{"title": "A Computer Modeling Approach to Understanding the Inferior Olive and Its Relationships to the Cerebellar Cortex in Rats", "book": "Advances in Neural Information Processing Systems", "page_first": 117, "page_last": 124, "abstract": null, "full_text": "A Computer Modeling Approach to Understanding \n\n117 \n\nA computer modeling approach to  understanding the \ninferior olive and its relationship to the cerebellar \n\ncortex in rats \n\nMaurice Lee and James M. Bower \n\nComputation and Neural Systems Program \n\nCalifornia Institute of Technology \n\nPasadena, CA 91125 \n\nABSTRACT \n\nThis paper presents the results of a simulation of the spatial relationship \nbetween the inferior olivary nucleus and  folium  crus IIA of the lateral \nhemisphere  of  the  rat  cerebellum.  The  principal  objective  of  this \nmodeling effort was to resolve an apparent conflict between a proposed \nzonal organization of olivary projections to cerebellar cortex suggested \nby  anatomical  tract-tracing  experiments  (Brodal  &  Kawamura  1980; \nCampbell & Armstrong  1983) and a more patchy organization apparent \nwith physiological mapping (Robertson  1987).  The results suggest that \nseveral unique features  of the olivocerebellar circuit may contribute to \nthe appearance of zonal organization using anatomical  techniques, but \nthat  the  detailed  patterns  of  patchy  tactile  projections  seen  with \nphysiological  techniques  are  a  more  accurate  representation  of  the \nafferent organization of this region of cortex. \n\n1  INTRODUCTION \n\nDetermining  the  detailed anatomical  structure  of the  nervous  system  has  been  a major \nfocus  of neurobiology  ever since anatomical  techniques  for  looking at the fine  structure \nof individual neurons were developed more than 100 years ago (Ram6n y Cajal 1911). In \nmore recent times, new  techniques that allow labeling of the distant targets of groups of \nneurons  have  extended  this  investigation  to  include  studies  of  the  topographic \nrelationships between different brain regions.  In  general, these so-called \"tract-tracing\" \ntechniques have greatly extended our knowledge of the interrelationships between neural \nstructures,  often  guiding  and  reinforcing  the  results  of  physiological  investigations \n(DeYoe  &  Van  Essen  1988).  However,  in  some  cases,  anatomical  and  physiological \ntechniques  have  been  interpreted as  producing conflicting results.  One  case,  considered \nhere, involves the pattern of neuronal projections from  the inferior olivary nucleus to the \n\n\f118 \n\nLee and Bower \n\ncerebellar cortex.  In  this paper we  describe  the results of a computer modeling effort, \nbased on the structure of the olivocerebellar projection, intended to resolve this conflict. \n\na \n\nc \n\nb \n\ne \n\nFigure 1.  a:  Profile of the rat brain,  showing three areas (Cx, cerebral cortex; \nPo, pons; Tr, spinal trigeminal  nucleus) that project to the cerebellum (Cb)  via \nboth  climbing  fiber  (CF)  pathways  through  the  inferior  olive (10)  and  mossy \nfiber  (MF) pathways.  b:  Magnified.  highly  simplified  view  of the  cerebellar \ncortex,  showing  a  Purkinje cell  (P)  being  supplied  with  climbing  fiber  input, \ndirectly,  and  mossy  fiber  input.  through  the  granule  cells  (G). \nc:  Zonal \norganization  of  the  olivocerebellar  projection.  Different  shading  patterns \nrepresent  input  from  different  areas  of  the  inferior  olive.  Adapted  from \nCampbell  &  Armstrong  1983.  Circled area  (crus  llNcrus  UB)  is enlarged  in \nFigure  1d; bracketed area (anterior lobe)  is enlarged in Figure  Ie.  d:  Detail of \nzonal  organization.  Dark  areas  represent  bands  of Purkinje  cells  that  stain \npositive  for  monoclonal  antibody  Zehrin  I.  According  to  Gravel  et al.  1987, \nthese  bands  have  boundaries  similar  to  those  resulting  from  partial  tracer \ninjections in the inferior olive.  Adapted from  Gundappa-Sulur et al.  1989.  e: \nPatchy  organization  of the  olivocerebellar  projection  (partial  map).  Different \nshading patterns represent input through the olive from different body surfaces. \nThe  horizontal  and  vertical  scales  are  different.  Adapted  from  Logan  & \nRobertson  1986. \n\n\fA Computer Modeling Approach to Understanding \n\n119 \n\n2  THE OLIVO CEREBELLAR SYSTEM \n\nPurlcinje cells, the principal neurons of the cerebellar cortex, are influenced by two major \nexcitatory afferent projections to the cerebellum, the mLJSSY fiber system and the climbing \nfiber system  (palay &  Chan-Palay  1973).  As shown in Figures  la and  Ib, mossy fibers \narise from many different nuclei and influence Purkinje cells through granule cells within \nthe  cortex.  Within  the  cortex  the  mossy  fiber-granule  cell-Purkinje  cell  circuit  is \ncharacterized  by  enormous  divergence  (a  single  mossy  fiber  may  influence  several \nthousand  Purkinje cells) and convergence  (a single Purkinje cell  may  be  influenced by \nseveral hundred thousand mossy fibers).  In contrast, as also shown in Figures la and Ib, \nclimbing fibers arise from  a single source, the inferior olive, and exhibit severely limited \ndivergence (10-15 Purkinje cells) and convergence (I Purkinje cell). \n\nBecause the inferior olive is the sole source of the climbing fiber projection to the entire \ncerebellar  cortex,  and  each  Purkinje  cell  receives  only  one  climbing  fiber,  the  spatial \norganization of the olivocerebellar circuit has been the subject of a  large research effort \n(Brodal  &  Kawamura  1980).  Much  of this  effort has  involved anatomical tract-tracing \ntechniques  in  which  injections  of neuron ally  absorbed  substances  are  traced  from  the \ninferior olive  to  the cerebellum or vice versa.  Based on  this  work it has  been proposed \nthat  the  entire  cerebellum  is  organized  as  a  series  of  strips  or  zones,  oriented  in  a \nparasagittal plane (Figures Ic, Id: Campbell & Armstrong 1983; Gravel et al.  1987). This \nprinciple of organization has served as the basis for several functional speculations on the \nrole  of  the  cerebellum  in  coordinating  movements  (Ito  1984;  Oscarsson  1980). \nUnfortunately, as suggested in  the introduction, these anatomical results are somewhat at \nodds  with  the pattern of organization revealed by  detailed electrophysiological mapping \nstudies  of olivary  projections  (Robertson  1987).  Physiological  results,  summarized  in \nFigure  Ie, suggest that rather than being strictly  zone-like, the olivocerebellar projection \nis organized more as a mosaic of parasagittally elongated patches. \n\n3  THE MODEL \n\nOur specific interests are with the tactilely responsive regions of the lateral  hemispheres \nof  the  rat  cerebellum  (Bower  et  al.  1981;  Welker  1987),  and  the  modeling  effort \ndescribed  here  is  a  first  step  in  using  structural  models  to  explore  the  functional \norganization of this  region.  As  with  previous modeling efforts  in the  olfactory  system \n(Bower 1990), the current model  is  based on features of the anatomy and physiology of \nthe real system.  In the following section we will briefly describe these features. \n\n3.1  ANATOMICAL ORGANIZATION \n\nStructure  of  the  inferior  olive.  The  inferior  olive  has  a  complex,  highly  folded \nconformation  (Gwyn  et  al.  1977).  The  portion  of the  olive  simulated  in  the  model \nconsists of a  folded  slab  of 2520 olivary neurons  with  a  volume of approximately 0.05 \nmm3  (Figure 2a). \n\nAfferent projections to the olive.  While inputs of various  kinds and origins converge \non  this  nucleus,  we  have  limited  those  simulated  here  to  tactile  afferents  from  those \n\n\f120 \n\nLee and Bower \n\nperioral  regions  known  to  influence  the  lateral  cerebellar hemispheres  (Shambes  et al. \n1978).  These  have  been  mapped  to the olive  following  the  somatotopically  organized \npattern suggested by several previous experiments (Gellman et al.  1983). \n\nStructure or the cerebellum.  The cerebellum is represented in the model by a flat sheet \nof 2520 Purkinje cells  with  an  area of approximately  2  mm1  (Figure 2a).  Within  this \nregion. each Purkinje cell receives input from  one. and only one. olivary neuron.  Details \nof Purlcinje cells at the cellular level have not been included in the current model. \n\na \n\nb \n\nFigure 2.  a:  Basic structure of the  model.  Folia crus I1A  and crus lIB  of the \ncerebellum and a cross section of the inferior olive are shown, roughly to  scale. \nThe  regions  simulated  in  the  model  are  outlined.  Clusters  of  neighboring \nolivary neurons project to parasagittal strips of Purkinje cells as indicated.  This \nfigure also shows simulated correlation results similar to those in Figure lb.  b: \nSpatial structure of correlations among records of climbing fiber activity in crus \nIIA.  Sizes of filled circles represent cross-correlation coefficients  with  respect \nto the \"master\" site (open circle).  Sample cross-correlograms are shown for two \nsites  as  indicated.  The  autocorrelogram  for  the  \"master\"  site  is  also  shown. \nAdapted from Sasaki et al.  1989. \n\n3.2  PHYSIOLOGICAL ORGANIZATION \n\nSpatially  correlated  patterns  or activity.  When  the  activities  of multiple  climbing \nfibers are recorded from  within cerebellar cortex, there is a  strong tendency for climbing \nfibers  supplying  Purkinje  cells  oriented  parasagittally  with  respect  to  each  other  to  be \ncorrelated in their firing activity (Sasaki et al. 1989: Figure 2b). It has been suggested that \nthese correlations reflect the fact that direct electrotonic couplings exist between olivary \nneurons (Llinas & Yarom  1981a, b;  Benardo & Foster 1986). These physiological results \nare simulated in  two ways  in  the current model.  First. neighboring olivary neurons are \nelectrotonically  coupled,  thus  firing  in  a  correlated  manner.  Second.  small  clusters  of \nolivary  neurons  have  been  made  to  project to  parasagittally  oriented  strips  of Purkinje \n\n\fA Computer Modeling Approach to Understanding \n\n121 \n\ncells.  Under these constraints. the model replicates  the parasagittal pattern of climbing \nfiber activity found in certain regions of cerebellar cortex (compare Figures 2a and 2b). \n\nTopography or cerebeUar  afferents.  As  discussed  above.  this  model  is  intended  to \nexplore  spatial  and  functional  relationships  between  the  inferior  olive  and  the  lateral \nhemispheres of the rat cerebellum.  Unfortunately. a  physiological map of the climbing \nfiber  projections  to  this  cerebellar region  does  not  yet exist  for  the  rat.  However.  a \ndetailed map of mossy fiber tactile projections to this region is available (Welker 1987). \nAs in the climbing fiber map in the anterior lobe (Robertson  1987; Figure Ie) and mossy \nfiber  maps  in  various areas  in  the  cat (Kassel  et al.  1984). representations  of different \nparts of the body surface are grouped into patches with adjacent patches receiving input \nfrom  nonadjacent  peripheral  regions.  On  the  assumption  that  the  mossy  fiber  and \nclimbing  fiber  maps  coincide.  we  have  based  the  modeled  topography  of the  olivary \nprojection to  the cerebellum on the well-described mossy fiber map (Figure 3a).  In  the \nmodel,  the  smoothly  varying  topography  of  the  olive  is  transformed  to  the  patchy \norganization  of  the  cerebellar  cortex  through  the  projection  pathways  taken  to  the \ncerebellum by different climbing fibers. \n\na \n\nb \n\n.-. -.:;:\":. \n\nFigure 3.  a:  Organization  of receptive  field  map  in  simulated region  of crus \nIIA.  Different shading patterns represent input from  different perioral surfaces. \nb:  Simulated tract-tracing experiment.  Left, tracer visualization (dark areas)  in \nthe cerebellum.  Right. tracer uptake (dark areas) in the inferior olive. \n\n\f122 \n\nLee and Bower \n\n4  RESULTS:  SIMULATION OF ZONAL ORGANIZATION \n\nHaving  constructed  the  model  to  include  each  of the  physiological  features  described \nabove. we proceeded to replicate anatomical tract-tracing experiments.  This was done by \nsimulating  the chemical labeling of neurons  within restricted areas of inferior olive and \nfollowing their connections to the cerebellum.  As in the biological experiments. in many \ncases simulated injections included several folds of the olivary nucleus (Figure 3b).  The \nresults (Figure 3b) demonstrate patterns of labeling remarkably similar to those seen with \nreal olivary injections in the rat (compare Figures Id and 3b). \n\n5  CONCLUSIONS AND FURTHER WORK \n\nThese simulation results  have demonstrated that a  broadly parasagittal organization can \nbe generated in  a model system which is actually based on a fine-grained patchy pattern \nof afferent projections. Further, the simulations allow us  to  propose  that the appearance \nof parasagittal zonation  may result from  several unusual features  of the olivary nucleus. \nFirst.  the  folding  characteristic of the inferior olive likely places  neurons  with  different \nreceptive  fields  within  a  common  area  of  tracer  uptake  in  any  given  anatomical \nexperiment.  resulting  in  co-labeling  of  functionally  different  regions.  Second.  the \ntendency for local clusters of olivary neurons  to project to parasagittal strips of Purkinje \ncells  could  serve  to  extend  tracer  injection  in  the  parasagittal  direction.  enhancing  the \nimpression  of  parasagittal  zones.  This  is  further  reinforced  by  the  tendency  of  the \npatches  themselves  to  be  somewhat  elongated  in  the  parasagittal  plane.  Finally,  the \nrestricted resolution of the anatomical techniques could very well contribute to the overall \nimpression of parasagittal zonation by obscuring small, unlabeled regions more apparent \nusing physiological procedures.  Modeling efforts currently under way will extend these \nresults to more than one cerebellar folium  in an attempt to account for the appearence of \ntransfolial zones in some preparations. \n\nIn  addition  to  these  interpretations  of  previous  data,  this  model  also  provides  both \ndirections  for  further  physiological  experiments  and  predictions concerning  the results. \nFirst, the model assumes that mossy fiber and climbing fiber projections representing the \nsame regions of the rat's body surface overlap in the cerebellum.  We take the similarity \nin modeled and real  tract-tracing results  (Figures  Id and 3b) as suggesting strongly  that \nthis  is,  in  fact.  the case;  however. physiological experiments  are currently underway  to \ntest this hypothesis.  Second, the  model predicts that the parasagittal pattern of climbing \nfiber correlations found  in a particular cerebellar region will be dependent on the pattern \nof tactile patches found  in  that region.  Those regions containing large  patches (e.g.  the \ncenter of crus  IIA)  should  clearly  show parasagittal  strips  of correlated climbing  fiber \nactivity.  However,  in cortical  regions  containing  smaller,  more diverse  sets  of patches \n(e.g.  more  medial regions of crus IIA),  this correlation structure should not be as  clear. \nExperiments are also under way to test this prediction of the model. \n\n\fA Computer Modeling Approach to Understanding \n\n123 \n\nAcknowledgements \n\nThis model has been constructed using GENESIS, the Caltech neural simulation system. \nSimulation code  for  the  model  presented  here can  be  accessed  by  registered  GENESIS \nthis  model  can  be  obtained  from \nusers. \ngenesiS@caltech.bitnet.  This work was supported by NIH grant BNS 22205. \n\nthe  simulator  or \n\nInformation  on \n\nReferences \n\nBenardo,  L.  S.,  and  R.  E.  Foster  1986.  Oscillatory  behavior  in  inferior olive  neurons: \n\nMechanism. modulation. cell aggregates.  Brain Res. Bull.  17:773-784. \n\nBower.  J.  M.  1990.  Reverse  engineering  the  nervous  system:  An  anatomical. \nIn  An  introduction  to  neural  and \nphysiological.  and  computer  based  approach. \nelectronic  networks.  ed.  S.  Zornetzer.  J.  Davis.  and  C.  Lau,  pp.  3-24.  Academic \nPress. \n\nBower, J. M .\u2022  and J.  Kassel  1989.  Variability in tactile projection patterns to crus ITA of \n\nthe Norway rat.  J. Neurosci.  (submitted for publication). \n\nBower,  J.  M.,  D.  H.  Beermann,  J.  M.  Gibson.  G.  M.  Shambes.  and  W.  Welker  1981. \nPrinciples  of  organization  of  a  cerebro-cerebellar  circuit.  Micromapping  the \nprojections from  cerebral  (SI)  to  cerebellar (granule cell layer)  tactile areas  of rats. \nBrain Behav. Evol.  18:1-18. \n\nBrodal.  A .\u2022  and  K.  Kawamura  1980.  Olivocerebellar projection:  A review.  Adv.  Anat. \n\nEmbryol. Cell Bioi. 64:1-140. \n\nCampbell,  N.  C.,  and  D.  M.  Armstrong  1983.  Topographical  localization  in  the \nolivocerebellar  projection  in  the  rat:  An  autoradiographic  study.  Brain  Res. \n275:235-249. \n\nDeYoe.  E.  A.,  and  D.  C.  Van  Essen  1988.  Concurrent processing  streams  in  monkey \n\nvisual cortex.  Trends Neurosci.  11:219-226. \n\nGellman.  R, J.  C.  Hook,  and  A.  R  Gibson  1983.  Somatosensory  properties  of  the \n\ninferior olive of the cat.  J. Compo Neurol.  215:228-243. \n\nGravel.  C .\u2022  L.  M.  Eisenman.  R  Sasseville,  and  R.  Hawkes  1987.  Parasagittal \norganization  of  the  rat  cerebellar  cortex:  Direct  correlation  between  antigenic \nPurkinje cell bands revealed by mabQ 113 and the organization of the olivocerebellar \nprojection.  J. Compo Neurol. 265:294-310. \n\nGundappa-Sulur. G., H. Shojaeian. M. Paulin, L. Posakony, R. Hawkes, and J.  M. Bower \n1989.  Variability  in  and  comparisons  of:  1)  tactile  projections  to  the  granule  cell \nlayers of cerebellar cortex; and 2) the spatial distribution of Zebrin I-labeled Purkinje \ncells.  Soc. Neurosci. Abstr.  15:612. \n\nGwyn, D.  G., G.  P. Nicholson, and B.  A.  Flumerfelt 1977.  The inferior olivary nucleus \n\nof the rat:  A light and electron microscopic study.  J.  Compo Neurol.  174:489-520. \n\nIto. M.  1984.  The cerebellum and neural control.  Raven Press. \nKassel, J .\u2022  G.  M.  Shambes. and W.  Welker 1984.  Fractured cutaneous projections to the \ngranule  cell  layer of the  posterior  cerebellar  hemispheres  of the  domestic  cat.  J. \nCompo Neurol.  225:458-468. \n\nLlinas,  R.,  and  Y.  Yarom  1981a.  Electrophysiology  of  mammalian  inferior  olivary \nneurones  in  vitro.  Different  types  of voltage-dependent  ionic  conductances.  J. \n\n\f124 \n\nLee and Bower \n\nPhysiol. (Lond.) 315:549-567. \n\nLlinas,  R.,  and  Y.  Yarom  1981b.  Properties  and  distribution  of  ionic  conductances \ngenerating electroresponsiveness of mammalian inferior olivary neurones in vitro.  J. \nPhysiol. (Lond.) 315:568-584. \n\nLogan,  K.,  and  L.  T.  Robertson  1986.  Somatosensory  representation  of the cerebellar \n\nclimbing fiber system in the rat.  Brain Res. 372:290-300. \n\nOscarsson,  O.  1980.  Functional  organization  of  olivary  projection  to  the  cerebellar \nIn  The  inferior  olivary  nucleus:  Anatomy  and  physiology,  ed.  J. \n\nanterior  lobe. \nCourville, C. de Montigny, and Y. Lammare, pp. 279-289.  Raven Press. \n\nPalay,  S.  L.,  and  V.  Chan-Palay  1973.  Cerebellar  cortex:  Cytology  and organization. \n\nSpringer-Verlag. \n\nRam6n  y  Cajal,  S.  1911.  Histologie  du  systeme  nerveux de  l' homme  et des  vertebres. \n\nMaloine. \n\nRobertson, L. T.  1987.  Organization of climbing fiber representation in the anterior lobe. \nIn New  concepts in  cerebellar  neurobiology,  ed.  J.  S.  King, pp.  281-320.  Alan  R. \nLiss. \n\nSasaki, K., J.  M.  Bower, and R. Llinas  1989.  Multiple Purkinje cell recording  in rodent \n\ncerebellar cortex.  Eur. J. Neurosci.  (submitted for publication). \n\nShambes, G.  M., J.  M.  Gibson, and W.  Welker 1978.  Fractured  somatotopy  in granule \ncell  tactile  areas  of rat  cerebellar  hemispheres  revealed  by  micromapping.  Brain \nBehav. Evol.  15:94-140. \n\nWelker,  W.  1987.  Spatial  organization  of  somatosensory  projections  to  granule  cell \ncerebellar cortex:  Functional and connectional implications of fractured  somatotopy \n(summary of Wisconsin studies).  In New concepts in  cerebellar neurobiology, ed. J. \nS. King, pp. 239-280.  Alan R. Liss. \n\n\f", "award": [], "sourceid": 196, "authors": [{"given_name": "Maurice", "family_name": "Lee", "institution": null}, {"given_name": "James", "family_name": "Bower", "institution": null}]}