{"title": "Mechanisms for Neuromodulation of Biological Neural Networks", "book": "Advances in Neural Information Processing Systems", "page_first": 18, "page_last": 27, "abstract": null, "full_text": "18 \n\nHarris-Warrick \n\nMECHANISMS  FOR  NEUROMODULATION \n\nOF BIOLOGICAL NEURAL  NETWORKS \n\nRonald  M.  Harris-Warrick \n\nSection of Neurobiology and Behavior \n\nCornell University \nIthaca, NY  14853 \n\nABSTRACT \n\nThe pyloric Central Pattern Generator of the crustacean stomatogastric \nganglion is a  well-defined  biological neural  network.  This  14-neuron \nnetwork  is  modulated  by  many  inputs.  These inputs reconfigure  the \nnetwork  to  produce  multiple  output  patterns  by  three  simple \nmechanisms:  1) detennining which  cells are active; 2) modulating the \nsynaptic  efficacy;  3)  changing  the  intrinsic  response  properties  of \nindividual  neurons.  The importance of modifiable  intrinsic  response \nproperties of neurons for network function and modulation is discussed. \n\n1  INTRODUCTION \nMany neural network models aim to understand how a particular process is accomplished \nby  a  unique  network  in  the  nervous  system.  Most  studies  have aimed at circuits  for \nlearning or sensory processing; unfortunately, almost no biological data are available on \nthe actual anatomical structure of neural networks serving these tasks, so the accuracy of \nthe theoretical  models is  unknown.  Much  more is known  concerning  the structure and \nfunction  of motor circuits generating simple rhythmic  movements,  especially in simpler \ninvertebrate nervous systems (Getting,  1988).  Called Central Pattern Generators (CPGs), \nthese are rather small circuits of relatively  well-defined composition.  The output of the \nnetwork  is  easily  measured  by  monitoring  the  motor  patterns  causing  movement. \nResearch on cellular interactions in CPGs has shown that simple models of fixed circuitry \nfor  fixed outputs are oversimplified.  Instead, these neural  networks have evolved with \nmaximal flexibility  in  mind, such that modulatory inputs to the circuit can reconfigure it \n\"on the fly\"  to  generate an almost infinite variety of motor patterns.  These modulatory \ninputs,  using  slow  transmitters  such  as  monoamines  and  peptides,  can  change  every \ncomponent of the  network,  thus constructing multiple functional  circuits from  a single \nnetwork (Harris-Warrick,  1988).  In this paper, I will describe a model biological system \nto demonstrate the types of flexibility that are built into real neural networks. \n\n\fMechanisms for Neuromodulation of Biological Neural Networks \n\n19 \n\n2  THE  CRUSTACEAN  STOMATOGASTRIC  GANGLION \nThe pyloric CPG in  the stomatogastric ganglion (STG) of lobsters and crabs is the OO8t(cid:173)\nunderstood neural circuit (Selverston and Moulins, 1987). The STG is a tiny ganglion of \n30 neurons that controls rhythmic movements of the foregut.  The pyloric CPG controls \nthe peristaltic pumping and filtering  movements of the pylorus, or posterior part of the \nforegut.  This network contains  14  neurons, each of which is unambiguously assignable \nto one of 6 cell types (Figure  lA).  Since each neuron can be identified from  preparation \nto preparation, detailed studies of the properties of each cell are possible. Thanks to  the \ncareful  work  of Selverston  and  Marder and  their  colleagues,  ~e anatomical  synaptic \ncircuitry is completely known (Fig.1A). and consists of chemical synaptic  inhibition and \nelectrotonic coupling; there is no chemical excitation in the circuit (Miller. 1987). \n\nDespite  the complete knowledge of the synaptic  connections  within  this  network.  the \nmajor question of \"how it works\" is still  an  important topic of neurobiological research. \nEarly modelling efforts (summarized in Hartline. 1987) showed that. while the pattern of \nmutual  synaptic  inhibition  provided important  insights  into  the phase relations  of the \nneurons active in  the three-phase motor pattern. pure connectionist models with  simple \nthreshold elements for neurons were insufficient to explain the motor pattern generated by \nthe network.  It has been necessary to understand the intrinsic response properties of each \nneuron  in  the  circuit.  which  differ  markedly  from  one  another  in  their  responses  to \nidentical  stimuli.  Most  importantly.  as  will  be described  below.  all  14  neurons  are \nconditional oscillators. capable (under the appropriate conditions) of generating rhythmic \nbursts of action potentials in  the absence of synaptic  input (Bal et al.  1988).  This and \nother  intrinsic  properties of the  neurons.  coupled  with  the  pattern  of mutual  synaptic \ninhibition within the circuitry. has generated relatively good models of the pyloric motor \npattern under a specified  set of conditions (Hartline, 1987). \n\nA.  Pyloric circuit \n\nB.  Combined \n\nPDN \n\nI~I \n\n1111 \n\n1.1 \n\nII \n\nII \n\nII \n\nt \n\nLI'.I'Y \n~lr.  I \n\n1111  I \n\nIII I I \n\nIII \n\n.' I \n\n.\" \n\nIill \n\nI; \n\nlilt. \n\n111 \n\nC.  Sucrose block \n\nI \n\nI \n\nI \n\nI \n\nI \n\nt \n\ni \n\nD.  Dopamine \n\nI \n\nI \n\nI \n\nI \n\nE.  Octopamlne \n\n111111 \n\nIII \n\n111111 \n\n111111 \n\n11111 1111 \n\n~ W III \n\nI\n\nF. \n\nSerotonin \n\n1111 \n\n1111 \n\n1111 \n\nFigure 1:  Multiple motor patterns from  the pyloric  network in  ~e presence of different \nneurotransmitters.  A.  Synaptic  wiring  diagram  of the  pylonc  CPG.  B.-F. Motor \npatterns  observed  under  different  cond~tions (s~ t~xt).  PDN.LP-PY.~VN traces: \nextracellular recordings  of action potentIals  from  mdlcated  neurons.  AB.  mtracellular \nrecording from the AB  interneuron.  From Harris-Warrick and Flamm (1987a). \n\n\f20 \n\nHarris-Warrick \n\n3  MULTIPLE  MOTOR  PATTERNS  PRODUCED  BY  AN \nANATOMICALLY  FIXED  NEURAL  NETWORK \nWhen  the  STG  is  dissected  with  intact  inputs  from  other  ganglia.  the  pyloric  CPG \ngenerates a stereotyped motor pattern (Miller.1987).  However. in vivo,  the network gen(cid:173)\nerates a widely varying motor pattern. depending on the feeding state of the animal (Rezer \nand  Moulins.  1983).  The motor pattern  varies  in  the  cycle frequency  and regularity. \nwhich cells are active. the intensity of cell firing, and phase relations. \nThis variability can be mimicked in  vitro. where experimental control over the system is \nbetter.  Two  major experimental  approaches  have  been  used.  First.  transmitters  and \nmodulators that are present in the input nerve to the STG can be bath-applied. producing \nunique variants on the basic motor theme.  Second. identified modulatory neurons can be \nselectively stimulated. activating and altering the ongoing motor pattern. \n\nAs  an  example.  the effects of the  monoamines  dopamine (DA).  serotonin  (SHT)  and \noctopamine (OCT)  on  the  pyloric  motor pattern  are shown  in  Figure  1.  When  modu(cid:173)\nlatory inputs from  other ganglia are present, the pyloric rhythm cycles strongly. with all \nneurons active (Combined).  Removal of these inputs usually causes the rhythm to cease. \nand cells are either silent or fire  tonically (Sucrose Block).  Bath application of some of \nthe transmitters present in  the input nerve can restore rhythmic cycling.  However.  the \nmotor pattern  induced  is  different and  unique  for  each  transmitter  tested:  clearly  the \npatterns induced by DA. SHT and OCT differ markedly in frequency. intensity. active cells \nand phasing (Flamm and Harris-Warrick. 1986a).  The conclusion is that an anatomically \nfIXed network can generate a variety of outputs in the presence of different modulatory in(cid:173)\nputs: the anatomy of the network does not determine its output. \n\n4  MECHANISMS  FOR  ALTERATION  OF  NEURAL \nNETWORK  OUTPUT  BY  NEUROMODULATORS \nWe have studied the cellular mechanisms  used  by  monoamines  to  modify  the  pyloric \nrhythm.  To do this. we isolate a single neuron or single synaptic interaction by selective \nkilling  of other neurons or pharmacological blockade of synapses (Flamm  and Harris(cid:173)\nWarrick. 1986b).  The amine is then added and its direct effects on the neuron or synapse \ndetermined.  Nearly every neuron in the network responded directly to all three amines we \ntested.  However. even in this simple 14-neuron circuit. different neurons responded differ(cid:173)\nently to a single amine.  For example. DA  induced rhythmic oscillations and bursting in \none cell type. hyperpolarized and silenced two others, and depolarized the remaining cells \nto  fire  tonically  (Fig.2).  Thus.  one  cannot  use  the  knowledge  of  the  effects  of  a \ntransmitter on one neuron  to infer  its actions  on  other neurons in  the same circuit.Our \nstudies of the actions of DA. SHT and OCT on  the pyloric  network have demonstrated \nthree simple mechanisms for altering the output from a network. \n\n\fMechanisms for Neuromodulation of Biological Neural Networks \n\n21 \n\nControl \n\nDopamine \n\nVDJ~ ___  _ \n\nLP  ----------LL- JJJ.UUillUW \n\npy  - - - - JJJJllJllillLlUJj  I \nJllillllilillll  I \nIe \n\n------\n\n-\n\nFigure  2:  Actions  of  dopamine  on  isolated  neurons  from  the  pyloric  network. \nControl:  Activity  of  each  neuron  when  totally  isolated  from  all  synaptic  input. \nDopamine: Activity of isolated cell during bath application of 10-4M dopamine. \n\n4.1  ALTERATION  OF  THE  NEURONS  THAT  ARE  ACTIVE  PARTICI\u00b7 \nPANTS  IN  THE  FUNCTIONAL  CIRCUIT \n\nBy simply exciting a silent cell or inhibiting an active cell. a neuromodulator can deter(cid:173)\nmine  which  of the cells in  a  network will  actively  participate in  the generation of the \nmotor  pattern.  Some  cells  thus  are  physiologically  inactive.  even  though  they  are \nanatomically present. \nHowever. in some cases. unaffected cells can make a significant contribution to the motor \npattern.  Hooper and Marder (1986) have shown  that the peptide proctolin activates the \npyloric rhythm and induces rhythmic oscillations in one neuron.  Proctolin has no effect \non three other neurons that are electrically coupled to the oscillating neuron;  these cells \nimpose an electrical drag on the oscillator neuron. causing it to cycle more slowly than it \ndoes when  isolated from  these cells.  Thus. the unaffected cells cause  the whole  motor \npattern to cycle more slowly. \n\n4.2  ALTERATION  OF  THE  SYNAPTIC  EFFICACY  OF \nCONNECTIONS  WITHIN  THE  NETWORK \n\nThe flexibility of synaptic interactions is well-known and is used in  virtually all models \nof plasticity in neural networks.  By changing the amount of transmitter released from the \npre-synaptic  tenninal or the post-synaptic  responsiveness (either by altering  the mem(cid:173)\nbrane resistance  or the number of receptors). the strength of a synapse can be altered over \nan order of magnitude.  Obviously. this will have important effects on the phase relations \nof neurons firing in the network. \n\nIn the STG. the situation is complicated by the fact that graded synapses are the primary \nfonn of chemical communication: the cells release transmitter as a continuous function of \nmembrane potential. and do  not require action potentials  to  trigger release (Graubard. \n\n\f22 \n\nHarris-Warrick \n\n1978).  Some neurons even release transmitter at rest and must be hyperpolarized to block \nrelease.  We have shown that graded synaptic transmission is also strongly modulated by \nmonoamines, which can completely eliminate some synapses while strengthening others \n(Fig.3; Johnson and Harris-Warrick, 1990).  Amines can change the apparent threshold for \ntransmitter release or the functional strength of the synapse.  Modulation of graded trans(cid:173)\nthe \nmission \nmotorpattern,  which  is  often  detennined  by  synaptic  interactions.  Graded  synaptic \ntransmission  occurs  in  many  species,  so  this  could  turn  out  to  be a  general  fonn  of \nplasticity. \n\nthus  allows  delicate  adjustments  of  the  phasing  between  cells  in \n\n6--0 \nlO\u00b7SM  Oct \n\nlO\u00b74M  DA \n\nControl \n\nPD~ ~ ~ \n\nLP~ ~ -1 %tIllV \n\nJ  IIIV \n\nI  I\u00ab \n\nFigure  3:  Modulation  of graded  synaptic  transmission  from  the PD neuron  to  the LP \nneuron  by octopamine and dopamine.  Experiment done in the presence of tetrodotoxin to \nabolish action potentials.  Other synaptic inputs to these cells have been eliminated. \n\nIn  one case, modulation of graded transmission results in  a sign reversal of the synaptic \ninteraction between two cells (Johnson and Harris-Warrick,  1990).  In the pyloric CPG, \nthe PD neurons  weakly  inhibit the  IC neuron by a graded chemical  mechanism, but in \naddition the two cells are weakly electrically coupled.  This mixed synapse is weak and \nvariable.  Dopamine weakens the chemical inhibition:  the electrical coupling dominates \nand the IC cell depolarizes upon PD depolarization.  Octopamine strengthens the chemical \ninhibition, and the IC cell  hyperpolarizes upon PD depolarization.  Combined chemical \nand electrical synaptic interactions have been detected in many other preparations, and thus \ncan undecly flexibility in the strength and sign of synaptic interactions. \n\n4.3  ALTERATION  OF  THE  INTRINSIC  RESPONSE  PROPERTIES \nOF  THE  NETWORK  NEURONS \n\nThe physiological response properties of neurons within a network are not fixed, but can \nbe extensively altered by neuromodulators.  As a consequence, the response to an identical \nsynaptic input can vary radically in the presence of different neuromodulators. \n\n4.3.1 \n\nInduction  of  bistable  firing  properties \n\nMany  neurons  in  both  vertebrates  and  invertebrates  are  capable of firing  in  \"plateau \npotentials\", where a brief excitatory stimulus triggers a  prolonged depolarized plateau, \nwith  tonic  spiking  for  many  seconds,  which  can  be  prematurely  truncated  by  a  brief \nhyperpolarizing input (Hartline et al,  1988).  Thus, the neuron shows bistable properties: \nbrief synaptic  inputs can  step it between  two  relatively  stable resting potentials which \ndiffer  markedly  in  spike  frequency.  This property  is  plastic,  and  can  be  induced  or \n\n\fMechanisms for Neuromodulation of Biological Neural Networks \n\n23 \n\nsuppressed by neuromodulatory inputs.  For example, Fig. 4 shows the DG neuron in the \nSTG.  Under control  conditions,  a  brief depolarizing  current injection  causes a  small \ndepolarization  that  is  subthreshold  for  spike  initiation.  However,  after  stimulating  a \nserotonergic/cholinergic modulatory neuron (called GPR), the same brief current injection \ninduces a prolonged burst of spikes on a depolarized plateau potential (Katz and Harris(cid:173)\nWarrick. 1989).  Similar results have been obtained in  turtle and cat spinal motor neurons \nafter  application  of  monoamines  such  as  serotonin  or  its  biochemical  precursor \n(Hounsgaard et aI.1988;  Hounsgaard and  Kiehn.1989).  Stimulation of a  modulatory \nneuron can also disable the plateau potentials that are normally present in a neuron (Nagy \net aI.  1988). \n\nDG.-J~ ~ 110mv \n\n_~,-___ -,,---_____ \"--- 11  nA \n\n~ \n\n5 see \n\n-GPR stirn. \n\nFigure  4:  Induction  of plateau  potential  capability  in  DG  neuron  by  stimulation  of a \nserotonergic/cholinergic sensory neuron, GPR. \n\n4.3.2 \n\nInduction  of  endogenous  rhythmic  bursting \n\nA more extreme fonn  of modulation can occur where  the modulatory stimulus induces \nendogenous rhythmic  oscillations in membrane potential  underlying rhythmic  bursts of \naction potentials.  For example. in Figure 4. the pyloric  AB  neuron shows  no intrinsic \noscillatory capabilities when  it is isolated  from  all  synaptic  input.  Bath application of \nmonoamines such as  DA.  5HT and OCT induce rhythmic  bursting in  this  isolated cell \n(Flamm and Harris-Warrick.  1986b). Brief stimulation  of the  serotonergic/cholinergic \nGPR neuron can also induce or  enhance  rhythmic bursting  that outlasts the stimulus by \n\nControl \n\nDopamine \n\n.J \n\nFigure  S:  Induction of rhythmic  bursting in  a synaptically  isolated  AB  neuron  by bath \napplication of dopamine (104 M). \n\nseveral  minutes.  The quantitative details  of the  bursting  (cycle  frequency.  oscillation \namplitude.  spike frequency,  etc.)  are different with  each  amine.  due  to different  ionic \nmechanisms  for  burst generation  (Harris-Warrick and  Flamm.  1987b).  Since the  AB \nneuron is the major pacemaker in the pyloric CPG, these differences underly  the marked \ndifferences in  pyloric  rhythm  frequency  seen with  the  amines  in Fig.I.  Induction  of \nrhythmic  bursting by  neuromodulators has  been  observed  in  vertebrates  (for example. \nDekin et aI.1985). and this is likely to be a general mechanism. \n\n\f24 \n\nHarris-Warrick \n\n4.3.3  Modulation  or  post-inhibitory  rebound \nMost neurons show post-inhibitory rebound, a period of increased excitability following \nstrong inhibition.  This  is  probably  due  in  part  to  the activation  of prolonged  inward \ncurrents during hyperpolarization (Angstadt and Calabrese,  1989).  This property can be \nmodified  by  biochemical  second  messengers  used  by  neuromodulators.  For example. \nelevation  of cAMP by  forskolin  enhances  post-inhibitory  rebound  in  the  pyloric  LP \nneuron (Figure 5; Flamm et al,  1987).  As a consequence of this modulation,  the cell's \nresponse to a simple inhibitory input is radically changed to a biphasic response, with an \ninitial inhibition followed by delayed excitation. \n\nControl \n\n~ ~IUUU!!lUU~Wl \n\nLP \n\n' rr - - -\ni : \n\n, r- ~ ~WJllllUllilllllliUllUUlU \n\n50 ~ Forskolin \n\n---u- U \n\nFigure  6:  Induction  of post-inhibitory  rebound  by  forskolin,  which  elevates  cAMP \nlevels, in  the LP neuron.  Control:  Hyperpolarizing current injection  does  not induce \npost-inhibitory  rebound,  measured  at  two  different  resting  potentials.  Forskolin: \nElevation of cAMP depolarizes LP and  induces tonic  spiking  (left).  At all  membrane \npotentials, a hyperpolarizing pulse is followed by an enhanced burst of action potentials. \n\nS  ENDOGENOUS  RELEASE  OF  NEUROMODULATORS \nFROM  IDENTIFIED  NEURONS \nMost of the results I have described were obtained with bath application of amines or pep(cid:173)\ntides, a method that can be criticized as being non-physiological.  To test this, a number \nof neurons containing identified neuromodulators have been found, and the action of the \nnaturally  released  and  bath-applied  modulator  directly  compared.  An  immediate \ncomplication arose  from  these  studies:  the  majority  of the known  modulatory  neurons \ncontain  more  than  one  transmitter.  All  possible  combinations  have  been  observed, \nincluding a slow transmitter with a fast transmitter, two or more slow  transmitters, and \nmultiple fast transmitters.  To fully  understand the complex changes in network function \ninduced  by  activity  in  these neurons,  it is  necessary  to  study  the actions  of all  the co(cid:173)\ntransmitters  on  all the neurons in  the network.  This has been recently accomplished  in \nthe  STG.  Here,  serotonin  is  released  by  a  set of sensory  cells  responding  to  muscle \nstretch (Katz et aI,  1989).  These cells also contain  and  release acetylcholine (Katz  et \nal,1989).  In  studying  the  actions  of the  two  transmitters,  remarkable  flexibility  was \nuncovered (Katz and Harris-Warrick, 1989,1990).  First, not all target neurons responded \n\n\fMechanisms for Neuromodulation of Biological Neural Networks \n\n2S \n\nto both released transmitters: some responded only to 5HT, while one cell responded only \nto ACh.  Second, the responses to released 5HT were all modulatory, but varied markedly \nin different cells, mimicking the bath application  studies described earlier.  Finally, the \ntwo transmitters acted over entirely different time scales.  ACh induced rapid EPSPs  last(cid:173)\ning tens to hundreds of msec via nicotinic receptors, while 5HT induced slow prolonged \nresponses lasting many seconds to minutes (for example, Fig.4). \n\nIt is now clear that neural networks are targets for multiple neuronal inputs using many \ndifferent transmitters and  modulators.  For example, the STG contains only  30 neurons, \nbut is innervated by over 100 axons from  other ganglia.  Twelve neurotransmitters have \nthus far been identified in these axons (Marder and Nusbaum,1989), and these are probably \na  minority of the total  that are present.  In recordings from  the input nerve to the gan(cid:173)\nglion, many axons are spontaneously active.  Thus, the pyloric network is continuously \nbathed with a  varying mixture of transmitters and  modulators, allowing for  very subtle \nchanges in the firing pattern.  In  vivo,  we expect that each modulator plays a small role \nin the overall mixture that determines the final motor pattern. \n\n6  CONCLUSION \nThe work described here shows conclusively that an anatomically fixed neural network can \nbe modulated to produce a large variety of output patterns.  The anatomical connections in \nthe  network are  necessary but not sufficient to  understand  the  output  of the  network. \nIndeed, it is best to  think of these networks as  libraries of potential components, which \nare then  selected and activated by  the modulatory inputs.  In  addition to altering which \nneurons are active and altering the synaptic strength in  the circuits, I have emphasized the \nimportant role of modulation of the intrinsic response properties of the network neurons \nin determining the final pattern of output.  Indeed, if this aspect of modulation is ignored, \npredictions of the actions of modulators on the final motor pattern are grossly in error. \n\nMany  modellers claim  that  this  emphasis on  the  intrinsic  computational properties  of \nsingle neurons is  unique to the invertebrates, which have few  cells to  work with.  In the \nvertebrates, they argue, the enormous increase in  numbers of cells changes the computa(cid:173)\ntional rules such  that each cell is a simple threshold element, and complex transforma(cid:173)\ntions  only  take  place  with  changes  in  synaptic  efficacy  in  the  circuits.  There  are \nabsolutely no data to  support this hypothesis of \"simple cells\" in vertebrates.  In fact, a \ngreat deal of careful work has shown that vertebrate neurons are dynamic elements that \nshow all the complex intrinsic response properties of invertebrate neurons (Llinas,1988). \nThese properties can be changed by neuromodulators, just as in the crustacean STG, such \nthat  vertebrate  cells  can  have  radically  different  physiological  \"personalities\"  in  the \npresence  of  different  modulators.  Network  models  which  ignore  the  complex \ncomputational properties of single neurons thus do not reflect the richness and variability \nof biological neural networks of both invertebrates and vertebrates alike. \n\nAcknowledgments:  Supported by NIH Grant NS17323  and Hatch Act NYC-19141O. \n\n7  BIBLIOGRAPHY \nAngstadt, J.D., Calabrese, R.L.  (1989)  A hyperpolarization-activated inward current in \n\nheart interneurons of the medicinal leech. 1. Neurosci. 9:  2846-2857. \n\n\f26 \n\nHarris-Warrick \n\nBal, T., Nagy, F., Moulins, M. (1988) The pyloric central pattern generator in Crustacea: \n\na set of conditional neuronal oscillators.  J.  Compo  Physiol.  A  163:  715-727. \n\nDekin,  M.S.,  Richerson. G.B .\u2022  Getting, P.A.  (1985) Thyrotropin-releasing honnone in(cid:173)\nduces rhythmic bursting in neurons of the nucleus tractus solitarius. Science 229:67-\n69. \n\nFlamm, R.E.,  Harris-Warrick. R.M.  (1986a)  Aminergic  modulation  in  lobster stomato(cid:173)\n\ngastric ganglion. I. The effects on motor pattern and activity  of neurons within  the \npyloric circuit. J. Neurophysiol.  55:  847-865. \n\nFlamm, R.E.,  Harris-Warrick. R.M. (1986b)  Aminergic  modulation in  lobster stomato(cid:173)\n\ngastric ganglion. II. Target neurons of dopamine, octopamine. and serotonin within \nthe pyloric circuit.  J.  Neurophysiol.  55:  866-881. \n\nFlamm. R.E .\u2022  Fickbohm,  D.,  Harris-Warrick. R.M.  (1987)  cAMP elevation  modulates \nphysiological activity of pyloric neurons in  the lobster stomatogastric ganglion.  J. \nNeurophysiol.  58:  1370-1386. \n\nGetting, P.A. (1988). Comparative analysis of invertebrate central pattern generators.  in: \nCohen,  A.H.,  Rossignol.  S.,  Grillner,  S.  (eds.),  Neural  Control  of  Rhythmic \nMovements in  vertebrates. John Wiley and Sons, New York, pp.  101-127. \n\nGraubard, K.  (1978)  Synaptic transmission without action potentials: input-output prop(cid:173)\n\nerties of a non-spiking presynaptic neuron.  J. Neurophysiol.  41:  1014-1025. \n\nHarris-Warrick, R. M.  (1988) Chemical modulation of central pattern generators. in: Co(cid:173)\nhen, A.H., Rossignol, S., Grillner. S.(eds.) Neural Control of Rhythmic Movements \nin vertebrates,  John Wiley &  Sons. New York. pp 285-331. \n\nHarris-Warrick.  R.M .\u2022  Flamm,  RE.  (1987a)  Chemical  modulation  of a  small  central \n\npattern generator circuit.  Trends in Neurosci.  9: 432-437. \n\nHarris-Warrick,  R.M.,  Flamm.  R  E.  (1987b)  Multiple  mechanisms  of bursting  in  a \n\nconditional bursting neuron.  J. Neurosci.  7:  2113-2128. \n\nHartline, D.K.  (1987)  Modeling stomatogastric ganglion. in:  Selverston, A.I .\u2022  Moulins. \nM. (eds.), The Crustacean Stomatogastric System. Springer-Verlag, Berlin. pp.  181-\n197. \n\nHartline, D.K., Russell. D.K .\u2022 Raper. J.A .\u2022 Graubard. K.  (1988)  Special cellular and sy(cid:173)\n\nnaptic  mechanisms  in  motor  pattern  generation.  Compo  Biochem.  Physiol. \n91C:115-131. \n\nHooper, S.L., Marder. E (1987) Modulation of the lobster pyloric rhythm by  the peptide \n\nproctolin.  J. Neurosci.  7:2097-2112. \n\nHounsgaard. J .\u2022  Kiehn, O.  (1989)  Serotonin-induced bistability of turtle motoneurones \ncaused by a nifedipine-sensitive calcium plateau potential.  J.  Physiol. 414:265-282. \nHounsgaard, J., Hultborn.  H., Jespersen, B.,  Kiehn.  O.  (1988)  Bistability of alpha-mo(cid:173)\ntoneurones  in  the  decerebrate  cat and  in  the  acute  spinal  cat after intravenous  5-\nhydroxy tryptophan.  J.  Physiol.  405:345-367. \n\nJan. L.Y., Jan, Y.N. (1982)  Peptidergic transmission in sympathetic ganglia of the frog. \n\nJ.  Physiol.  327:  219-246. \n\nJohnson, B. R., Harris-Warrick. R.M.  (1990) Aminergic  modulation of graded synaptic \n\ntransmission in the lobster stomatogastric ganglion.  J. Neurosci., in press. \n\nKatz.  P.S .\u2022  Eigg,  M.H.,  Harris-Warrick. R.M.  (1989)  Serotonergic/cholinergic  muscle \nreceptor  cells  in  the  crab  stomatogastric  nervous  system.  I.  Identification  and \ncharacterization of the gastropyloric receptor cells.  J. Neurophysiol. 62:  558-570. \n\n\fMechanisms for Neuromodulation of Biological Neural Networks \n\n27 \n\nKatz, P.S.,  Harris-Warrick, R.M. (1989) Serotonergic/cholinergic muscle receptor cells \nin  the  crab  stomatogastric  nervous  system.  II.  Rapid  nicotinic  and  prolonged \nmodulatory effects on neurons in the stomatogastric ganglion.  J. Neurophysiol.  62: \n571-581. \n\nKatz,  P.S.,  Harris-Warrick, R.  M.  (1990) Neuromodulation  of the crab pyloric  central \npattern generator by serotonergic/cholinergic proprioceptive afferents.  J. Neurosci., \nin press. \n\nLlinas, R.R. (1988) The intrinsic electrophysiological properties of mammalian neurons: \n\ninsights into central nervous function.  Science 242:  1654-1664. \n\nMarder, E., Nusbaum, M.P. (1989) Peptidergic  modulation of the motor pattern genera(cid:173)\ntors  in  the stomatogastric ganglion.  in:  Carew, T.I., Kelley,  D.B.  (eds.), Perspec(cid:173)\ntives in Neural Systems and Behavior, Alan R. Liss, Inc., New York. pp 73-91. \n\nMiller, J.P.  (1987) Pyloric  mechanisms. \n\nin:  Selverston,  A.I.,  Moulins,  M.  (eds.) ~ \n\nCrustacean Stomatogastric System, Springer-Verlag, Berlin, pp.  109-136. \n\nNagy,  F.,  Dickinson,  P.S.,  Moulins,  M.  (1988)  Control  by  an  identified  modulatory \nneuron of the sequential expression of plateau properties of, and synaptic inputs to, a \nneuron in a central pattern generator. J. Neurosci. 8:2875-2886. \n\nRezer, E., Moulins, M.  (1983)  Expression of the crustacean pyloric pattern generator in \n\nthe intact animal.  J.  Compo  Physiol.  153:17-28. \n\nSelverston,  A.I.,  Moulins,  M.  (eds.)  (1987)  The  Crustacean  Stomatogastric  System \n\nSpringer-Verlag, Berlin, 338 pp. \n\n\f", "award": [], "sourceid": 220, "authors": [{"given_name": "Ronald", "family_name": "Harris-Warrick", "institution": null}]}