre system,the Schaffer collateral/commissural system and the dentateassociational/commissural system, are enriched in glutamate-likeimmunoreactivity (Storm-Mathisen et al., 1983; Liu et al., 1989; Bramham et al.,1990; Ottersen, 1991). These fibre systems also take up radiolabelledglutamate or aspartate (Taxt and Storm-Mathisen, 1984) and release glutamateand/or aspartate in a calcium dependent manner (Nadler et al., 1978;Hamberger et al., 1978; Ottersen, 1991). It has been shown that the reversalpotential for glutamate-induced responses is similar to that for the naturalXIE 45transmitter both in the Schaffer collateral system (Hablitz and Langmoen, 1982)and in the perforant paths (Crunelli et al., 1984).As the major excitatory neurotransmitter in the hippocampus, glutamatehas diverse actions in the hippocampal neurons and plays a very important rolein the synaptic transmission of the hippocampus. The actions of glutamate aremediated by five different subtypes of glutamate receptors defined by thestructural analogues of glutamate which have the highest specificity for eachreceptor type (Watkins et al., 1990; Zorumski and Thio, 1992; Barnes andHenley, 1992). These analogues of glutamate are N-methyl-D-aspartate(NMDA), c-amino-3-hydroxy-5-methyl-4-isoxazole propionate (AMPA), kainate,(±)-1-aminocyclopentane-trans-1 ,3-dicarboxylic acid (trans-ACPD) and L-2-amino-4-phosphonobutanoate (L-AP4). The ACPD receptor is referred to as themetabotropic glutamate receptor because it is linked to G-proteins andphosphoinositide metabolism (P1 turnover). The AMPA and kainate receptorstogether are referred to as the non-NMDA receptors because these tworeceptors act very similarly in many brain areas. Autoradiographic studies haveshown that the NMDA, AMPA, kainate and metabotropic receptors are dense inthe molecular layer of the dentate gyrus, and the stratum oriens and the stratumradiatum of the CAl and CA3 areas of the hippocampus (Young et al., 1991).NMDA and non-NMDA have been shown to be co-localized on the subsynapticmembrane (Bekkers and Stevens, 1989).4.1.1. NMDA receptors and their role in synaptic transmissionNMDA receptor is the most well characterized glutamate receptor subtypebecause of its physiological properties and the development of NMDA receptorspecific antagonists. This receptor is a ligand-activated receptor-channelcomplex, which can be activated by glutamate or NMDA. L-Glutamate, Laspartate and several sulphur amino acids, including L-homocysteic acid, areXIE 46potent endogenous NMDA receptor agonists (Mayer, 1991). The most potentand selective competitive NMDA antagonists are D-APV (Davies and Watkins,1982), D-CPP and CPP (Davies et al., 1986; Olverman and Watkins, 1989).Noncompetitive antagonists, which apparently plug the NMDA channels in avoltage-dependent manner, include dissociative anaesthetics such as ketamine,PCP, TCP and MK-801, divalent cations such as Mg2 and Zn2, and tricyclicantidepressants such as desipramine (Watkin et al., 1990; Mayer, 1991,MacDonald et al., 1991; Zorumski and Thio, 1992).Unlike other ionotropic glutamate receptor channels which are permeableonly to Na and K, NMDA receptor channels are highly permeable to Ca2.Activation of the NMDA receptor-channel complex allows Na and Ca2 influx,and K efflux through the NMDA receptor channels (Mayer, 1991; Mayer andMiller, 1990). Single-channel studies have shown that NMDA receptor channelsopen for 5-10 ms on an average, with a main conductance state of 50 pS, andsubconductance states of 10, 20, 30 and 40 pS (Cull-Candy and Usowicz, 1987;Jahr and Stevens, 1987; Ascher et al., 1988b, Cull-Candy et al., 1988). Themain conductance state of the ion channel is preferentially opened by NMDAand shows significant permeability to Ca2 (Ascher and Nowak, 1988b; Jahr andStevens, 1987). Under the normal physiological conditions, the NMDA receptor-channel complex is regulated in a voltage-dependent manner by extracellularMg2 (Nowak et al., 1984; Mayer et al., 1984; Jahr and Steven, 1990; Mayer,1991). The current-voltage (l-V) relationship for NMDA is highly nonlinear suchthat very little current flows at potentials negative to -70 mV, whereas beyondthis point the current increases with depolarization. The I-V curve for NMDAbecomes linear after removing Mg2 from the recording medium (Nowak et al.,1984; Mayer et al., 1984; Jahr and Steven,1990). At single channel level, Mg2has been found to block only the main conductance state (50 pS) in a voltage-XIE 47dependent manner (Nowak et al., 1984; Jahr and Stevens, 1987). The voltage-dependent blockade of NMDA receptor channels by extracellular Mg2 may bedue to the binding of the ion within the channel pore when the channel is open(Nowak et al., 1984; Mayer et al., 1984, Jahr and Stevens, 1990a). Recentevidence suggests that Mg2 probably blocks the channels in both the open andclosed conformations (Jahr and Stevens, 1990a,b). Since physiological level ofMg2 (approximately 1 mM) in cerebrospinal fluid is much higher than theconcentration that is required to have significant effect on NMDA channels, thesynaptic transmission mediated through the NMDA receptors in thehippocampus is highly affected by the extracellular Mg2 blockade of the NMDAchannels under the normal physiological conditions. Johnson and Ascher(1990) have recently suggested that intracellular Mg2 also blocks the NMDAreceptor channels. However, the block would probably not influence the NMDAresponses under normal physiological conditions because the concentrationwhich is similar to the physiological intracellular Mg2 levels (1-3 mM) has verylittle effect on the NMDA channels at voltages below 0 mV (Johnson, 1991).Extracellular zinc, another divalent cation, has been shown to reversiblyinhibit the NMDA-evoked responses. Unlike Mg2, Zn2 non-competitivelyblocks the NMDA receptor-channel complex in a voltage-independent mannervia a site presumed to be located on the extracellular domain of the complex(Peter et al., 1987; Westbrook and Mayer, 1987; Mayer et al., 1989b). Singlechannel analysis shows Zn2 induces two effects on NMDA channels (Christineand Choi, 1990; Legendre and Westbrook, 1990). At low concentration (<10mM), Zn2 decreases the channel open probability by decreasing the channelopening frequency. At higher concentration, Zn2 decreases the channel opentime and burst duration, and reduces the single channel current amplitude. Theeffect of Zn2 on the single channel current amplitude is voltage-dependent, ItXIE 48has been suggested that Zn2 may act at two separate sites in the NMDAcomplex (Christine and Choi, 1990; Legendre and Westbrook, 1990). SinceZn2 is present in nerve terminals and synaptic vesicles in the hippocampus(Perez-C lausell and Danscher, 1985), it is possible that the ion can be released,together with glutamate, from the nerve terminals (Aniksztejn, 1987). Therefore,this ion may play a very important role in the hippocampal synaptic transmission.Zn2 has been shown to reduce the NMDA receptor-mediated EPSPs as well asGABA receptor-mediated IPSPs in the hippocampus (Forsythe et al., 1988;Mayer and Vyklicky, I 989b).Johnson and Ascher (1987) first reported that glycine strongly enhancesthe NMDA response, probably via an allosteric action at a site on the NMDAreceptor-channel complex. Strychnine, a powerful antagonist of the inhibitoryglycine receptor, did not inhibit the effect of glycine on the NMDA responses.Evidence has shown that the glycine binding sites located on the NMDAreceptor-channel complex are distinct from the inhibitory glycine receptors(Langosch et al., 1990; Barnes and Henley, 1992). More recent studies suggestthat glycine not only facilitates the NMDA responses, but also is essential for theNMDA receptor activation (Kleckner and Dingledine, 1988; Lerma et al., 1990;Sekiguchi, 1990; Vyklicky et al., 1990; Zorumski and Thio, 1992).Pharmacological and kinetic studies show that there are two binding sites forglycine on the NMDA receptor-channel complex. Occupation of these bindingsites by glycine is required for the activation of the NMDA receptor by NMDAagonists (Benveniste et al., 1990; Javitt et al., 1990; Benveniste and Mayer,1991; Clements and Westbrook, 1991). The glycine action on the NMDAreceptor-channel complex can be antagonized by several drugs, including 7-chlorokynurenic acid (Kemp et al., 1988), CNQX, 3-amino-l-hydroxypyrrolid-2-XIE 49one (HA-996) and a series of derivatives of indole-2-carboxylic acid (Huettner,1989). HA-996 is also a partial agonist (Foster and Kemp, 1989).The NMDA receptor-channel complex can be modulated by other agents,such as polyamines, sulthydryl redox reagents and H ions. The endogenouspolyamines, spermines and spermidine enhance NMDA responses in a voltage-independent fashion by binding a distinct site on the complex (Ransom andStec, 1988; Williams et al., 1990, Scott, et al., 1993). Polyamines can enhancemaximal responses to NMDA and glycine by increasing the affinity of the NMDAcomplex for glycine (McGurk et al., 1990). Polyamines also enhance the bindingof non-competitive antagonists, such as MK-801 and TCP to the NMDA receptorchannels (Ransom and Stec, 1988), and the binding of a competitive antagonist,CPP (Pullman and Powel, 1991). The disulfide reducing reagent dithiothreitolpotentiates the NMDA responses whereas the disulfide oxidizing reagent 5, 5-dithio-bis-2-nitrobenzoic acid reduces NMDA responses in a voltage-independent manner by acting at a site that is distinct from the NMDA, glycine,polyamine binding sites (Aizenman et al., 1989; Reynold et al., 1990). Changesin the extracellular pH value can affect the NMDA receptor-channel complex(Tang et al., 1990; Vyklicky et al., 1990; Traynelis and Cull-Candy, 1990, 1991).Extracellular pH values greater and less than physiological pH enhance anddepress NMDA currents, respectively. These effects are voltage-independent.Recent evidence suggests that changes in the extracellular pH value alsochange the potency of NMDA receptor competitive antagonists (Benveniste andMayer, 1992). Other than the extracellular modulators, whole cell patch clampstudies show that intracellular factors are essential for maintaining the normalNMDA receptor function. During whole cell recording, the NMDA responsesgradually “run down”. The run down could be prevented if the recordingelectrode contained an ATP regenerating solution, suggesting that the loss ofXIE 50similar factors is responsible for the irreversible decline in response (MacDonaldetal., 1989).Desensitization is a property shared by many ligand-gated ion channel-receptor complexes, including glutamate receptors. Three different modes ofNMDA receptor desensitization have been observed. One type of NMDAreceptor desensitization is voltage- andCa2-dependent (Clark et al., 1990;Zorumski et al., 1989; Vyklicky et al., 1990). The rate and degree ofdesensitization increases with hyperpolarization and higher extracellular Ca2concentration (Clark et al., 1990). Intracellular injection of BAPTA reduces theCa2-dependent desensitization (Mayer et al., 1991). With low extracellulara second type of NMDA receptor desensitization, which is voltage-dependent and modulated by extracellular glycine, develops (Mayer, 1989b,Benveniste et al., 1990). The glycine-dependent NMDA receptor desensitizationis reduced or inhibited by high concentrations of extracellular glycine (Mayer etal., 1991). The third form of NMDA receptor desensitization is Ca2- andglycine-independent (Sather et al., 1990; Shirasaki et al., 1990).The degree which NMDA receptors contribute to synaptic transmission inthe hippocampus is highly dependent on the extracellular Mg2 concentration,the membrane potential and the state of synaptic inhibition. Under normalphysiological conditions, the EPSP or EPSC mediated by NMDA receptors at theresting membrane potential in the hippocampal CAl neurons is very small, if nottotally absent (Collingridge et al., 1988a; Collingridge and Lester, 1989). TheNMDA receptor-mediated component of the EPSP, which is APV-sensitive,appears when the cell is depolarized. The NMDA receptor-mediated componenthas a slow rise time (8-20 ms), a long duration (approximately 100-200 ms) anda reversal potential close to 0 mV (Collingridge et al., 1988a). The voltagedependent NMDA component is due to the blockade of extracellular Mg2. TheXIE 51NMDA component of the EPSP is affected by synaptic inhibition. The NMDAcomponent can be observed at potentials near rest when the synaptic inhibitionis blocked by GABAA antagonists (picrotoxin or bicuculline) or reduced usingvoltage-clamp techniques (Collingridge et al., 1988a; Collingridge and Lester,1989). The NMDA component is also frequency-dependent. Under normalphysiological conditions, the NMDA component can be recorded during high-frequency stimulation (Herron et al., 1986; Collingridge, 1988b). It is suggestedthat during high-frequency stimulation, a neuron may become depolarized for asufficient time to reduce the Mg2 block of NMDA channels (Collingridge et al.,I 988b).4.1.2. Non-NMDA glutamate receptorsAMPA and kainate receptors together are generally referred to as nonNMDA receptors, which are readily distinguished pharmacologically from NMDAreceptors. AMPA receptors are also called quisqualate receptors. However,quisqualate acts not only on the AMPA receptors, but also on the metabotropicglutamate receptors. AMPA appears to be much more selective to the AMPAreceptors. Therefore, the ionotropic quisqualate receptors are named afterAMPA in order to distinguish from the quisqualate-activated metabotropicglutamate receptors (Watkins et al., 1990; Zorumski and Thio, 1992).Autoradiographical studies show that the distribution of AMPA sites resemblesthat of NMDA sites; however, a higher level of binding is observed in thepyramidal cell layer (Monaghan et al., 1984a). Kainate binding sites areprimarily confined to the stratum lucidum of CA3, the termination zone of thedentate granule cell-mossy fibres (Foster et al., 1981; Collingridge and Lester,1989). L-Glutamate is the major endogenous neurotransmitter for the nonNMDA receptors. The concentrations of glutamate needed to activatephysiological responses at non-NMDA receptors are 100 times greater thanXIE 52those required at NMDA receptors (Patineau and Mayer, 1990). The mostpotent and selective competitive antagonists of the non-NMDA receptors areCNQX, DNQX and NBQX (Blake et al., 1988; Honore et al., 1988; Sheardown etal., 1990). At higher concentration, CNQX and DNQX may act as non-competitive NMDA antagonists by interacting with the glycine site (Birch et al.,1988; Harris and Miller, 1989; Yamada et al., 1989). NBQX is specific for nonNMDA receptors and also shows a 30 fold selectivity for AMPA receptorscompared to kainate receptors (Sheardown et al., 1990).The ion channels coupled to non-NMDA receptors are permeablenonselectively to monovalent cations (Na and Kj, but exhibit a poorpermeability to Ca2 (Ascher and Nowak, 1988; Vylicky et al., 1988). The nonNMDA receptor channels are insensitive to Mg2 (Nowak et al., 1984).Therefore, the I-V relationship for the two agonists is nearly linear over therange -90 to +30 mV. The reversal potential for AMPA or kainate isapproximately 0 mV.Non-NMDA receptors can be readily differentiated from NMDA receptors,but AMPA and kainate receptors are more difficult to differentiate from oneanother experimentally. AMPA and kainate show some competitive interactions,which may be explained by both agonists acting through the same receptor-channel complex with a common binding site (Lodge and Johnson, 1990, Barnesand Henley, 1992). However, independent actions of AMPA and kainate havealso been observed (Perouansky and Grantyn, 1989; Mayer and Westbrook,1987). It has been suggested that the diverse heterogeneous populations ofbinding sites for kainate and AMPA exist (Barnes and Henley, 1992). Severalcloned non-NMDA receptor-channel complex subunits, when expressed alone orin combination with others, respond to both AMPA and kainate (Boulter et al.,1990; Dawson et al., 1990; Keinanen et al., 1990; Nakanishi et al., 1990).XIE 53However, the kainate selective non-NMDA receptor subunit has also beencloned (Egebjerg et al., 1991).The channels coupled to AMPA and kainate receptors have multipleconductance states like those opened by NMDA (Jahr and Steven, 1987; Ascherand Nowak, 1988b; Cull-Candy et al., 1988; Cull-Candy and Usowicz, 1989).Activation of AMPA receptors preferentially opens a channel with conductancestates of 8-1 5pS, while kainate receptor activation primarily opens the channel tosmaller conductance states of 1-5 pS (Jahr and Stevens, 1987). Its mean opentime of these conductances range from 0.5 to 5.3 ms. A high conductance (35pS) activated by quisqualate, an AMPA receptor agonist, has been described inthe hippocampal neurons by Tang et al. (1989). The mean open time is 3 to 8ms. This conductance may underlie the generation of the fast AMPA-receptormediated EPSCs.Desensitization occurs to both AMPA and kainate-activated responses.In dissociated and cultured hippocampal neurons, the response to glutamateand quisqualate desensitizes rapidly, while the response to kainate does not(Kiskin et al., 1986; MacDonald et al., 1987; Mayer and Vyklicky, 1989b; Trusselland Fischbach, 1989; Trussell et al., 1988). However, pretreatment withglutamate or with quisqualate substantially reduced the subsequent response tokainate (Kiskin et al., 1986; Trussell et al. 1988). Aniracetam (nootrophic agent),concannavalin A and wheat germ agglutinin (WGA) can reduce thedesensitization of the non-NMDA receptors. Aniracetam appears to act byprolonging the AMPA channel open time (Tang et al., 1991; Vyklicky et al.,1991).The non-NMDA responses have demonstrated “rundown” during wholecell patch clamp recordings. This rundown can be prevented by the inclusion ofXIE 54phosphorylation factors and activators of cAMP-dependent protein kinase in thepatch electrode (Greengard et al., 1991; Wang et al., 1991).Under normal physiological conditions, synaptic transmission at theresting membrane potential in the hippocampal CAl pyramidal cells is primarilymediated by non-NMDA receptors. The non-NMDA receptors mediate a fastEPSP or EPSC component, which is voltage-independent over the range -100 to+40 mV and CNQX-sensitive (Hestrin et al., 1990a,b) . The reversal potential ofthe EPSP or EPSC component is around 0 mV. The rise and the decay timeconstant of the non-NMDA component are 1-3 ms and 7-8 ms, respectively(Hestrin et al., 1990a,b). There is the possibility that the decay time constant ofthe fast EPSC may reflect the mean open time of the AMPA channel.4.1.3. AP4 receptorsA fourth type of glutamate receptor has been proposed based on theantagonist actions of the L-glutamate analogue L-2-amino-4-phosphonobutyrate(L-AP4). The distribution of L-AP4 receptors is highly localized. In thehippocampus, L-AP4 receptors are located in the lateral perforant pathway(Koerner and Cotman, 1981) and in the guinea pig mossy fibre pathway(Yamamoto et al., 1983 Lanthorn et al., 1984). L-AP4 depresses the amplitudeof the EPSC or EPSP in hippocampal neurons (Cotman et aL, 1986; Forsytherand Clements, 1990). It has been suggested that the activation of presynapticAP4 receptors by L-AP4 or L-glutamate depresses the release of glutamate fromthe presynatic terminals (Cotman et al., 1986; Crowder et aL, 1987; Gannon etal., 1989; Forsyther and Clements, 1990). The AP4 receptors may act asautoreceptors at the synaptic terminals. The mechanisms of this presynapticinhibition are still not very clear. Recent evidence suggests that L-AP4depresses glutamate release by inhibiting calcium influx via G-protein-coupledpresynaptic AP4 receptors in cultured olfactory bulb neurons (Trombley andXIE 55Westbrook, 1992). Data on elucidating the precise binding site through which LAP4 exerts its inhibitory effect on excitatory synaptic transmission are limited.This is because no selective agonists and antagonists are known to havesufficient affinity at the L-AP4 receptor that would enable the development ofradioligand binding assays.4.1.4. ACPD receptorsACPD receptors are a distinct family of glutamate receptors, members ofwhich are coupled to various second messenger systems through GTP bindingproteins (G-proteins) (Schoepp and Conn, 1993; Tanabe et al., 1992, Desai etal., 1992). The G-protein-linked glutamate receptors, also referred to asmetabotropic glutamate receptors (mGluRs), are selectively activated by (±)1-aminocyclopentane-trans-1 ,3-dicarboxylic acid (trans-ACPD) or its active isomerls,3R-ACPD (Schoepp and Conn, 1993; Conn and Desai, 1990; Palmer et al.,1989). ACPD receptors has been found in the CAl, CA3 and the dentate gyrusof the hippocampus (Young et al., 1991). Glutamate, quisqualate and ibotenatecan activate ACPD receptors nonselectively. ACPD receptors were firstdescribed as the glutamate receptors that are coupled to phospholipase C (PLC)and phosphoinositide hydrolysis (Sladeczek et al, 1985; Nicoletti et al., 1985),ACPD receptors now have been demonstrated to couple with other secondmessenger systems which include the activation of phospholipase D (Boss andConn, 1992) and changes in cAMP formation (Schoepp et al., 1992; Schoeppand Johnson, 1993; Casabona et al., 1992; Winder and Conn; 1992). Recentmolecular studies have shown that ACPD receptors are a highly heterogeneousclass of glutamate receptors that are coupled to multiple second messengersystems (Tanabe et al., 1992; Schoepp and Conn; 1993). L-AP3 is able toinhibit the ACPD-stimulated phosphoionsitide hydrolysis in the hippocampalslices (Schoepp et al., 1991). However, it fails to prevent the inhibition ofXIE 56forskolin-stimulated cAMP formation by I s,3R-ACPD (Schoepp and Johnson,1993; Casabona et al., 1992). Therefore, the ACPD receptors that arenegatively linked to cAMP formation are distinct from the phosphoinositidehydrolysis-linked ACPD receptors.ACPD receptors play a significant role in the modulation of ion channelsand synaptic transmission in the hippocampus (Desai et al., 1992; Schoepp andConn et al., 1993). Activation of ACPD receptors by ACPD receptor agonistsdepolarizes hippocampal neurons associated with an increase in the inputresistance by reducing the voltage-dependent K current (IM). In addition, theactivation of ACPD receptors blocks the spike frequency accommodation andthe slow after-hyperpolarizing potential (AHP) by blocking theCa2-dependentK current (lAHp) and delays the spike repolarization (Charpak et al., 1990;Stratton et al., 1989; Desai and Conn, 1991; Hu and Storm, 1991; Hu and Storm,1992). These ACPD receptor-mediated actions enhance excitability ofhippocampal neurons. Furthermore, the activation of ACPD receptors results ininhibition of a high-threshold Ca2 current in cultured hippocampal neurons(Lester and Jahr, 1990, Swartz and Bean, 1992).Several lines of evidence have shown that the activation of ACPDreceptors by ACPD receptor agonists reduces both the NMDA receptor- andnon-NMDA receptor-mediated EPSPs, and also suppresses the GABAAreceptor-mediated fast IPSP and the GABAB receptor-mediated slow IPSP in thehippocampal CAl neurons (Baskys and Malenka, 1991; Pacelli and Kelso,1991). The mechanisms of the inhibition of synaptic transmission by ACPDreceptor agonists are not clear. It has been suggested that the reduction ofEPSPs by ACPD receptor agonists is probably due to the activation ofpresynaptic ACPD receptors which act as the autoreceptor, and the subsequentreduction of glutamate release (Baskys and Malenka, 1991). Recent evidenceXIE 57suggests that the reduction of hippocampal IPSPs by ACPD agonists is at leastpartially mediated by a reduction in synaptic excitation of GABAergicinterneurons (Desai and Conn, 1992).L-AP3 selectively inhibits the ACPD-stimulated phosphoinositidehydrolysis, but fails to block both the ACPD agonist-induced actions on theexcitability of hippocampal neurons (Hu and Storm, 1992) and the ACPDagonist-induced depression of hippocampal synaptic transmission.(Goh andMusgrave, 1993). Boss et al. (1992) have recently shown that the ACPDagonist-induced phosphoinositide hydrolysis and the modulation of hippocampalpyramidal cell excitability do not undergo parallel developmental regulation. It isbelieved that the ACPD receptors that mediate the ACPD agonist-inducedactions on hippocampal neuron excitability and synaptic transmission are eitherL-AP3 insensitive or distinct from the phosphoinositide hydrolysis-linked ACPDreceptors. It is possible that different subtypes of ACPD receptors mediatedifferent types of physiological responses by coupling to different secondmessenger systems. The discovery of more selective and potent antagonists ofACPD receptors will help to establish the functions of specific ACPD receptorsubtypes. Recently, a new ACPD receptor antagonist, (+)-cw-Methyl-4-carboxyphenyiglycine ((+)-MCPG) has been suggested to be more selectivethan L-AP3 (Ito et al., 1992; Jane et al., 1993). However, the selectivity of thisdrug has been questioned by others (Chinestra et at., 1993).Recent evidence has shown the involvement of ACPD receptors insynaptic plasticity (Behnisch et al., 1991; Ito and Sugiyama, 1991; Aniksztejn etat., 1992; Bortolotto and Collingridge, 1993). This action of the ACPD receptorswill be further discussed in the chapter on LTP.XIE 584.2. Inhibitory neurotransmitters in the hippocampusy-Aminobutyric acid (GABA) is the major inhibitory neurotransmitter in themammalian brain (Nicoll et al., 1990; Sivilotti and Nistri, 1992).Immunohistochemical studies show that the GABA-synthesizing enzymeglutamic acid decarboxylase (GAD) is concentrated in the nerve terminals ofinterneurons in the hippocampus (Ribak et al., 1978; Somogyl et al., 1983;Schwartzkroin and Mueller, 1987). The interneuron terminals containing GAD-like immunoreactivity make synaptic contacts with pyramidal and granule cellsomata and dendrites (Ribak et al., 1978; Somogyi et al., 1983). There are atleast two types of GABA receptors in the hippocampal neurons. They areGABAA and GABAB receptors. In the hippocampal CAl field, GABAA receptorsare found in somata and dendrites of pyramidal cells while GABAB receptors aremainly located in pyramidal dendrites and presynaptic terminals (Nicoll et al.,1990; Sivilotti and Nistri, 1992).4.2.1. GABAA receptorsThe GABAA receptor-channel complex comprises binding sites for GABAand for its allosteric modulators (Sivilotti and Nistri, 1992). Activation of GABAAreceptors by GABA opens C[ channels. Single channel analysis has shown thatGABA activates a channel with a main conductance state of 20-30 pS andmultiple subconductances (Segal and Barker, I 984a; Gary and Johnson, 1985;Edward et al., 1989; Sivilotti and Nistri, 1992). In addition to GABA, muscimoland isoguvacine, GABAA agonists, can also selectively activate GABAAreceptors thereby opening the GABAA channels (Segal and Barker, 1984b;Sivilotti and Nistri, 1992). Bicuculline and picrotoxinin are selective GABAAantagonists. While bicuculline acts as a competitive GABAA antagonist,picrotoxinin does not block GABA responses competitively. The binding site ofpicrotoxinin appears to be closely associated with the C[ channel of the GABAAXIE 59receptor (Sieghart, 1992). It has been shown that picrotoxinin reduces themean open time and the frequency of channel opening (Twyman et al., 1989).Penicillin also antagonizes GABAA-mediated responses in a non-competitivemanner (MacDonald and Barker, 1977; Pickles and Simmonds, 1980). TheGABAA receptor channels can be allosterically modulated by several classes ofdrugs, such as benzodiazepines, barbiturates and steroids. Benzodiazepines,such as diazepam, enhance the actions of GABA at the GABAA receptor byincreasing the frequency of the Cl channel opening (Study and Barker, 1981).Barbiturates, such as pentobarbital, enhance the actions of GABA by increasingthe mean channel open time (Study and Barker, 1981; Sigel et al., 1990). Athigher concentrations, barbiturates are able to enhance C[ conductance in theabsence of GABA (Sigel et al., 1990; Sieghart, 1992). Like barbiturates,steroids, such as the anaesthetic alfaxalone, also enhance GABA-stimulated Clconductance by prolonging the open time of the Cl- channel (Barker et al., 1987;Peters et al., 1988). A run down of the GABAA responses has been observedduring whole cell recording. The run down can be prevented by including Mg2-ATP in the low Ca2 intracellular medium (Chen et al., 1990). It appears that theGABAA receptor function is regulated by a phosphorylation process. Thisphosphorylation is not mediated by cAMP because analogues of cAMP decreaseGABAA responses (Tehrani et al., 1989).4.2.2. GABAB receptorsIt has been found that GABA can presynaptically inhibit the release ofother neurotransmitters. This action of GABA is insensitive to bicuculline andmimicked by the GABA analogue baclofen. The receptors mediating this GABAaction have been classified as GABAB receptors (Hill and Bowery, 1981).Baclofen is a selective agonist of this receptor. Electrophysiological evidencehas shown that GABAB receptors mediate not only the presynaptic actions ofXIE 60GABA and baclofen, but also the postsynaptic actions of these drugs in thehippocampus (Newberry and Nicoll, 1984; Dutar and Nicoll, 1988a, b). Whetherthe postsynaptic and the presynaptic GABAB receptors are homogenous,however, is still not clear (Dutar and Nicoll, 1988a,b; Thompson and Gahwiler,1992; Solis and Nicoll, 1992a,b; Lambert et al., 1993; Potier and Dutar, 1993).CGP35348 and 2-hydroxy-saclofen can selectively antagonize both thepresynaptic and the postsynaptic GABAB responses. On the other hand,phaclofen, which can completely block the postsynaptic GABAB responses, haslittle effect on the presynaptic GABAB responses (Dutar and Nicoll, 1988a,b;Davies et al., 1990; Olpe et al., 1990; Solis and Nicoll, 1992a,b).Activation of GABAB receptors in the hippocampal pyramidal cells resultsin an increase in the membrane conductance for K (Newberry and Nicoll, 1984;1985). The increase in K conductance by GABA8 agonists is blocked bypertussis toxin and the hydrolysis-resistant GDP analog, GDPI3S (Nicoll, 1988;Nicoll et al., 1990; Thompson and Gahwiler, 1992; Potier and Dutar, 1993). Itappears that the GABAB receptor-mediated increase in K conductance is linkedto a pertussis toxin-sensitive G protein. Evidence has shown that this G proteindoes not couple to intracellular second messengers, such as Ca2 or cAMP, butis directly linked to the K channels (Andrade et al., 1986). Activation of proteinkinase C with phorbol esters blocks the GABAB receptor-mediated increase inK conductance (Baraban et al., 1985; Andrade et al., 1986). The effects ofGABAB agonists on voltage-dependent K currents have also been reported(Saint et al., 1990; Gage, 1992). GABA8 agonists can modulate a transientvoltage-dependent K current, the A current, in hippocampal neurons Theeffects of GABAB agonists on Ca2 currents are less clear than those on Kcurrents in the hippocampal neurons. It has been suggested that the activationof GABAB receptors by baclofen does not change sustained Ca2 current inXIE 61hippocampal neurons (Gahwiler and Brown, 1985). However, Scholz and Miller(1991) have recently reported that activation of GABAB receptors by baclofeninhibits a o-CgTX-sensitive Ca2 current in the soma of cultured hippocampalpyramidal cells through the activation of a pertussis toxin-sensitive G protein.More recently, Mintz and Bean (1993) have shown that the activation of GABABreceptors in cerebellar Purkinje neurons by baclofen suppresses P-type Ca2currents through a G protein-mediated mechanism, It is not clear whetherbaclofen has similar effect on P-type Ca2 channels in hippocampal pyramidalneurons. At single channel level, data on GABAB receptors are limited.Premkumar et al. (1990) have recently reported that the GABA- or baclofeninduced K’ channels in cultured neurons have multiple conductance states withmany smaller conductance states of 5-6 pS and a maximum conductance of 70pS. The GABA8 agonist-induced K currents are blocked by GABABantagonists and suppressed by pertussis toxin. Whether the properties of thechannels activated by applied GABAB agonists on the soma of cultured neuronsare the same as those properties of the channels coupled to GABAB receptors inthe dendrites remains to be determined.The GABAB receptors located in the presynaptic terminals are not as wellcharacterized as the postsynaptic GABAB receptors. While the actions of GABAand baclofen which are mediated by the presynaptic GABA8 receptors areblocked by GABAB antagonists, such as 2-hydroxy-saclofen and CGP35348,pertussis toxin and protein kinase C, it is not very clear whether the actions ofGABA and baclofen are mediated through the activation of K conductance,depression of Ca2 current or other mechanisms(Nicoll et al., 1990; Sivilotti andNistri, 1992; Thompson and Gahwiler, 1992; Thompson et al., 1993).XIE 624.2.3. GABAA and GABAB receptor-mediated responsesGABA can elicit multiple types of responses in hippocampal pyramidalneurons (Andersen et al., 1980; Alger and Nicoll, 1982b; Nicoll et al., 1990).GABA evokes a hyperpolarization with a reversal potential of -70 mV in thesoma. This hyperpolarization is due to an outward C[ current. The secondGABA-evoked response is a depolarization in the dendrites. The induction ofthis depolarization requires a higher concentration of GABA. This depolarizingresponse is due to an inward C1 current and has an approximate reversalpotential of -50 mV. GABA also evokes a bicuculline-sensitive fasthyperpolarization in the dendrites (Alger and Nicoll, 1982b; Newberry and Nicoll,1984a,b). This hyperpolarization has a reversal potential of -70 mV, which issimilar to the somatic hyperpolarization. The GABA-induced somatichyperpolarization, dendritic depolarization and fast hyperpolarization areblocked by bicuculline and picrotoxinin, and are Cl--mediated. It has beensuggested that GABAA receptors located on the main dendritic shafts mediatethe dendritic fast hyperpolarization while the GABAA receptors on the finedendritic arbors are responsible for the dendritic depolarization. Nevertheless,the difference in the direction of C1 movement between the soma and thedendrites raises the questions whether GABAA receptors are homogenous or adendritic Cl--uptake mechanism that reverses the electrochemical Cl- gradientacross dendritic membranes exists (Alger and Nicoll, 1982b; Misgeld et al.,1986; Sieghart, 1989). These questions remain to be answered. In addition tothe GABAA receptor-mediated fast hyperpolarization in the dendrites, a dendriticslow hyperpolarization can be elicited by GABA. This slow hyperpolarization isresistant to GABAA antagonists and blocked by GABAB antagonists. Thishyperpolarization is K-dependent. It is believed that this dendritic slowXIE 63hyperpolarization is mediated by GABAB receptors. Baclofen can evoke asimilar hyperpolarization (Dutar and Nicoll, I 988a, b).4.2.4. GABAA and GABAB receptor-mediated IPSPsUnder normal physiological conditions, GABA receptors mediate threetypes of IPSPs in the hippocampal pyramidal neurons (Nicoll et al., 1990). A fastIPSP can be evoked by antidromic stimulation of pyramidal cell axons in thealveus. This fast antidromic IPSP with a latency to peak of 30-50 ms and aduration of 200-300 ms is mediated mainly by somatic GABAA receptors (Algerand Nicoll, 1982a). This fast antidromic IPSP has a reversal potential of around-70 mV and is changed by altering the Cl gradient across the membrane(Spencer and Kandel, 1961c, Alger and Nicoll, 1982a, Dingledine andLangmoen, 1980). Orthodromic stimulation of afferents in the stratum radiatuminduces a fast IPSP and a slow IPSP. The latency and duration of theorthodromic fast IPSP is very similar to the antidromic fast IPSP. Theorthodromic IPSP is generated mainly on the dendrites of the pyramidal cells byactivation of GABAA receptors, but also has somatic component elicited viafeedback and/or feedforward mechanisms (Alger and Nicoll, I 982a). Theorthodromic slow IPSP has a slow latency to peak of 150-250 ms and a durationof 400-1500 ms (Alger and Nicoll, 1982a; Davies et al., 1990). This slow IPSPhas a reversal potential of approximate -90 mV. This slow IPSP is due to anincrease in K conductance by activation of GABAB receptors. In the presenceof 4-AP and pentobarbital, orthodromic stimulation of the stratum radiatum canevoke a depolarizing IPSP. This depolarizing IPSP has a reversal potential of -50 mV and is blocked by GABAA antagonists (Alger and Nicoll et al., 1982a).The GABAA receptors mediating this depolarizing IPSP are located in thedendrites area. The reasons for the absence of this depolarizing IPSP undernormal conditions are not very clear. One hypothesis is that the powerful fastXIE 64and slow IPSPs mask the depolarizing IPSP because it has been shown thatblockade of the slow IPSP by pertussis toxin reveals a depolarizing IPSP(Thalmann, I 988a,b).Similar to GABA-induced GABAA responses, the antidromic IPSP, theorthodromic fast IPSP and the depolarizing IPSP are blocked by GABAAantagonists and altered by a change in extracelluar or intracellular Cl-concentration. The slow IPSP and the GABAB agonist-induced dendritichyperpolarization seem to be mediated by the same type of GABAB receptorsthrough similar mechanisms. Both are sensitive to pertussis toxin and blockedby the same type of GABAB antagonists, Cs, QX-314, and protein kinase C(Dutar and Nicoll, 1988a,b; Andrade, 1991; Lambert et aL, 1989; Nathan et al.,1990). These GABAB responses are also not affected by cAMP and intracellularinjection of Ca2 chelators(i.e., BAPTA and EGTA) (Alger, 1984; Newberry andNicoll, 1984b; Hablitz and Thalmann, 1987).4.2.5. Presynaptic GABAB receptor-mediated actionsGABAB receptors have been found not only at postsynaptic sites but alsoat both inhibitory and excitatory nerve terminals in the hippocampus (Bowery etal., 1990; Nicoll et al., 1990; Thompson et al., 1993). While the postsynapticGABAB receptor-mediated responses are primarily due to an increase in Kconductance, the mechanisms underlying the presynaptic GABAB receptor-mediated the depression of both inhibitory and excitatory synaptic transmissionare not clear, It is possible that the activation of presynaptic GABAB receptorsdecreases Ca2 influx by directly blocking voltage-gated conductances (Scholzand Miller, 1991) and/or by increasing K1’ conductances (Thompson andGahwiler, 1992; Saint et al., 1990). An reduction in Ca21’ influx in presynapticterminals may lead to a depression in synaptic transmission. However, thecurrent evidence is not sufficient to make a conclusion, It has been reported thatXIE 65presynaptic and postsynaptic GABAB receptors have distinct pharmacologicaland physiological properties (Dutar and Nicoll, 1988b). Phaclofen, a weakGABAB antagonist, blocks the postsynaptic GABAB receptor-mediatedresponses. However, this agent does not block the presynaptic GABABresponses (Dutar and Nicoll, 1988b; Davies et al., 1990). Intraventricularinjection of pertussis toxin blocks the hippocampal slow IPSP without affectingthe presynaptic action of baclofen (Dutar and Nicoll, 1988b).Tetrahydroaminoacriciine (THA), a K channel blocker, inhibits the actions ofbaclofen and GABA mediated through postsynaptic GABAB receptor, and has noeffect on the presynaptic GABAB receptor-mediated responses (Lambert andWilson, 1993). These lines of evidence seem to support the idea thatpresynaptic and postsynaptic GABA8 receptors are different subtypes of GABABreceptors. However, the idea is not supported by several recent reports(Thompson and Gahwiler, 1992; Solis and Nicoll, 1992a). CGP35348 and 2-hydroxy-salcofen, more potent GABAB antagonists, are able to block thepostsynaptic GABAB responses as well as presynaptic GABAB responses(Davies et al., 1990; Thompson and Gahwiler, 1992; Solis and Nicoll, 1992a).These antagonists not only block the paired-pulse depression of IPSPs, which isbelieved to be mediated by the GABAB receptors at the inhibitory nerveterminals, but also prevent the baclofen-induced EPSP depression, which ismediated by the GABA8 receptors at the excitatory nerve terminals. Directinjections of pertussis toxin into the hippocampus prevents both the postsynapticand the presynaptic GABA8 receptor-mediated responses (Potier and Dutar,1993). This result contradicts the previous finding reported by Dutar and Nicoll(1 988b). The discrepancy between these two findings may be due to the site ofpertussis toxin injection since the toxin diffuses poorly in nervous tissue (Van derPloeg et al., 1991). Thus, petussis toxin directly injected into the hippocampusXIE 66may cause a better effect on GABAB receptors than that injectedintraventricularly. Furthermore, protein kinase C activator, phorbol esters, alsoinhibits both the postsynaptic and the presynaptic GABAB receptor-mediatedresponses (Thompson and Gahwiler, 1992; Dutar and Nicoll, 1988). It isapparent that the recent evidence seems to more favour the idea that thepresynaptic and the postsynaptic GABAB receptors are pharmacologicallyindistinguishable.4.2.6. Spontaneous IPSPsIn addition to stimulation-induced IPSP5 or IPSCs, spontaneous IPSPs orIPSCs have been observed in the hippocampal neurons (Alger and Nicoll, 1980;Collingridge et al., 1984; Miles and Wong, 1984). These spontaneous IPSPsare blocked by bicuculline and picrotoxin and are not sensitive to GABABantagonists. The spontaneous IPSPs are enhanced by pentobarbitone andsuppressed by tetrodotoxin (TTX) (Alger and Nicoll, 1980). It is apparent thatinterneurons discharge is needed to generate these spontaneous IPSPs. Otisand Mody (1992) have recently showed that spontaneous IPSCs recorded in thepresence of glutamate receptor blocker, CNQX and APV, are not affected by theapplication of baclofen and CGP35348. This evidence indicates that undernormal conditions spontaneous transmitter release does not activate bothpresynaptic and postsynaptic GABA8 receptors.Another type of spontaneous IPSPs or IPSCs, called miniature IPSPs orIPSCs, are observed in the presence of TTX and Cd2 substitution of Ca2(Ropert et al., 1990; Otis et al., 1990). These miniature IPSCs are also blockedby bicuculline and picrotoxinin. The miniature IPSCs are much smaller and lessfrequent than the TTX-sensitive spontaneous IPSCs. It appears that GABArelease continues even in the absence of firing interneurons (Ropert et al.,1990).XIE 674.2.7. Inhibitory CircuitryWhile the major afferents produce EPSPs in the hippocampal principlecells, the inhibitory potentials are very powerful and have ubiquitous features ofhippocampal cell physiology (Kandel et al., 1961; Spencer and Kandel, 1968;Purpura et al., 1968). In the CAl region, IPSPs can be generated through twodifferent types of inhibitory pathways: a feedback (recurrent) and a feedforwardcircuit (Kandel et al., 1961; Dunwiddie et al., 1980; Buzsaki, 1984; Alger andNicoll, 1982a).The feedback inhibition is a circuit consisting of three elements. Firstly,an excitatory input directly arrives at the pyramidal cells. Secondly, theexcitatory output of the pyramidal cells excites the inhibitory cells throughrecurrent axon collaterals. These inhibitory cells are mainly basket cells and0/A interneurons. Thirdly, the inhibitory cells may discharge and inhibit thepyramidal cells, including those which initially activated the interneurons(Buzsaki, 1984). The feedback inhibition is supported by studies that antidromicstimulation of alvear fibres produces prominent IPSPs in the pyramidal cells(Kandel et al., 1961; Andersen et al., 1964a). The feedback IPSP initiated by anantidromic stimulation is mediated by GABAA receptors and has a reversalpotential of approximately -70 mV, resulting from an increase in Cl conductance(Spencer and Kandel, 1961c, Alger and Nicoll, 1982a, Dingledine andLangmoen, 1980). It has been suggested that feedback inhibitory fibresterminate primarily on the soma and the initial segment region of the pyramidalcells (Kandel et al., 1961; Andersen et al., 1964a; Gottlieb and Cowan, 1972;Schwartzkroin et al., 1982). However, these GABAergic terminals also makecontact with the apical and basal dendrites of pyramidal neuron although thedensity of contacts on the dendrites is lower than that on the soma (Somogyi etal., 1983).XIE 68In a feedforward inhibition, orthodromic stimulation of the stratumradiatum or the stratum oriens directly excites the inhibitory cells, which in turninduces IPSPs (Alger and Nicoll, 1982a). Orthodromic stimulation produces afast GABAA receptor-mediated IPSP and a slow GABAB receptor-mediated IPSP(Alger and Nicoll, I 982a; Alger, 1984). The fast IPSP has characteristics similarto the antidromic IPSP. The slow IPSP appears to be generated on thepyramidal cell dendrites (Leung, 1978; Lacaille et al., 1989). The inhibitory cellsexcited by orthodromic stimulation include basket cells, 0/A interneurons and LM interneurons.L-M interneurons make contact primarily with the dendrites of thepyramidal cells while basket cells and 0/A interneurons make synapses not onlyon the somata of pyramidal cells, but also on the basal and proximal apicaldendrites (Lacaille et al., 1989). Basket cells and 0/A interneurons mediateboth feedforward and feedback inhibition whereas L-M interneurons mediatesolely feedforward inhibition in the CAl field of the hippocampus. It has beensuggested that 0/A interneurons and basket cells mediate the fast IPSP while LM interneurons mediate the slow IPSP in the pyramidal cells (Alger and Nicoll,1982a, b; Alger, 1984; Newberry and Nicoll, 1985; Lacaille et aL, 1989; Williamsand Lacaille, 1992).4.3. Other neurotransmitters in the hippocampusSeveral other neurotransmitters have been found in the hippocampus(Nicoll et al., 1990; Schwartzkroin and Mueller, 1987; Swanson et al., 1987).These neurotransmitters are primarily localized in the nerve terminals of extrinsicprojections from other areas of the brain. Most of these neurotransmitters arenot as well studied as glutamate and GABA in the hippocampus.XIE 694.3.1. AcetylcholineAcetyicholine (ACh) is a major neurotransmitter in the CNS. The majorcholingergic fibres in the hippocampus are from the septum (Lewis et al., 1967,Swanson et al., 1987). ACh plays a very important role in hippocampal function,such as 8 rhythm activity (Petsche et al., 1962; Andersen et al., 1979).Both muscarinic and nicotinic receptors have been found in thehippocampus (Kuhar and Yamamura, 1976; Marks and Collins, 1982;Schwartzkroin and Mueller, 1987; Swanson et al., 1987). However, most of thecholinergic responses in the hippocampus seem to be mediated by muscarinicreceptors. Application of ACh or carbachol, an ACh analogue, generally inducesa slow depolarization associated with an increase in the input resistance ofhippocampal pyramidal cells (Benardo and Prince, 1982; Dodd, et al., 1981;Segal, 1982; Cole and Nicoll, 1983, 1984). It has been suggested that the AChevoked depolarization is due to the blockade of a voltage-independent leak Kcurrent (Benardo and Prince, 1982; Benson et al., 1988; Madison et al., 1987a)and a voltage-dependent K current, the M current (Halliwell and Adams, 1982).The ‘M is probably mediated by M2 receptors (Nicoll et al., 1990). Activation ofmuscarinic receptors also decrease aCa2-dependent K current, the ‘AHP(Benardo and Prince, 1982; Madison et al., 1987a; Muller et al., 1988) and atransient K, the ‘A (Nakajima et al., 1986). A decrease in Ca2 currents by theactivation of muscarinic receptors has also been reported (Gahwiler and Brown,1987b). A transient hyperpolarization induced by ACh or carbacol has alsobeen reported (Benardo and Prince, 1982; Haas, 1982). Muscarinic receptorshave been suggested to mediate a presynaptic depression in the hippocampus(Hounsgaard, 1978; Valentino and Dingledine, 1981; Dutar and Nicoll; 1988c;Williams and Johnston, 1988, Pitler and Alger, 1992). The mechanisms of thispresynaptic inhibition are not very clear.XIE 704.3.2. NoradrenalineNoradrenergic innervation is denser in the hilus of the dentate gyrus thanin the CAl and CA3 fields (Swanson et al., 1987). The noradrenergic terminalsin the hippocampus are largely from the pontine nucleus locus coeruleus(Blackstad et al., 1967). In hippocampal pyramidal cells, one of the mostsignificant effects of noradrenaline (NA) is the blockade of the slow AHP (1AHp)(Haas and Konnerth, 1983; Madison and Nicoll, 1982,1986). This action of NAis mediated by activation of f31 receptors, which is probably involved in a cAMP-dependent protein kinase (Nicoll et al., 1990). Application of NA produce ahyperpolarization of CAl neurons and sometime a depolarization(Langmoen etal., 1981; Segal, 1981). The hyperpolarization is mediated by an activation of creceptors and the depolarization is mediated by 13 receptors (Nicoll et al., 1990).Andreasen and Lambert (1991) have recently reported that NA receptorsparticipate in the regulation of GABAergic inhibition in hippocampal CAlneurons.4.3.3. SerotoninSerotonergic fibres from the raphe nuclei innervate all fields of thehippocampus (Kohler, 1982; Swanson et al., 1987). Serotonin, also called 5-hydroxytryptamine (5-HT), primarily induces a hyperpolarization associated witha decrease in the input resistance of the hippocampal neurons (Segal, 1980;Jahnsen, 1980). This hyperpolarization is due to an increase in a Kconductance by the activation of 5-HTIA receptors (Segal and Gutnick, 1980;Andrade and Nicoll; 1987; Colino and HaIliwell, 1987; Ropert, 1988). 5-HTIAreceptors are coupled to K channels through a pertussis toxin-sensitive Gprotein. These K channels probably also couple to GABAB receptors (Andradeet al., 1986; NicolI, 1988). In addition to the hyperpolarization, 5-HT-induces andepolarization which is probably due to a blockade of a leak K conductance,XIE 71has also been reported (Jahnsen et al., 1980, Andrade and Nicoll, 1987, Colinoand Halliwell, 1987). 5-HT can reduce 1AHP in hippocampal neurons (Andradeand NicolI, 1987; Colino and Halliwell, 1987). The receptor subtype whichmediates the depolarization and the blockade of AHP is not 5-HT1A and remainsto be determined.4.3.4. DopamineIt has been reported that hippocampus receives dopaminergic fibres fromother areas of the brain such as substantia nigra (Swanson, 1982; Swanson etal., 1987). Dopamine plays a significant role in the functioning of the brain. Ithas been suggested that dopamine Dl receptors can facilitate release oftransmitters like GABA in the midbrain. However, the functions of dopamine inthe hippocampus are not very clear (Nicoll et at., 1990).4.3.5. HistamineHippocampus receives histaminergic projections from thesupramammillary region of the midbrain (Segal and Landis, 1974; Schwartzkroinand Mueller, 1987). Histamine evokes a hyperpolarization of hippocampalneurons. This effect is probably mediated by H2 receptors and activation of a Kconductance (Haas, 1981).4.3.6. AdenosineAdenosine is known to have multiple effects in the hippocampus.Adenosine induces a hyperpolarization of hippocampal pyramidal cells as aresult of an enhancement of a K conductance (Okada and Ozawa, 1980; Segal,1982; Gerber et al., 1989; Dunwiddie and Fredholm, 1989). Pertussis toxin canprevent the adenosine-induced hyperpolarization (Fredholm et al., 1989). Itappears that adenosine, GABAB and 5-HTIA receptors are coupled to the sametype of K conductance through a pertussis toxin-sensitive G protein. Anotherpostsynaptic effect of adenosine is to enhance ‘AHP (Greene and Haas, 1985;XIE 72Haas and Greene, 1984). Adenosine also depresses synaptic transmissionpresynaptically (Dunwiddie, 1984; Dunwiddie and Haas, 1985). Themechanisms for this presynaptic depression are still unknown. It has beenreported that the presynaptic action of adenosine is not blocked by pertussistoxin, but inhibited by phorbol esters, PKC activators. However, postsynapticactions of adenosine are blocked by both agents (Thompson and Gahwiler,1992). Therefore, the mechanisms for presynaptic actions of adenosine may bedifferent from those for postsynaptic actions. Adenosine can also modulateadenylate cyclase activity. However, the link between the biochemical andelectrophysiological actions of adenosine are not clear (Nicoll et al., 1990).4.4. Neuropeptides in the hippocampusA number of neuropeptides have been found in the hippocampus. Theseneuropeptides include opioid peptides, somatostatin, substance P, neuropeptideY, cholecystokinin, vasoactive intestinal peptide, thyrotropin releasing hormone,neurotensin, and oxytocin/vasopressin (Swanson et al., 1987). The functionalrole for most of these peptides in the hippocampus is unknown. However, someof these peptides such as opioid peptides and somatostatin have been found toplay a significant role as a neurotransmitter or neuromodulator in thehippocampus.4.4.1. Opiold peptidesSeveral oploid peptides, including enkephalins, 3-endorphin anddynorphins, have been identified in the hippocampus (Swanson et al., 1987).Binding studies suggest the existence of several subtypes of opioid receptors inthe brain. Three major opioid receptor subtypes are designated as j.t (Mu), ö(Delta) and K (Kappa) (Swanson et al., 1987). The antagonist naloxone hashigh, but variable, affinity to all of these receptors. Opioid peptides do not causesignificant change in the membrane properties of hippocampal pyramidalXIE 73neurons (Dingledine, 1985). On the other hand, opioid peptides induce ahyperpolarization of hippocampal interneurons, which is due to an increase in aK conductance (Nicoll, 1986). This hyperpolarization results in an inhibition ofinterneuronal firing (Madison and Nicoll, 1988; Wimpey et al., 1990). This effectof opioid peptides is mediated by j.t receptors, which couple to the K channelsthrough a pertussis toxin-sensitive G protein (Nicoll et al., 1990; Thompson etal., 1993). The inhibition of interneurons by opioids reduces GABA release fromthe GABAergic interneurons, which provide major inhibitory input to thepyramidal cells (Nicoll et al., 1990; Madison and Nicoll, 1988; Nicoll, 1986).Recent evidence has shown that opioids produce a decrease in miniature IPSPfrequency and monosynaptic IPSP amplitude without changing the meanminiature IPSP amplitude in the pyramidal cells (Thompson et al., 1993). Thisaction of opioid is prevented by the treatment of cultured neurons with pertussistoxin. Since opioid peptides reduce the GABAergic IPSPs by inhibiting theinterneurons, these peptides indirectly increase the EPSP of the pyramidal cells(Nicoll et al., 1990). Opioid peptides increase the amplitude and duration, butnot the slope of the EPSP. In CA3 field, dynorphin has been found in theglutamatergic mossy fibres (McGinty et al., 1983). Therefore, dynorphin may actas a co-transmitter in the mossy fibres. Actions of dynorphin are mediated by icreceptors. Dynorphin action at ic receptors inhibits a voltage-dependent,dihydropyridine-insensitive N-type current in dorsal root ganglion neurons(MacDonald and Wertz, 1986). Recently, Weisskopf et al. (1993) havedemonstrated that dynorphin is co-released with glutamate from the mossyfibres. The release of dynorphin requires high frequency stimulation of mossyfibres. The synaptic release of dynorphin induced by tetanus causesheterosynaptic depression of the EPSP which can be reversed by application ofnaloxone. The released dynorphin may act on presynaptic ic receptors on theXIE 74mossy fibre terminals. The activation of these presynaptic ic receptors bydynorphin reduces glutamate release from the mossy fibres (Gannon andTerrian, 1991; Weisskopf et al., 1993). Wagner et aL, (1992) have reported thatic-receptors located on perforant path terminals mediate the inhibition ofexcitatory synaptic transmission by ic-opioid agonists.4.4.2. SomatostatinSomatostatin was first found as a peptide that inhibits growth hormonesecretion from the anterior pituitary (Brazeau et al., 1973). Recent evidence hasshown that somatostatin plays a significant role as a neurotransmitter orneuromodulator in the brain (Epelbaum, 1986; Olpe et al., 1980). In thehippocampus, somatostatin-immunoreactivity is present in the CAl and CA3interneurons. Many of these interneurons are GABAergic interneurons whichmake contacts with pyramidal cells (Somogyi et al., 1984; Kunkel andSchwartzkroin, 1988). Somatostatin has diverse physiological actions in theCNS. Somatostatin induces a hyperpolarization associated with a reduction inthe input resistance of the hippocampal pyramidal neurons (Pittman and Siggin,1981). This hyperpolarization is due to an increase in a K conductance by theactivation of somatostatin receptors, which is probably coupled to pertussistoxin-sensitive G-protein. An increase in hippocampal ‘M by somatostatin hasalso been reported (Moore et al., 1988). It has been suggested that differentactions of somatostatin may be mediated by different receptor subtypes. Moredetails will be discussed in the chapter on experimental tools.4.4.3. CholecystokininCholecystokinin (CCK) was originally isolated from the gut.Cholecystokinin octapeptide (CCK8) was later identified throughout the CNS(Dockray, 1976; Rehfeld, 1978). Immunohistochemical studies have shown thepresence ofCCK8-immunoreactivity in hippocampal neurons (Greenwood et al.,XIE 751981). Some of the CCK8-immunoreactivity appear in the GABAergicinterneurons which synapse on the pyramidal cells of the hippocampus(Somogyi et al., 1984; Kosaka et al., 1985). CCK8 has been reported toincrease the excitability of hippocampal pyramidal neurons associated with adepolarization and a decrease in the input resistance of the neurons (Dodd andKelly, 1981). Buckett and Saint have reported that CCK8 blocks a voltage-dependent K conductance. This blockade of the voltage dependent Kconductance may explain the increase of the hippocampal neuron excitability byCCK8. CCK8 has been found to increase synaptic transmission while CCK8antagonists have depressant effect on synaptic transmission (Jaffe et al., 1987).The authors have suggested that endogenous CCK8 is released directly orindirectly by stimulation of Schaffer collateral-commissural fibres, therebyincreasing the excitability of CAl pyramidal cells. An inhibition of synaptictransmission in the hippocampus has also been reported. The discrepancy inthese results is not clear. Nonetheless, further studies are needed to determinethe physiological role of this endogenous peptide in the hippocampus.4.4.4. Neuropeptide YNeuropeptide Y has been detected in the hippocampus (Swanson et al.,1987). Neuropeptide Y has no significant effect on the intrinsic properties ofpyramidal cells (Colmers et al., 1987). However, neuropeptide Y can reduce theEPSP in the CAl pyramidal cells through a presynaptic mechanism, It has beenreported that neuropeptide Y inhibits a Ca2 conductance by activatingneuropeptide Y receptors at the presynaptic terminals that release glutamate(Colmers et al., 1988; Klapstein and Colmers, 1992). The neuropeptide Yreceptors, which mediate the presynaptic inhibition, couple to a pertussis toxininsensitive G protein (Colmers and Pittman, 1989).XIE 764.4.5. Other neuropeptidesOther neuropeptides such as substance P, vasopressin/oxytocin,vasoactive intestinal peptide, angiotensin II and neurotensin have beendetected in the hippocampus (Swanson et al., 1987). The functional role ofthese peptides in the hippocampus remains to be determined.5. LONG-TERM POTENTIATION IN THE HIPPOCAMPUSLong-term potentiation (LTP) was first reported in the hippocampaldentate area of anaesthetized rabbits by Lomo (1966) and later described indetail by Bliss and Lomo (1973), Bliss and Gardner-Medwin (1973). Theydemonstrated that brief trains of high-frequency stimulation of the perforant pathinduced a long-lasting potentiation of excitatory synaptic transmission in thedentate area of the rabbit hippocampus in vivo. Since then, LTP has been foundin all excitatory pathways of the hippocampus in vivo as well as in vitropreparations. LTP has also been observed in other regions of the brain,including the neocortex (Tsumoto, 1992; Kirkwood et al., 1993), but LTP is mostwidely studied and best understood in the hippocampus, particularly in the CAlfield. Since LTP is thought to be involved in the mechanisms underlying at leastsome forms of learning and memory, it has been attracting researchers fromaround the world for the past two decades. Even though much progress hasbeen made in characterizing the properties of LTP, the mechanisms underlyingLTP remain controversial at present. The following discussion will be focusedmore on LTP in the CAl area of the hippocampus since all the data in thepresent studies has been obtained from that area.5.1. Characteristics and properties of LTP5.1.1. Definition and classification of LTPBrief trains of high-frequency stimulation to monosynaptic excitatorypathways in the hippocampus can induce an abrupt and persistent enhancementXIE 77of synaptic transmission in vivo as well as in vitro. This activity-dependentsynaptic plasticity was first described in detail by Bliss and his colleagues in1973. They observed the long-lasting potentiation of the population responses inthe dentate area following repetitive stimulation (tetanus: 10-20 Hz for 10-15 secor 100 Hz for 3-4 sec) of the perforant path fibres in rabbit hippocampus in vivo.The amplitude of the population spike and the field EPSP was increased toapproximately 200-300 percent and 150-200 percent of the pre-tetanicresponses, respectively. The latency of the population spike was reduced aftertetanus. The potentiation of these field potentials could last for several hours inanaesthetized rabbits (Bliss and Lomo, 1973) and several weeks in theunanaesthetized rabbits (Bliss and Gardner-Medwin, 1973). The time course oftetanus-induce LTP seems to consist of two phases, an initial phase whichdeclines within 20 mm after tetanus and the second phase which occurs 20 to 30mm after tetanus and maintains stable. The first and second phases are alsocalled decremental LTP and non-decremental LTP, respectively. Bliss and hiscolleagues also reported a phenomenon that the population spike did not alwayspotentiate with corresponding changes in the field EPSP after tetanus. Theyexplained that their observation was probably due to an increase in excitability ofthe granule cells by tetanus. Andersen et al. (1980) analyzed this phenomenonin the hippocampal slices and found that repetitive stimulation resulted in anincrease in the amplitude of the population spike evoked by a given size of thefield EPSP in addition to an increase in the size of the field EPSP. They calledthis form of LTP as EPSP-spike (E-S) potentiation. The E-S potentiationindicates an enhancement of cell body excitability or a reduction of spikethreshold (Abraham et al., 1985, 1987; Kairiss et al., 1987; Tomasulo et al.,1991). The mechanisms underlying the E-P potentiation are unknown andbelieved to be distinct from the mechanisms of the synaptic LTP which theXIE 78potentiation of population spike is directly attributable to the EPSP potentiation,but not the change in neuronal excitability. The synaptic LTP has attractedmuch greater interest since it is clearly identified with potentiation at a particulargroup of tetanized synapses. The present discussion is also concentrated onthe synaptic LTP, unless otherwise stated.In addition to the increase in the amplitude of the field EPSP during LTPdescribed by Bliss and his colleagues (1973), the slope of the field EPSP is alsoincreased following tetanic stimulation (Alger and Teyler, 1976). In fact, theslope of the EPSP is a more efficient index for measuring the synapticpotentiation.In a single hippocampal neuron, the slope and amplitude of the EPSPand/or the probability of spike discharge are increased following tetanicstimulation of the afferents (Schwartzkroin and Wester, 1975; Andersen et al.,1977; Andersen et al., 198Cc). LTP of the EPSP is not associated with changesin the membrane potential, input resistance and excitability of the neuron(Andersen et al., 1977,198Cc; Barrionuevo et al., 1986). Barrionuevo et al.(1986) using the voltage clamp technique showed that the LTP of EPSP isassociated with an increase in synaptic conductance. In some case, theprobability of spike discharge is increased without the potentiation of the EPSP(Andersen et al., 1977). This is similar to the E-S potentiation in extracellularrecording.5.1.2. Distribution of LTP in the hippocampusLTP occurs in all excitatory pathways in the hippocampus, which includethe medial perforant path (Bliss and Lomo, 1973), the lateral perforant path(McNaughton et al., 1978), the mossy fibres (Alger and Teyler, 1976), theSchaffer collaterals (Schwartzkroin and Wester, 1975) and the commissuralprojection to the CAl neurons (Buzsaki, 1980). A projection of perforant path toXIE 79the distal apical dendrites of CAl also displays LTP (Doller and Weight, 1985).Excitatory afferents form synapses not only with pyramidal and granule cells butalso with interneurons. Evidence has shown that tetanic stimulation of excitatoryafferents can also induce LTP in the inhibitory interneurons in the CAl field(Buzsaki and Eidelberg, 1982) and the dentate gyrus (Kairiss et al., 1987).5.1.3. Homosynaptic and heterosynaptic LTPIn the CAl field and the dentate gyrus, LTP occurs only in the tetanizedpathway, but not in the non-tetanized pathways which converge on the samepopulation of cells (Bliss et al., 1973; Andersen et al., 1977, Lynch et al., 1977;Sastry et al., 1986). This input-specificity of LTP is called homosynaptic LTP(McNaughton, 1983). A heterosynaptic depression has been reported in thenon-tetanized pathways (Lynch et al., 1977; Alger et al., 1978). Unlike the CAlfield and the dentate gyrus where only homosynaptic LTP occurs, in the mossyfibre pathway of the CA3 field, LTP can occur in both tetanized and nontetanized pathway (Yamamoto and Chujo, 1978; Misgeld et al., 1979). This isthe so called heterosynaptic LTP. This heterosynaptic LTP is due to thepolysynaptic component of the evoked responses in the CA3 field (Higashimaand Yamamoto, 1985). Recently, Zalutsky and Nicoll (1992) have shown thatLTP in the mossy fibre pathway also displays input specificity if the non-mossyfibre inputs are blocked.5.1.4. Cooperativity of LTPCooperativity indicates the existence of an intensity threshold for theinduction of LTP. The threshold for the induction reflects the need for a certainnumber of presynaptic fibres to be activated simultaneously (McNaughton et al.,1978). Weak tetanic stimulation induces a post-tetanic potentiation (PTP) thatlasts for 2-5 mm or short-term potentiation (STP) that lasts for 15-20 mm, but notLTP (McNaughton et al., 1978; Malenka, 1991). Both high frequency and highXIE 80intensity of stimulation are required to induce LTP. However, frequency andstimulus strength can interact such that increasing one relatively decreases therequirement for the other. For example, a train with higher frequency andrelative lower stimulus strength is as efficient as a train with higher stimulus andrelative lower frequency to induce LTP. In the CA3 region, LTP in the mossyfibre pathway shows no apparent cooperativity (Zalutsky and Nicoll, 1992).5.1.5. Associativity of LTPMcNaughton and his colleagues (1978) first examined associativity ofLTP using two stimulating electrodes to activate two completely separatepathways converging on the same population of postsynaptic cells inhippocampus. One pathway was designated as the weak pathway which, whentetanized alone, could not produces LTP. The other pathway was designated asthe strong pathway that produced LTP when tetanized. When both pathwayswere tetanized at the same time or almost at the same time, LTP occurred notonly in the strong pathway but also in the weak pathway. This is so-calledassociative LTP. Sastry et al. (1986) showed that pairing single stimulation toafferents repetitively with depolarization of postsynaptic cells eventually inducesLTP, further confiming associativity of LTP. Similar results have been reportedby other investigators (Kelso et al., 1986; Wigstrom et al., 1986). Theassociative LTP is maximal when the weak pathway is activated at the sametime as the conditioning trains to the strong input. However, LTP does not occurif the trains to the weak pathway are delayed longer than 100-200 msec (Kelsoand Brown, 1986). This associative LTP can occur even when the two pathwaysare spatially separated by several hundred microns such as the two afferentsterminating on the dendritic tree originating from opposite sides of the cell body(Gustafsson and Wigstrom, 1986). The interaction between the two pathwaysmay be through the postsynaptic cell. The LTP induced by pairing the afferentXIE 81activation with postsynaptic depolarization is facilitated if picrotoxinin, a GABAAantagonist, is present in the recording medium (Sastry et al., 1986). It ispossible that the blockade of GABA inhibition allows the cell to be depolarizedmore readily. The induction of this LTP is not due to the high-frequency sodiumspikes induced by depolarizing pulses because the LTP is not affected evenwhen the sodium spikes are blocked by QX-222 or QX-314 (Kelso et al., 1986;Gustafsson et al., 1987). Malinow and Miller (1986) found that pairinghyperpolarization with the activation of afferents prevents the induction of LTP.Based on these results, associative LTP may be due to pairing the activation ofthe weak pathway with the postsynaptic cell depolarization which is induced by atetanic stimulation to the strong pathway. Tetanizing to the weak pathway alonecannot depolarize the cell sufficiently to induce LTP. On the other hand, theinteractions between the two pathways may also occur in presynaptic terminals(Goh and Sastry, 1985a; Sastry et al., 1986). A pairing of the conditioning trainsof one pathway with the test stimulation of another pathway causes apostconditioning decrease in the excitability of the presynaptic terminals in thetest pathway (Sastry et al., 1986). The associative effects are larger if twoseparate pathways terminate on the same dendritic tree. In contrast, theassociative effects are smaller if two pathways terminate on the dendritic treefrom opposite sides of the soma (Gustafsson and Wigstrom, 1986). Twopathways may interact more readily if they terminate on the same dendritic tree.5.2. The induction of LTPIt is generally accepted that LTP consists of at least two parts: inductionand maintenance. The induction is the initial sequence of events that triggersthe process. The maintenance is the process underlying the persistence of thepotentiated response. At present, the mechanisms underlying the induction ofLTP seem to be better understood than those underlying the maintenance. TheXIE 82associative property of LTP indicates that the both presynaptic and postsynapticactivity are required for the induction of LTP.5.2.1. Involvement of glutamate receptors in LTPThe involvement of glutamate receptors in the induction of LTP has beendetermined basically by the use of antagonists of glutamate receptor subtypes.Collingridge et al. (1983) first suggested that activation of NMDA receptors playsa critical role in the induction of LTP because they showed that the induction ofLTP in the CAl field is blocked by APV, a selective NMDA antagonist. Sincethen, in addition to APV, several other NMDA antagonists, such as MK-801 (aNMDA channel blocker) and 7-chlorokynurenic acid (an antagonist at theallosteric glycine site), have also been reported to block the induction of LTP(Harris et al., 1984; Wigstrom and Gustafsson, 1984; Coan et al., 1987; Bashiret al., 1990). In the dentate gyrus, APV also blocks the induction of LTP in theperlorant path (Morris et al., 1986). In the CA3 region, two forms of LTP occur.While the LTP in the associational/commissural pathway is sensitive to APV, theLTP in the mossy fibre pathway is not affected by APV (Harris and Cotman,1986). The excitatory synaptic transmission in the mossy fibre pathway isprimarily mediated by non-NMDA receptors (Monaghan and Cotman, 1985;Neuman et al., 1988). Other neurotransmitters may also be involved in theexcitatory synaptic transmission in this pathway. APV has little effect on thebasal excitatory synaptic responses. The properties of NMDA receptors makethe involvement of NMDA receptors a suitable explanation for the properties ofLTP. Strong tetantic stimulation or depolarizing current pulse is required for thedepolarization of the neuron which removes the Mg2 blockade. At the sametime, activation of afferents releases glutamate which together with thedepolarization of the postsynaptic cell can open up the NMDA channels. Theactivation of the weak pathway alone is not sufficient to remove the voltage-XIE 83dependent blockade of Mg2 in the NMDA channels; the depolarization of thepostsynaptic cell alone is also not able to activate the NMDA receptors.Therefore, only if both these postsynaptic and presynaptic events occursimultaneously, does the activation of NMDA receptors take place. It has beensuggested that the activation of NMDA receptors allows Ca24 influx whichprobably triggers a series of biochemical events and subsequently induces LTP.Brief exposure of the slices to Mg2 free medium during low frequencystimulation of afferents can induce LTP (Coan and Collingridge, 1985; Wigstromand Gustafsson, 1988). This potentiation induced by Mg2-free medium isblocked by APV. The evidence mentioned above appears to support the role ofNMDA receptors in the induction of LTP. However, application of NMDA aloneinduces STP but not LTP (Collingridge et al., 1983; Kauer et al., 1988a). Recentevidence has shown that the activation of NMDA receptors by the application ofNMDA (Huang et al., 1992b; lzumi et al, 1992) or weak tetanus (Huang et al.,I 992b) before tetanic stimulation inhibits the induction of LTP in the CAl region.It is obvious that the activation of NMDA receptors plays an important role in theinduction of LTP. The different timing of this activation can lead to promotion orsuppression of the induction of LTP. Furthermore, factors in addition toactivation of NMDA receptors are needed for the induction of LTP. Thesefactors can be presynaptic, postsynaptic or both. Recently, metabotropicglutamate receptors have been suggested to be involved in the induction ofLTP. ACPD, a metabotropic glutamate receptor (ACPD receptor) agonist,induces a slowly developing LTP (Otani and Ben-An, 1991; Bortolotto andCollingridge, 1992, 1993). The ACPD-induced LTP is not blocked by APV(Bortolotto and Collingridge, 1993). ACPD fails to potentiate the EPSP if atetanus-induced LTP established before drug application, and vice versa(Bortolotto and Collingridge, 1993). This result suggests that these two types ofXIE 84LTP may share common mechanisms. AP3, an ACPD receptor antagonist, hasbeen reported to block the induction of LTP in the hippocampus (Izumi et al.,1991; Behnisch et al., 1991). However, the effects of AP3 on LTP seem to becontroversial because AP3 has also been reported to have no effect on theinduction of LTP (Stanton et al., 1991; Musgrave et al., 1993). Moreover, AP3 isnot a very selective and potent antagonist (Schoepp and Conn, 1993).Recently, Bashir et al. (1993) have shown that a more selective ACPD receptorantagonist, (RS)-cx-methyl-4-carboxyphenyglycine (MCPG), blocks both tetanus-and ACPD-induced LTP. This result further supports the involvement of ACPDreceptors in the induction of LTP. Musgrave et al. (1993) have demonstratedthat co-application of ACPD and NMDA induces a LTP, which develops fasterthan the ACPD-induced LTP. The induction of LTP by ACPD alone or withNMDA is not required for the activation of afferents (Bortolotto and Collingridge,1992; Musgrave et al., 1993). Whether these LTP are the same as tetanus-induced LTP remains to be determined. It appears that ACPD receptors play amajor role in the induction of LTP. Synaptic activation of both NMDA and ACPDreceptors is required for the induction of LTP.In the CAl region, an APV-insensitive LTP has also been reported(Grover and Teyler, 1990, 1992). Induction of this LTP, however, requires ahigher frequency tetanic stimulation and is blocked by dihydropyridine, avoltage-dependent Ca2 channel blocker.In the CA3 region, the induction of LTP in the mossy fibre pathway is notblocked by APV, but by MCPG (Bashir et al., 1993). Activation of ACPDreceptors is probably also needed for the induction of LTP in this pathway.Activation of non-NMDA receptors is not required for the induction of LTPin the hippocampus because CNQX, a selective non-NMDA receptor blocker,does not block the induction of LTP (Kauer et al., 1988b).XIE 855.2.2. The role ofCa2 in the induction of L TPSeveral lines of evidence suggest that Ca2 plays an important role in theinduction of LTP (Dunwiddie et al., 1978; Wigstrom et al., 1979; Izumi et al.,1987; Huang et al., 1988; Lynch et al.1 1983; Malenka et al., 1988; Morishita andSastry, 1991). Tetanic stimulation of afferents during a short period of exposureof slices toCa2-free (Dunwiddie et al, 1978; Wigstrom et al., 1979) or low Ca2medium (Dunwiddie and Lynch, 1979) fails to induce LTP. lzumi et al. (1987)has shown that extracellular Ca2 is required not only during tetanus but alsoimmediately after tetanus. An exposure to high Ca2 medium has been reportedto induce LTP (Turner et al., 1982). However, Huang et al. (1988) demonstratedthat perfusion of high Ca2 medium could only cause a short term potentiationwhich reverted following the return to control medium. The explanation for thedifference between these two studies is not known. Nevertheless, these resultssupport that the change in extracellular Ca2 concentration can affect theinduction of LTP. The mechanisms can be presynatic, postsynaptic or both.Direct injection of Ca2 chelators such as EGTA, nitr-5 or BAPTA into thepostsynaptic cell can block the induction of LTP in the drug-injected neuronwithout changing basal synaptic responses of that neuron (Lynch et al., 1983;Malenka et al., 1988; Morishita and Sastry, 1991). These results suggest thattetanic stimulation may induce a large increase in postsynaptic Ca2concentration which is critical for the induction of LTP. The increase inpostsynaptic Ca2 concentration can be due to a Ca2 influx or a release ofintracellular stores. It has been suggested that tetanic stimulation induces alarge postsynaptic Ca2 entry. Using Ca2-imaging techniques, it has beenshown that tetanic stimulation elevates Ca2 within dendrites and spines(Regehr and Tank, 1990; Muller and Connor, 1991). The tetanus-induced rise inpostsynaptic Ca2 lasts for several minutes (Muller and Connor, 1991).XIE 86Whether this long period of increase in intracellular Ca2 concentration bytetanus is required for the induction of LTP remains to be determined. Izumi etal. (1987) have reported that exposure of slice to Ca2-free medium for 5 mmafter tetanus inhibits the induction of LTP. In contrast, Malenka et al. (1992)have shown that the rise of intracellular Ca2 for the induction of LTP is onlyneeded for a few seconds by using photolabile Ca2 buffer diazo-4. Whether anincrease in postsynaptic Ca2 concentration alone can induce LTP is anotherquestion needed to be answered. An release of Ca2 into postsynaptic neuronsby the caged Ca2 compound nitr-5 has been suggested to be sufficient toproduce a potentiation of synaptic transmission (Malenka et al., 1988).However, their conclusion has been questioned. First, this potentiation can onlylast for 20 to 70 mm. Second, whether this potentiation by photolysis of nitr-5 isthe same as the tetanus-induced LTP is not clear. Occlusion experiments havenot been done.The rise in postsynaptic Ca2 concentration can be generated by threemechanisms. (1) The activation of NMDA receptors, which have been found inthe dendrites and spines of CAl neurons, by tetanus allows Ca2 entry throughNMDA channels (McDermott et. al. 1986; Regher and Tank, 1990); (2) Strongdepolarization induce the opening of voltage-dependent Ca2 channels, whichare also present in the dendrites, (Westenbroek et al., 1990); (3) Ca2releases from intracellular stores. In the CAl region, synaptic activation ofNMDA channels are apparent to allow Ca2 entry which is critical for theinduction of LTP. Under certain circumstances, such as strong tetanicstimulations which induce LTP in the presence of APV, voltage-gated Ca2channels mediate the Ca2 entry (Grover and Teyler, 1990; 1992; Kullmann etal., 1992). Moreover, a brief application of tetraethylammoniun (TEA), a Kchannel blocker, induces aCa2-dependent NMDA-independent form of LTPXIE 87(Aniksztejn and Ben-An, 1991). This form of LTP is blocked by Ca2 channelblockers such as flunarizine and nifedipine, or intracellular injection of Ca2chelator, BAPTA (Aniksztejn and Ben-An, 1991; Huang and Malenka, 1993).The voltage-dependent Ca2 channels appear to mediate the induction of thisTEA-induced LTP. Whether this TEA-induced LTP and tetanus-induced LTPhave similar mechanisms are not very clear. In addition to the Ca2 entrythrough NMDA or voltage-dependent channels, the release of Ca2 fromintracellular stores may also contribute to the rise of intracellular Ca2concentration. Thapsigargin, which depletes intracellular Ca2 stores, alsoblocks the induction of LTP (Harvey and Collingnidge, 1992). Furthermore,ACPD-induced LTP is probably also due to the increase in release of Ca2 fromintracellular stores (Bortolotto and Collingnidge, 1993).In the CA3 region, LTP in the associational-commissural pathway, but notthe one in mossy fibre pathway, isCa2-dependent (Zalutsky and Nicoll, 1990,but also see Williams and Johnston, 1989).In summary, evidence has clearly shown that postsynaptic Ca2 play acritical role in the induction of LTP in most of the pathways in the hippocampus.However, there is no convincing evidence to support that a rise in postsynapticCa2 concentration is sufficient to trigger the induction of LTP. Whetherchanges of Ca2 homeostasis in presynaptic terminals occur in the induction ofLTP is unknown.5.2.3. The role of protein kinases in the induction of LTPThe involvement of protein kinase C (PKC) was first reported byRouttenberg et al. (1985). They showed that a 47-kDa protein (referred to asFl, B50 or GAP-43) as a PKC substrate undergoes a LTP-associatedphosphorylation (Akers and Routtenberg, 1985; Lovinger et al., 1985). Resultsfrom early studies suggest that the activation of PKC plays an important role inXIE 88the maintenance of LTP (Akers et al., 1986; Lovinger et al., 1987; Malinow et aL,1988; Malenka et al. 1989; Reymann et al., 1988a,b). Most of these studies arebased on the use of PKC inhibitors such as H-7, polymyxin B, sphingosine.These inhibitors are able to facilitate the decay of LTP. Application of H-7 hasbeen reported reversibly block the established-LTP (Malinow et al., 1988).However, the blockade of the established-LTP by H-7 has been questioned byothers who demonstrated that H-7 can suppress the control responses as well(Muller et al., 1990; but also see Huang et al., 1992).Recent studies have shown that PKC activity plays a significant role in theinduction of LTP. Direct injection of PKC into the postsynaptic neurons inducesa long-lasting potentiation (Hu et al., 1987). More recently, direct injection of asynthetic peptide PKC (19-31) which acts as a potent PKC antagonist, or of acombination of different types of PKC inhibitors into the postsynaptic cells hasbeen found to block the induction of LTP (Malinow et al., 1989; Wang and Feng,1992). Our results have also shown that intracellular injection of K-252b, apotent and relatively selective PKC inhibitor, blocks the induction of LTP (Xieand Sastry, 1991). Biochemical studies showed that an increase in cytosolicactivity of PKC occurs in the first 2 mm after the induction of LTP and disappearsin 5 mm (Otani et al., 1992). These results support the role of PKC in theinduction of LTP.Extracellular application of phorbol ester, a PKC activator, can induce along-lasting potentiation (Malenka et al., 1986). However, this phorbol esterinduced long lasting potentiation is different from the tetanus-induced LTPbecause the latter does not occlude the former (Gustafsson et al. 1988; Kamiyaet al., 1988; Muller et al., 1990). The phorbol ester-induced long-lastingpotentiation is not sustained after complete washout of the drug. Therefore, theXIE 89experiment of phorbol ester cannot conclude the exact role of PKC in the LTPprocess.The difficulty of studies on PKC involvement in LTP is due to the lack ofselective PKC inhibitors. Controversy on this subject will not be over until aselective inhibitor is available. The available evidence suggests thatpostsynaptic PKC plays an important role in the induction of LTP. Activation ofpostsynaptic PKC probably converts STP to LTP. The role of PKC on theinduction of LTP may be linked to its effect on NMDA receptor-mediatedresponses. A recent report has shown that PKC is able to potentiate the NMDAcurrent by removing the voltage-dependent Mg2 block of the NMDA channels incultured trigeminal neurons (Chen and Huang, 1992). Activation of ACPDreceptors has been shown to potentiate NMDA current through PKC in the CAlpyramidal neurons (Aniksztejn et al., 1992). The role of PKC in the maintenanceof LTP is no longer as certain as predicted previously. Whether PKC ispersistently activated after the induction of LTP is not clear. A recent reportsuggests that activation of presynaptic PKC may last longer but no more than 60mm (Huang et al., 1992). Whether presynaptic PKC really contributes to themaintenance of LTP remains to be determined.Ca2/calmodulin-dependent protein kinase II (CaM-KIl), which is foundabundantly in postsynaptic dendrites (Kelly et al., 1984; Kelly, 1991), has beenshown to contribute to the induction of LTP (Mody et al., 1984; Reymann et al.,1988b; Malenka et al., 1989; Malinow et al., 1989). Injection of selective CaMKll inhibitors, the synthetic peptides, into the postsynaptic cell blocks theinduction of LTP with the STP intact (Malenka et al., 1989; Malinow et al., 1989).Bath application of calmidazolium, a CaM-KIl inhibitor, blocks the induction ofboth STP and LTP (Reymann et al., 1988b). It is possible that presynaptic CaMKIl also plays a role in LTP. Recently, Silva et al. (1992a) have shown that LTPXIE 90is not observed in the hippocampus of cx-CaM-Kll mutant mice. This resultfurther supports the role of CaM-Ku in the induction of LTP. CaM-Ku canregulate itself through an autophosphorylation mechanism whereby it becomesCa2-independent following its initial activation (Saitoh and Schwartz, 1985;Miller and Kennedy, 1985; Kelly, 1991). Because of this property of CaM-KIlactivation of CaM-KIl may contribute to the later phase of LTP.5.2.4. The role of GABA receptors in the induction of LTPUnder normal conditions, the activation of NMDA receptors is critical forthe induction of LTP (Collingridge et al., 1983). Depolarization of postsynapticcells by tetanic stimulation or injection of depolarizing current pulses removesthe voltage-dependent Mg2 block on NMDA channels. This is crucial for theNMDA channel opening as well as the induction of LTP. In contrast,hyperpolarization of postsynaptic neurons blocks the NMDA channel openingand the induction of LTP (Mayer et al., 1984; Malinow and Miller, 1986).Stimulation of afferents activates not only glutamatergic principle cells, but alsoGABAergic interneurons which in turn make synapses on the principle cells inthe hippocampus. Activation of GABA receptor-mediated synaptic inhibitionwhich hyperpolarizes the postsynaptic neurons has a significant effect on theactivation of NMDA receptors and the induction of LTP. In the presence ofGABAA antagonists such as picrotoxinin, the induction of LTP can be facilitated(Wigstrom and Gustafsson, 1983). Disinhibition by picrotoxinin enhances theNMDA responses by removing the shunting effect produced by the GABAAreceptor-mediated hyperpolarization. In contrast, application of GABA preventsthe induction of LTP (Scharfman and Sarvey, 1985). This effect of GABA isprobably due to the GABA receptor-mediated hyperpolarization which makes theNMDA channels more difficult to open.XIE 91Recent evidence has shown that GABAB receptors play an important rolein the induction of LTP in both CAl (Davies et al., 1991) and dentate gyrus ofthe hippocampus (Mott and Lewis, 1991). Baclofen, a GABAB agonist, has beenreported to facilitate the development of LTP of the population spike (Mott et al.,1990; Olpe and Karlsson et al., 1990). GABAB receptor antagonists, 2-hydroxy-saclofen and CGP35348, have been found to block the induction of LTP (Davieset al., 1991; Mott and Lewis, et al., 1991). Presynaptic GABAB receptors in theinhibitory terminals are believed to act as autoreceptors that appear to mediatethe paired-pulse depression of the fast IPSP (Davies et al., 1990). This issupported by the ability of GABAB antagonists such as 2-hydroxy-saclofen toreduce the paired-pulse depression of IPSPs (Davies et al., 1990). Duringtetanic stimulation, GABA release can be down-regulated by the activation ofpresynaptic GABAB receptors. Subsequently, the postsynaptic GABA receptor-mediated hyperpolarization is reduced, the NMDA currents are enhanced, andthe induction of LTP is facilitated. Application of phaclofen, a weak GABABantagonist, or a low concentration (100 tiM) of CGP-35348 only suppresses thepostsynaptic GABAB receptor-mediated responses with no significant effect onthe presynaptic GABAB receptor-mediated responses such as paired-pulsedepression. However, phaclofen and the low concentration of CGP-35348 canfacilitate the induction of LTP. This result suggests that postsynaptic GABABreceptors, like postsynaptic GABAA receptors, can modulate the induction ofLTP as well.5.3. The maintenance of LTPThe non-decremental phase of LTP is usually referred to as themaintenance of LTP which is stable and lasts for hours in vitro and days in vivo.The mechanisms underlying the maintenance of LTP are much less clear thanthose underlying the induction of LTP. Debates have been on the loci and theXIE 92nature of the maintenance of LTP. Whether the maintenance of LTP is due topresynaptic or postsynaptic changes is controversial. The review will be focusedon the current progress on this subject.5.3.1. Postsynaptic versus presynaptic mechanismsPostsynaptic synaptic mechanisms include changes in postsynapticreceptor number or sensitivity, or changes in morphology of synaptic spines. Anincrease in postsynaptic glutamate receptor numbers has been suggested(Lynch and Baudry, 1984). Tetanic stimulation activates calpain, the Ca2-dependent proteolytic enzyme, which degrades cytoskeleton proteins andunmasks the hidden postsynaptic receptors (Baudry et al., 1980, 1981; Lynch etal. 1982). However, whether LTP is associated with the activation of calpain isnot very clear (Sastry, 1985; Oliver et al., 1990). In contrast, several studieshave shown that LTP is not associated with an increase in transmitter binding(Sastry and Goh, 1984; Lynch et al., 1985; Kessler et al., 1991). Furthermore,an immunohistochemical study has shown the absence of calpain I and II in theregion of axo-dendritic synapses in rat brain (Hamakudo et al., 1986). Increasesin receptor sensitivity during the maintenance of LTP has been suggested.AMPA receptor sensitivity has been found to increase 20-30 mm after tetanicstimulation. (Davies et al., 1989). PKC inhibitor, K-252b, prevents the increasein AMPA receptor sensitivity (Reymann et al., 1990). It is possible that proteinkinases directly phosphorylate the receptors or ion channels. Furthermore,Kauer et al. (1988a,b) showed that the non-NMDA component of the EPSP, butnot the NMDA component of the EPSP, increases during the maintenance ofLTP. If an increase in neurotransmitter release from presynaptic terminalsoccurs during the maintenance of LTP, both NMDA and non-NMDA componentsof EPSPs should increase in a similar way. These results seem to support theidea that postsynaptic modification is responsible for the maintenance of LTP.XIE 93However, recent studies have shown that tetanic stimulation induces a persistentincrease in both non-NMDA component and NMDA component of the EPSP(Bashir et al., 1991; Xie et al., 1992; Asztely et al., 1992). These resultschallenge the previous suggestion that postsynaptic mechanisms areresponsible for the maintenance of LTP. Among these studies, Bashir et al.,(1991) have demonstrated that the LTP of the NMDA component of the EPSP issimilar in size to the LTP of the non-NMDA component. Asztely et al. (1992),however, have observed that the increase in the NMDA component is relativelysmaller than that in the non-NMDA component. The explanation for thediscrepancy between the two findings is unknown. It is more likely that bothNMDA and non-NMDA components of the EPSP are potentiated during themaintenance of LTP. Since postsynaptic modifications occur during themaintenance of LTP, the degrees of increase in the NMDA and the non-NMDAcomponents are different.Another hypothesis for postsynaptic mechanisms is that changes insynaptic spine neck resistance after the induction of LTP may be responsible forthe potentiation. An enlargement of synaptic spines after repetitive afferentstimulation has been reported (Chang and Greenough, 1984; Fifkova and vanHareveld, 1977; Desmond and Levy, 1983, 1988). Widening of the spine neckwould decrease the resistance of spine neck and allow more current flow into thedendrites (Brown et al., 1988). So far, there is a lack of experimental evidenceto support changes in spine shape having functional significance in LTP. Incontrast, Jung et al. (1992) have recently shown that changes in spine neckresistance are not responsible for the maintenance of LTP.Presynaptic changes have been suggested to be responsible for themaintenance of LTP. An increase in glutamate release during LTP in thehippocampus has been reported (Dolphin et al., 1982; Bliss et al., 1986, LynchXIE 94et al., 1989a; Ghijsen et al., 1992). These studies showed that the glutamateconcentration of the extracellular fluid increases after tetanic stimulation andremains elevated as long as LTP lasts. Although these results indicate thatglutamate release is probably associated with LTP, they are not entirelyconclusive that the increase of glutamate in the extracellular fluid is due to therelease of glutamate from the tetanized terminals. It cannot be ruled out thatglutamate is released from sources like glia, and glutamate uptake is decreased.Changes in excitability of presynaptic terminals during LTP have been reported(Sastry, 1982; Goh and Sastry, I 985a; Sastry et al. 1986). It is speculated thatthese changes in the excitability of presynaptic terminals may lead to anincrease in neurotransmitter release. However, more direct evidence is neededto conclude that the increase of glutamate in the extracellular fluid is releasedfrom the presynaptic terminals.The observation of the increase in glutamate release during LTP is notwithout controversy. A few other studies have detected no change or onlytransient increase in glutamate release after the induction of LTP in both thedentate gyrus and the CAl region (Aniksztejn et al., 1989; Roisin et al., 1990).More recently, Klancnik et al.(1992) reported that high-frequency stimulation ofSchaffer collateral-commissural fibres results in a significant increase inconcentrations of cysteine suiphinic acid and homocysteic acid, but not otheramino acids such as glutamate and aspartate. These discrepancies areprobably due to the differences in the preparation and methods such as thefrequency of tetanus used in the studies. If a sustained increase in glutamaterelease during LTP, indeed, occurs, it remains to be determined whether theincrease is responsible for the maintenance of LTP or unrelated to LTPQuantal analysis is developed and well-accepted now in studies onsynaptic transmission at neuromuscular junctions (Del Castillo and Katz,XIE 951954a,b; Katz, 1969). In general, the analysis provides estimates of theamplitude of response to a single quantum (v, quantal amplitude), the probabilityof release (p) and the number of release sites (N) based on the binomialdistribution. The mean number of quanta released by a nerve impulse (m, meanquantal content) is equal to N.p. The mean amplitude of a single-fibre (unitary)postsynaptic potential (E) is thus equal to v.m.. In general, an increase in mindicates a presynaptic mechanism while a change in v suggests a postsynapticmechanism.Several methods of quantal analysis are commonly used. They are thedirect method, the histogram method including deconvolution technique, thevariance method and the method of failure (Redman, 1990; Voronin, 1993). Thedirect method is the technique that measures the quantal amplitude and quantalcontent directly. It is assumed that the amplitude of spontaneous miniaturepostsynaptic potentials or currents is equal to v. This method requires highsignal resolution recording in order to measure small miniature currentaccurately. The histogram method is to estimate v by measuring the averageinterpeak distances. The histogram must include a sufficient number of trials(N>500). The noise standard deviation (Sn) to v should be low. To minimize theproblems created by background noise, the deconvolution technique based on acomputer optimization algorithm is used to separate the underlying statisticalfeatures of the amplitude fluctuation from the background noise and the effectsof random sampling (Edwards et al., 1976a, b; also see Voronin, 1993 forreview). In the variance method, the mean amplitude of postsynaptic current orpotential (E) and its variance (S2) are measured. Based on a binomial model,the coefficient of variance (V) of the mean synaptic current is independent ofpostsynaptic parameters such as v using the equation 1N2=NpI(1-p), where1N2=E/S (Malinow and Tsien, 1990; Bekkers and Stevens, 1990). A shift inXIE 96IN2 implicates a presynaptic change. The method of failure is to study smallsynaptic responses evoked with minimal stimulation which is only slightlystronger than the highest stimulus that gave all failures. This method works forthe Poisson distribution. The equation is m=ln(N/N0), where “N” and “N0” arenumber of trials and number of failure, respectively. A change in failure rate isinterpreted as a presynaptic locus. Since all these methods discussed aboveare based on certain assumptions originally made for the neuromuscularjunction, application of these methods to synapses in the CNS has met withmany difficulties. Some of the assumptions for the neuromuscular junction, suchas identical release probabilities for all release sites and identification of a singlesynaptic vesicle as the quantum, cannot be readily applied to synapses in thebrain. Neurons in the CNS receive multiple inputs from many axons. It is notcertain whether release probabilities from different sites are identical.Spontaneous miniature potentials, usually called minis which define the quantain neuromuscular junctions, cannot be clearly characterized in the centralneurons because they are too small to be well resolved with the noise level ofthe membrane potential. The variability of miniature potential size is muchgreater for the central neurons because the synaptic activities occur at a varietyof location over the synaptic tree. The sites of origin of minis are always unclear(Redman, 1990; Stevens, 1993). Despite these problems of applying quantalanalysis to central synapses, this technique has been used to analyze centralsynaptic transmission such as LTP (Voronin, 1983) and paired-pulse facilitation(Hess et al., 1987). The results of these studies have been largely compromisedby the high noise level caused by the high resistance of microelectrode and thebackground synaptic activity. With the recent development of whole cellrecording technique, which greatly improves the signal resolution, raises a newinterest in applying quantal analysis to determine the locus of LTP.XIE 97Unfortunately, the findings obtained with this new technique do not resolve theissue on the locus of LTP, but rather make it more complicated because of theinconsistency of the results. Malinow and Tsien (1990) have observed changesin the synaptic variability and a decrease in the proportion of synaptic failureusing the variance and failure methods in the slices. Bekkers and Stevens(1990) have shown an increase in release probability using the histogram,variance and failure methods in cultured neurons as well as in the slices. Theseauthors have suggested that the locus of LTP is almost purely presynaptic eventhough a postsynaptic change is not completely ruled out. Malinow (1991) hasrecorded synaptic transmission between individual pre- and postsynapticneurons in slices and his conclusion is in accord with those reports discussedabove. However, these results have not convinced the other investigators whohave come out with different results (Foster and McNaughton, 1991; Kullmannand Nicoll, 1992; Manabe et al., 1992). Foster and McNaughton (1991) havefound a significant increase only in quantal size using the failure, variance andhistogram methods and suggested a purely postsynaptic locus of LTP.Meanwhile, three different groups (Kullmann and Nicoll, 1992; Liao et al., 1992;Larkman et al., 1992) have shown that an increase in both quantal size andquantal content occurs during LTP. The latter three reports support the role ofboth presynaptic and postsynaptic modifications during LTP. There are someinteresting findings reported by Voronin et al. (1992a, b, c) and Kuhnt et al.,(1992). They have found that an increase in quantal content (m) is primarilyresponsible for the large LTP of EPSPs while an augmentation in quantalamplitude (a) is accounted for the small LTP of EPSPs. These authors suggeststwo types of synaptic mechanism for the maintenance of LTP. The change inquantal amplitude is saturated at about 10 to 30% increase in the post-tetanicamplitude above the pre-tetanic EPSP amplitude. The increase in quantalXIE 98content, on the other hand, contributes to the rest, the major part of LTP.Analyzing spontaneous miniature synaptic currents, Manabe et al. (1992) havefound a significant increase in the spontaneous EPSC amplitude but not in thefrequency during LTP or the application of NMDA in the hippocampal slices.The results implicate a postsynaptic modification in LTP. In contrast, Malgaroliand Tsien (1992) have shown that the application of glutamate induces long-term potentiation of the frequency of spontaneous EPSC without observing anychanges in the amplitude of spontaneous EPSC in cultured hippocampalneurons. Their findings imply a presynaptic locus of LTP.In summary, the application of quantal analysis for resolving the locus ofLTP has not succeeded because of conflicting results and interpretations.Whether the technique designed for the study of neuromuscular junction issuitable for synapses in the CNS remains an opening question. Edwards,Larkman and their colleagues (Edwards et al., 1990; Edwards 1991; Larkman etal., 1991) hypothesized that the number of receptors available in thepostsynaptic membrane of central synapses is limited and can be saturated byreleased transmitter from a single vesicle. Meanwhile, there are relativelyunlimited postsynaptic receptors for released transmitters to act on inneuromuscular junctions. Therefore, the quantization of responses in centralsynapses is determined not only by the amount of transmitter released, but alsoby the availability of postsynaptic receptors. If this is indeed the case, theassumptions used in neuromuscular junctions cannot directly apply for synapsesin the CNS. A more complete quantal analysis and a better understanding ofcentral synapse structure and function are required to resolve the issue on thelocus of LTP.XIE 995.3.2. Signal transduction mechanismsThe transient increase in intracellular Ca2 concentration during theinduction of LTP may activate several second messenger systems which causea series of biochemical events. These biochemical events may be responsiblefor the maintenance of LTP. Protein kinasesThe involvement of protein kinase C (PKC) and Ca2/calmodulin-dependent protein kinase (CaMKII) in LTP has been previously discussed insection 5.2.3. An increase in cytosolic activity of PKC occurs and lasts for a fewminutes after tetanus (Otani et al., 1992). It has been suggested thatpresynaptic PKC activation may last for long periods (30 mm to 1 hour) but stillfar shorter than the time course of LTP (Huang et al., 1992a). It is believed thatactivation of PKC after tetanus converts STP to LTP (Ben-An et al., 1992). It ispossible that the activation of PKC after tetanus leads to a permanent change inreceptors or ion channels, which causes LTP. However, it is not certain whetherPKC activity is directly involved in the maintenance of LTP. An increase in Ca2-independent phosphorylation of a 17 kDa PKC substrate protein occurs 45-60mm after the induction of LTP (Klann et al., 1991). Recently, a Ca2-independent PKC-like activity in hippocampal cytosol fractions has been foundto increase during LTP (Suzuki et al., 1992). The enhancement is blocked bycalpain inhibitor and PKC inhibitor. However, how this kinase contributes to LTPis not known. Whether the activator-independent forms of PKC are involved inthe maintenance of LTP remains to be determined.CaMKII can be activated by the transient cytosolic Ca2 increase duringthe induction of LTP (Malenka et al., 1989, Malinow et al., 1989). The kinasecan remain active through an autophosphorylation mechanism. Whether theXIE 100autophosphorylated form of this kinase contributes to the maintenance of LTP isnot known.Protein kinase A (PKA), a cyclic adenosine monophosphate (cAMP)dependent protein kinase, has recently been reported to play a role in the latestage of LTP (Frey et al., 1993; Matthies and Reymann, 1993). Application ofPKA inhibitors such as H-8 or KT5720, or cAMP antagonist such as Rp-cAMPSsuppresses the late stage (3 hours after the induction) of LTP. These agentsonly have a small effect on the early stage (30-60 mm) of LTP. Application ofSp-cAMPS, a membrane-permeable analog of cAMP which activates PKA,induces an initial depression followed by a slowly developing LTP which issimilar to the late stage of the tetanus-induced LTP (Frey et al. 1993). Thetetanus-induced late stage LTP and the Sp-cAMP-induced LTP occlude onanother. This result suggests that the two types of LTP may share a commonmechanism. The Sp-cAMPS-mnduced LTP is blocked by anisomycmn, a proteinsynthesis inhibitor, which also block the late stage of the tetanus-induced LTP.This finding implies that the activation of PKA may be related to proteinsynthesis. According to the report (Frey et al., 1993), cAMP level increases Imm, but not 10 mm, after tetanic stimulation (100 Hz for I 5, 3 trains) whichinduces a longer lasting (more than 3 hours) LTP. A single tetanus (100 Hz, forIs), which only induces early stage (1-3 hours) LTP, does not raise the cAMPlevel. A similar increase in cAMP level after tetanus in both the CAl (Chetkovichet al., 1991) and the dentate gyrus (Stanton and Sarvey, 1985) has also beenobserved by others. The increase in cAMP is NMDA receptor-(Frey et al., 1993;Chetkovich et at., 1991) and dopamine (Dl) receptor-dependent (Frey et al.,1993). The results demonstrate a possible role of cAMP and PKA in themaintenance of LTP. However, a number of questions remain to be answered.Whether a sustained increase in the activity of PKA is needed for LTP remainsXIE 101to be determined. Whether the PKA inhibitors also block other protein kinases isnot clear. Some inhibitors seem to suppress the basal synaptic transmission(Frey et al., 1993).Protein tyrosine kinases (PTK5) have also been reported to be involved inLTP based on an inhibitor study (O’Dell et al., 1991). The PTK inhibitors,lavendustin A and genistein, block the induction of LTP with no effect on thenormal synaptic transmission and established LTP. Recently, Grant et al.(1992)have reported the deficiency of LTP in mice with mutation in a tyrosine kinasegene, fyn. Protein synthesisInvolvement of protein synthesis in LTP has been reported (Stanton andSarvey, 1984; Krug et al., 1984; Otani et al., 1989; Frey et al., 1988, 1989;Frazeli et al., 1993). Most of these studies are based on the use of proteinkinase inhibitors. Only the late stage of LTP is suppressed by anisomycin, aprotein synthesis inhibitor (Frey et al., 1988: Otani et al., 1989). However,another protein synthesis inhibitor, cycloheximide, if applied for 30 mm or more,blocks not only the maintenance but also the induction of LTP (Stanton andSarvey, 1984; Deadwyler et al., 1987). Cycloheximide is probably more potentthan anisomycin. Anisomycin is ineffective if applied 15 mm after tetanicstimulation (Otani et al., 1989). Therefore, protein synthesis may occur veryrapidly from already existing messenger RNAs (mRNAs), probably inpostsynaptic sites such as cell bodies, shortly after tetanus (Otani and Abraham,1989; Frey et al., 1989). The newly synthesized proteins may exert their effectslater on. However, a recent report using gel electrophoresis has shown thatprotein synthesis after tetanic stimulation continues for more than 1 hour (Fazeliet al., 1993). It is apparent that the time course of protein synthesis requiresfurther study. A similar fashion decay of LTP has been observed when LTP isXIE 102induced in synapses that are surgically isolated from the major site of proteinsynthesis in the cell body layer (Frey et al., 1989). Taken together, proteinsynthesis is probably required for the maintenance and/or the induction of LTP.Classes of newly synthesized protein during LTP have not been identified yet. Immediate early genesImmediate early genes (lEGs) are a class of genes that show rapid andtransient but protein synthesis-independent increase in expression toextracellular signals such as growth factors, neurotransmitters or depolarization.Many lEGs code for transcription factors and may regulate the expression ofother target genes. Several lEGs, including a c-fos-related gene (Dragunow etal., 1989, Jeffery et al., 1990), Jun-B and zif/268 (Cole et al., 1989; Wisden etal., 1990), have been shown to transiently increase transcription in the dentategyrus following the induction of LTP by perforant path stimulation. In thesestudies, multiple trains of tetanic stimulation are used to induce long lasting(several hours or days) LTP. The increase in c-fos expression occurs only ifanimals are not under central anaesthesia during tetanic stimulation.Pentobarbital, which blocks the induction of long lasting LTP (Krug et al., 1984;Nikolaev et al., 1991), also blocks c-fos expression (Nikolaev et al., 1991).Dragunow et al. (1989) suggest that c-fos is likely to be related to the late stageof LTP. However, the correlation between the increase in c-fos and LTP hasbeen questioned because c-fos expression does not always occur during LTPand vice versa. (Nikolaev et al., 1991; Kaczmarek, 1992).An increase in zif/268 expression after LTP-inducing tetanic stimulation inanaesthetized and awake animals has been reported (Cole et al., 1989;Richardson et al., 1992). The increase in zif/268 probably occurs atpostsynaptic sites. This increase is blocked by APV, a NMDA blocker (Wisdenet al., 1990). The increase in zif/268 expression correlates better with theXIE 103maintenance than with the induction of LTP (Abraham et al., 1991; Richardsonet al., 1992). Although the correlation between the expression of zif1268 andLTP is better than the one between the c-fos and LTP, direct evidence for theincrease in zif1268 expression responsible for the maintenance of LTP is notavailable at present. A dissociation between LTP induction and zif1268expression has also been reported (Schreiber et al., 1991). The increase inzif1268 occurs shortly after multiple train tetanus, reaches the peak inapproximately 20 mm and then returns to baseline within 2 hours (Richardson etal., 1992). It has been proposed that the lEG may not be directly involved in thelate stage of LTP, but serves as a third messenger in the cascade of cellular andnuclear events that are responsible for the maintenance of LTP. In general,most of the current available data indicate that an increase in lEGs expression isassociated with the late stage (several hours and days) of LTP. Whether thisincrease is responsible for LTP requires more experimental evidence. Release of proteinsAn increase in the release of proteins, including newly synthesized ones,into extracellular space in vitro (Duffy et al., 1981) and in vivo (CharriautMarlangue et al., 1988; Fazeli et al., 1988) has been observed after theinduction of LTP. The presence of proteases in push-pull perfusates followingthe induction of LTP has been reported (Fazeli et aL, 1990). Otani et al. (1992)have recently shown that the increase in protein release lasts for hours and isblocked by cycloheximide, a protein synthesis inhibitor. It is possible that thereleased proteins are involved in the maintenance of LTP. Fluid collected fromguinea pig hippocampal (Chirwa and Sastry, 1986) or rabbit neocortical surface(Sastry et al., 1988a) during tetanic stimulation induces LTP and enhancesneurite growth in PC-12 cells. Some components released into the fluid may beproteins (Sastry et al. 1988a,b; Xie et al., 1991), and this issue will be discussedXIE 104later. At present, there is clear evidence that the increase in protein release intothe extracellular fluid is associated with LTP. Whether these proteins aredirectly involved in LTP is unknown.5.3.3. Possible retrograde messengersBased on the current available data, the induction of LTP is mainlypostsynaptic while the maintenance of LTP is, at least in part, presynaptic. Ifthis is true, some forms of communication between the postsynaptic neuron andthe presynaptic terminal must occur. Therefore, the hypothesis of retrogrademessengers has been proposed (Sastry et al., 1986; Bliss et al., 1986; Bliss etal. 1988). Malinow and Tsien (1990) have found that LTP occurs only if pairingpresynaptic stimuli with postsynaptic depolarization is given within 30 mm aftergaining whole-cell access. The finding suggests that some kinds of diffusiblecytoplasmic substances which are critical for the induction of LTP, are lostduring the recording and these substances may be retrograde messengers.Ideally, retrograde messengers should be released from the postsynaptic cellsduring tetanic stimulation. The released substances should be highly diffusibleand be able to act rapidly on the presynaptic terminals to induce transmitterrelease. There are not many substances which meet these requirements. Inrecent years, several substances, including arachidonic acid, nitri oxide, andcarbon monoxide, have been proposed as retrograde messengers. In addition,K, amino acids and proteins released from postsynaptic cells during tetanicstimulation are also candidates for retrograde messengers. Arachidonic acidArachidonic acid, an unsaturated fatty acid, has been found to induce aslowly developing LTP (Williams et al., 1989). This LTP is activity-dependent(Williams et al., 1989; also see Drapeau et al., 1990). The arachidonic acidinduced LTP is not blocked by APV (Williams et al., 1989). However, recentXIE 105studies have shown that the arachidonic acid-induced LTP is blocked by APV(O’Dell et al., 1991) and arachidonic acid can potentiate NMDA currents (Milleret al. 1992). Nordihydrogualaretic acid (NDGA), an inhibitor of arachidonic acidproduction and metabolism, has been reported to block the induction (Lynch etal., 1989; Williams and Bliss, 1989) and the maintenance of LTP (1989). Anincrease in glutamate release from hippocampal synaptosomes by arachidonicacid (Lynch and Voss, 1990) and an increase in arachidonic acid concentrationin the postsynaptic membrane after LTP (Clements et al., 1991) have been seen.These findings further support the involvement of arachidonic acid in LTP.However, whether arachidonic acid acts as a retrograde messenger is doubtful.The effect of arachidonic acid on EPSP5 is slow to develop. In contrast, thetetanus-induced LTP develops very rapidly. Furthermore, arachidonic acid candirectly potentiate NMDA currents (Miller et al., 1992) and inhibit glutamateuptake in glial cells (Barbour et al., 1989; Martin et al., 1991a) suggesting thatthe actions of arachidonic acid on postsynaptic and glial cells may also beresponsible for the arachidonic acid-induced LTP. Nitric oxide and carbon monoxideNitric oxide (NO), which is derived from arginine in a reaction catalyzedby NO synthase has recently been reported to be involved in LTP (Bohme et al.,1991; Schuman and Madison, 1991; O’Dell et al., 1991). NO is highly diffusibleand considered to be a strong candidate as a retrograde messenger for LTP.Bath application of NO synthase inhibitors such as N-methyl-L-arginine (NMLA)and L-nitroarginine (LNA), or NO scavenger, hemoglobin (Hb) inhibits theinduction of LTP (Bohme etal., 1991; O’Dell et al., 1991; Schuman and Madison,1991a; Haley et al, 1992; also see Errington et al., 1991). However, NOinhibitors have no effect on the established LTP (Haley et al., 1992). Injection ofNO synthase inhibitors into the postsynaptic neuron also prevent the LTPXIE 106induced by pairing postsynaptic depolarization with presynaptic stimulation(Schuman and Madison, 1991). On the other hand, sodium nitroprusside, whichrelease NO, can induce LTP (Bohme et al., 1991). Furthermore, directapplication of NO to hippocampal slices paired with a weak tetanus induces aninput-specific LTP, which is blocked by hemoglobin, but not by APV andnifedipine, a Ca2 channel blocker (Zhuo et al., 1993). This LTP developsrapidly and occludes the tetanus-induced LTP. Tetanic stimulation and NMDAhave been reported to activate NO synthase in the hippocampus (East andGarthwaite, 1991). Most of the data discussed above seem to favour the notionthat NO, indeed, is a retrograde messenger for LTP in the hippocampus.However, several lines of evidence undermine the hypothesis of NO as aretrograde messengers for LTP in the hippocampus. First, histochemicalevidence for NO synthase (NOS-l) in CAl pyramidal cells is not available (Bredtet al., 1991; Vincent and Kimura, 1992, but also see Schweizer et al., 1993).Second, it is not clear whether NO can directly increase glutamate release frompresynaptic terminals. Third, injection of NO inhibitor into the postsynapticneuron does not prevent the tetanus-induced LTP (Schuman and Madison,1991). Fourth, the application of NO paired with low frequency stimulation doesnot induce LTP, but LTD (Zhuo et al., 1993). Fifth, the selectivity of NOinhibitors and hemoglobin has also been questioned. Therefore, conclusionshould not be made until the answers for these questions are found.Recently, carbon monoxide (CO) has been brought forward as a potentialretrograde messenger for LTP (Zhuo et al., 1993; Stevens and Wang, 1993).Much like NO, C’O is a highly diffusible agent. Moreover, the heme oxygenasewhich produces CO is clearly present in hippocampal pyramidal cells (Verma etal., 1993). Bath application of CO to hippocampal slices when paired with aweak tetanus or even low frequency stimulation induces an input-specific LTPXIE 107which is not blocked by APV and nifedipine (Zhuo et al., 1993). The hemeoxygenase inhibitors such as zinc protoporphyrin IX (ZnPP) prevent theinduction of LTP induced by a strong tetanus (Zhuo et al, 1993; Stevens andWang et al., 1993). Furthermore, ZnPP can reverse the established LTP withoutaffecting the control pathway (Stevens and Wang, 1993). At present, it is notknown whether tetanic stimulation is able to activate heme oxygenase. Basedon the available results, CO appears to be a very attractive candidate to serveas a retrograde messenger for LTP in hippocampus. However, it is too early toconclude the role of CO in LTP because more studies on this agent are needed. Neurotrophic factorsNeurotrophic factors have been suggested to be involved in LTP (Sastryet al., 1988b). Application of nerve growth factor (NGF) when paired with weaktetanus induces LTP (Sastry et al., 1988b; also see Tancredi et al., 1993).Recently, LTP has been reportedly associated with an increase in neurotrophinlevels in the stimulation region in hippocampal slices (Patterson et al., 1992).The evidence further supports the involvement of neurotrophic factors in LTP.Other neurotrophic factors such as epidermal growth factor (EGF) and fibroblastgrowth factor (FGF) have also been shown to convert STP to LTP or to facilitateLTP (Terlau and Seifert, 1989; Abe et al., 1991; 1992; lshiyama et al., 1991;Hisajima et al., 1992). It is believed that neurotrophic factors such as NGF arereleased from postsynaptic neurons and induce biochemical and morphologicalchanges in the presynaptic terminals (Hendry et al., 1974; Springer and Loy,1985; Thoenen, 1991; Patterson and Nawa, 1993). Substances collected fromguinea pig hippocampus (Chirwa and Sastry, 1986) or rabbit neocortical surface(Sastry et al., 1988a) during tetanic stimulation, when applied to guinea pighippocampal slices, induce LTP. These substances also enhance neuritegrowth in cultured PC-12 cells (Sastry et al., 1988a). Saccharin which interferesXIE 108with NGF binding and decreases neurite growth (lshii, 1982) prevents not onlythe tetanus-induced LTP, but also the substances-induced LTP (Morishita et al.,1992). Saccharin also blocks the enhancement of neurite growth in PC-12 cellsproduced by the substances. It is possible that these substances released frompostsynaptic cells during tetanic stimulation act as neurotrophic factors andenhance transmitter release from presynaptic terminals. Further details on theeffect and characteristics of these substances will be discussed in the resultsand discussion of this thesis. Released KReleased K from postsynaptic cells during tetanic stimulation is also acandidate to serve as a retrograde messenger for LTP. This ion is highlydiffusible and induces effects very rapidly. Elevation of extracellular K in theabsence (May et al., 1987) and the presence of extracellular Ca2 (Ballyk andGoh, 1992; Fleck et al., 1992) has been demonstrated to induce LTP. Themechanisms for the K-induced LTP are not known. lontophoretic application ofK or baclofen to dendritic zone in the CAl field when paired with weak tetanusalso induces LTP (Ballyk and Goh, 1992). It is possible that an increase insynaptic K level induced by tetanic stimulation causes further changes aroundthe synapses, including the presynaptic terminals. ACPD receptors have beenimplicated to be involved in the maintenance of LTP. Presynaptic ACPDreceptors mediate a positive feed back of glutamate exocytosis (Herrero et al.,1992). The coupling between phospholipase C (PLC) and ACPD receptors ispotentiated by extracellular K (Irving et al., 1992). However, it is not sufficientevidence to decide what kind of role extracellular K plays in LTP. Furthermore,K can be released from sources other than postsynaptic cell such as glia. Infact, involvement of glia in LTP has been suggested (Sastry et al., 1988c)XIE 1095.4. Modulatory factors on LTPBoth endogenous glutamate and GABA play a critical role in controllingLTP in hippocampus. However, many other neurotransmitters have modulatoryeffects on LTP. Therefore, a brief discussion is provided here on the modifyingrole of neurotransmitters in LTPAcetylcholine (ACh) has been reported to induce a NMDA receptor-dependent LTP in the CAl region (Markram and Segal, I 990a). The effect ofACh is probably due to the NMDA potentiation by ACh (Markram and Segal,I 990b). Carbacol, an ACh receptor agonist, has shown to potentiate NMDAcurrents (Otani and Ben-An, 1993).Noradenaline (NA) has been shown to induce LTP in the dentate gyrus(Neuman and Harley, 1983). This LTP is linked to NMDA receptor activation(Burgard et al., 1989; DahI and Sarvey, 1990). In the CAl region, NA inducesLTP of the population spike, but not LTP of the EPSP (Heginbotham andDunwiddie, 1991).Serotonin has been recently demonstrated to block the LTP induced byprimed burst stimulation but not the LTP induced by strong tetanus (Corradetti etal., 1992, also see Bliss et al., 1983).Dopamine has been found to increase after tetanic stimulation (Frey etal., 1990). Furthermore, Frey et al. (1991) have shown that Di receptorantagonist blocks the maintenance of LTP.In summary, LTP can be modulated by other neurotransmitters. Themechanisms for the effect of these neurotransmitters on LTP are far less knownthan those for glutamate and GABASeveral neuropeptides, including somatostatin (Matsuoka et al., 1991a,b), arginine-vasopressin (AVP4..8)(Rong et al., 1993) and dynorphin (Weisskopfet al., 1993) have been shown to be involved in LTP. Somatostatin has beenXIE 110found to be co-localized with GABA in the CAl and CA3 interneurons. Since thepeptide and GABA can be co-released from the inhibitory terminals by afferentstimulation, it is of interest to understand how the peptide interacts with GABAand modulates LTP.Recently, free radicals have also been shown to facilitate the decay ofLTP in the CAl area of the hippocampus (Peilmar et al., 1991). Both freeradicals and antioxidants have been shown to modulate NMDA receptorfunction. Whether antioxidants, which remove free radicals, can also modulateLTP will be discussed later.5.5. LTP in hippocampalinterneuronsOrthodromic stimulation in CAl field activates not only the afferents topyramidal cells, but also the afferents to interneurons which, in turn, provideinhibition to pyramidal cells through the feedforward system. Therefore,changes in interneurons after tetanus directly affect the LTP in CAl neurons.LTP in CAl interneurons has been reported in vitro (Taube and Schwartzkroin,1987; Abraham et al., 1987) and in vivo (Buzsaki and Eidelberg, 1982; Kairiss etal., 1987). However, IPSP5 recorded from CAl neurons show inconsistentchanges after tetanic stimulation of the stratum radiatum (Abraham et al., 1987).IPSPs of CAl neurons have been shown to increase, decrease or not changeafter tetanus (Abraham et al., 1987). The IPSPs in the CAl neurons usuallyrepresent a mixture of feedforward and feedback inhibition. Changes inpyramidal neurons properties directly alter the expression of IPSPs. Therefore,the changes in the IPSPs of CAl neurons do not directly reflect the changes ininterneuron EPSP. Recently, Reece and Redman (1992) and Rai et al. (1993)have shown that LTP in CAl interneurons can be induced by pairing weaktetanus and postsynaptic depolarization. This result is consistent with theprevious findings reported by others (Buzsaki and Eidelberg, 1982; Taube andXIE 111Schwartzkroin, 1987). However, the mechanisms for interneuron LTP is littleknown. This is partly due to the difficulties of studies on interneurons.5.6. The physiological significance of LTPLTP has attracted the interest of many neuroscientists because it is aphenomenon of use-dependent modification in synapses of mammalian brainthat is thought to be the cellular mechanism of learning and memory. Hebb(1949) postulated that the synapses linking two cells are strengthened whencoincident activity of two cells occurs. This synaptic strengthening, hesuggested, forms the foundation for learning and memory. Several lines ofevidence support LTP as the cellular mechanism for learning and memory. First,the characteristics of LTP such as input-specificity, cooperativity andassociativity are consistent with Hebb’s rule. LTP can be induced by 9 rhythms(Larson et al., 1986; Larson and Lynch, 1989), which occur during exploratory orlearning activities (O’Keefe and Nadel, 1978; Vanderwolf, 1969). Second,behavioural studies have shown that some learning behaviours induce LTP orLTP-like changes in synaptic responses in the hippocampus (Sharp et al., 1985;Green and Greenough, 1986; Wilson et al., 1986; Skelton et al., 1987). Priorlearning enhances the LTP induced later in the same brain (Bergis et al., 1990).A prior induction of LTP also affects the animal’s later spatial learning (Castro etal., 1989; McNaughton et aL, 1986; Berger, 1984). Third, recent studies havedemonstrated that pharmacological agents such as APV, which block theinduction of LTP in hippocampus, also impair spatial learning and memory ofanimals (Davis et aL, 1992; Morris et al., 1986; Bolhuis and Reid, 1992).Genetic studies have shown that mice with mutations in certain genes, exhibitimpaired spatial learning as well as impaired LTP (Silva et al., 1992a,b; Grant etal., 1992). These results suggest that LTP is correlated with certain types oflearning and memory. However, although these lines of evidence are consistentXIE 112with the idea that LTP underlies the cellular mechanisms for certain forms oflearning and memory, there is lack of direct evidence to make any firmconclusion at present. Experimentally, it is difficult to study isolated “learningand memory” in animals. Pharmacological agents and gene mutation not onlyaffect learning and memory, but also may induce general changes in thestructure and the function of the brain, If LTP underlies learning and memory atthe cellular level, further understanding in the mechanisms of LTP will help todevise rational treatments for neurological disorders such as Alzheimer’sdisease, which are related to learning and memory.5.7. SummaryLTP has been extensively studied by neuroscientists around the worldsince its discovery in hippocampus two decades ago. Among LTPs in differentparts of the brains, the LTP in hippocampus is most well studied. Much progresshas been made in elucidating the mechanisms underlying the induction andmaintenance of LTP in recent years. However, many fundamental issues onLTP have not been resolved. The induction of LTP is primarily mediated throughpostsynaptic mechanisms. In most cases, the activation of NMDA is necessaryfor the induction. The Ca2 influx through NMDA channels activates a numberof enzymes which trigger a series of biochemical and morphological changes inthe synapses. This would result in the induction of LTP. In addition to theNMDA receptor activation, a presynaptic element is obviously involved in theinduction of LTP because the induction of LTP is presynaptic activity-dependent.Based on currently available data, the maintenance of LTP is likely due to bothpresynaptic and postsynaptic changes. The questions are what are thesechanges and how these changes occur. Many investigators favour the notionthat LTP is due to an increase in neurotransmitter release from presynapticterminals, changes in postsynaptic receptor sensitivity and/or number, and/orXIE 113changes in spine structure. These changes can be caused by the activation ofprotein kinases which induce phosphorylations of protein molecules in receptorsor ion channels. However, more convincing evidence is required to concludethat these changes, indeed, occur and are responsible for the maintenance ofLTP. Increases in protein synthesis and gene expression seem to be associatedwith LTP. Whether these increases are responsible for LTP remains to bedetermined. The search for retrograde messengers is one of the hottest topicsin the field of LTP for the last few years. Nitric oxide, arachidonic acid,neurotrophic factors are among the strongest candidates. So far, none of themhas been proved. GABAergic input is known to have influence in LTP, but therole of GABA8 receptors in LTP has been basically unknown until recently. Inmany studies of LTP, GABAergic inhibition is blocked to minimize the influenceof the IPSPs on the EPSP. However, under normal physiological conditions,changes in GABAergic inhibition cannot be ignored. Understanding the changesof GABA receptor-mediated IPSPs during LTP will help to resolve some of theissues involved in LTP. Some endogenous neuropeptides are co-released withGABA or glutamate from presynaptic terminals. Endogenous peptides, such assomatostatin, AVP4..8 and dynorphin, have been shown to have modulatoryeffects on LTP. Interactions between neuropeptides and neurotransmitters mayoccur during LTP. What kinds of roles do endogenous neuropeptides play inLTP is largely unknown. Finally, the physiological significance of LTP is not veryclear. LTP is thought to underlie the cellular mechanism for learning andmemory. If this is the case, agents, which affect learning and memory, shouldalso affect LTP. Studies of the effects of these agents on LTP will lead to abetter understanding of the physiological significance of LTP.XIE 1145.8 Rationale and specific aimsAs noted in the review on LTP, many questions on LTP remainunanswered. The main goal in the present studies is to examine variousmechanisms that modulate LTP. Briefly, the present investigations focus onthree major subjects: (1) the role of endogenously released substances in LTP,(2) the influence of changes in GABAergic inhibition on LTP of the EPSP, (3) therole of x-tocopherol in LTP.5.8.1. The role of released substances in LTPIt has been reported that proteins are released during tetanic stimulationor LTP (Duffy et al., 1981; Charriault-Marlangue et al., 1988). It is, however, notknown whether these released proteins are involved in LTP. Previous studies inour laboratory found that substances collected from the guinea pig hippocampus(Chirwa and Sastry, 1986) and rabbit neocortex (Sastry et al., 1988a) duringtetanic stimulation, when applied on guinea pig hippocampal slices, could induceLTP of the population spike in these slices. Since these substances arediffusible, it is possible that they act as retrograde messengers for LTP. It hasbeen hypothesized that retrograde messengers are released from postsynapticcells and cause presynaptic modifications which lead to LTP (Bliss et al., 1986;Sastry et al., 1986). If some of these substances are retrograde messengers forLTP, they could be released from postsynaptic cells through a NMDA receptor-dependent mechanism, and subsequently cause presynaptic modifications.However, the substances are not characterized and the mechanisms underlyingtheir release and actions have not been examined. Experiments were, therefore,designed to address the following questions: (1) Does the release of thesesubstances require the activation of NMDA receptors? (2) Do these substancesenhance the excitatory transmission or do they also affect the inhibitorytransmission? (3) Do the LTP-inducing actions of the substances involveXIE 115presynaptic or/and postsynaptic elements? (4) What are the molecular weightsof the LTP-inducing substances and are proteins present?5.8.2. The influence of GABAergic inhibition on LTPStudies based on the effect of GABA receptor antagonists on LTPindicate that GABAergic inhibition has significant influence on the induction ofLTP (Wigstrom and Gustafsson, 1983; Davies et al., 1991; Mott and Lewis,1991). In many studies of LTP of the EPSP, the fast and the slow IPSPs aretherefore blocked to minimize the effects of IPSP on the EPSP. However, underphysiological conditions, stimulation of the stratum radiatum evokes not onlyEPSP but also GABA receptor-mediated fast and slow IPSPs in the CAl region.It is known that tetanic stimulation induces LTP of the EPSP. However, posttetanic changes in the IPSPs are controversial. If post-tetanic changes in theIPSPs occur, these changes may affect the expression of LTP of the EPSP.Therefore, the main objective of this study is to examine whether tetanicstimulation induces long-term changes in the IPSP5, and if so, whether thesechanges modulate the expression of LTP of the EPSP. Since tetanic stimulationis believed to induce LTP of the EPSP through an increase in postsynaptic freeCa2 and PKC activity (Lynch et al., 1983; Malinow et al., 1989), and sinceincreases in intracellular free Ca2 and PKC activity have been shown to havesignificant effects on the IPSPs (Baraban et al., 1985; Dutar and Nicoll, 1988b;Chen et al., 1990), BAPTA (Ca2 chelator) and K-252b (PKC inhibitor) wereused as tools to determine whether changes in intracellular free Ca2 and PKCactivity would affect post-tetanic changes in the IPSPs. These experiments weredesigned to answer the following questions: (1) Is there any long-term changein the lPSPs after tetanic stimulation of the stratum radiatum, and if so, are thereany differences between the fast and the slow IPSPs? (2) Do changes inpostsynaptic free Ca2 and PKC activity affect post-tetanic IPSPs? (3) Are post-XIE 116tetanic changes in the IPSPs dependent on glutamatergic transmission? and (4)Do post-tetanic IPSPs have any effect on the expression of LTP of the EPSP?Somatostatin (SS), a peptide which is co-localized in some of GABAergicneurons and their terminals in the CAl area, may be co-released with GABAfrom the same terminals by tetanic stimulation. At present, it is unknownwhether this peptide has effects on LTP of the EPSP in CAl neurons.Application of SS has been shown to depress both the fast and slow IPSPsthrough an unknown mechanism (Scharfman and Schwartzkroin, 1989). Ifinteractions between SS and GABA occur at the same synapses, the peptidemay modify LTP of the EPSP as well. If SS, like picrotoxinin (GABAA blocker) orphaclofen (GABAB blocker), blocks the fast and slow IPSPs, this peptide mayfacilitate the induction of LTP of the EPSP. On the other hand, SS may preventthe induction of LTP of the EPSP since it causes a hyperpolarization of theneurons (Pittman and Siggins, 1981). Therefore, it is important to determinehow SS interacts with GABAergic inhibition and whether SS can influence LTPof the EPSP. Experiments were conducted to address the following issues: (1)How does SS modulate GABAergic inhibition? and (2) Does SS affect LTP of theEPSP?5.8.3 The role of a-tocopherol in LTPFree radicals and antioxidants have been implicated in adverselyaffecting the memory process in diseases such as Alzheimer’s. cc-Tocopherol, amajor lipid soluble antioxidant in the biological systems, has been shown to beinvolved in spatial learning in rats (Moriyama et al., 1990) and its levels in thebrain are lower in Alzheimer’s disease (Jeandel et al., 1989). In culturedneurons, x-tocopherol can act as a neurite-inducing factor (Nakjima et al., 1991);and such factors have been implicated in LTP (Sastry et al., 1988b). Meanwhile,free radicals have been shown to facilitate the decay of LTP (Pellmar et al.,XIE 1171991). These lines of evidence raise the possibility that x-tocopheroI may beinvolved in learning and memory as well as LTP. Therefore, the role of cxtocopherol in LTP was examined. Experiments were conducted to resolve thefollowing issues: (1) Does cL-tocopherol induce LTP of the EPSP? (2) If theagent induces LTP, is this LTP similar to the tetanus-induced LTP? (3) Is theability to induce LTP impaired in hippocampal slices obtained from vitamin Edeficient animals?6. PHARMACOLOGICAL TOOLS FOR EXPERIMENTS6.1. SomatostatinSomatostatin (SS or SS-14) is a 14-amino acid containing peptideoriginally isolated from the hypothalamus and inhibits growth hormone secretionfrom the anterior pituitary (Brazeau et al., 1973). Several lines of evidence haveshown that this peptide acts as a neurotransmitter in the CNS (Olpe et al., 1980;Epelbaum, 1986). SS and its precursor (SS-28) can induce multiplephysiological actions in the brain, including the modulation of Ca2 and Kconductances, neuronal cell firing and neurotransmitter release (Epelbaum,1986; lnoue and Yoshii, 1992). The discussion here is focused on the actions ofSS in the hippocampus.6.1.1. Distribution of somatostatin in hippocampusImmunohistochemical studies have revealed the presence of SS bindingsites and SS-like immunoreactivity in mammalian hippocampus and severalother areas of the brain (Bennet-Clarke et al., 1980; Kohler and Chan-Palay,1982). In the hippocampus, SS-immunoreactivity are present in someinterneurons in the stratum oriens and the stratum radiatum of the CAl and CA3fields, and the hilus of the dentate gyrus (Johansson et al., 1984; Somogyi et at.,1984; Kosaka et al., 1988). Most of the SS-containing neurons are asubpopulation of GABAergic interneurons (Somogyi et al., 1984; Kosaka et al.,XIE 1181988). The highest densities of SS receptors are found in the stratum oriensand stratum radiatum of the CAl field. Somewhat lower densities are found inthe CA3 field. The pyramidal cell layer and the stratum lacunosum-moleculareof the CA field have very few SS receptors (Swanson et al., 1987). Recentstudies have revealed that SS receptor subtypes exist in the brain (Raynor andReisine, 1992; Bell and Reisine, 1993). Four types of SS receptors have beencloned. (Bell and Reisine, 1993). But the function of these cloned SS receptorsremains to be defined. Two of the better known SS receptors are the SS1 andSS2 receptors. The SS1 receptors have selectively high affinity for the agonistMK678 while the SS2 receptors have high affinity for the agonist CGP 23996with no affinity for MK678. Both SS1 and SS2 receptors have been found in theCAl field. The dentate gyrus expresses primarily 5S1 receptors (Martin et al.1991 b; Raynor and Reisine, 1992).6.1.2. Actions of somatostatin on hippocampal neuronsThe electrophysiological data on the effects of SS have been quiteconfusing in the past (Delfs and Dichter, 1985). This is probably due to the dataobtained from different types of cells and the existence of multiple SS receptorsubtypes. In the hippocampus, SS has been reported to induce a depolarization(Dodd and Kelly, 1978) or a hyperpolarization (Pittman and Siggins, 1981) of theCAl neurons. Scharfman and Schwartzkroin (1988) have shown that SSdepolarizes the pyramidal cells when applied directly on the soma. Whenapplied to the dendrites, particularly the distal dendrites, SS can induce ahyperpolarization of the CAl neurons. These results implicate that differentsubtypes of SS receptors may exist in different area of the neurons. Bathapplication of SS consistently induces a hyperpolarization of the CAl neurons,associated with a reduction in the input resistance (Pittman and Siggins, 1981;Schariman and Schwartzkroin, 1988; Moore et al., 1988). The SS-inducedXIE 119hyperpolarization is due to an increase in K conductance (Pittman and Siggins,1989; Twery et al., 1991). In addition, SS also enhances the noninactivating,voltage-dependent K current (IM) in the hippocampal CAl neurons. The SSinduced increase in ‘M is probably mediated by arachidonic acid metabolites(Schweitzer et al., 1990). Whether the effects of SS are mediated throughdifferent subtypes of SS receptors is not certain. It has been reported that theeffects of somatostatin on the resting K conductance are pertussis toxin-sensitive while the augmentation of ‘M is pertussis toxin-insensitive (Schweitzeret al., 1989). It is known that SS1 receptors couple to G proteins whilereceptors do not (Raynor and Reisine, 1992). Therefore, it is possible thatdifferent subtypes of SS receptors mediate the effects of SS on the resting Kconductance and ‘M The effects of SS on Ca2 currents in hippocampalneurons are not clear. However, SS has been reported to reduce Ca2 currentsin neocortical neurons (Wang et al., 1990) and other types of cells (Chen et al.,1990). Since SS and GABA are co-localized in some interneurons in the CAlarea of the hippocampus (Somogyi et al., 1984), interactions between SS andGABA may be relevant to the role of SS in epilepsy as well as LTP. It have beenreported that SS depresses the GABA receptor-mediated IPSPs in hippocampalCAl neurons (Schariman and Schwartzkoin, 1989). The mechanisms of theinteractions between SS and GABA are however not very clear.6.1.3. The functional roles of somatostatinSeveral lines of evidence implicate that somatostatin may be involved inlearning and memory as well as LTP (Schettini, 1991; Haroutunian et al., 1987;Matsuoka et al., 1991a, b). The intracerebroventricular (i.c.v.) administration ofSS significantly increases the active avoidance behaviour of the animals (Vecseiand Widerlov, 1988). Cysteamine can produce a rapid, relatively selective, butreversible, depletion of central SS stores in animal (Sagar et al., 1982).XIE 120Cysteamine-induced depletion of SS in the CNS results in a significantattenuation of passive avoidance retention test performance of the drug-treatedrats (Haroutunian et al., 1987). LTP of the mossy fibre pathway in thehippocampus obtained from cysteamine-treated guinea pigs is significantlysmaller than the LTP in the hippocampus obtained from the control guinea pigs(Matsuoka et al., 1991b). SS has also been reported to enhance LTP in themossy-fibre pathway of the CA3 field. The mechanisms of the effects of SS onLTP are not clear. However, these reports implicates that somatostatin may playan important role in some forms of learning and memory. The effects of SS onLTP of the EPSP in the CAl field have not been reported. SS may play a role inLTP in the CAl field because of the interactions between SS and GABA.Involvement of SS in Alzheimer’s disease has been reported (Davies andTerry, 1981; Ferrier et al., 1983; Beal, 1990). In patients with Alzheimer’sdisease, SS levels are significantly decreased in several areas of the brain(Schettini, 1991; Bissette and Myers, 1992). The decrease in SS levels is moresignificant than the decrease in the levels of other neuropeptides in the brains ofpatients with Alzheimer’s disease (Ferrier et al., 1983). The SS receptor bindingin several parts of the brain, including the hippocampus, is significantlydecreased with aging (Sato et al., 1991). These results indicate that SS is likelyto be involved in the memory process. However, the exact role of SS inAlzheimer’s disease and aging is not clear.SS has been reported to be involved in epilepsy (Riekkinen and Pitkanen,1990). Changes in SS levels in the frontal cortex and the hippocampus followingsingle seizures have been reported (Sperk and Widman, 1985). Incerebrospinal fluid (CSF), SS level is elevated in the kindling model of epilepsy(Kato et al., 1983; Pitkanen et al., 1987). In epileptic patients, SS binding siteschange in number (Riekkinen and Pitjanen, 1990). The real role of SS inXIE 121epilepsy is not clear. The effects of SS on GABAergic inhibition may play a rolein epilepsy.6.2. a-Tocopherolc-Tocopherol (vitamin E) is also called the “antisterility vitamin” becauseit was originally found to be required to sustain a normal pregnancy in femalerats (Evans and Bishop, 1922). x-Tocopherol is distributed throughout thetissues of animals and humans. Further studies have indicated that citocopherol, as an integral part of the cell membrane, is essential for themaintenance of normal structure and function of many organ systems, includingthe human nervous system (Tappel, 1962; Sokol, 1988, 1989). A prolongeddeficiency of ci-tocopherol results in a number of degenerative neurologicalsyndromes in humans and experimental animals (Muller et al., 1983; Sokol,1988, 1989). ci-Tocopherol is a major lipid-soluble antioxidant which canscavenge free radicals attacking from outside of the membrane and within themembrane, and stabilize the biological membrane (Tappel, 1962; Witting, 1980;Erin et al., 1984; Lucy, 1972). ci-Tocopherol can also modulate the metabolismof arachidonic acid cascade (Reddy et al., 1988; Tran and Chan, 1988).6.2.1. Physicochemical propertiescL-Tocopherol, 2,5,7, 8-tetramethyl-2-(4’, 8’, 1 2’-trimethyltridecyl )-6-chrom-anol or 5,7,8-trimethyl tocol, the most active vitamin E factor, was first isolatedfrom wheat germ oil (Evans et al., 1936). ci-Tocopherol is a liposolublecompound and liquid at room temperature. The melting point of ci-tocopherol is2.5-3.5 °C. ci-Tocopherol is insoluble in water. It is soluble in oils, ethanol,acetone and dimethyl sulfoxide (DMSO). However, the disodium salt of citocopherol phosphate is soluble in water while possesses the activity of citocopherol. cc-Tocopherol is stable to heat and alkalis in the absence of oxygen.It can be slowly oxidized by atmospheric oxygen.XIE 1226.2.2. Antioxidant propertiesA free radical is a molecule with an unpaired electron in its outer orbit.The unpaired electron makes the molecule unstable and highly reactive,especially towards biological molecules such as lipids, protein and DNA. Freeradicals can be produced in a variety of cellular processes in biological systems.Increases in oxygen-derived free radicals have been implicated in pathogenesisof many diseases.Free radicals and active oxygen compounds such as superoxide (02i,hydrogen peroxide (H20), hydroxyl radicals (HO) are normally generated bycellular metabolism. Under normal physiological conditions, intrinsic enzymesystems and antioxidants keep the free radical levels in check (Demopoulos etal., 1982; Fridovich, 1978; Halliwell, 1987). Increases in these free radicalsoccur when intrinsic enzymes and antioxidants decrease or/and production offree radicals increase under certain pathological conditions such as ischemicinjury and Alzheimer’s disease.cc-Tocopherol is a major lipid-soluble antioxidant present in biologicalmembranes and prevents lipid peroxidation of membranes (Tappel, 1962;Witting, 1980; Burton and Ingold, 1989). cL-Tocopherol reacts directly withperoxyl radicals, an intermediate of lipid peroxidation, and forms a very stabletocopheroxyl radical. Subsequently, it stops the free radical chain reaction.Therefore, c-tocopherol successfully protects reactive substrates, such aspolyunsaturated fat, in membranes (Tappel, 1962). c-Tocopherol can alsoscavenge other reactive oxy-radicals, such as 0j, H0. It is apparent that atocopherol plays a critical role in the maintenance of the structure and functionof the membrane. Deficiency of a-tocopherol leads to membrane damage.- XIE 1236.2.3. The functional role of a-tocopherol in the CNSA severe and prolonged deficiency of cL-tocopherol in humans andanimals results in a series of neurological syndromes, including ataxia,hyporeflexia, proprioceptive loss, ophthalmoplegia and retinal pigmentation(Sokol, 1989). Neuropathological lesions include a degeneration of the axons inseveral parts of the brain. Treatment of patient with pharmacological doses of ctocopherol prevents the development of neurological syndromes or causesimprovement (Sokol, 1989).Several diseases, including Alzheimer’s disease, and normal aging(Wartanowicz et al., 1984) have been reported to be associated with decreasesin cL-tocopherol and other antioxidant levels, and increases in several freeradicals contents (Wartanowicz et al., 1984; Carney et al, 1991; Jeandel et al.,1989; Smith et al., 1991). Whether these changes in cL-tocopherol and freeradicals are directly related to Alzheimer’s disease and normal aging is not clear.Both Alzheimer’s disease and aging are associated with deterioration ofmemory. Behavioural changes of x-tocopherol deficient rats have beenobserved (Ichitani et al., 1991; Moriyama et al., 1989, 1990). In cL-tocopheroldeficient rats, impairment of spatial learning ability associated with an increasein lipid peroxide content in hippocampus has been observed (Moriyama et al.,1990). In these reports, the experimental rats showed no significant changes inbody weight, locomotor activity when the learning tests were performed. Theseresults implicate that the learning impairment in x-tocopherol rats may be partlyrelated to the increase of lipid peroxide content in hippocampus. x-Tocopherolhas been shown to support the survival of cultured neurons and to enhanceneurite growth (Nakajima et al., 1991; Sato et al., 1993). Recently, free radicalshave been reported to facilitate the decay of LTP in the hippocampus (Pelimar etal., 1991). These findings demonstrate that x-tocopherol and free radicals areXIE 124probably related to certain types of learning and memory. However, clearevidence is needed to clarify this matter.7. METHODS AND MATERIALS7.1. AnimalsMale Hartley guinea pigs, male Wistar rats and New Zealand Whiterabbits were obtained from the Animal Care Centre of the University of BritishColumbia. Once a week, 5-6 guinea pigs (150-200 g, approximately 20 day old)were received from the centre and placed in a wire cage in the animal room ofthe Department of Pharmacology & Therapeutics for less than a week beforebeing used for experiments. These guinea pigs had free access to guinea pigchow and water. Two rabbits (either sex; approximately 2-3 kg; 42-56 days old)were received each time and 2-3 times a week. Each of these rabbits was keptin a big wire cage and fed with rabbit chow in the animal room. Three sets ofsixteen male rats (50-100 g, approximately 10-14 days old) were obtained forvitamin E deficient experiments. Each set of rats was divided into two groups (8in each group) and placed in separate cages. One group was fed on a vitamin Econtrol diet while the other group was fed on a vitamin E deficient diet. Thesediets were obtained from Harlan Teklad (Madison, Wisconsin). With theexception of vitamin E, the components of vitamin E deficient diet are the sameas those of vitamin E control diet. The first set of rats was fed on these diets forone month before being used for experiments. The second and third sets of ratswere fed for two and three months, respectively. Different rooms were used forguinea pigs, rats and rabbits. The temperature and humidity of these animalrooms were controlled at 22-23 °C and 50-55%, respectively.7.2. Perfusion mediaThe standard physiological medium was prepared daily. This mediumcontained (in mM) NaCl 120, KCI 3.1, NaHCO3 26, NaH2PO4 1.8, MgCl2 2.0,XIE 125CaCI2 2.0 and dextrose 10.0. The pH of the medium was 7.4 when bubbled withcarbogen (95% 02 and 5% CC2). The hippocampal slices were perfused withthis standard medium unless otherwise stated.In some experiments in which picrotoxinin (20 j.tM) was added to themedium, the concentration of CaCI2 and MgCl2 was increased to 4 mM andNaH2PO4was omitted. The increase in the concentration of Ca2 and Mg2 inthis picrotoxinin-containing medium was to minimize epileptiform activity whichcan be induced by adding picrotoxinin. Exclusion of NaH2PO4in this mediumwas to avoid precipitation that may result from the increase in CaCl2. Themodification did not change the pH of the medium.In the Mg2-free medium, Mg2 was omitted from the standard mediumwhile the other components of the standard medium were unchanged.Drugs used in the experiments were usually first prepared asconcentrated stock solutions once or twice a week. These stock solutions werestored in the refrigerator. The concentrated stock solutions were diluted with thestandard medium to the desired concentrations just before a experiment. Drugsprepared in this manner included D(-)2-amino-5-phosphonovalerate (D-APV), 6-cyano-7-nitroquinoxaline-2,3-dione (CNQX), N-methyl-D-aspartate (NMDA), Lglutamate, picrotoxin in (PTX), tetrodotoxi n (TTX), (-)-baclofen, 2-0H-saclofen,phaclofen, sphingosine, K-252b, somatostatin-14 (SS) and cc-tocopherolphosphate (disodium salt). In some experiments, NaH2PO4was omitted fromthe standard medium when cL-tocopherol phosphate was added. CNQX, 2-OH-saclofen and phaclofen were purchased from Tocris Neuramin. K-252b wasobtained from Dr. A. Mon of Kyowa Hakko Kogyo Ltd. Japan. All other drugswere from Sigma.The standard medium and the drug-containing media were contained inseparate 50 ml barrels (Fig. 3). A tube from each of the barrels was connectedXIE 126to a common manifold, and the single outlet of the manifold was connected tothe slice chamber via a connecting tube. A reservoir (1 L) containing thestandard medium was placed above the barrels. This reservoir was connectedto the main barrel which also contained the standard medium. The media in allthe barrels and the reservoir were oxygenated via separate lines which wereconnected to a carbogen cylinder. The lines from the reservoir to the mainbarrel and from the main barrel to the slice chamber were always open exceptduring the period of drug application. The flow rate was 1.5-2 mI/mm. A suctionline was connected to the outlet of the slice chamber so that the flow rate in thechamber was always constant. When drugs were applied, the line from the mainbarrel was closed quickly with a clip, and the line from the drug-containing barrelwas opened simultaneously. The same procedure was applied for thetermination of the application.7.3. Slice ChamberThe slice chamber was fully described by Panaboina and Sastry (1984)from this laboratory and is illustrated in Figure 3. The chamber contained threemain components. They are as follows: a plexiglass stage, a circular chamberdrilled into the top surface of the plexiglass stage, and an aluminumtemperature-regulating bar which was located underneath the chamber andcould control the temperature of the chamber according to the need of theexperiments.7.4. Preparation of slicesTransversely sectioned hippocampal slices (450 j.tm thick) were obtainedfrom male guinea pigs, with a few exceptions in which male rats were used. Theanimal was gently placed on the top of an ice pack (to lower the animal’smetabolic demand) in a dessicator. The animal was anaesthetized with amixture of halothane (2%) and carbogen (95% °2’ 5% C02) introduced into theXIE 127Media barrels0000000000000000000000000000000 000 00000 0000000000000000cO 000CC 000C O0000000000 0000000 00 0000ManifoldFigure 3 Slice chamber and perfusion system for maintaining hippocampalslices.The slice chamber and perfusion system used in the present study is shown inthis sketch. Note that the transversely sectioned hippocampal slices wereplaced between the meshes of the inner and outer rings which were inserted inthe slice chamber. Oxygenated standard medium contained in one of the mediabarrels coursed through the manifold to the slice chamber where the slices wereincubated. A suction line was used to maintain the constant flow rate andmedium volume in the chamber.XIE 128dessicator. The animal was put on the operation table once the stage of surgicalanaesthesia was reached. The skull of the animal was exposed by sagittal cutof the skin covering the top of the head with a scalpel. A small hole was made atthe base of the skull. A pair of small scissors was inserted into the hole and cutthrough the skull along the sagittal suture. Then, the plates of the skull and durawere carefully removed and the brain was exposed. The brain was carefullytaken out, placed on dissecting paper and quickly soaked with cold (4 °C)oxygenated standard medium. The hippocampus from one side of the brainwere dissected free and placed on the cutting platform of a Mcllwain tissuechopper where the hippocampus was transversely cut (450 jim thick of eachsection). The sliced hippocampus was then transferred to a plate containingcold oxygenated medium, where each section of the hippocampus wasseparated with a fine stainless steel spatula. Two to three slices obtained fromthe middle-portion of the hippocampus were placed between two nylon meshesto minimize movement and quickly transferred to the slice chamber, where theslices were submerged in the oxygenated medium. A carbogen line was placedat the edge of the slice chamber to introduce carbogen to the atmosphere abovethe slices. The procedure from the start of surgery to the placement of the slicesin the chamber was completed within 3 mm. The slices in the chamber wereallowed to equilibrate with the standard medium for at least 1 hour prior torecording. Only those slices with well-defined border for cell body layers wereused for experiments. The slices prepared in this manner could usually supportnormal physiological responses for more than 12 hours. However, mostexperiments lasted no more than 4 hours. Only one slice was used per animal inmost experiments. In some experiments in which picrotoxinin was present in theperfusing medium, the CA3 region of the hippocampal slices was cut to minimizeXIE 129the epileptiform activity of CAl neurons caused by spontaneous activity of CA3neurons.7.5. Recording and stimulating systems7.5.1. AmplifiersExtracellular responses, including field EPSP and population spike, wereamplified 1000 times with a DAM-5A differential pre-amplifier (World PrecisionInstruments). Intracellular responses were recorded with an Axoclamp 2A (AxonInstruments) in bridge current clamp mode. Intracellular EPSP5 and IPSPs wereamplified 10 times. The amplified field potentials and intracellular responseswere fed into the Data Precision 6000 waveform analyzer. The current andvoltage outputs from Axoclamp 2A were also connected to a chart recorder andan oscilloscope.7.5.2. Recording electrodesExtracellular recording electrodes were prepared from standard-wallborosilicate tubing (outer wall diameter, 1.20 mm; inner wall diameter; 0.69 mm:Sutter Instrument Co.). The microelectrodes were pulled on a PE-2 verticalpuller (Narishige Scientific Instruments) and filled with 4 M NaCI. Theresistances of these electrodes were I to 3 M). Intracellular recordingelectrodes were made from standard-wall borosilicate tubing (outer walldiameter, 1.00 mm; inner wall diameter, 0.58 mm: Sutter Instrument Co.). Theseelectrodes were pulled on a Flaming/Brown (model P-87) micropipette puller.The recording electrodes were filled with one of the following: 4 M potassiumacetate; 4 M potassium acetate pIus 0.1 M 1 ,2-bis(2-aminophenoxy)etheane-N’,N’,N’,N’-tetraacetic acid (BAPTA) (Simga); 4 M potassium acetate plus 100mM QX-314 (Astra Pharmaceuticals, Sweden); 4 M potassium acetate plus 5 jiMK-252b (Kyowa Hakko Kogyo LTD.). The resistances of these electrodes were80-1 20 M.XIE 1307.5.3. Stimulating unitsA Grass S88 stimulator (Grass Instrument Co.) was used to generatecurrents. This stimulator had two channels and each channel was connected toa photoelectric stimulus isolation unit (Grass Instrument Co.). One unit coupledto the Axoclamp was used as an external DC command generator. The DCcurrent could be injected through recording electrodes. Another unit wasconnected to a stimulating electrode.Bipolar concentric stimulating electrodes (SNEX-1 00) were obtained fromRhodes Medical Instruments. These electrodes had a small tip (0.1 mm indiameter), which could minimize the tissue damage caused by the placement ofthe electrodes. The resistances of these electrodes were approximately I M,and electrodes were replaced if their resistances were higher than 57.5.4. Recording unitsExtracellular field potentials were recorded with a DATA Precision 6000waveform analyzer, where the slope and amplitude of the field potentials weremeasured. The intracellular EPSPs and IPSPs were recorded and analyzed withthe DATA 6000 analyzer. The membrane potentials and the input resistances ofneurons were continuously monitored on an oscilloscope (Tektronix Inc.) and onthe chart paper of a polygraph (Grass Instrument Co.) or a HP7470-A graphicsplotter (Hewlett-Packard). In some experiments, responses were recorded witha video cassette recorder (Toshiba) and used for later analysis.7.6. Application of drugsBath application was used in most experiments of the present studies(see the section 7.2). K-252, BAPTA and QX-314 were applied into the CAlneurons through recording electrodes. Hyperpolarizing current pulses (0.2-0.5nA, 200 ms) were used to inject K-252b (an inhibitor of PKC) and BAPTA (aCa2 chelator) into the neurons. The blockade of afterhyperpolarization and theXIE 131loss of spike accommodation were used as the indices of chelation of theintracellular [Ca2] in the neurons. Depolarizing current pulses (0.1-0.5 nA, 200ms) were employed to aid the release of QX-314 into the cells. The blockade ofthe Na spike and the slow IPSP were used to assess the effects of QX-314 inthe neurons (Nathan et al., 1990).7.7. Collection of endogenous samplesEach rabbit was anaesthetized with 1.5% halothane and its neocortexexposed (see Fig. 4). One 8-mm diameter cup was placed on the neocorticalsurface of each cerebral hemisphere. A ring-shaped stimulating electrode with7-mm loop diameter and a suction line were placed in each cup.Micromanipulators were used to control both suction lines which collectedsamples into a common collection bottle that was kept on dry ice. Oxygenatedmedium (0.1 ml), used for perfusing guinea pig hippocampal slices, was addedinto each cup every 5 mm. At the end of each 5 mm incubation period, theneocortical surface was tetanized (bipolar pulses of 0.5 ms duration at 50 Hz for5s, 30V) and the tetanized neocortical sample (TNS) was collected from thesuction lines. This procedure was repeated until the desired volume of the TNSwas collected. The same procedure was used to collect the untetanizedneocortical sample (UNS) except the collection took place without any tetanicstimulation. In some experiments, a single tetanic stimulation was given to therabbit neocortical surface prior to any collection. After the termination of thistetanus, samples were collected at 5 mm intervals without further tetanicstimulation. These samples will be referred to as TUNS to indicate that theywere from previously tetanized neocortex that was untetanized during thecollection of the sample. Some TNS were collected from rabbits pretreated withMK-801 ([(+)-5-methyl-1 0,11 -dihydro-5H-dibenzo(a, b)cyclohepten-5, I 0-iminemaleate]) (Merck, Sharp & Dohme), a non-competitive NMDA receptor blocker.XIE 132In these rabbits, MK-801 (I jig/kg) was injected i.p. in two divided doses beforethe collection of samples (0.5 ÷ 0.5 jig/kg, with a I h interval between injections).All samples were stored at -80 °C in the freezer immediately after eachcollection. The samples were taken out from the freezer and allowed to melt atroom temperature just before the application of the samples.7.8. Separation of endogenous samplesIn some experiments, the tetanized neocortical samples (TNS) wereseparated into different fractions according to their molecular weight. Fivedifferent fractions were obtained: <3,000, 3,000-10,000, 10,000-25,000, 25,000-50,000 and >50,000 kDa. Centriflo membrane cones type CF 25 and type CF50A from Amicon were used for fractionating molecular weights of 25,000 and50,000 kDa, respectively. The samples were centrifuged at a relative centrifugalforce of 1000 g at 4°C for 30 mm. For the 10,000 mcI. wt fraction, contricon-lOmicroconcentrators from Amicon were used. The samples were centrifuged at arelative centrifugal force of 5000 g at 4 °C for 1 hour. For the 3000 mol. wtfraction, centricon-3 microconcentrators from Amicon were used. Thesesamples were centrifuged at a relative centrifugal force of 7500 g at 4°C for 2hours.7.9. Gel electrophoresisOne-dimensional (1 -D) sodium dodecyl sulfate (SDS) gel electrophoresiswas performed to analyze the presence of peptides in the neocortical samples.The procedure used was similar to that reported by Charriault-Marlangue et al.(1988). Briefly, 100 jil of samples were mixed with 50 jil of sample bufferconsisting of 3% SDS, 10% sucrose, 3 mM EDTA, and 20 jig/mI BromophenolBlue in 30 mM triza hydrochloride (Tris-HCI) buffer (pH 8.0). The mixture wasthen incubated at 37°C for 20 mm to reduce sulfhydryl groups. TheXIE 133Figure 4 Preparation for collection of samples from rabbit neocortical surface invivo.h. --___XIE 134polyacrylamide gel was composed of 95% acrylamide and 5% N,N’-methlenebis-acrylamide with 0.1% SDS. Then, 50 pi samples were applied to the gelslots and electrophoresis was performed at 200 mA in 50 mM Tri-HCI buffercontaining 0.1% SDS pH (8.0). After electrophoresis, Coomassie Blue dye wasused for gel staining. Six protein standards (Sigma) were used: namely, clactalbumin (mol. wt 14.400), trypsin inhibitor (20,000), carbonic anhydrase(30,000), ovalbumin (43,000), albumin (67,000) and phosphorylase B (94,000).The two-dimensional (2-D) gel electrophresis was performed by ProteinDatabase Inc., Huntington Station, NY. Acrylamide gel slab (15%) was used. Inthese experiments, paired UNS and TNS were run with a sample population ofn=4.7.10. Extracellular recordingsPopulation spike and field EPSPs were recorded in the CAl cell bodylayer and the dendrites area, respectively, in response to 0.1-0.2 Hz stimulationof the stratum radiatum at the CAI-CA2 junction with a bipolar stimulatingelectrode (Fig. 5 & 6). The amplitude and slope of the field potentials wererecorded and analyzed with a DATA 6000 waveform analyzer. The stimulusstrength was set so that the control population spike and field EPSP were about50% of the maximum size, leaving room for potentiation. The control responseshad to be stable for 20 to 30 mm before any drug applications or tetanicstimulation. The amplitudes of control population spike and field EPSP wereusually 1.0-1.5 mV and 0.5-1.0 mV, respectively.7.11. Intracellular recordingsIntracellular responses were recorded in the CAl neurons in response to0.03-0.2 Hz stimulation of stratum radiatum near the CAI/CA2 junctions with abipolar stimulating electrode (Fig. 5 & 6). Only those CAl neurons with stableresting membrane potentials of -55 to -65 mV and input resistances greater thanXIE 13525 were used for these studies. The input resistances of neurons weremonitored by injecting rectangular hyperpolarizing current pulses (0.1-0.2 nA,200ms) through the recording electrode and measuring the plateau of thechange in the membrane potential. The membrane voltage responses todifferent rectangular current commands (0.05-0.75 nA of hyperpolarizing and0.05-0.1 nA of depolarizing current pulses of 200 ms) were measured to plot thecurrent-voltage curves. The slope and amplitude of EPSPs and the amplitude ofIPSPs were recorded and analyzed using a DATA 6000 waveform analyzer. Thestimulation strengths were adjusted so that the control EPSPs or/and IPSPswere about half of their maximum size, leaving adequate room for potentiation.The control responses had to be stable for at least 20 mm prior to any drugapplication or tetanic stimulation. The amplitudes of control responses wereusually 5-10 mV for EPSPs and 2-5 mV for IPSPs. If drugs changed themembrane potential, DC currents were sometimes injected into the neurons tobring the membrane potential back to the control level when the records ofEPSPs and IPSPs were taken so that comparisons could be made between thecontrol and test synaptic responses at similar membrane potential.7.12. Induction of LTPLTP could be induced by tetanic stimulations or single stimulation pairedwith depolarization of the neurons. Repetitive direct injections of depolarizingcurrent pulses (3-4 nA, 100-200 ms) into the CAl neurons paired withstimulation (0.2 Hz) of the stratum radiatum could induced LTP (Sastry et al.,1986). The induction of this LTP usually required the presence of picrotoxinin inthe perfusing medium. In the present studies, tetanic stimulations were used toinduce LTP in most of the experiments (Fig. 6 & 7). The following three types oftetanic stimulations were used to induce LTP; (1) 400 Hz for I sec; (2) 100 Hzfor 5 sec; (3) two trains of 100 Hz for I sec, 20 sec interval. During tetanicXIE 136stimulations, the stimulation strength was maintained the same as the controlone unless otherwise stated.7.13. Data analysisPopulation spike, field EPSP, intracellular EPSP and IPSPs wererecorded and analyzed in the DATA 6000 wavform analyzer. The initial slopesof the field EPSP and intracellular EPSP were computed by the DATA 6000waveform and expressed as mV/msec. The amplitude of population spike wasmeasured (Fig. 6). The amplitudes of EPSPs and IPSPs were the lengthsmeasured from the baseline to each peaks of the responses (Fig. 6). Individualrecords in each experiment were quantitied by averaging 4-8 consecutiveresponses. Individual records were taken every 5 to 10 mm throughout eachexperiment. All the control records were averaged. The individual control andthe post-drug or post-tetanic responses were then normalized as a percentageof this averaged record. Data were displayed in graphs or histograms, andexpressed as mean ± S.E.M. The following statistical tests were employed toanalyze the data. Paired Student’s t-test was used to compare two correlatedsamples such as the pie-drug and the post-drug responses. Two-tailed testswere used unless otherwise stated. For multisample comparisons in which onlya single criterion was involved (e.g. mean of EPSP slope), one-way ANOVA wasused. Duncan’s multiple comparison test was applied as “a posterior” test, ifone-way ANOVA first rejected a multisample hypothesis of equal means. The Pvalue was chosen arbitrarily at <0.05 for all the statistical tests.8. RESULTS8.1. LTP and IPSPsStimulation of the stratum radiatum which evokes EPSP5 in the CAlneurons of the hippocampus, also induces the fast and slow inhibitorypostsynaptic potentials (IPSP5) mediated through GABAA and GABAB receptors,øcXIE 137Figure 5 Schematic illustration for positioning of stimulating and recordingelectrodes in hippocampal slices.Extracellular and intracellular responses in the CAl area were evoked with thestimulating electrode (STIM) placed in the stratum radiatum near the CAI-CA2junction. Extracellular EPSPs and population spike were recorded with theelectrodes positioned in the apical dendritic area (REC2) and the cell body layer(RECI) of CAl pyramidal cells, respectively, Intracellular EPSP and IPSPs wererecorded with the electrode impaled in the soma of the CAl neurons (REC3).REC (1)I?EC (2)30mm 60mmFigure 6 Representative field potentials and LTP in hippocampal CAl area.In A, population spikes recorded in CAl cell body layer in response tostimulation of the stratum radiatum before and after tetanic stimulation areillustrated. The measurement of population spike amplitude is shown in thecontrol record. Note the increase in the amplitude of post-tetanus records at 5,30 and 60 mm represents LTP of the population spike. In B, the control andpost-tetanus extracellular EPSPs are illustrated. Note the positive spike (*)shown in the 5 mm post-tetanus record represents a spike recorded in the apicaldendritic area. The increase in the slope of extracellular EPSPs after tetanusindicates the occurrence of LTP of extracellular EPSPs.ControlXIE 138Tet 5 mmVAB2mV20 msControl TetV5 ruin 60 mmJ0.5mV20 msXIE 139AControl 60 mm post4etanus5mV200 msB1j 5mVlOmsFigure 7 Characteristics of intracellular synaptic potentials and LTP inhippocampal CAl cells.In A, intracellular EPSP, fast and slow IPSPs evoked by stimulation of thestratum radiatum were recorded in the CAl pyramidal neurons. In B, the EPSPsare shown in slow recording speed. Amplitude of EPSP, fast and slow IPSPs ismeasured from the baseline to the peak of each response. Note that theincrease in the amplitude or slope of EPSP after tetanus indicates LTP ofintracellular EPSP. Evaluation of LTP of fast and slow IPSP is based on thechange in amplitude of these responses. Records in A and B were obtainedfrom the same neuron with a resting membrane potential of -63 mV.XIE 140respectively (Fig. 7a). Tetanic stimulation of the stratum radiatum is known toinduce LTP of the EPSP in the CAl neurons of the hippocampus (Schwartzkroinand Wester, 1975). Transient depression or no change in IPSPs after a tetanicstimulation has been reported (McCarren and Alger, 1985; Thompson andGahwiler, 1989). Inconsistent long-term changes of the IPSPs after tetanicstimulation have been reported by others (Misgeld et aI., 1979; Abraham et al.,1987; Taube and Schwartzkroin, 1988). Following a tetanic stimulation of thestratum radiatum, IPSPs have been shown to be unchanged, increased, ordecreased (Abraham et al., 1987). Most of these studies were focused onchanges in the fast IPSP. In the present studies, experiments were conducted toexamine whether long-term changes in the IPSPs, especially the slow IPSP,occur following tetanic stimulations. Whether such changes in the IPSPs affectLTP of the EPSP was also tested.8.1.1. Tetanic stimulation and IPSPsIntracellular EPSP, fast IPSP and slow IPSP in the CAl neurons wererecorded with electrodes containing 4 M potassium acetate in response to thestimulation of the stratum radiatum at 0.03-0.1 Hz. The IPSPs would get smallerat frequencies higher than 0.1 Hz (Ben-An et al., 1979; Davies et al., 1990). Incontrol experiments, the EPSP, the fast and slow IPSPs were stable whenevoked at 0.03-0.1 Hz and recorded for 60 mm (responses at 60 mm after thestart of recording were expressed as a percentage of the averaged record takenduring the first 10 mm: EPSP: 105.15 ± 3.10; fast IPSP: 92.75 ± 6.67; slow IPSP:97.87 ± 6.67; n=6).After obtaining stable control responses for 20 mi tetanic stimulation(400 Hz for Is; or 2 trains of 100 Hz for I s, 20 s interval) was applied to thestratum radiatum. Since the two protocols of tetanic stimulation induced similarresults, data obtained from different tetanic stimulations were pooled together forXIE 141discussion in this and the subsequent sections. Tetanic stimulation induced LTPof the EPSP and the fast IPSP, but not of the slow IPSP in the CAl neurons (Fig.8B &C, n=8). The potentiation of the fast IPSP began at the first post-tetanicrecord and the enhancement persisted for at least 30 mm when the recordingwas stopped. In three experiments in which the post-tetanus responses werefollowed for 60 mm after tetanic stimulation, the potentiation of the EPSP and thefast IPSP was present throughout the 60 mm period. In slices exposed topicrotoxinin (20 j.tM), a tetanic stimulation produced LTP of the EPSP, butslightly depressed the slow IPSP (Fig. 9, n=8). The LTP of the EPSP was largerin the picrotoxinin-exposed slices than in the control slices (Fig. 8 & 9). Theseresults indicate that the tetanic stimulation of the stratum radiatum not onlyinduces LTP of the EPSP but also causes long-term changes in the fast IPSPs inCAl neurons. It is possible that these long-term changes in the IPSPs maymodify the expression of LTP of the EPSP.8.1.2. Effects ofCa2 chelator on IPSPsTetanic stimulation is known to induce a Ca2 influx through membranechannels coupled to NMDA receptors on the CAl neurons (Collingridge et al.,1983; Huang et al., 1988). The Ca2 influx is believed to trigger a series ofchanges which are involved in the induction of LTP of the EPSP (Lynch et al.,1983; Malenka et al., 1988). A rise of intracellular Ca2 concentration wassuggested to destabilize the phosphorylation of the GABAA receptor complexwhich decreases the fast IPSP (Chen et al., 1990). Experiments were, therefore,conducted to examine whether the tetanus-induced Ca2 influx leads to longterm changes in the IPSPs using the Ca2 chelator, BAPTA.Intracellular EPSP, fast IPSP and slow IPSP were recorded withelectrodes containing 4 M potassium acetate and 0.1 M BAPTA. BAPTA is afast Ca2 chelator. BAPTA was injected into the CAl neurons by applyingXIE 14220 zM CQX + 40 pM APV 50 pM Piaotoxinin 500 pM COP 35348A__Ba b a+b I a+bControl Post-tet200fl’Is SmV250 —O 0 EPSP-L fIPSP loomsC) 200 0sIPSPT*o 0O 150T 1(/) Th- 4—* _*_Qo jjjl “t i —k-,U) D—..._, T rg 100 —------. IA500 10 20 30 40 50Time (mm)Figure 8 Pharmacologically isolated components of intracellular synapticresponses and the effects of tetanic stimulation on these components.In A, an EPSP followed by a fast IPSP and a slow IPSP recorded in the CAlcells in response to stimulation of the stratum radiatum is illustrated in thecontrol record. The EPSP was abolished by 40 tM APV and 20 jiM CNQX. Thefast IPSP was blocked by 50 jiM picrotoxinin, and the slow IPSP was inhibited by500 jiM CGP-35348. The membrane potential of this neuron was -60 mV. In B,changes in the EPSP and the fast and slow IPSPs induced by tetanic stimulationare shown. The record in Bb was taken 20 mm after tetanic stimulation. Theresting membrane potential of the cell was -61 mV. In C, graphs obtained from 8neurons illustrate the effects of tetanic stimulation (arrow) on the EPSP, the fastand slow IPSPs. Note that the increase in the EPSP and the fast IPSP, but notin the slow IPSP, occurred after tetanic stimulation. All points on the graphs inthis figure and subsequent figures represent the mean ± S.E.M.. Recordingswere made from one neuron per slice. Asterisk (*) indicates that post-tetanicresponses are significantly different from the control responses using one-wayANOVA and Duncan’s multiple comparison test. The P value was chosenarbitrarily at <0.05.XIE 143AaC0nW0101 bpOSttettet a+blOmV200 msB2001501000C0C)0CDCi)CD(I)C00U)o sIPSP• EPSPII IIA0 10 20 30 40 5050Time (mm)Figure 9 Changes in EPSP and slow IPSP induced by tetanic stimulation in thepresence of picrotoxinin.Changes in the EPSP and the slow IPSP caused by tetanic stimulation (arrow)are shown in A & B. The fast IPSP was blocked by 20 jtM picrotoxinin in theseexperiments. A was from an individual experiment. Post-tetanic responses wererecorded 20 mm after tetanic stimulation. The resting membrane potential of theneuron in A was -65 mV. The graphs shown in B were from data obtained from8 neurons. Note that tetanic stimulation only induced LTP of the EPSP but notof the slow IPSP.XIE 144hyperpolarizing rectangular current pulses (0.2-0.5 nA, 200 ms) through therecording electrodes for 10 to 20 mm. The blockade of after-spikehyperpolarization and the loss of spike accommodation were used as the indicesof chelation of intracellular Ca2 in the CAl neurons (Fig. 10) (Strom, 1987c).Intracellular injection of BAPTA into the CAl neurons did not change the controlEPSP, fast and slow IPSPs. However, tetanic stimulation of the stratumradiatum did not induce LTP of the EPSP in BAPTA-injected CAl neurons (Fig.10, n=6). In contrast, the EPSP was slightly depressed after tetanic stimulationin 2 of 6 BAPTA-injected neurons. LTP of the fast IPSP and the slow IPSPoccurred in these CAl neurons after tetanic stimulation (Fig. 10, n=6). The LTPof the fast IPSP was significantly greater in BAPTA-injected CAl neurons (Fig.lOB) than in control neurons (Fig. 8C). Like the potentiation of the fast IPSP, thepotentiation of the slow IPSP, which did not occur in control CAl neurons, beganat the first post-tetanic record and was present until the experiment wasterminated.The slow IPSP was sometimes difficult to estimate accurately because thelargely enhanced fast IPSP overlapped with the early phase of the slow IPSP.Therefore, some experiments were performed in slices exposed to picrotoxininwhich completely blocked the fast IPSP. When picrotoxinin was present in theperfusing medium throughout the experiments, stimulation of the stratumradiatum evoked only EPSP and slow IPSP. Tetanic stimulation induced LTP ofthe slow IPSP but not of the EPSP in the BAPTA-injected neurons (Fig. 11, n=6).The input resistance and the resting membrane potential of the CAlneurons were not significantly changed during LTP of the fast and the slowIPSPs. The equilibrium potentials of the fast and the slow IPSPs measured 20mm post-tetanus were -70 to -75 mV and -90 to -95 mV, respectively (n=4, Fig.15). These equilibrium potentials were not significantly different from the1300.6nA400 insXIE 145Figure 10 Effects of tetanic stimulation on evoked synaptic responses inBAPTA-injected neurons.Changes in EPSP, fast and slow IPSPs caused by tetanic stimulation (arrow) inneurons injected with BAPTA are illustrated in A & B. A shows the records of anindividual experiment. The post-tetanic responses were recorded 20 mm aftertetanic stimulation. The control and post-tetanic responses recorded with fastand slow speed were superimposed separately. The resting membrane potentialof the neuron in A was -58 mV. In B, data obtained from 6 neurons are shown.Note that tetanic stimulation caused LTP of the fast and slow IPSPs butprevented LTP of the EPSP in neurons injected with BAPTA. C shows the abilityof a depolarizing pulse to generate action potentials in CAl neuron before (Ca)and after (Cb) BAPTA injection. Note that chelation of postsynaptic free Ca2reduced the spike accommodation in the CAl neurons.a+bPost15mV200 ins____J5my50 inso EPSPL fIPSPC sIPSPABCa°— 2500C00 2000150C!)CG)100I0C!)a)50a0 10 20 30 40 50Time (mm)Before BAPTA injectionbAfter BAPTA injectionXIE 146ABaf\0fl00C0C)0CC/)C0U)C0U)ci)blOmV200 msPost-tet a+b0 sIPSP• EPSP*A0 101501005040 50Figure 11 Effects of tetanic stimulation on EPSP and slow IPSP in BAPTAinjected neurons in the presence of picrotoxinin.Post-tetanic changes of EPSP and slow IPSP in neurons injected with BAPTAare illustrated in A & B. In these experiments, the fast IPSP was blocked bypicrotoxinin (20 ElM). Records from one experiment are shown in A. In B, dataobtained from 6 neurons are shown. The resting membrane potential of theneuron in A was -67 mV, and post tetanic record was obtained 20 mm aftertetanic stimulation. Note that LTP of the slow IPSP but not of the EPSP inneurons injected with BAPTA occurred after tetanic stimulation.20Time30(mm)XIE 147equilibrium potentials of the control IPSPs. It appears that the potentiation ofIPSPs was not due to changes in the equilibrium potentials of IPSPs.The results are in agreement with those in literature that chelation ofpostsynaptic intracellular Ca2 prevents the induction of LTP of the EPSP(Lynch et al., 1983; Malenka et al., 1988). These results also confirm the findingreported by Morishita and Sastry (1991). Morishita and Sastry suggested thatthe Ca2 influx induced by tetanic stimulation suppresses the induction of LTP ofthe IPSPs and that the suppression of the IPSPs may lead to a better expressionof LTP of the EPSP in the CAl neurons. LTP of the slow IPSP occurs aftertetanic stimulation only if the influx Ca2 is sufficiently chelated.8.1.3. Effects of protein kinase C inhibitor on IPSPsTetanic stimulation induces Ca2 influx through the NMDA channels. Therise of intracellular Ca2 concentration can trigger activation of several proteinkinases which are involved in the LTP process. Protein kinase C (PKC) is oneof the most studied protein kinases involved in LTP. Tetanic stimulationincreases PKC activity (Otani et al. 1992), presumably through the increase ofpostsynaptic Ca2 concentration. Activation of PKC is believed to be requiredfor the induction and/or the maintenance of LTP (Lovinger et at., 1987; Malinowet al., 1989; Malenka et al., 1989). Drugs that increase the activation of PKCdecrease the GABAB receptor-mediated slow IPSP in CAl neurons (Baraban etal., 1985; Dutar and Nicoll, 1988). It would be of interest to determine whetherthe increase in PKC activity leads to long-term changes in the IPSPs followingtetanic stimulation.PKC inhibitors like H-7, sphingosine and polymyxin B have been reportedto block the induction of LTP (Malinow et al., 1988, 1989). These drugs are lipidsoluble and can cross cell membranes making it difficult to decide their sites(presynaptic or postsynaptic) of the action. A potent PKC inhibitor, K-252b, wasXIE 148used in the present experiments. K-252b cannot readily cross cell membranes(Kuroda et al. 1992). This property of K-252b was confirmed by the followingexperiments. In one set of experiments, K-252b (5 tM) was added to theextracellular medium for 15-30 mm and the stratum radiatum was tetanized 5mm after the start of drug application. K-252b did not block the induction of LTPof the EPSP in these experiments (Fig. 12B, n=6). The LTP induced in thepresence of K-252b was not significantly different from the control LTP (Fig. 12C, n=6). In another set of experiments, K-252b was injected into the CAlneurons, through electrodes containing 4 M potassium acetate and 5 i.tM K252b, prior to the tetanic stimulation (by applying hyperpolarizing rectangularcurrent pulses, 0.2-1 nA and 200 ms for 2 mi. K-252b did not change themembrane potential, the input resistance and the control evoked synapticresponses. In these experiments, K-252b clearly blocked the induction of LTP ofthe EPSP (Fig. 12A, n=6). These results demonstrate that postsynaptic PKCactivity is critical for the induction of LTP of the EPSP. Therefore, injections ofK-252b into the CAl neurons was used in subsequent experiments.In K-252b-injected CAl neurons, tetanic stimulation did not induce LTP ofthe EPSP but caused LTP of the fast and the slow IPSPs (Fig. 13, n=1 6). LTP ofthe fast IPSP was significantly greater in K-252b-injected CAl neurons than incontrol neurons. The potentiation of the fast and the slow IPSPs was seen in thefirst post-tetanic record. In the slices exposed to picrotoxinin (20 .iM), tetanicstimulation clearly induced LTP of the slow IPSP, but not of the EPSP (Fig. 14,n=6). The equilibrium potentials of the control and the 20 mm post-tetanic fastand slow IPSPs were not significantly different (n=4, Fig. 15).Chelation of intracellular Ca2 or inhibition of postsynaptic PKC activityblocked the induction of LTP of the EPSP but increased LTP of the fast and slowIPSPs. These results suggest that the increase in postsynaptic PKC activityFigure 12 Actions of K-252b on the induction of LTP of the EPSP in CAlneurons.In A, K-252b was injected into CAl neurons through recording electrodescontaining 5 1iM K-252b. Effects of intracellularly injected K-252b on LTP of theEPSP induced by tetanic stimulation (arrow) are shown (n=7). A represents thetime when K-252b was injected. Note that intracellularly injected K-252b did notaffect control EPSP but blocked the induction of LTP of the EPSP. In B, theeffects of extracellular application of K-252b (5 jtM, the long horizontal barabove the abscissa) on LTP are illustrated (n=6). In the presence of K-252 inthe medium, tetanic stimulation (arrow) induced LTP of the EPSP which is notsignificantly different from the control LTP obtained without the presence of K252b which is shown in C (n=5).200150A0C0C-)0CO00Cr,0wBXIE 14925010200C-)01500CO0(F)0Li5010 40 5020 30Time (mm)C0 10 20 30Time (mm)40 502502001500C0C-)0aU)a(1)LU500 10 20 30 40 50Time (mm)0I.C000DU)0ci)U)C0aC’)ci)Time (mm)Figure 13 Actions of K-252b on changes in EPSP, fast and slow IPSPs inducedby tetanic stimulation.Post-tetanic changes of the EPSP, the fast and slow IPSPs in neurons injectedwith K-252b are illustrated in A & B. Tetanic stimulation (arrow) induced LTP ofthe fast and slow IPSPs but not of the EPSP. A shows the records from oneexperiment. B illustrates data obtained from 8 neurons in graphs. The restingmembrane potential of the neuron in B was -60 mV and post-tetanic record wasobtained 20 mm after tetanic stimulation.ABXIE 150a+b5mV100 msa b a+bCoMroI Post—TeLJ5mV200ms2500 EPSPfIPSPc::i sIPSP200150100 I—50A0 10 20 30 40 50XIE 151Aa4\cor00C0(-)0ciC’)C(IDC0(I)b\PPT\ThS mYlOOmsB0 sIPSP• EPSP15010050J 10—0I iIAI I I I0 10 20 30 40 50Time (mm)Figure 14 Effects of K-252b on post-tetanic changes of EPSP and slow IPSP inthe presence of picrotoxinin.Post-tetanic changes of the EPSP and the slow IPSP in neurons injected with K252b are shown in A & B. In these experiments, the fast IPSP was blocked bypicrotoxinin (20 iiM). A shows the records of an individual experiment, and Billustrates data obtained in 8 neurons injected with K-252b. Tetanic stimulation(arrow) induced LTP of the slow IPSP but not of the EPSP in neurons injectedwith K-252b. The resting membrane potential of the neuron in A was -67 mVand post-tetanic record was obtained 20 mm after tetanic stimulation.XIE 152C Membrane Potential Cmv)>E0(1)..—10z fIPSP (pre—tet)A fIPSP (post—tet)o sJPSP (pre—tet)• sIPSP (post—tet)Figure 15 Equilibrium potentials for fast and slow IPSPs in control, BAPTA- andK-252b-injected neurons.A, B, and C show equilibrium potentials for the fast IPSPs before and aftertetanic stimulation, and for the slow IPSPs before and after tetanic stimulation incontrol (A), BAPTA- (B) and K-252b- (C) injected neurons. The post-tetanicdata were obtained 20 mm after tetanic stimulation. Note that equilibriumpotentials for the fast and slow IPSPs in control, BAPTA- and K-252b injectedneurons did not significantly change after tetanic stimulation. Equilibriumpotentials for the fast and slow IPSPs were approximately -70 to -75 mV and -90to -95 mV, respectively.A Membrane Potential Cmv)—90 —80B Membrane Potential (mV)—100 —90 —80 —70 —60 —500 —..>E0C,)-5Control—10o>E0C,.)-5BAPTA—100 —90 —80 —70 —600K—252bXIE 153Table I Effects of intracellular injection of BAPTA and K-252b on post-tetanicEPSP, fast and slow IPSPsControl neurons BAPTA-injected K-252b-injectedneurons neuronsEPSP 138.8 ± 10.0* 94.2 ± 11.4 99.8 ± 5.4fIPSP 123.6 ± 8.4* 179.5 ± 17.0* 166.0 ± 7.8*sIPSP 98.5 ± 11.3 155.6 ± 15.5* 155.0 ± 12.8*No. of neurons 8 6 16Data shown in this table and tables 2, 4 & 5 are responses (in height) 20 mmpost-tetanus as a percentage of control record (100%) and represent the mean ±S.E.M. Asterisk (*) indicates that the post-tetanus responses are significantlydifferent from the control responses at a P-value of <0.05 using paired Student’st-test in this table and tables 2, 4 & 5.Table 2 Effects of intracellular injection of BAPTA andEPSP and slow IPSP in picrotoxinin-treated slicesK-252b on post-tetanicControl neurons BAPTA-injected K-252b-injectedneurons neuronsEPSP 149.8 ± 9•7* 105.0 ± 12.0 99.9 ± 12.2sIPSP 73.5 ± 12.6* 125.9 ± 10.1* 126.4 ± 10.1*No. of neurons 8 6 6Table 3 Effects of BAPTA and K-252b on Dost-tetanic EPSP durationControl neurons BAPTA-injected K-252b-injectedneurons neuronswithout with PTX without with PTX without with PTXPTX PTX PTXEPSP 108.8± 114.0± 67.3± 103.3± 79.8± 87.5±duration 5.0 3.9 7.2 4.9 9.1 6.8No. of8 8 6 6 16 6neuronsIn slices exposed with picrotoxinin (PTX, 20 tM), fast IPSP was blocked.XIE 154depresses LTP of the IPSPs and allows a better expression of the LTP of theEPSP. LTP of the slow IPSP occurred only if postsynaptic PKC activity wasinhibited or if intracellular Ca2 was chelated. Tables I & 2 summarize the long-term changes of the IPSPs induced by tetanic stimulation and the effects ofBAPTA and K-252b on the IPSPs in CAl neurons.8.1.4. Effects of IPSPs on EPSPIt is logical for me to think that the blockade of the fast IPSP cansignificantly increase the height and duration of the EPSP. In most BAPTA- orK-252b injected CAl neurons, the potentiation of the fast IPSP induced by thetetanus is associated with a reduced height and duration of the EPSP (Fig. 10 &13, Table 3). In control neurons, the slow IPSP was slightly depressed or notchanged by tetanic stimulation. The potentiation of the post-tetanic slow IPSPoccurred only in the BAPTA- and K-252b-injected neurons. Whether a blockadeof the potentiation of the slow IPSP would change the shape of the EPSP wasexamined. In K-252b- (Fig. 16, n=8) and BAPTA- (Fig. 17, n=8) injected CAlneurons, phaclofen (0.5-1 mM, applied for 10 mm), a GABAB antagonist,blocked the slow IPSP when applied before and after tetanic stimulation. Theblockade of the potentiated slow IPSP by phaclofen is associated with a moresignificant change in the shape of the EPSP compared to the pre-tetanic slowIPSP (Fig. 16 & 17). The results indicate that the potentiated IPSPs significantlydistort the shape of the EPSP in BAPTA- and K-252b-injected CAl neurons.8.1.5. Effects ofAPV on IPSPsMultiple trains of tetanic stimulation (kindling) induce epileptic activity inCAl neurons associated with NMDA receptor-mediated suppression of theIPSPs. APV, a NMDA antagonist, blocks the kindling-induced epileptic activityas well as the suppression of the IPSPs (Stelzer et al., 1987). It is possible thatkindling causes a large Ca2 influx through NMDA receptor-gated channels andXIE 155a+b c+dEPSP sIPSPAa b C dControl Phoclofen Post—Tet Phoclofen5 mV200msB m’sp2500200C-,0150000• 1000000C)0Figure 16 Effects of phaclofen on EPSP in K-252b-injected neurons before andafter tetanic stimulation.Phaclofen (500 aiM) was applied to slices before and 20 mm after tetanicstimulation. A shows the records of an individual experiment. In B, dataobtained from 8 neurons are summarized in histograms. Note that before tetanicstimulation, phaclofen suppressed the slow IPSP with no significant effect on theEPSP in K-252b injected neurons. Application of phaclofen 20 mm after tetanicstimulation not only depressed the slow IPSP but also enhanced the amplitudeof the EPSP in these neurons. Records shown in B were from the same neuronC C2.2 T2— Q 0c (noo-(jo-a-0C CC) C) C)CO Cl) 00 .C 0 .CC)0Q0-C CC) C) C)- o’SCO U5 00 .C 0 .CC) 0-0-0-with a resting membrane potential of -63 mV.XIE 1560C0000aVC00.U,VaControlb PhoclolenCPost—letjPhenFigure 17 Effects of phaclofen on the EPSP in BAPTA-injected neurons beforeand after tetanic stimulation.A and B show effects of phaclofen (500 p.M) on the EPSP in neurons injectedwith BAPTA. Phaclofen was applied before and 20 mm after tetanic stimulation.In these experiments, slices were exposed to picrotoxinin (20 1iM) to block thefast IPSP. The records obtained from one individual experiment are shown in A,and data from 8 neurons are expressed in histograms in B. The membranepotential was -68 mV. Note that before tetanic stimulation, phaclofen blockedthe slow IPSP without significantly affecting the EPSP. After tetanic stimulation,phaclofen not only suppressed the slow IPSP but also increased the amplitudeof the EPSP.ABa+bmVc+d 200EPSP sIPSP300250200150100500C C C C— v_u — U UCO U, 0 C 0O o: o.co0aa.a.XIE 157subsequently triggers the activation of protein kinases. The suppression of theIPSPs induced by kindling may be due to the increase in postsynaptic Ca2concentration and the activation of PKC. In the present experiments, APV (40 iM) was applied to the picrotoxinin-free (Fig. 18) or the picrotoxinin-treated (Fig.19) slices 5 mm before tetanic stimulation and the drug application terminatedafter the tetanic stimulation. In control CAl neurons, tetanic stimulation failed toinduce LTP of the EPSP and caused a slight depression of the slow IPSP in thepresence of APV (Fig. 1 8A, I 9A). In BAPTA- and K-252b- injected neurons, LTPof both the EPSP and the slow IPSP did not occur after tetanic stimulation inpresence of APV (Fig. 1 8B & C, 198 & C). In slices not exposed to picrotoxinin,tetanic stimulation in presence of APV caused a LTP of the fast IPSP while APVprevented LTP of both the EPSP and the slow IPSP in control, BAPTA- or K252b-injected neurons (Fig. I 8A, B & C). This LTP of the fast IPSP is smaller inamplitude than the LTP obtained in BAPTA- and K-252b-injected neurons testedin the absence of APV. These results indicate that APV not only blocks theinduction of LTP of the EPSP but also suppresses the potentiation of the IPSPs.8.1.6. IPSPs in the absence of glutamatergic transmissionUnder normal conditions, IPSPs evoked by tetanic stimulation of thestratum radiatum consist of both polysynaptic and monosynaptic components.The polysynaptic component of IPSPs sometimes was more prominent than themonosynaptic one although the stimulating electrode was placed in the stratumradiatum near the recording electrode. When glutamatergic transmission in thehippocampus was blocked by APV and CNQX, monosynaptic IPSPs could berecorded (Fig. 20A) (Davies et al., 1990). Under these conditions, changes inthe monosynaptic IPSPs after tetanic stimulation were examined. When APV(40 1iM) and CNQX (20 jiM) were present in the perfusing medium throughoutthe experiment, tetanic stimulation induced a LTP of the fast IPSP andC’)0C’,0C,)a)c,)CU-CC.)C)CCUa)4-,U)0a0>00U)0LiicoU)Li..015000U)00c-n00U)•0D0EXIE 158ABCA’B’A’B’A’B’I a+bLZ: ‘2mV200 ms150 ControlC0()00U) Uo 0-o 100 —Ô-—0--’.en -9:_____A00)-ø2- 50 I< 0 10 20 30 40 50Time (mm)b\:> LVmV200 mso BAPTA150C001oU) 0/100 —O—0U, 09:A0)•0250< 0açI10010 20 30 40 5CTime (mm)b>‘jSrnV200 mUK—252b1 1—o—-o I Ib1*T *T0I ‘/1-o 1A0 10 20 30 40 50Time (mm)XIE 159Figure 18 Effects of APV on post-tetanic changes of IPSPs.Effects of APV on postsynaptic changes of the EPSP, fast and slow IPSPs areillustrated in AB’ (control neurons, n=5), BB’(n=BAPTA-injected neurons, n=6)and CB’(K-252b-injected neurons, n=6). Tetanic stimulation (arrow) was given inthe presence of APV (40 jiM, the horizontal line above the arrow). Individualrecords obtained from a control neuron (AA’), a BAPTA-injected neuron (BA’)and a K-252b injected neuron (CA’) are demonstrated. Post-tetanic responseswere recorded 20 mm after tetanus. Note that tetanic stimulation induced LTP ofthe fast IPSP but not of the EPSP and the slow IPSP in the presence of APV. Insome BAPTA-injected neurons (BA’) and K-252b-injected neurons (CA’), theEPSP was suppressed after tetanic stimulation.Figure 19 Effects of APV on post-tetanic changes of slow IPSP in the presenceof picrotoxinin.The ability of APV to block post-tetanic changes of the EPSP and the slow IPSPis illustrated in A, B & C. In these experiments, slices were exposed to 20 iMpicrotoxinin. In the presence of APV (the horizontal bar above the abscissa),tetanic stimulation (arrow) did not cause LTP of the EPSP and the slow IPSP incontrol (A, n=6), BAPTA- (B, n=8) and K-252b- (C, n=8) injected neurons.A Control0C0(-)00(n0C)C00U,C)50XIE 160BAPTA150 Q sIPSP• EPSPI.100 I 1 i 1 .—.—•jMi TIA500 10 20 30Time (mn)B0C0C-)00Co0‘a)U,C00anC)K—252b0 10 20 30 40 50lime (mira)40 50o sIPSP• EPSPC1500C0C)0o iooVa0U)anC00.U)500 10 20 30 40 50Time (mira)XIE 161an insignificant change in the slow IPSP of the control neurons (Fig. 20B, n=6).Similar results were obtained in BAPTA-injected (n=6) and K-252b-injectedneurons (n=6) (Fig. 20C & D; see table 4). The tetanus-induced potentiation ofthe fast IPSP obtained in APV- and CNQX-treated slices was much smaller thanthe one observed in drug-free slices. In picrotoxinin-treated hippocampal slices,the isolated slow IPSP did not significantly change after tetanic stimulation incontrol neurons or BAPTA- and K-252b-injected neurons (Fig. 21, n=6 for eachgroup; see table 5). These results indicate that long-term changes in the IPSPsfollowing tetanic stimulation require glutamatergic transmission.8.2. Effects of somatostatin on GABAergic inhibition and LTPSomatostatin and GABA are co-localized in some interneurons of the CAlarea of the hippocampus. If the peptide and the amino acid are co-releasedfrom the same presynaptic terminals, interactions between the actions ofsomatostatin and the GABAergic responses on CAl neurons may occur.Somatostatin has been reported to depress GABA receptor-mediated IPSP5 inthe hippocampal CAl neurons (Scharfman and Schwartzkroin, 1989). Undernormal conditions, changes in GABAergic inhibition can affect the induction ofLTP of the EPSP (Wigstrom and Gustafsson, 1983; Davies et al., 1991; Mott andLewis, 1991). The present experiments were conducted to examine the actionsof somatostatin on the GABAergic synaptic transmission and LTP.8.2.1. Effects of somatostatin on CA I neuronsDifferent concentrations of somatostatin-14 (SS, 0.2-5 jiM) were appliedto determine the appropriate concentration to induce significant and consistenteffects in CAl neurons. A concentration of 2 p.M SS, applied to slices for 2 mm,was found to have reliable effects on the membrane potential, input resistanceand evoked synaptic responses (n=8). This concentration was therefore used inthe quantitative studies.XIE 162A Control neurona Control b CNQX+APV a+bB Control neurona Control b Post-tet a+bC BAPTA-injected neurona Control b Post-tet bD K-252b-injected neurona Control b Post-tet200 msFigure 20 Post-tetanic changes of IPSPs in the presence of APV and CNQX.Monosynaptic IPSPs were obtained in the presence of APV (40 jIM) and CNQX(20 iiM). In A, the activation of monosynaptic IPSPs in CAl neurons isillustrated. In B, C & D, post-tetanic changes of monosynaptic IPSPs in control(B), BAPTA- (C) and K-252b- (D)-injected neurons are shown. Note that tetanicstimulation induced an increase in the fast IPSP without significantly altering theslow IPSP in these neurons. A and B are from the same neuron with a restingmembrane potential of -59 mV. C and D were from different neurons with aresting membrane potentials of -57 mV and -63 mV, respectively.XIE 163A Control neurona Control b Post-tet a+b----B BAPTA-injected neurona Control b Post-tet a÷bC K-252b-injected5 mVZOOmsa Control b Post-tet÷Figure 21 Post-tetanic changes of monosynaptic slow IPSP.Isolated slow IPSP was obtained in the presence of APV (40 p.M), CNQX (20 p.M) and picrotoxinin (20 p.M). Tetanic stimulation did not cause any significantchanges in the monosynaptic slow IPSP in control (A), BAPTA- (B) and K-252b-(C) injected neurons. A, B and C were from different neurons with restingmembrane potentials of -58, -66 and -61 mV, respectively.XIE 164Table 4 Effects of intracellular injection of BAPTA and K-252b on post-tetanicfast and slow IPSPs in the presence of APV and CNQXControl neurons BAPTA-injected K-252b-injectedneurons neuronsfIPSP 142.6 ± 6.3* 130.8 ± 9•9* 135.5 ± 11.1*sIPSP 110.4 ± 10.9 115.0 ±9.1 112.7 ± 10.1No. of neurons 6 6 6(see the footnote in table 1).Table 5 Effects of intracellular injection of BAPTA and K-252b on post-tetanicslow IPSP in the presence of APV, CNQX and picrotoxininControl neurons BAPTA-injected K-252b-injectedneurons neuronssIPSP 90.5±9.1 88.1 ±9.1 86.2±8.2No. of neurons 8 6 8(see the footnote in table 1).XIE 165Application of SS induced a hyperpolarization of 7.49 ± 0.73 mV (n=16) ofthe hippocampal CAl neurons associated with a decrease in the inputresistance by 12.46 ± 1.10 M (n=16) (Fig. 22A). SS significantly depressed thefast IPSP (response as a percentage of control: 31.50 ± 2.73, n=16) and theslow IPSP (response as a percentage of control: 35.33 ± 2.10, n=16) withoutsignificantly changing the EPSP (response as a percentage of control: 111.2 ±7.83, n=16) (Fig. 22B). These results are consistent with the reports in literature(Pittman and Siggins, 1981) that SS, when applied to the slices, consistentlyhyperpolarizes CAl neurons and reduces the input resistance of the neurons.8.2.2. Effects of somatostatin on GABA receptorsIn order to determine whether SS interacts with GABAA receptors, actionsof SS were examined in the presence of picrotoxinin. In slices exposed topicrotoxinin (20 jiM), SS hyperpolarized the CAl neurons by 7.32 ± 0.73 mV(n=24) and reduced the input resistance by 11.3 ± 1.13 M2 (n=24) (Fig. 23A).SS also depressed the slow IPSP (response as a percentage of control: 36.50 ±2.30, n=24) without significantly changing the EPSP (response as a percentageof control: 105.20±6.90, n=24) (Fig. 23B).To determine whether SS interacts with GABAB receptors, SS wasapplied in the presence of phaclofen. Phaclofen (1 mM), which blocked theGABAB receptor-mediated slow IPSP, significantly reduced the baclofen-inducedhyperpolarization (n=4), but not the SS-induced hyperpolarization (n=16) (Fig.24A & B). In the presence of phaclofen, SS induced a hyperpolarization of 7.58± 0.67 mV and a decrease in the input resistance by 12.6 ± 1.20 M2 (n=16).The current-voltage (l-V) curves showed a decrease in the slope resistance inthe presence of somatostatin (Fig. 24C) and a reversal potential ofapproximately -90 mV for the action of SS. Phaclofen did not change the SSinduced decrease in the slope resistance (Fig. 24C). 2-OH-Saclofen, whichXIE 166blocks both presynaptic and postsynaptic GABAB receptors (Davies et al., 1990;Dutar and NicolI, 1988), did not change the actions of SS on the membranepotential, input resistance and fast IPSP in the CAl neurons (n=4, Fig. 25).These results indicate that SS does not interact with GABAA and postsynaptic &presynaptic GABAB receptors.8.2.3. Effects of somatostatin during the activation of GABABreceptorsBaclofen-induced hyperpolarization is thought to be mediated by GABABreceptors coupled to specific K channels through a pertussis toxin-(PTX)sensitive G protein (Andrade et al., 1986). It has been suggested that SSreceptors are also coupled to K channels through a PTX-sensitive G protein(Schweitzer et al., 1989; Yatani et al., 1990). Whether GABAB and SS receptorsshare the same type of K channels is, however, unclear. It has been reportedin the literature that the hyperpolarizing action of baclofen is suppressed by SS(Twery and Gallagher, 1990). Experiments were conducted to determinewhether the actions of SS would be affected by the activation of GABABreceptors. Baclofen (20 jiM) was applied to the slices prior to the application ofSS. Baclofen induced a prolonged hyperpolarization of 15.5 ± 1.67 mV and adecrease in the input resistance by 20.3 ± 2.12 M2 (n=6) of the CAl neurons.SS was applied to the slices at the peak of the baclofen-inducedhyperpolarization. SS did not induce further changes in the membrane potentialand the input resistance of the CAl neurons (n=6) (Fig. 26A &C). Even if themembrane potential of the neurons was adjusted to the pre-baclofen level byinjecting a depolarizing current, SS did not have any further hyperpolarizingaction (n=6) (Fig. 26B). These results indicate that the activation of GABA8receptors results in a decrease in the action of SS on the CAl neurons.XIE 167Atftfl?lttti;iiib C 120mv4 miiiBakbmY200 msFigure 22 Effects of somatostatin on the membrane potential, input resistanceand evoked synaptic responses in CAl neurons of hippocampal slices.Somatostatin (SS, 2 1iM) was applied for 2 mm. The input resistance wasmonitored by injecting hyperpolarizing current pulses throughout the experiment.A shows a hyperpolarization and a reduction of input resistance caused bysomatostatin. B shows the effects of somatostatin on the EPSP, fast and slowIPSPs. The records of synaptic responses shown were taken during the control(a), at the peak of somatostatin-induced hyperpolarization while the membranepotential was adjusted to the control level by injection of a depolarizing current(b), and during a recovery from the actions of the peptide (c). Note thatsomatostatin suppressed the fast and slow IPSPs without significantly alteringthe height of the EPSP. In this figure and in figure 23, the small upward anddownward deflections are due to the EPSP and the IPSPs, respectively, and thelarger downward deflections are the voltage responses due to thehyperpolarizing intracellular current injections. The resting membrane potentialof this neuron was -60 mV.AXIE 168ss6OmV10 mmbCd___j4mV200 msFigure 23 Actions of somatostatin in the presence of picrotoxinin.In the experiment, the hippocampal slice was superfused with picrotoxinin (201M)-containing medium to block the fast IPSP. A shows a hyperpolarization anda decrease in input resistance of the CAl neuron induced by somatostatin (SS,2 p.M, applied for 2 mm). In B, records of the synaptic transients taken duringcontrol (a), during the hyperpolarization induced by somatostatin (b), during acurrent-clamp of the neuron back to control membrane potential (C), and duringa recovery from the action of the peptide (d). Note that somatostatin depressedthe fast and slow IPSPs, but not the EPSP in the picrotoxinin-treated slices. Theresting membrane potential of this neuron was -62 mV.XIE 169A Baclofen Baclofen Baclofen— b Phaclofen c10 mm 60 mVB SS SS SSb Phaclofen cI %C Current (nA)—1.500 —1.000 —0.500 0.000 0.50050 CD0-3aD—70 a-o0C..CDD—90 9<—110Figure 24 The effects of phaclofen on the hyperpolarizing actions of baclofenand somatostatin in picrotoxinin-treated slices.In Aa: baclofen (20 jiM, I mm) induced a hyperpolarization and a decrease inthe input resistance of the CAl neuron. In Ab, the effects of baclofen on themembrane potential and the input resistance were significantly blocked in thepresence of phaclofen (1 mM). In Ac, 20 mm after the washout of phaclofen,baclofen’s action recovered from the antagonism by phaclofen. B shows asimilar interaction between somatostatin and baclofen; Note that phaclofen didnot block the action of somatostatin. A and B were recorded from the sameneuron with a resting membrane potential of -61 mV. In C, current-voltagecurves for control, somatostatin and somatostatin in the presence of phaclofen,were plotted from a different CAl neuron with the resting membrane potential of-58 mV. Note that the decrease in the slope resistance by somatostatin that wasnot significantly affected by phaclofen. The reversal potential for thehyperpolarizing action of somatostatin was close to -90 mV. Phaclofen alone didnot significantly change the membrane potential and the input resistance of theneuron.o — Controlc- ssA — SS & PhaclofenXIE 170Control 2-OH-Sac 2-OH-Sac & SSZOO msFigure 25 Effects of somatostatin on IPSPs in the presence of 2-OH-saclofen.Effects of 2-OH-saclofen (1 mM) on action of somatostatin are illustratedApplication of 2-OH-saclofen (1 mM) blocked the slow IPSP in the CAl neuron.In the presence of 2-OH-saclofen, somatostatin (2 1iM) still suppressed the fastIPSP. Note that 2-OH-saclofen has been reported to block the presynapticGABAB receptor-mediated paired-pulse depression of the IPSPs in CAlneurons (Davies at al., 1990).XIE 171BC ssBaclofen60 mV10 mmFigure 26 The interactions between baclofen and somatostatin.A shows a hyperpolarization and a reduction in the input resistance induced bysomatostatin (2 p.M). B shows the blockade of the action of somatostatin bybaclofen (20 p.M) recorded from the same neuron as in A. Baclofen induced aprolonged hyperpolarization associated with a decrease in the input resistance.At the peak of the baclofen-induced hyperpolarization, the neuron was current-clamped to the pre-baclofen membrane potential when somatostatin wasapplied. Note that somatostatin did not induce any further hyperpolarization or adecrease in the input resistance of the neuron. C was taken from a differentneuron, there was a lack of effect of somatostatin during the hyperpolarizationinduced by baclofen (20 iiM). The resting membrane potentials of the neuronsin A & B were -57 and -59 mV, respectively.ssBaclofenCXIE 172A BAPTAa__________ _______bSS__130 mV6mmB ss ssSphingosineaFigure 27 Possible role of protein kinase C and postsynaptic intracellular Ca2on hyperpolarizing action of somatostatin.The hyperpolarization induced by somatostatin in control neurons (Aa, Ba), aneuron injected with BAPTA (Ab), and a neuron exposed to sphingosine (30 pM)(Bb). Note neither a chelation of the Ca2 nor an inhibition of PKC interferedwith the hyperpolarizing action of the peptide. Ba and Bb were from the sameneuron. Aa (resting membrane potential: -64 mV), Ab (resting membranepotential: -62 mV) and B (resting membrane potential: -59 mV) were from threedifferent neurons.XIE 1738.2.4. The role ofCa2 and PKC in the actions of somatostatinIt has been suggested that changes in intracellular Ca2 concentrationand PKC activity can affect the GABAB receptor-mediated responses (Barabanet aL, 1985). Whether actions of somatostatin are affected by changes inintracellular Ca2 concentration and PKC activity was examined in the presentstudy.BAPTA was injected into the CAl neurons through recording electrodes,as previously described. SS induced a hyperpolarization of 7.13 ± 0.65 (n=6), adecrease in the input resistance by 12.9 ± 1.33 Mc (n=6), a depression of thefast IPSP (response as a percentage of control: 33.50 ± 2.1, n=6) and asuppression of the slow IPSP (response as a percentage of control: 36.66 ±2.33, n=6) while not significantly affecting the EPSP (response as a percentageof control: 107.00 ± 8.33, n=6), in BAPTA-injected neurons (Fig. 27A). Theeffects of SS in the BAPTA-injected neurons were similar to those in the controlneurons. The results indicate that chelation of postsynaptic Ca2 does not affectthe actions of SS.In slices exposed to sphingosine (30 pM), a PKC inhibitor, SShyperpolarized the CAl neurons by 7.77 ± 0.81 mV (n=4) and reduced the inputresistance by 12.3 ± 1.33 M2 (n=4). SS also depressed the fast IPSP (responseas a percentage of control: 34.33 ± 2.99, n=4) and the slow IPSP (response as apercentage of control: 37.11 ± 3.33, n=4) without significantly changing theEPSP (response as percentage of control: 110.0 ± 10.33, n=4) (Fig. 27B).Sphingosine did not change the membrane potential, input resistance, EPSPand IPSPs in the CAl neurons. The findings suggest that the actions of SS arenot due to changes in PKC activity.XIE 1748.2.5. Effects of QX-314 on the actions of somatostatinQX-314, a derivative compound of lidocaine, has been found to block notonly the GABAB receptor-mediated slow IPSP (Nathan et al., 1990) andhyperpolarization (Andrade, 1991), but also the 5-HT1A receptor-mediatedhyperpolarization in the hippocampal CAl neurons. These effects might be dueto the blockade of a G protein-regulated K channels. Experiments were,therefore, conducted to determine whether QX-314 also affects the actions of SSin CAl neurons.QX-314 was allowed to diffuse into the CAl neurons through recordingelectrodes. Since it usually took 20-30 mm after the penetration of the cell forQX-314 to block the Na spike and suppress the slow IPSP, it was possible toexamine the actions of SS during the early (5-15 mm) and the late (35-65 mm)stages of QX-314 diffusion into the cells. During the early stage of recording,when the Na spike and the slow IPSP were not blocked by QX-314, SS wasable to cause a hyperpolarization of 6.98 ± 0.61 mV (n=12), a reduction in theinput resistance by 10.03 ± 0.93 M12 (n=12) (Fig. 28Aa) and a depression of thefast IPSP (response as a percentage of control: 44.19 ± 3.1, n=12) withoutsignificantly changing the EPSP (response as a percentage of control: 107.23 ±7.81, n=12) (Fig. 28Bb). During the late stage of recording, SS could onlyinduce very small hyperpolarization of 1.21 ± 0.10 mV (n=12). and decrease inthe input resistance by 1.67 ± 0.07 MQ (n=12) (Fig. 28Ab). More interestingly,QX-314 reduced the depressant effect of SS on the fast IPSP (response in SSas a percentage of control: 94.15 ± 5.5, n=12) (Fig. 28Be). The baclofeninduced hyperpolarization of 1.72 ± 0.08 mV (n=8) and decrease in the inputresistance by 1.88 ± 0.07 M2 (n=8) were also much smaller during the latestage of recording in QX-31 4-injected cells (Fig. 28Ac). Baclofen, however, stillsignificantly depressed the fast IPSP (response in baclofen as a percentage ofXIE 175control: 22.13 ± 1.80, n=8) during the late stage of recording (Fig. 28Bf). Inthese cells, the effect of SS on the slow IPSP could not be examined becauseQX-314, alone suppressed the slow IPSP. As reported in literature (Nathan etal., 1990), QX-314 by itself sometimes caused a slight increase in the inputresistance and a small depolarization of the neurons. These resultsdemonstrate that QX-31 4 blocks not only the baclofen-induced hyperpolarizationbut also the SS-induced hyperpolarization. However, QX-314 reduces thedepressant effect of SS, but not of baclofen, on the fast IPSP, suggesting a post-synaptic locus of action for SS on the fast IPSP.8.2.6. Effects of somatostatin on the induction of L TPIt is apparent that SS interacts with the GABAergic transmission in theCAl neurons. GABAergic transmission is known to play an important role inmodulating the induction of LTP. Experiments were, therefore, performed toexamine whether SS affects the induction of LTP of the EPSP.After stable control intracellular EPSP and field EPSP were obtained for20 to 30 mm, tetanic stimulation was given to the stratum radiatum during anapplication of SS (2 j.tM for 2 mm). The post-tetanus responses were monitoredfor 30-60 mm. Two protocols of tetanic stimulations (2 trains of 100 Hz for I s,20 s interval; 100 Hz for 0.5 s) were used in these experiments.Since SS can induce a hyperpolarization of the CAl neurons, the first setof experiments was conducted to determine whether SS blocks the induction ofLTP by hyperpolarizing the neurons. In 6 slices, tetanic stimulation (2 trains of100 Hz for 1 s, 20 s interval) in the presence of SS induced LTP of the fieldEPSP (field EPSP slope 30 mm post-tetanus as a percentage of control: 190.75± 20.37, n=6) (Fig. 29). This LTP was not significantly different from the LTP(EPSP slope 30 mm post-tetanus as a percentage of control: 174.00 ± 12.53,n=6) obtained in control experiments in which the tetanic stimulation was givenXIE 176A 8mm 40mm 60mmSS SS Baclofena_______ _____uifl120mv4 mmB 5mm 8mm 20mmc BaclofenabJ 5 mVlOOms35 mm 40 mm 60 mmBaclofeneffFigure 28 Effects of QX-314 on the actions of somatostatinXIE 177Figure 28 Effects of QX-314 on the actions of somatostatin.The responses were recorded with a QX-314 (100 mM)-containing recordingelectrode. QX-314 was passively diffused into the cell through the recordingelectrode, It usually took 20-30 mm to observe the blocking effects of QX-314on the Na spike and the slow IPSP. Aa: the hyperpolarizing action ofsomatostatin (2 1iM) was intact 8 mm after the penetration of the neuron (whenthe Na spike and the slow IPSP were not blocked by QX-314). Ab:somatostatin was applied 40 mm after the penetration of the cell when the Naspikes were blocked and the slow IPSP was suppressed by QX-314; Note thatthe hyperpolarizing action and the change in the input resistance caused by SSwere greatly attenuated. Ac: the actions of baclofen (20 tM) 60 mm after thepenetration of the neuron. Note that baclofen produced little hyperpolarizationof this neuron although it suppressed the synaptic transient (by activating thepresynaptic GABAB receptors?). The resting membrane potential of the neuronin A was -60 mV. B was taken from a different neuron recorded with a QX-314-filled electrode. Somatostatin (Bb) and baclofen (Bc) caused a depression ofthe synaptic transients during the early stages of recording. However, during thelate stages of recording, baclofen (Bf) but not somatostatin (Be) continued todepress the EPSP and the fast IPSP. Bb and Bc were taken when themembrane potential of the neuron was current-clamped at the control level (-56mV). In Be and Bf, the drugs did not significantly change the membranepotential. QX-314 started to suppress the action potential of the neuron in B 25mm after the penetration of the cell. The resting membrane potential of theneuron in B was -56 mV. The resting membrane potentials of the neurons in A &B were not significantly changed by QX-314.XIE 178without SS. Similar results were obtained for intracellularly recorded EPSPs.Tetanic stimulation (2 trains of 100 Hz for Is, 20 s interval) in the presence of SScaused LTP of the EPSP (intracellular EPSP slope 30 mm post-tetanus as apercentage of control: 162.48 ± 13.47, n=7) (Fig. 30).SS not only hyperpolarizes CAl neurons, but also depresses the fast andslow IPSPs in these neurons. The blockade of the fast IPSPs facilitates theinduction of LTP of the EPSP (Wigstrom and Gustafsson, 1983). In theexperiments performed by Wigstrom and Gustafsson (1983), a weak tetanuswas used to induce a short-term potentiation (of approximately 5 mm duration)under normal conditions. However, this weak tetanus caused LTP when theGABAA receptor-mediated inhibition was blocked by picrotoxinin (Wigstrom andGustafsson, 1983). Similar protocol was used in the present study to determinewhether SS can facilitate the induction of LTP. A weak tetanus (100 Hz, 0.5 s)was given in the presence of SS. During the application of SS, the membranepotentials of the CAl neurons were current-clamped to control levels. Underthis condition, this weak tetanus did not induce LTP but caused STP(intracellular EPSP slope 30 mm post-tetanus as a percentage of control: 97.5 ±8.02, n=6) (Fig. 31). This STP was not different from the STP (98.20 ± 4.25,n=6) obtained in the absence of SS. It, therefore, appears that SS doe