{"title": "Feedback Synapse to Cone and Light Adaptation", "book": "Advances in Neural Information Processing Systems", "page_first": 391, "page_last": 398, "abstract": null, "full_text": "FEEDBACK SYNAPSE TO CONE AND LIGHT ADAPTATION \n\nJosef Skrzypek \n\nMachine Perception Laboratory \n\nUCLA - Los Angeles, California 90024 \n\nINTERNET: SKRZYPEK@CS.UCLA.EDU \n\nAbstract \n\nLight  adaptation  (LA)  allows  cone  vIslOn  to  remain  functional  between \ntwilight  and  the  brightest  time  of day  even  though,  at  anyone  time,  their \nintensity-response (I-R)  characteristic  is  limited  to  3  log  units of the  stimu(cid:173)\nlating  light.  One mechanism  underlying  LA, was localized  in  the outer seg(cid:173)\nment of an isolated cone (1,2). We found that by adding annular illhmination, \nan  I-R  characteristic  of a  cone  can  be  shifted  along  the  intensity  domain. \nNeural network involving feedback  synapse from  horizontal cells to cones is \ninvolved  to  be  in  register  with  ambient  light  level  of  the  periphery.  An \nequivalent  electrical  circuit  with  three  different  transmembrane  channels \nleakage,  photocurrent  and  feedback  was  used  to  model static  behavior of a \ncone.  SPICE simulation showed  that interactions between feedback  synapse \nand  the  light  sensitive  conductance  in  the  outer  segment  can  shift  the  I-R \ncurves along  the  intensity domain, provided  that phototransduction  mechan(cid:173)\nism is not saturated during maximally hyperpolarized light response. \n\n1  INTRODUCTION \n\n1.1  Light response in cones \n\nIn the vertebrate retina, cones respond to a small spot of light with sustained hyperpolari(cid:173)\nzation  which  is  graded  with  the  stimulus over three log  units of intensity  [5].  Mechan(cid:173)\nisms underlying this I-R relation was suggested to result from  statistical superposition of \ninvariant  single-photon,  hyperpolarizing  responses \ninvolvnig  sodium  conductance \nchanges that are gated by cyclic nuclcotides (see 6). The shape of the response measured \nin cones depends on the size of the stimulating spot of light, presumably because of peri(cid:173)\npheral  signals mediated  by a  negative  feedback  synapse  from  horizontal  cells  [7,8];  the \nhyperpolarizing response to the spot illumination in the central portion of the cone recep(cid:173)\ntive field  is  antagonized by  light in the  surrounding periphery  [11,12,13]. Thus the cone \n\n391 \n\n\f392 \n\nSkrzypek \n\nmembrane is influenced by two antagonistic effects;  1)  feedback, driven by peripheral il(cid:173)\nlumination and 2) the light sensitive conductance, in the cone outer segment Although it \nhas  been shown that key aspects of adaptation can be observed in isolated cones  [1,2,3], \nthe effects of peripheral illumination on adaptation as related  to feedback input from hor(cid:173)\nizontal  cells  have  not  been  examined.  It  was  reported  that  under  appropriate  stimulus \nconditions  the  resting  membrane potential for  a  cone can  be reached  at two drastically \ndifferent intensities for a spot/annulus combinations [8,14]. \n\nWe  present  here  experimental  data  and  modeling results  which  suggests  that  results  of \nfeedback  from  horizontal  cells  to cones  resemble the  effect of the  neural  component of \nlight adaptation  in cones.  Specifically, peripheral signals mediated via feedback  synapse \nreset  the  cone sensitivity  by  instantaneously  shifting  the  I-R  curves  to  a  new  intensity \ndomain.  The  full  range  of  light  response  potentials  is  preserved  without  noticeable \ncompression. \n\n2  RESULTS \n\n2.1 \n\nIdentification of cones \n\nPreparation  and the general  experimental procedure  as  well as criteria for  identification \nof cones has been detailed in  [15,8].  Several criteria were used to distinguish cones from \nother cells in  the OPL such as:  1)  the depth of recording in the retina [II, 13],2) the se(cid:173)\nquence  of  penetrations  concomitant  with  characteristic  light  responses,  3)  spectral \nresponse curves [18],4) receptive field diameter [8],  5) the fastest time  from dark poten(cid:173)\ntial to the peak of the light response [8,  15], 6)domain of I-R curves and 7)  staining with \nLucipher Yellow  [8,  11,  13].  These values represent averages derived  from  all intracel(cid:173)\nlular recordings in 37 cones, 84 bipolar cells, more  than  1000 horizontal cells, and more \nthan 100 rods. \n\n2.2  Experimental procedure \n\nAfter identifying a cone, its I-R curve was recorded.  Then, in a presence of center illumi(cid:173)\nnation  (diameter =  100  urn)  which elicited  maximal  hyperpolarization  from  a  cone,  the \nperiphery of the receptive  field was stimulated with an annulus of inner diameter (ID) = \n750 urn  and the outer diameter (OD) = 1500 urn. The annular intensity was adjusted to el(cid:173)\nicit depolarization  of the  membrane  back  to the  dark  potential  level.  Finally, the center \nintensity was  increased again in a stepwise manner to antagonize the effect of peripheral \nillumination, and this new I-R curve was recorded. \n\n2.3 \n\nPeripheral illumination shifts the I-R curve in cones \n\nSustained  illumination  of a  cone  with  a  small  spot  of light,  evokes  a  hyperpolarizing \nresponse, which after transient peak gradually repolarizes  to some steady level (Fig.  1a). \nWhen  the  periphery  of the  relina  is  illuminated  with  a  ring  of light  in  the  presence  of \ncenter spot,  the  antagonistic  component of  response  can  be recorded  in a  form  of sus(cid:173)\ntained  depolarization.  It has  been  argued  previously  that in  the  tiger  salamander cones, \nthis  type  of response  in  cones  is  mediated  via synaptic  input from  horizontal cells.  [11, \n12]. \n\n\fFeedback Synapse to Cone and Light Adaptation \n\n393 \n\nThe  significance of this result is  that the resting membrane potential for this cone can be \nreached  at two drastically  different  intensities  for  a  spot/annulus combinations; The ac(cid:173)\ntion  of an  annular  illumination is  a  fast depolarization  of the  membrane;  the  whole pro(cid:173)\ncess  is completed  in  a fraction  of a second  unlike  the  previous  reports  where  the course \nof light-adaptation lasted for seconds or even minutes. \nResponse  due  to  spot  of light  measured  at  the  peak  of hyperpolarization,  increased  in \nmagnitude with  increasing intensity over three log units (fig.  l.a). The same data is plot(cid:173)\nted  as open circles  in  fig.  l.b.  Initially, annulus presented during the central illumination \ndid  not produce  a noticeable response.  Its amplitude reached  maximum  when  the center \nspot intensity was  increased to 3 log units. Further increase of center intensity resulted in \ndisappearance of the annulus- elicited depolarization. Feedback action is graded with an(cid:173)\nnular  intensity  and  it  depends  on  the  balance  between  amount  of light  falling  on  the \ncenter and  the  surround of the cone receptive field.  The change in  cone's membrane po(cid:173)\ntential, due to combined effects of central and annular illumination is plotted as filled cir(cid:173)\ncles  in  fig.  lb.  This  new  intensity-response curve  is  shifted along  the  intensity  axis  by \napproximately two log units. Both I-R curves span approximately three log units of inten(cid:173)\nsity.  The  I-R  curve  due  to  combined center and  surround  illumination  can  be described \nby  the  function  VNm = I/(I+k)  [16]  where  Vm  is a  peak  hyperpolarization  and  k  is  a \nconstant intensity generating half-maximal response. This relationship [x/(x+k)]  was sug(cid:173)\ngested  to  be an  indication  of the  light adaptation  [2].  The  I-R  curve  plotted  using  peak \nresponse  values  (open  circles),  fits  a  continuous  line  drawn  according  to  equation  (1-\nexp(-kx\u00bb.  This  has  been  argued  previously  to  indicate absence of light adaptation  [2,1]. \nThere  is  little if any  compression or change  in  gain  after the  shift of the cone operating \npoint to  some  new  domain  of intensity.  The  results  suggest  that  peripheral  illumination \ncan  shift  the  center-spot  elicited  I-R  curve  of  the  cone  thus  resetting  the  response(cid:173)\ngenerating mechanism in cones. \n\n2.4 \n\nSimulation of a cone model \n\nThe  results  presented  in  the  previous sections  imply that maximal hyperpolarization  for \nthe  cone  membrane  is  not  limited  by  the  saturation  in  the  phototransduction  process \nalone.  It seems reasonable  to assume  that such a limit may be  in  part detennined by the \nbatteries of involved  ions.  Furthennore,  it appears  that shifting I-R  curves along the  in(cid:173)\ntensity domain is  not dependent solely on the light adaptation mechanism localized to the \nouter segment of a  cone.  To  test these  propositions we  developed  a simplified compart(cid:173)\nmental  model  of a  cone  (Fig.2.)  and  we  exercised  it  using  SPICE  (Vladimirescu et al., \n1981). \nAll  interactions  can  be  modeled  using  Kirchoffs  current  law;  membrane  current  is \ncm(dv/dt)+lionic '  The  leakage  current  is  lleak = Gleak(V m-EleaJ,  light  sensitive  current  is \nIlight = Glight *(V m-Elight)  and  the  feedback  current is  lib = Gfb *(V m-Efb).  The left branch \nrepresents  ohmic  leakage  channels  (Gleak)  which  are  associated  with  a  constant battery \nEleak  (  -70  mY).  The middle branch  represents  the  light sensitive conductance  (Glight)  in \nseries with + 1 m V ionic battery (Elight)  [18]. Light adaptation effects could be incorporat(cid:173)\ned here by making Glight time varying and dependent on internal concentration of Calci(cid:173)\num  ions.  In  our preliminary  studies  we  were  only  interested  in  examining  whether  the \nshift of I-R  is  possible and  if it would explain  the  disappearance  of depolarizing FB  re(cid:173)\nponse with hyperpolarization  by the center light. This can be done with passive measure(cid:173)\nments of membrane potential amplitude. The right-most branch represents ionic channels \nthat are  controlled by  the  feedback  synapse.  With,  Efb = -65  mV  [11]  Gfb  is  a  time  and \nvoltage independent feedback conductance. \n\n\f394 \n\nSkrzypek \n\nThe  input  resistance  of an  isolated  cone  is  taken  to  be  near  500  Mohm  (270  Mohm \nAttwell,  et  al.,  82).  Assuming  specific  membrane  resistance  of 5000 Ohm*cm*cm  and \nthat a cone is 40 microns long and has a 8 micron diameter at the base we get the leakage \nconductance  G1eak  =  1/(lGohm).  In  our  studies  we  assume  G1eak  to be  linear  altghouth \nthere  is evidence that cone membrane  rectifies (Skrzypek. 79). The Glight and Gfb are as(cid:173)\nsumed to be equal and add up to l/(lGohm). The Glight varies with light intensity in pro(cid:173)\nportion of two to three log units of intensity for a tenfold change in conductance. This re(cid:173)\nlation  was  derived  empirically,  by  comparing  intensity  response  data  obtained  from  a \ncone  (V m=f(LogI)}  to  {V m=f(LogGli$hJ}  generated  by  the  model.  The  changes  in  Glb \nhave  not  been  calibrated  to  changes  In  light intensity  of the  annulus.  However,  we  as(cid:173)\nsume that G lb can not undergo variation larger that Glight. \nFigure  3  shows  the  membrane  potential  changes  generated  by  the  model  plotted  as  a \nfunction  of Rlighh  at different  settings  of the  \"feedback\"  resistance  Rib.  With  increasing \nRfb,  there  is  a  parallel  shift along  the  abscissa  without any  changes  in  the  shape of the \ncurve.  Increase in Rlight  corresponds to increase in light intensity and the increasing mag(cid:173)\nnitude of the light response  from  Om V (Eligh0  all the way down to  -65  m V (Efb).  The in(cid:173)\ncrease  in  Rfb  is associated  with  increasing  intensity  of the  annular  illumination,  which \ncauses additional hyperpolarization  of the  horizontal cell and consequently a decrease  in \n\"feedback\"  transmitter released  from  HC  to cones.  Since we assume  the Etb =--65mV,  a \nmore negative level than the nonnal resting membrane potential. a decrease in Gfb would \ncause  a  depolarizing  response  in  the  cone.  This can  be  observed  here  as  a  shift of the \ncurve  along  the  abscissa.  In  our  model,  a  hundred  fold  change  in  feedback  resistance \nfrom  O.OlGohm to  IGohm, resulted in shift of the  \"response-intensity\" curve by approxi(cid:173)\nmately two log units along the abscissa. The relationship between changes in Rfb and the \nshift of the  \"response-intensity\"  curve  is  nonlinear and  additional increases  in  Rfb from \n1 Gohm to lOOGohm results in decreasing shifts. \n\nMembrane  current  undergoes  similar  parallel  shift  with  changes  in  feedback  conduc(cid:173)\ntance.  However, the photocurrent (lligh0  and the feedback current (lfb),  show only satura(cid:173)\ntion with increasing Glight  (not shown). The limits of either llight or Ifb currents are defined \nby  the  batteries of the model. Since these currents are associated with batteries of oppo(cid:173)\nsite polarities,  the  difference  between  them  at various  settings  of the  feedback  conduc(cid:173)\ntance Gfb  determines  the amount of shift for Ileak  along the abscissa.  The compression in \nshift  of  \"response  intensity\"  curves  at  smaller  values  of Glb  results  from  smaller  and \nsmaller  current  flowing  through  the  feedback  branch  of the  circuit.  Consequently.  a \nsmaller Gib changes are required to get response in the dark than in the light. \n\nThe shifting of the \"response-intensity\" curves generated by our model is not due to light \nadaptation as described by [1,2]  although it is possible that feedback effects could be in(cid:173)\nvolved in  modulating  light-sensitive channels.  Our model suggests  that in order to gen(cid:173)\nerate additional light response after the membrane of a cone was fully  hyperpolarized by \nlight,  it  is  insufficient  to  have  a  feedback  effect  alone  that  would  depolarize  the  cone \nmembrane.  Light sensitive channels  that  were  not previously  closed  [18]  must also  be \navailable. \n\n\fFeedback Synapse to Cone and Light Adaptation \n\n395 \n\n3  DISCUSSION \n\nThe results presented  here  suggest that synaptic feedback  from  horizontal cells to cones \ncould contribute to the process of light adaptation at the photoreceptor level.  A complete \nexplanation of the  underlying  mechanism requires further studies but the results seem  to \nsuggest that depolarization of the cone membrane by a peripheral illumination, resets the \nresponse-generating process in the cone. This result can be explained withing the frame(cid:173)\nwork of the current hypothesis of the light adaptation, recently  summarized by [6]. \nIt  is  conceivable  that  feedback  transmitter  released  from  horizontal  cells  in  the  dark, \nopens channels  to  ions with reversal potential near -65  mV  [11].  Hence, hyperpolarizing \ncone  membrane by  increasing  center spot intensity  would reduce  the  depolarizing  feed(cid:173)\nback response as  cone nears  the battery of involved ions.  Additional increase in annular \nillumination, further  reduces  the  feedback  transmitter and  the  associated feedback  con(cid:173)\nductance  thus  pushing  cone's  membrane  potential  away  from  the  \"feedback\"  battery. \nEventually, at some values of the center intensity, cone membrane is so close to -65 mV \nthat no change in feedback conductance can produce a depolarizing response. \n\nACKNOWLEDGEMENTS \n\nSpecial gratitude to Prof. Werblin for providing a superb research environment and  gen(cid:173)\nerous support during early part of this project We acknowledge  partial support by  NSF \ngrant ECS-8307553, ARCO-UCLA Grant #1, UCLA-SEASNET Grant KF-21, MICRO(cid:173)\nHughes  grant #541122-57442,  ONR  grant #NOOOI4-86-K-0395,  ARO  grant  DAAL03-\n88- K-00S2 \n\nREFERENCES \n1.  Nakatani,  K.,  &  Yau.  K.W.  (1988).  Calcium  and  light adaptation  in  retinal rods  and \ncones. Nature. 334,69-71. \n\n2.  Matthews, H.R.,  Murphy, R.L.W., Fain, G.L.,  &  Lamb,  T.D.  (1988).  Photoreceptor \nlight adaptation is mediated by cytoplasmic calcium concentration. Nature, 334, 67-69. \n\n3.  Normann, R.A.  &  Werblin,  F.S.  (1974).  Control of retinal  sensitivity.  I. Light and \ndark-adaptation of vertebrate rods and cones. J. Physiol. 63, 37-61. \n\n4.  Werblin, F.S.  &  Dowling, J.E.  (1969).  Organization  of the  retina of the mudpuppy, \nNecturus maculosus. II. Intracellular recording. J. Neurophysiol. 32, (1969),315-338. \n\n5.  Pugh, E.N. & Altman, J. Role for calcium in adaptation. Nature 334, (1988),  16-17. \n\n6.  O'Bryan P.M., Properties of the dpolarizing  synaptic potential evoked by  peripheral \nillumination in cones of the turtle retina.  J.Physiol. Lond. 253, (1973), 207-223. \n\n7.  Skrzypek J., Ph.D. Thesis, University of California at Berkeley, (1979). \n\n8.  Skrzypek,  J.  &  Werblin,  F.S.,(1983).  Lateral  interactions  in  absence  of feedback  to \ncones. J. Neurophysiol. 49, (1983), 1007-1016. \n\n\f396 \n\nSkrzypek \n\n9. Skrzypek. J. & Werblin. F.S., All horizontal cells have center-surround antagonistic re(cid:173)\nceptive fields. ARVO Abstr., (1978). \n\n10. Lasansky. A.  Synaptic action mediating cone responses to annular illumination in the \nretina of the larval tiger salamander. J. Physiol.  Lond. 310, (1981), 206-214. \n\n11.  Skrzypek  J .\u2022  Electrical  coupling  between  horizontal  vell  bodies  in  the  tiger \nsalamander retina.  Vision Res. 24, (1984), 701-711. \n\n12. Naka, K.I.  & Rushton, W.A.H.  (1967).  The generation and spread of S-potentials in \nfish  (Cyprinidae) J. Physiol., 192, (1967),437-461. \n\n13. AttweU, D., Werblin, F.S. & Wilson. M.  (1982a).  The properties of single cones iso(cid:173)\nlated from  the tiger salamander retina. J. Physiol.  328.259-283. \n\n\fFeedback Synapse to Cone and Light Adaptation \n\n397 \n\nI \nI  C \n\n--L-\n--r-\n\nm \n\nE'Uk \n-70mv \n\nG \n\nIUk \n\nG \n\nIfgnt \n\nG fa \n\n+ I mv \n\nFig.  2  Equivalent circuit model of a  cone based on  three  different transmembrane chan(cid:173)\nnels.  The ohmic leakage channel consists of a  constant conductance GlcGJ:  in  series  with \nconstant battery ElcGJ:.  Light sensitive channels are represented  in  the  middle branch by \nG/i,J\".  Battery Eli,ltt. represents the reversal potential for light response at approximately \nOm V.  Feedback synapse  is  shown  in  the  right-most branch  as  a  series  combination of \nGfb  and  the  battery  Efb  = -65mV.  representing  reversal  potential  for  annulus  elicited, \ndepolarizing response measured in a cone. \n\n-> \n:! -c .. \n-0 a. \n\u2022 \nc \n\u2022 ~ \n.a e \u2022 :I \n\n0.02 \n\n0.00 \n\n-0.02 \n\n\u00b7o.Ot \n\n.0.01 \n\n--0- 'Vmlat Rfbo.Ol Q \n~ 'Vmlat RIb-.' Q \n\n'Vm lor At., Q \n\n---\n\n'Vmlat Rlb-I OG \n\n.0.01 \n\n.\u2022 \n\n\u00b72 \n\n0 \n\n2 \n\nLog Alight \n\n\u2022 \n\nFig.  3  Plot of the membrane potential  versus  the logarithm of light-sensitive  resistance. \nThe  data  was  synthesized  with  the  cone  model  simulated by  SPICE.  Both  current  and \nvoltage curves can be fitted by x/(x+k) relation (not shown) at all different settings of Gtb \n(Rfb)  indicated  in  the  legend.  The shift of the  curves.  measured at  1/2  maximal  value \n(k=x) spans about two log units.  With increasing settings of Rtb (10 Gohms). curves be(cid:173)\ngin to cross (Vm at -6SmV) signifying decreasing contribution of \"feedback\" synapse. \n\n\ft.IJ \n\\D \n00 \n\ntil \"\" ~ '0 \nn> \"\" \n\n\u2022 \n\nb \n\nto, \n\n-> \nE -11 \n\n0 \n\n! \n> \n\u2022 \nE  \u00b710 \n> \n\n, --\n\n.........\"\". \n\nIOnN \n\n'lie  I \n,.., ___ _ \n\n~ \n\n-.~-------\n\ns \nC \n\n\u00b720 \n\n-31, \n\u00b71 \n\n.... ,,-...... \n\n'l--q \n\n,...... \n\n'Q , \n'\\0 \n, \n\" 6, \n\nb.e.Q_~.DO \n\ni '  \n\u00b71 \n\n\u2022\u2022 \n\ni . . . .  \n\u00b72 \n\n2 \n\n\u2022 \n.......... \n\nFig.  1  (a)  Series  of responses  to  a  combination of center  spot  and  annulus.  SurroWld  illumination  (S)  was  fixed  at  -3.2  l.u. \nthroughout the experiment.  Center spot intensity (C) was increased in 0.5 I.u. steps as indicated by the numbers near each trace. \nIn the dark (upper-most trace) surround illumination had no measurable effect on the cone membrane potential.  Annulus-elicited \ndepolarizing response increased with  intensity in  the center up 10  about -3  l.u.  Further increase of the  spot intensity diminished \nthe surround response.  Plot of the peak hyperpolarizing response versus center spot intensity in log units in shown in (b) as open \ncircles.  It  fits  the  dashed  curve  drawn  according  to  equation  l-exp(-kx).  The curve  indicated  by  filled  circles  represents  the \nmembrane potential measurements taken in the middle of the depolarizing response.  This data can be approximated by a continu(cid:173)\nous curve  derived  from  x/(x+k).  All  membrane  potential  measurement are  made  with  respect 10  the resting  level  in  the  dark. \nThis result shows that in the presence of peripheral illumination,  when  the  feedback is  activated, membrane potential follows  the \nintensity-response curve which is shifted along the Log I axis. \n\n\f", "award": [], "sourceid": 379, "authors": [{"given_name": "Josef", "family_name": "Skrzypek", "institution": null}]}