Impulse: The Premier Journal for Undergraduate Publications in the Neurosciences.June 2004, 1 (1): 1-58

Xenopus laevis RetinalGanglion Cell DendriticArbors Develop Independentlyof Visual Stimulation

Rebecca L. Rigel and Barbara LomBiology Department and Neuroscience Program,Davidson College, Davidson, NC, 28035-7118

Newly formed neurons must locate theirappropriate target cells and then form synapticconnections with these targets in order toestablish a functional nervous system. In thevertebrate retina, retinal ganglion cell (RGC)dendrites extend from the cell body and formsynapses with nearby amacrine and bipolar cells.RGC axons, however, exit the retina andsynapse with the dendrites of midbrain neuronsin the optic tectum. We examined how visualstimulation influenced Xenopus RGC dendriticarborization. Neuronal activity is known to bean important factor in shaping dendritic andaxonal arborization. Thus, we reared tadpoles indark and light environments then usedrhodamine dextran retrograde labeling toidentify RGCs in the retina. When we comparedRGC dendritic arbors from tadpoles reared indark and light environments, we found nomorphological differences, suggesting thatphysiological visual activity did not contributeto the morphological development of XenopusRGC dendritic arbors.

Key Words: dendrite; retina; activity; light

IntroductionIn the developing Xenopus laevis visual

system retinal ganglion cells (RGCs) extendaxons out from the retina to their target in themidbrain. While RGC axons are navigatingtoward the optic tectum where they arborize andform synapses with tectal neurons, dendrites arealso beginning to extend from the RGC soma inorder to synapse with other retinal neurons. Inmany higher vertebrates, the effect of neuronalactivity on development of RGC axonal arborshas been well characterized. However, in bothhigher and lower vertebrates considerably less isknown about RGC dendritic arbor development.

Consequently, we examined how visualstimulation influenced RGC dendriticarborization in Xenopus laevis by comparing themorphologies of RGCs in tadpoles that had beenreared in light versus dark environments.

Figure 1 Vertebrate Retinal Anatomy Diagram of avertebrate retina illustrating retinal connectivity, cellularand synaptic cell layers, and cell types in the retina. PR=photoreceptor layer; ONL=outer nuclear layer; INL=innernuclear layer; IPL=inner plexiform layer; GCL=ganglioncell layer; r=rod; c=cone; HC=horizontal cell; BP=bipolarcell; A=amacrine cell; RGC=retinal ganglion cell. Imageadapted from Cajal.

Retinal Ganglion CellsRetinal ganglion cells (RGCs) connect

the eye to the brain via their axons (,) whichmake up the optic nerve. Their cell bodies arelocated within the innermost ganglion cell layer(GCL) of the retina, and their dendrites extendinto the inner plexiform layer (IPL) where theyarborize and receive synaptic input from

Page 51 to 58Impulse: The Premier Journal for Undergraduate Publications in the Neurosciences.

June 2004, 1 (1): 1-58

amacrine cells and bipolar neurons (fig. 1).RGC axons extend from the cell body, cross theoptic chiasm, and navigate through the midbrainto reach the contralateral optic tectum. Therethey arborize and form synapses with tectalneuron dendrites to form a retinotopic map, ahighly ordered topographical map of the visualworld. RGCs form the only neuronalconnections between the eyes and the brain.Because RGCs convey all visual information tothe brain, if RGCs do not develop properly, anorganism’s vision may be significantlycompromised.

Xenopus: A Model System for RGC StudiesXenopus laevis, the African clawed frog,

is a model organism because these animals canbe bred in captivity, their eggs can be harvestedin abundance, they reproduce by externalfertilization, and they develop rapidly, whichmakes tadpoles accessible for study at all stagesof development (Nieuwkoop and Faber, 1967).Xenopus tadpoles have been used to study visualsystem development (Johnson and Harris, 2000;D i n g w e l l et al., 2000). Most of ourunderstanding of Xenopus RGC developmenthas focused on axon navigation and arborization(Cohen-Cory, 1999, 2002; Cline, 2001), but it isalso possible to study RGC dendriticdevelopment in Xenopus (Sakaguchi et al.,1984; Holt, 1989; Lom and Cohen-Cory, 1999;Lom et al., 2002;). Retrograde transport ofrhodamine dextran from the tectum tocontralateral RGCs makes it possible tovisualize the morphology of individualfluorescent RGC dendrites in vivo (Lom andCohen-Cory, 1999; Lom et al., 2002).

Xenopus RGC Dendritic DevelopmentAlthough more is known about the

development of Xenopus RGC axons, RGCdendritic development is also well characterized(fig. 2; Sakaguchi et al., 1984; Holt, 1989; Lomand Cohen-Cory, 1999; Lom et al., 2002).RGCs extend primary dendrites directly outfrom the cell body after the beginning ofaxonogenesis, and before their axons havereached their target at the optic tectum. Asaxonogenesis continues, more short, unbranchedprimary dendrites extend from the RGC cellbody and begin to lengthen toward their target,

the IPL of the retina. In the IPL, RGC dendritessynapse with amacrine and bipolar cells, whileRGC axons synapse with dendrites of tectalneurons in the contralateral optic tectum. Thefirst synaptic activity in the brain is detected atstage 39, after RGC axons synapse with thetectum (Sakaguchi et al., 1985; Holt, 1989).Xenopus RGC dendritic morphologies appearsimilar through developmental stage 45;however, after stage 46, RGC dendriticmorphologies begin to be divisible into threecategories based on soma size and dendriticbranching arrangement (Sakaguchi et al., 1984).

Figure 2 Time Course of RGC axonal and dendriticarborization in the Xenopus visual system. Transversediagram of the Xenopus brain and one eye depicts thecontralateral retinotectal projection as it develops. Imagesadapted from Sakaguchi et al. (1984), Nieuwkoop andFaber (1967), and Chien and Harris (1994).

Neuronal Activity Influences AxonArborization in the Developing VisualSystem

As the visual system develops, neuronalactivity is very important for establishing andrefining visual connections (Cohen-Cory, 2002).Studies of RGC development have examined theeffects of neuronal activity on axonal growth,arborization, and formation of the retinotopicmap on the midbrain. In several species,including cats and zebrafish, action potentials in

Page 51 to 58RGC Dendritic Arborization in Xenopus laevis; Rigel and Lom

the midbrain and/or retina have been shown tobe necessary for the proper development of RGCaxonal arbors. Fetal cat RGCs exposed to thesodium channel blocker tetrodotoxin (TTX) donot fire action potentials during the period whenRGC axons synapse with their target neurons inthe midbrain. Terminal axon arbors of theseRGCs that were deprived of all endogenousretinal activity were much larger than theterminal axon arbors of control RGCs (Sretavanet al., 1998). Size differences in axonal arborsof control and TTX-treated RGCs indicated thatneuronal activity played an important role in thedevelopment of terminal axon arbors. In asimilar study, TTX was used to inhibit actionpotentials in developing zebrafish RGCs(Gneugge et al., 2001). Zebrafish RGC terminalaxon arbors were much larger when exposed toTTX as compared to control RGC axon arborsalso indicating that neuronal activity isimportant in RGC axon arborization.

Neuronal Activity Influences Xenopus laevisRGC Axon Arborization

Studies of Xenopus RGCs have shownthat, like in mammals and in zebrafish, actionpotentials are also important for thedevelopment of axonal arbor morphology at thetectum. Neuronal activity stabilizes XenopusRGC axonal arbors. Time-lapse imaging ofindividual RGC axon arbor morphologiesrevealed that TTX suppression of retinal activityresulted in significantly more complex axonarbors (Cohen-Cory, 1999). More axonalbranches were added than were eliminated,which created unstable axonal arbors (Cohen-Cory, 1999).

In a related study the glutamate agonistNMDA was used to reduce postsynaptic activityin the Xenopus tectum. The number of stablebranches on RGC axons dramatically decreasedand significantly more branches were added andretracted (O’Rourke et al., 1994). These resultsindicate that developing RGC axons rapidlyextend short pioneer branches that probe theenvironment presumably in search of a place toform a stable synapse with a tectal neuronaldendrite. Thus, the increased extension andretraction of axon branches in RGCs deprived ofpost-synaptic activity may be an indication ofsynaptic instability.

Neuronal Activity Influences RGC DendriticArborization

Although most studies of neuronalactivity have examined RGC axon developmentat the target, neuronal activity is also importantin the development of dendritic arbors within theretina. In one study, TTX was used to blockaction potentials in the eyes of kittens (Wong etal., 1991). RGCs in eyes deprived of neuronalactivity showed reduced complexity of dendriticarbors, and dendrites with fewer branches andspines. These results indicated that retinalactivity participates in RGC dendriticarborization. In a reciprocal study, Campbell etal. (1997) treated kitten midbrains with TTX todetermine how absence of activity at the targetaffected growth of RGC dendritic arbors. TTXtreated animals had slightly more dendriticspines, but in general, RGC dendriticdevelopment in TTX treated animals was similarto untreated animals. Taken together the resultsof these studies suggest that neuronal activitywithin the retina may be more influential inregulating RGC dendritic arborization thanneuronal activity in the midbrain.

The effects of neuronal activity on thedevelopment of Xenopus RGC dendritic arborsare unknown. In our study, we investigated theeffects of visual stimulation on Xenopus RGCdendritic arborization to determine ifphysiological activity induced by light versusdark environments had any influence on RGCdendritic arborization.

MethodsWe examined the effects of neuronal

activity on Xenopus RGC dendritic morphology.Neuronal activity was modulated by controllingthe light environment in which the Xenopustadpoles were reared to control the amount oflight-evoked retinal activity. We conducted fourexperiments in which we evaluated the effects oflight on the growth of RGC dendrites inXenopus tadpoles (fig. 3). All reagents wereobtained from Fisher or Sigma unless otherwiseindicated. Tadpoles were reared in 20%modified Steinberg’s solution (60 mM NaCl,

Page 51 to 58Impulse: The Premier Journal for Undergraduate Publications in the Neurosciences.

June 2004, 1 (1): 1-58

Figure 3 Diagram of Experimental Methods. Tadpoleswere reared in light or dark environments during the periodwhen RGC dendritic arbors begin to form. RGC dendriteswere visualized by retrograde rhodamine dextran labelingand visualized with fluorescence microscopy.

0.67 mM KCl, 0.34 mM Ca(N03)2, 0.83 mMMgSO4, 10 mM HEPES, and 40 mg/lgentamycin, pH 7.4; Keller, 1991).Approximately 0.001% phenylthiocarbamidewas included to reduce pigmentation. Animalswere anesthetized during dye injection andbefore fixation with 0.05% tricanemethanesulfonate. All animal care and useprotocols were approved by the DavidsonCollege Institutional Animal Care and UseCommittee (protocols #8-03-01 and 8-03-03).

We obtained Xenopus embryos by invitro fertilization (Sive et al., 2000) and stagedembryos according to Nieuwkoop and Faber(1967). When the tadpoles reached stage 35/36(the onset of dendritogenesis) we placed twoplastic dishes of tadpoles under the twofiberoptics of a halogen lamp source. One dishwas covered with foil to shield the tadpoles fromlight. At stage 43 (when RGC axons haveinnervated the optic tectum), we microinjectedthe left optic tectum of each anesthetized tadpolewith the fluorescent dye rhodamine dextran

(Molecular Probes) and rapidly returned thetadpoles to their light or dark environment. Atstage 45 (before RGC morphologies subdivideinto three categories) we fixed both groups oftadpoles in 4% paraformaldehyde overnight at4° C. We then removed the right eye of thetadpoles and flattened the retina ontomicroscope slides in order to view thefluorescently stained RGC dendrites. BecauseRGC axons are the only anatomical connectionbetween the optic tectum and the retina, anyfluorescently labeled neurons in the retina arenecessarily RGCs (Lom and Cohen-Cory, 1999).After capturing digital images of the RGCdendritic arbors using fluorescence microscopy,we traced each dendrite’s morphology manually,then scanned the traced images into thecomputer for length analysis with Scion Imagingsoftware (www.scioncorp.com). All datacollection and analysis were performed underblind conditions to minimize any potentialbiases. For each RGC we determined cell bodyarea, total dendritic arbor length, and cell bodydiameter. Numbers of primary dendrites anddendrite branch tips were counted manually. Foreach of the four experiments, we normalized thedata to the average of the dark values for eachparameter. Because the average length of thedendrites and complexity of dendritic arborsvaried between each experiment, normalizingthe data was necessary for comparison of databetween individual experiments. We used PrizmSoftware (GraphPad) to perform an unpaired,two-tailed t-test for all data analysis.

Results and DiscussionOur results indicate that physiological

stimulation of RGCs with light does not affectthe morphology of developing RGC dendriticarbors in Xenopus laevis. RGC dendritic arborsfrom tadpoles reared in light and dark conditionsappeared morphologically similar (figure 4).For all parameters, including number of primarydendrites, number of dendritic tips, ratio of tipsto dendrites, length of dendrites, and area ofRGC somas, and the differences between light-reared and dark-reared RGC dendrites wereinsignificant (p > 0.05; figure 5). RGCs indark-reared tadpoles extended 4.1 + 0.2 (SEM)primary dendrites compared to an average of 4.5

Page 51 to 58RGC Dendritic Arborization in Xenopus laevis; Rigel and Lom

Figure 4 RGCs from tadpoles reared in the lightenvironment and dark environment. No obviousmorphological differences were observed between RGCdendritic arbors of tadpoles reared in light versus darkenvironments.

+ 0.3 primary dendrites in light-reared tadpoles.Similarly, RGCs in dark-reared tadpolesextended 14.9 + 1.3 branch tips compared to anaverage of 14.4 + 1.2 branch tips per RGC inlight-reared tadpoles.

While our results indicate that RGCdendritic morphogenesis is independent ofenvironmental light, it is possible that neuronalactivity in the retina does influence growth ofdendritic arbors in developing X e n o p u s .Because neurons may become desensitized whenexposed to a continuous stimulus, it is possiblethat tadpoles reared in the light becameaccustomed to the light and that retinal neuronsceased or altered their firing in response toconstant light stimuli. Retinal activity maytherefore influence developing dendritic arbors,but because the neurons became habituated tothe constant, bright light their firing patternswere altered and the light stimulation did nothave a significant influence on dendritedevelopment. Future studies in which thetadpoles would be exposed to intermittent lightinstead of continuous light would help resolvethis question. Using a strobe light set at a

Figure 5 Light stimulation does not alter themorphology of RGC dendritic arbors. The averagevalues of each parameter measured in light-reared versusdark-reared RGCs were ratioed for primary dendritenumber, branch tip number, branch tips per dendrite, totaldendrite arbor length, and soma area. Thus, a value at ornear 1.0 indicates that there was no difference in theaverage of the morphological parameter. (n = 161 RGCs)

specific frequency, the retinal neurons would notbecome accustomed to the light and would thusbe more likely to respond to light. RGCs oftadpoles reared under the strobe light could thenbe compared to RGCs of tadpoles shielded fromthe strobe light to determine if retinal activityintermittently evoked from an exogenous lightsource does in fact play a role in development ofRGC dendritic morphology.

In a study of Xenopus optic tectalneurons (the synaptic targets of RGC axons),visual stimulation significantly increased the rateof RGC dendritic growth, the number of newbranches, and the length of dendritic branches(Sin et al., 2002). The investigators used a panelof green light to stimulate the tadpole retinalneurons. The panel of light had rows that turnedon sequentially for one second, as in a wavemotion. Because tectal cells do not becomeadapted to repeated stimulation it is likely thatRGCs do not get adapted to repeated stimulationeither (Sin et al., 2002). If this is the case,exposing tadpoles to repeated intervals of lightand dark may have an affect on the growth ofRGC dendritic arbors not observable in ourstudy.

Spontaneous neuronal activity may alsoaffect development of RGC dendritic arbors inXenopus. In vitro recordings of electrical

Page 51 to 58Impulse: The Premier Journal for Undergraduate Publications in the Neurosciences.

June 2004, 1 (1): 1-58

activity in the retinas of chicks, neonatal rabbits,fetal rats, and embryonic ferrets has indicatedthat spontaneous activity is important for thedevelopment of morphologically correct RGCsand cells in the lateral geniculate nucleus (LGN;Wong et al., 1998; Wong, 1999). Spontaneousbursts of activity present in the developingretinas of higher and lower vertebrates suggeststhat endogenous retinal activity may be essentialfor development and refinement of primaryvisual connections (Wong, 1999). Futurestudies in X e n o p u s could also considermeasuring retinal activity in light and darkenvironments.

Wong and Wong (2000) have alsoshown that when endogenous activity in cellswith high spontaneous activity is inhibited,dendritic mobility decreases. Endogenousneuronal activity appears to be necessary fornormal postnatal maturation of RGC axons(Wong et al., 1991). In embryonic chicks andferret kits, spontaneous retinal activity isimportant for refinement of the retinotopic map.Spontaneous activity in chick and ferret retinasis dependent on excitatory cholinergicstimulation (Wong et al., 1998). In ferret kits,blockade of endogenous retinal activity bycholinergic blockers altered the pattern of axonallamination on the LGN. The axonal projectionfrom the active retina was greatly enlarged andformed connections on the LGN where theopposite eye normally synapses. Projectionsfrom the inactive eye were significantly smallerthan normal. The abnormal development ofaxonal projections indicates that spontaneousretinal activity helps drive development ofstereotypical connections with the brain beforevision begins (Penn et al., 1998).

Lohmann et al. (2002) studied theeffects of TTX on embryonic chick RGCs. Theyfound that in developing RGCs Ca+2 isspontaneously released both throughout theneuron and locally within small dendriticsegments. Ca+2 release helps stabilize RGCdendritic structure and blockade of local Ca+2

release causes immediate dendritic retraction.This work suggests that Ca+2 release may be away in which afferent activity regulatesdendritic structure in developing dendrites.

Spontaneous neuronal activity is alsoimportant in the development of axonal

connections at the visual cortex. Catalano andShatz (1998) injected TTX into the brain of catsduring the period when LGN axons werereaching their target at the visual cortex. At thisperiod of development spontaneous actionpotentials in the retinas are relayed to the LGNand likely help guide growing axons. The TTXinjections caused the axons to extend significantprojections within the subplate of the corticalareas where the neurons do not usually extend.LGN axons that did not reach the tectum weretopographically disordered. Axonal disarray atthe visual cortex indicates that neuronal activityis necessary for thalamic axons to locate thecorrect target and to establish appropriateconnections.

Whereas rearing tadpoles in the darkinhibits retinal activity provoked by externalsources, it does not silence spontaneous retinalactivity that might affect RGC dendritic growth.To determine if spontaneous retinal activityinfluences RGC dendritic growth futureexperiments could inject TTX into the eyes oftadpoles to silence any spontaneous retinalactivity. RGC dendritic arbors of tadpoles notexposed to TTX could then be compared to theTTX exposed RGCs to determine what effectsspontaneous activity has on developing RGCs.

ConclusionAlthough our results indicate that

physiological stimulation of RGCs by constantlight does not influence Xenopus RGC dendriticarborization, it is still possible that neuronalactivity plays a role in early dendritic arbordevelopment. Neuronal activity is important forRGC dendritic and axonal arborization in otherorganisms, and it is probable that some type ofneuronal activity may be important in thedevelopment of morphologically correctdendritic arbors in Xenopus. More research isnecessary to determine whether endogenousactivity or a different type of physiologicalstimulation influence Xenopus RGC dendriticarbor development and morphology.

ReferencesCampbell G, Ramoa AS, Stryker MP, Shatz CJ

(1997) Dendritic development of retinalganglion cells after prenatal intracranial

Page 51 to 58RGC Dendritic Arborization in Xenopus laevis; Rigel and Lom

infusion of tetrodotoxin. Vis Neurosci 14:779-788.

Catalano SM, Shatz CJ (1998) Activity-dependent cortical target selection bythalamic axons. Science 281: 559-562.

Chien CB, Harris WA (1994) Axonal guidancefrom retina to tectum in embryonic Xenopus.Curr Top Biol. 29: 135-169.

Cline HT (2001) Dendritic arbor developmentand synaptogenesis. Curr Opin Neurobiol11: 118-126.

Cohen-Cory S (1999) BDNF modulates, butdoes not mediate, activity-dependentbranching and remodeling of optic axonarbors in vivo. J. Neuroscience 19: 9996-10003.

Cohen-Cory S (2002) The developing synapse:the construction of synaptic structures andcircuits and its modulation by neuronalactivity. Science 298: 770-776.

Dingwell KS, Hold CE, Harris WA (2000) Themultiple decisions made by growth cones ofRGCs as they navigate from the retina to thetectum in Xenopus embryos. J Neurobiol 44.246-59.

Gneugge L, Schmid S, Neuhauss SC (2001)Analysis of the activity-deprived zebrafishmutant m a c h o reveals an essentialrequirement of neuronal activity for thedevelopment of a fine-grained visuotopicmap. J. Neurosci 21: 3542-3548.

Holt CE (1989) A single-cell analysis of earlyretinal ganglion cell differentiation inXenopus: from soma to axon tip. J Neurosci9: 3123-45.

Johnson KG, Harris WA (2000) Connecting theeye with the brain: the formation of theretinotectal pathway. Results Probl CellDiffer 31: 157-77.

Keller R (1991) Early embryonic developmentof Xenopus laevis. In: Xenopus laevis:

practical uses in cell and molecular biology(Kay BK, Peng HB, eds), pp 102-116. SanDiego: Academic.

Lohmann C, Myhr KL, Wong RO (2002)Transmitter-evoked local calcium releasestabilizes developing dendrites. Nature 418:177-181.

Lom B, Cohen-Cory S (1999) Brain-derivedneurotrophic factor differentially regulatesretinal ganglion cell dendritic and axonalarborization in vivo. J Neurosci 19: 9928-9938.

Lom B, Cogen J, Lontok AY, Vu T, Leung A,French A, Cohen-Cory S (2002) Local andtarget-derived BDNF differentially modulateretinal dendritic arborization in vivo. JNeurosci 22:7639-49.

Nieuwkoop PD, Faber J (1967) Normal table ofXenopus development. Amsterdam, Holland:Elsevier.

O’Rourke NA, Cline HT, Fraser SE (1994)Rapid remodeling of retinal arbors in thetectum with and without blockade of synaptictransmission. Neuron 12: 921-934.

Penn AA, Riquelme PA, Feller MB, Shatz CJ(1998) Competition in retinogeniculatepatterning driven by spontaneous activity.Science 279: 2108-2112.

Sakaguchi DS, Murphey RK (1985) Mapformation in the developing Xenopusretinotectal system: an examination ofganglion cell terminal arborizations. J.Neurosci 5: 3228-3245.

Sakaguchi DS, Murphey RK, Hunt RK,Tompkins R (1984) The development ofretinal ganglion cells in a tetraploid strain ofXenopus laevis: a morphological studyutilizing intracellular dye injection. J CompNeurol 224: 231-251.

Sin WC, Haas K, Ruthazer ES, Cline HT (2002)Dendrite growth increased by visual activity

Page 51 to 58Impulse: The Premier Journal for Undergraduate Publications in the Neurosciences.

June 2004, 1 (1): 1-58

requires NMDA receptor and Rho GTPases.Nature 419: 475-480.

Sive HL, Grainger R, Harland RM (2000) EarlyDevelopment of Xenopus laevis. Cold SpringHarbor, NY: Cold Spring Harbor Laboratory.

Sretavan DW, Shatz CJ, Stryker MP (1998)Modification of retinal ganglion cell axonmorphology by prenatal infusion oftetrodotoxin. Nature 336: 468-471.

Wong RO (1999) Retinal waves and visualsystem development. Annu Rev Neurosci 22:29-47.

Wong RO, Herrmann K, Shatz CJ (1991)Remodeling of retinal ganglion cell dendritesin the absence of action potential activity. J.Neurobiology 22: 685-697.

Wong WT, Wong RO (2000) Rapid dendriticmovements during synapse formation andrearrangement. Curr Opin Neurobio 10:118-124.

Wong WT, Sanes JR, Wong RO (1998)Developmentally regulated spontaneousactivity in the embryonic chick retina. JNeurosci 18: 8839-8852.

AcknowledgementsThe authors thank Amy Becton, Jordan Case, IanWilloughby, Sarah Tyndall, and James Barnes for animalcare, technical assistance, and/or comments on themanuscript. This research was supported by The NationalScience Foundation, The Whitehall Foundation, andDavidson College.

Corresponding AuthorDr. Barbara Lom, Davidson College, Department ofBiology and Neuroscience Program, Box 7118, Davidson,NC 28035. balom@davidson.edu