In Vivo Development of Retinal ON-Bipolar Cell Axonal Terminals Visualized in Nyx::MYFP Transgenic Zebrafish

In vivo development of retinal ON-bipolar cell axonal terminals visualized in nyx::MYFP transgenic zebrafish

ERIC H. SCHROETER,1 RACHEL O.L. WONG,1* and RONALD G. GREGG2* 1Department of Anatomy and Neurobiology, Washington University School of Medicine, St. Louis, Missouri 2Department of Biochemistry and Molecular Biology, and Center for Genetics and Molecular Medicine, University of Louisville, Louisville, Kentucky (Received February 13, 2006; Accepted May 12, 2006!

Abstract Axonal differentiation of retinal bipolar cells has largely been studied by comparing the morphology of these interneurons in fixed tissue at different ages. To better understand how bipolar axonal terminals develop in vivo,we imaged fluorescently labeled cells in the zebrafish retina using time-lapse confocal and two photon microscopy. Using the upstream regulatory sequences from the nyx gene that encodes nyctalopin, we constructed a transgenic fish in which a subset of retinal bipolar cells express membrane targeted yellow fluorescent protein ~MYFP!. Axonal terminals of these YFP-labeled bipolar cells laminated primarily in the inner half of the inner plexiform layer, suggesting that they are likely to be ON-bipolar cells. Transient expression of MYFP in isolated bipolar cells indicates that two or more subsets of bipolar cells, with one or two terminal boutons, are labeled. Live imaging of YFP-expressing bipolar cells in the nyx::MYFP transgenic fish at different ages showed that initially, filopodial-like structures extend and retract from their primary axonal process throughout the inner plexiform layer ~IPL!. Over time, filopodial exploration becomes concentrated at discrete foci prior to the establishment of large terminal boutons, characteristic of the mature form. This sequence of axonal differentiation suggests that synaptic targeting by bipolar cell axons may involve an early process of trial and error, rather than a process of directed outgrowth and contact. Our observations represent the first in vivo visualization of axonal development of bipolar cells in a vertebrate retina. Keywords: Zebrafish, Bipolar cell, Nyctalopin, Axonal development

Introduction Cepko, 2005!, Bhlhb4 ~Bramblett et al., 2004!, Vsx1 ~Chow et al., 2004!, and Irx5 ~Cheng et al., 2005!, affect the differentiation of Retinal bipolar interneurons form the essential link between photo- bipolar cells. Immunolabeling for vesicular glutamate transporters receptors and the output layer of the retina, the ganglion cell layer. also provided interesting insights into when ON and OFF bipolar Synaptic connections in the inner retina between bipolar cells and cell axonal terminals differentiate ~Sherry et al., 2003!. Ultrastruc- the retinal ganglion cells are organized broadly into two major tural studies across many species have revealed when bipolar cells laminae ~reviewed by Wassle, 2004!. Connectivity between retinal make synapses in the OPL and IPL ~e.g., Olney, 1968; Dubin, cells that are depolarized by increased illumination ~ON-center 1970; Nishimura & Rakic, 1987; Crooks et al., 1995; Schmitt & cells! is restricted approximately to the inner half to two-thirds of Dowling, 1999!. the inner plexiform layer ~IPL!. Conversely, connections between Although much is known about bipolar cell development across retinal neurons that are hyperpolarized by increased illumination many species, how immature bipolar cells target their axons to the ~OFF-center cells! occupy the outer third to half of the IPL. Axon appropriate ON or OFF sublamina requires further study. To do so, terminals of ON and OFF bipolar cells thus stratify in distinct complete labeling of bipolar axonal terminals across development layers within the IPL. is necessary. This can be achieved in part, by immunostaining for Studies in the past have investigated the molecular and cellular various proteins specific to bipolar cells, but such an approach factors regulating bipolar cell development. For example, loss of generally leads to labeling of populations of cells, which limit transcription factors Chx10 ~Burmeister et al., 1996; Rowan & resolution of the morphology of individual terminals ~e.g., Miller et al., 1999; Gunhan-Agar et al., 2000, 2002; Kay et al., 2004!. Also the axonal terminals of developing bipolar cells are already laminated at the earliest ages when immunolabeling reveals their Address correspondence and reprint requests to: R.O.L. Wong, Department of Biological Structure, University of Washington, Seattle, morphology, making it difficult to assess bipolar cell structure WA 98125. E-mail: [email protected] prior to axonal stratification. Golgi techniques applied to fixed *The authors contributed equally. retinas at different ages have, however, enabled comparison of the

833 834 E.H. Schroeter et al. axonal and dendritic morphology of bipolar cells across develop- et al., 2000; Gregg et al., 2003!, although only exons 2 and 3 ment ~Quesada et al., 1981; Quesada & Genis-Galvez, 1985!. But, contain coding sequence. In silico analyses of zebrafish se- direct comparison of developmental changes of individual cells or quences also identified a gene that generates a predicted cDNA within a subtype of bipolar cells is difficult with this approach. sequence encoding the zebrafish nyctalopin protein ~Accession # While much insight has been gained from studies utilizing XM_692177!. Experimental analyses of cDNA clones from zebra- fixed tissue, live imaging approaches are required to reveal the fish indicate the nyx gene in this species also contains 3 exons dynamic behaviors underlying structural changes resulting in the ~data not shown! encoding a predicted protein of 449 amino mature morphology. In recent years, it has become feasible to label acids that shares 51% identity with the mouse protein. A DNA and visualize live retinal cells by driving expression of fluorescent fragment that extended 1482 bp upstream of the 3' end of proteins using ubiquitous or cell-specific promoters ~reviewed by exon 2 ~position 25 in the ORF, Acc # XM_692177! was cloned Lohmann et al., 2005; Morgan et al., 2005; Mumm et al., 2005!. by PCR ~primers: AC-CGGCAATATTGATGATGA; GAAACG- Most recently, bipolar cell development has been studied in retinal CAAGAAATAAGCATGA! from genomic DNA. This fragment explants from transgenic mice in which ON bipolar cells express was cloned into a modified pCS2ϩ Gal4/VP16 plasmid ~Koster green fluorescent protein ~GFP! under the control of the mGluR6 & Fraser, 2001! resulting in pZNYX-GalVP16. The final con- promoter ~Morgan et al., 2006!. But, for technical reasons, it is not struct is shown schematically in Fig. 1A. Intron 1 was included yet possible to follow the in vivo development of retinal bipolar in the hope that providing a splice site would improve in vivo cells in mammals. Such observations, however, are readily achieved expression of the Gal-VP16 protein. The Gal40VP16 expression using zebrafish. system was used to amplify expression levels since nyx mRNA The combination of rapid embryonic development of the ze- levels in mouse retina are low ~Gregg et al., 2003!. This driver brafish retina and the transparency of the embryos permit visual- plasmid is also modular, allowing us to use it with reporter ization, and time-lapse imaging of retinal cells marked by fluorescent plasmids containing the 14X UAS E1b promoter ~Koster & proteins during the period when their circuitry is forming. For Fraser, 2001!. Three different reporter plasmids expressing dif- example, using stable transgenic zebrafish lines in which subsets ferent fluorescent proteins were used in the present study. pUAS- of amacrine cells express various spectral variants of green fluo- MYFP and pUAS-MCFP express membrane-targeted versions of rescent protein, recent studies have determined how amacrine cell EYFP and ECFP respectively and were created by cloning the neurites ramify within the IPL early in development ~Kay et al., 14XUAS E1b promoter ~Koster & Fraser, 2001! into pEYFP-N1 2004; Godinho et al., 2005!. Here, we explored how retinal bipolar or pECFP-N1 ~Clontech!, replacing the CMV promoter, along cells in zebrafish obtain their appropriate stratification level in the with the first 20 amino acids of the zebrafish GAP43 gene fused IPL and form bouton-like terminals with maturation. to the amino terminus of the FP coding sequence. pUAS- To label bipolar cells, we searched for a suitable promoter to DsRedExpress was a generous gift from Martin Meyer ~Stanford drive expression of the fluorescent proteins. The nyx gene encodes University! and has the 14XUAS E1b promoter cloned into a small leucine rich proteoglycan that is mutated in human X-linked pDsRed-Express-1 ~Clontech!. Congenital Stationary Night Blindness ~CSNB1, Bech-Hansen et al., 2000; Pusch et al., 2000!. Loss of nyctalopin in mice results Injections in loss of the ERG b-wave ~Gregg et al., 2003!, which is derived from signaling through ON bipolar cells. We show here that Injections of DNA and imaging were performed essentially as de- sequences from the zebrafish nyx promoter are able to drive scribed previously ~Lohmann et al., 2005!. Briefly, DNA was di- expression in a subset of ON bipolar cells. We generated a trans- luted in 1X Danieau’s solution ~58 mM NaCl, 7 mM KCl, 0.6 mM genic line in which morphologically defined ON-bipolar cells Ca~NO3!2, 0.4 mM MgSO4, 5 mM HEPES, pH 6.8! at a final con- express membrane targeted yellow fluorescent protein ~MYFP! centration of 10–20 ng0ml. Phenol Red ~;0.1%! was added to aid and used this line, in combination with transient expression exper- in visualization of the injection bolus. 1.0 mm capillaries with fil- iments, to follow the dynamic development of the axonal terminals ament ~WPI! were pulled on a Model P-87 needle puller ~Sutter of ON-bipolar cells in vivo. Inst.!. Tips were trimmed to a diameter of 15–20 mm and backfilled with DNA solution. Eggs were collected from population tanks, beginning at light onset, at intervals of 15–20 min. Eggs in the Materials and methods chorion were oriented cell side up on a silicone tray and injected Zebrafish ~Danio rerio! were maintained on a 14:10 light dark using a picospritzer. Eggs at the single cell stage were injected with cycle at 288C. Experimental protocols were approved by the 0.5–2 nl of DNA solution. Injected eggs were maintained in 0.3X Washington University Institutional Animal Care and Use Com- Danieau’s in an incubator at 28.58C. At 24 hpf, the medium was mittee. Unless otherwise noted, zebrafish homozygous for the roy replaced with that containing 0.2 mM PTU ~Phenyl-thiourea! to orbison ~roy! mutation ~Ren et al., 2002! were used for all prevent melanin synthesis, and fish were maintained in this me- experiments. Homozygous roy fish have reduced numbers of iri- dium until all imaging experiments were completed. dophores, a highly birefringent pigment cell that is responsible for the characteristic iridescence found in fish skin and eyes. Using Generation of TG(nyx::Gal4V16; UAS::MYFP)Q16 this more transparent line, it was possible to image much deeper into the retina in vivo and at much older ages ~Ն2 weeks or more, Fish were collected and injected as described for imaging. Super- post-fertilization!. coiled pZNYX-GalVP16 and pUAS-MYFP were co-injected into one cell stage eggs. F0 larvae showing especially high numbers of bipolar cells expressing YFP were reared to adulthood and mated Generation of the nyctalopin-directed transgene and with un-injected roy/roy adults. F progeny were screened for YFP reporter plasmids 1 expression using an Olympus SZX12 fluorescence stereo- In human and mouse, nyctalopin is encoded by the nyx gene, microscope fitted with appropriate excitation and emission filters. which is composed of 3 exons ~Bech-Hansen et al., 2000; Pusch One founder was identified that produced offspring with YFP Zebrafish Bipolar Cell Development 835

low melting temperature agarose on a coverslip. After solidifica- tion of the agarose, the coverslip was transferred to a 60 mm culture dish and covered with 0.3ϫ Danieau’s solution containing 0.2 mM PTU and 0.02% tricaine. In some experiments, prior to mounting, embryos were incubated for1hin100mM “Bodipy- Texas Red” ~CellTrace Bodipy TR methyl ester, Molecular Probes C34556! and 2% DMSO in 0.3ϫ Danieau’s solution. This lipo- philic dye accumulates in the plasma membrane of retina cells and is a useful counter stain for live imaging in zebrafish ~Cooper et al., 2005!. For time-lapse imaging the culture dish was placed in a custom made ITO glass heating system and maintained at 28.58C using a temperature controller ~TC2 bip, Cell Micro Controls!. Laser scanning confocal microscopy was performed using either an Olympus FV1000, FV500 or FV300 upright microscopes. Most images were acquired using an Olympus LUMFL 60X 1.1NA water immersion objective with refractive index correction collar. Multiphoton microscopy was performed using an in-house modified FVX microscope, using a Ti:Sapphire laser ~Tsunami, Spectra- Physics! at 840 nm ~CFP excitation! and 880 nm ~YFP-excitation!. For both confocal and two photon reconstructions, images were typically acquired with XY resolution of 0.1 mm and 0.5 mm Z-resolution for low magnification images. Higher magnification images ~e.g., Fig. 4G!, were typically acquired with XY resolution of 0.06 mm and Z-resolution 0.3 mm. Optical sections typically encompassed a depth of 10 to 50 mm of tissue. Three-dimensional ~3D! analysis of acquired image stacks were performed using Metamorph ~Universal Imaging! and AMIRA ~TGS Inc.!, and final figures were prepared using Photoshop and Illustrator ~Adobe!. To quantify the distribution of axonal tips for each age-group, confocal or two photon image stacks of the bipolar cells in the background of Bodipy-Texas Red labeling were recon- structed in three dimensions using AMIRA. The 3D stacks were rotated until the boundaries of the IPL appeared parallel to each other. The distance of each axonal tip to the inner boundary of the IPL at that location, was measured. Live slice preparation of adult zebrafish retina was performed as described previously ~Miller et al., 1999!. Briefly, retinas were removed from the eye-cup, flattened onto a piece of black filter paper ~Millipore!, placed in zebrafish Ringer’s solution ~Wester- field, 2000!, and slices cut using a razor-blade.

Fig. 1. ~A! Schematic diagram showing the major features of the pZNYX- Immunohistochemistry GalVP16 construct. The nyx gene fragment includes exons 1 and 2, and ϫ intron 1, and extends 1459 bp upstream of the predicted translation start Fish were euthanized in Tricaine ~2% in 0.3 Danieau’s solution!. site. ~B! Bipolar cells transiently expressing YFP under the control of the Eyes were removed and retinas dissected in 0.1M phosphate nyx promoter at 4.5 dpf after injecting nyx::Gal4VP16 and pUAS-MYFP buffered saline ~PBS!, pH 7.4, and then fixed in 4% paraformal- into one-cell stage embryos. ~C! YFP-expressing bipolar cells at 5.5 dpf in dehyde in PBS for 2 h. Tissue was equilibrated in 30% sucrose the background of TG(Pax6DF4::mCFP)Q01, revealing the depth at which overnight, and embedded and frozen in OCT ~Tissue-Tek!. Cryo- terminals of these cells stratify in the IPL. ~D! Example of a bistratified sections were incubated overnight in rabbit anti-PKC ~AB1610 220 bipolar cell in the TG(Pax6DF4::mCFP) background. IPL, inner plexi- Chemicon! diluted 1:1000 in PBS containing 5% normal donkey form layer; OPL, outer plexiform layer. serum ~NDS! and 0.5% Triton X-100. Sections were then rinsed twice for 10 min in PBS and incubated for1hwithsecondary antibody, Alexa Fluor 568 Goat anti-rabbit ~A11036 Molecular expression in the retina. These progeny were selected and used to Probes! diluted 1:1000 in 5% NDS in PBS. After two 10 min rinses Q16 establish the transgenic line TG(nyx::Gal4VP16; UAS::MYFP) , in PBS, sections were cover-slipped with Vectashield mounting hereafter referred to as nyx::MYFP. When the F1-F4 generation medium ~Molecular Probes!. transgenic fish were crossed to roy/roy fish, transgenic offspring were generated at the expected frequency of ;50%, consistent Results with co-integration of both plasmids at a single site in the genome. Transient expression driven by the nyx promoter is Imaging specific to ON bipolar cells Embryos at various stages were anaesthetized with tricaine ~0.02% Coinjection of pZNYX-GalVP16 and pUAS-MYFP consistently in 0.3ϫ Danieau’s solution! and mounted in a drop of molten 1% resulted in sparse labeling of cells within the inner nuclear layer of 836 E.H. Schroeter et al. the retina, with expression starting at around 3 days post fertiliza- 2005!. Injection of pZNYX-GalVP16 and either pUAS-MYFP or tion ~dpf!. For a given set of injections, 1–50% of injected pUAS-DsRed Express into eggs from TG(pax6::MCFP)Q02 and embryos showed expression, although the level of expression of TG(pax6::MCFP)220, respectively, allowed easy separation of YFP in individual cells was highly variable even within the same the cell types during confocal imaging of live embryos. In all fish. Ectopic transient expression was also seen in various other instances observed ~33 cells!, bipolar axon terminals were found tissues although this was infrequent and highly variable between proximal to the outer ~OFF! lamina of the GFP-positive ama- embryos. Careful analyses of the expression of MYFP in the eye crine substrata ~example, Fig. 1D!. showed that there was mosaic expression in cells with the classic morphology of bipolar cells ~Fig. 1B!. Cell bodies were found in ON bipolar cells are labeled in nyx::MYFP the outer half of the INL with dendrites ramifying in the OPL and transgenic fish axons extending into the IPL. The cells showed various axonal morphologies suggesting that more than one subtype of bipolar YFP expression in the transgenic embryos ~Fig. 2A! was first cell was being labeled ~Fig. 1B!. detected in the pineal gland at 2 dpf. Retinal expression began at Several techniques were used to determine if all the YFP- 2.5–3 dpf, in one or a few cells immediately adjacent to the ventral expressing bipolar cells were stratified within the presumptive furrow, with more weakly expressing cells subsequently appearing “ON” or “OFF” sublayers of the IPL. In the TG(pax6::MCFP)Q01 mainly within the dorsal half of the retina ~Fig. 2B!. The number transgenic fish line, CFP is ubiquitously targeted to cellular of cells expressing YFP gradually increased until around 4 dpf, at membranes, allowing the boundaries of the IPL to be readily which time expression is observed throughout the retina ~Fig. 2C!. discerned ~Godinho et al., 2005!. Eggs from TG(pax6::MCFP)Q01 Examination of isolated retina from adult nyx::MYFP fish showed injected with pZNYX-GalVP16 and pUAS-MYFP confirmed that that YFP expression persists, and remains specific to bipolar cells. YFP-expressing bipolar cells had axons that terminated within The density of YFP expressing cells was highest in central retina the lower half of the IPL ~Fig. 1C!. Line TG(pax6::MCFP)Q02 and declined towards the retinal periphery ~Fig. 2D!. and TG(pax6::MCFP)220 express membrane-targeted CFP or At higher magnification it was evident that the YFP-expressing GFP specifically in amacrine cells that stratify their arbors into cells resembled ON bipolar cells, based both on the location of two major bands: an inner ~presumed ON! and an outer their soma and the morphology and strata in which their axons ~presumed OFF! band ~Kay et al., 2004; Godinho et al., ramified ~Fig. 2E!. These cells had somata within the INL, termi-

Fig. 2. ~A! YFP expression in the 7 dpf nyx::MYFP fish. Expression is evident in the eye and in the pineal gland. ~B, C! Confocal sections of live nyx::MYFP embryos at 3 and 14 dpf, respectively. N, nasal; V, ventral. Arrowhead marks location of the ventral furrow. ~D! Distribution of YFP-positive cells across the retina ofa3month old nyx::MYFP fish imaged in a live slice preparation by confocal microscopy. The region indicated as central retina was located approximately 0.5 mm from the optic nerve head. ~E-G! Immunolabeling of a retinal cross-section from nyx::MYFP fish with ~E! anti-PKC, a marker for ON bipolar cells, ~F! anti-YFP and ~G! both anti-PKC and anti-YFP. Note that not all PKC-positive cells contain YFP. Arrow 1 shows a tip positive for both PKC and YFP. Arrow 2 indicates an example of a tip that was PKC-positive but YFP-negative. Zebrafish Bipolar Cell Development 837 nal processes or dendrites in the OPL, and axon-like projections mostly in the immature peripheral margin of the retina ~5 fish that stratified in the inner half of the IPL. Collectively, these sampled; 32 cells co-expressing CFP and YFP, 12 expressing CFP features suggested that they were likely to be ON bipolar cells only!. ~Connaughton & Nelson, 2000!. The majority of cells also were positive for both YFP and PKC ~Figs. 2E–G! a marker for ON bipolar cells in zebrafish ~Yazulla & Studholme, 2001!. However, Immature ON bipolar cell axonal terminals show some PKC positive cells were YFP negative, suggesting that the exploratory behavior throughout the IPL nyx::MYFP expression pattern was mosaic in ON bipolar cells. The inclusion of GAL4/VP16 in nyx::MYFP to amplify expres- Both the transgenic line and transient expression of fluorescent sion from the nyx promoter should also allow specific expression reporters driven by the nyx promoter enabled us to visualize and of any genetic construct containing a UAS promoter. To test this follow the development of ON bipolar cells in vivo. Imaging the functionality, eggs from nyx::MYFP positive fish were injected nyx::MYFP transgenic line at different days suggests that the axons with pUAS::MCFP. As expected, bipolar cells expressing CFP of ON bipolar cells undergo highly dynamic changes as they form were present in a mosaic pattern typical of transient expression axonal terminals. The earliest detectable expression of MYFP by experiments ~Fig. 3A!. Not surprisingly, most CFP-expressing the nyx promoter reveals that by 3 dpf, bipolar cells have acquired cells co-expressed YFP ~Figs. 3B–D!. However, some cells only typical bipolar cell morphology ~Fig. 1B!. Axons of these cells expressed CFP ~Figs. 3E–G!. The number of cells expressing only have filopodia that appear to originate at one or two specific sites CFP was variable from retina to retina, and they tended to be found along the axon shaft. To determine if we could find cells at a more immature stage we took advantage of the fact that as the retina grows, newly differentiating cells appear in the peripheral margin. In this region the density of labeled cells in the nyx::MYFP transgenic is very low, permitting us to image individual cells at the very earliest stages of differentiation. In contrast to cells found away from the marginal zone, these bipolar cells had filopodia extending at many sites along the primary axonal process ~Figs. 4A and 4B!. Higher magnification images also showed filopodia have no strong foci or point of origin and are distributed throughout the depth of the IPL ~Figs. 4C and 4D!. A bias in the distribution, however, was apparent at this early stage. The majority ~66 %!,of the terminals were located within the inner half of the IPL and the outer 10% were devoid of processes. ~3 animals, 11 cells; 180 tips.! Comparison of this measured distribution with a normal distribution generated using a random generator ~180 tips, 100,000 iterations; MatLab! suggested that these two distributions were statistically different ~p Ͻ 2.0 ϫ 10Ϫ9, t-test!. Also, the measured distribution was skewed towards the inner half of the IPL ~p Ͻ 0.0001!. Although functional separation of the IPL into ON and OFF sublaminae cannot be determined without electrophysiology, especially at 3 dpf, previous studies suggest that the IPL can be roughly divided into inner and outer halves at this early age. Expression of GFP in subpopulations of amacrine cells in the TG(pax6::MGFP)220 line ~Fig. 1D! is restricted to two major bands as early as 3 dpf ~Kay et al., 2004!. Thus, even at 3 dpf, a larger proportion of tips appeared to be distributed in the presumed ON sublamina. By 6 dpf, the axonal terminals of the YFP-positive bipolar cells were clearly restricted to the inner half of the IPL ~Fig. 4E!. At this age, the density of bipolar cells expressing YFP in the peripheral regions we imaged was too high to allow us to assign filopodia to individual bipolar axons. Therefore we transiently expressed MCFP ~see Fig. 3! to visualize individual bipolar cells at this age, and then measured their axonal filopodial tip distribution ~Fig. 4F!. Statistical analyses confirmed that at 6 dpf the tips were biased towards the inner half of the IPL ~p Ͻ 0.0001!. Low magnification images at 6 dpf also revealed that the major foci along the primary axon appeared as discrete terminals ~Figs. 1C, 1D, 4E!. Many filopodia, however, still extended from these foci Fig. 3. ~A! Confocal reconstruction of a 7 dpf retina demonstrating in vivo and were readily seen at higher magnification ~Fig. 4G!. In addi- expression of CFP in bipolar cells after injection of pUAS-MCFP into the tion, serial confocal sections revealed that the central portion of nyx::MYFP transgenic line. ~B–D! Higher magnification image of a bipolar these terminals developed areas devoid of YFP labeling that are not cell ~arrows! expressing both YFP and CFP. ~E–G! Examples of cells at the evident in terminals from younger fish ~Supplemental Movie 1!. peripheral margin that express CFP but not YFP. Note also that at 6 dpf, despite the high density of cell labeling, 838 E.H. Schroeter et al.

Fig. 4. ~A! Morphology of immature YFP-expressing bipolar cells at the retinal periphery at 3 dpf. ~B! Distribution of axonal filopodial tips in the developing IPL at 3 dpf ~3 animals, 11 cells, 180 tips!. ~C, D! Higher magnification view of individual axonal terminals from a3dpfnyx::MYFP fish. Filopodia are distributed along the primary axon throughout most of the depth of the IPL. ~E! Confocal reconstruction of bipolar cells in the nyx::MYFP fish at 6 dpf. Axon terminal branches are strongly biased to the lower half of the IPL. Arrows mark gaps in YFP-expression consistent with mosaic expression of the transgene. ~F! Distribution of axonal filopodial tips in the developing IPL at 6 dpf ~2 animals, 22 cells, and 237 tips!~G! High magnification two-photon reconstruction of axon terminals from a 6 dpf nyx::MYFP fish. Arrow marks a single discrete terminal.

there were occasional gaps ~Fig. 4E, arrows!, consistent with magnification, the irregular shape of the immature axonal terminal mosaic expression of the transgene, as suggested by immunolabel- from which filopodia extended and retracted is apparent. At 5 dpf ing with anti-PKC ~Fig. 2E–2G!. ~n ϭ 38 cells!, the exploratory behavior of the axonal filopodia is Using time-lapse imaging, we next examined the dynamic be- still evident ~Fig. 5C!. This particular cell also demonstrates that havior of the axonal filopodia in individual YFP-expressing cells although changes occurred largely at the terminal, ectopic filopodia from 3 to 7 dpf. Eighty cells were recorded at 1–2 h time intervals occasionally extended and retracted from the axonal shaft outside for up to 12 h at different points during this 5 day period. At 3 dpf of the terminal structure ~Fig. 5C, arrows!. These rapid structural ~n ϭ 25 cells!, axonal filopodia were observed to extend and retract changes were detected even at 7 dpf ~n ϭ 17 cells; Supplemental largely from individual foci along the axon shaft ~Fig. 5A!. To cap- Movie 2!, an age at which the developing fish are free swimming ture the highly dynamic nature of the filopodia at this age, we ac- and able to actively hunt live food, and well past when the retina is quired images at 15 min intervals ~Fig. 5B!. At this higher first responsive to light at 3 dpf ~Branchek, 1984; Easter & Nicola,

Supplemental Movies 1 and 2 Supplemental Movies 1 and 2 can be viewed in this issue of VNS by visiting journals.cambridge.org Supplemental Movie 1: High magnification serial confocal sections ~0.5 mm steps! through the IPL ofa6dpfnyx::MYFP fish. Axon terminals display many filopodia but contain a core within which YFP is absent. Scale bar ϭ 5 mm. Supplemental Movie 2: Time-lapse series of an individual axon terminal from 7 dpf nyx::MYFP fish ~15 minute intervals!, showing rapid extension and retraction of filopodia in an immature terminal. Scale bar ϭ 5 µm. Zebrafish Bipolar Cell Development 839

Fig. 5. Time-lapse recording of individual bipolar cells showing dynamic changes in morphology ~A! Cell at 3 dpf from fish injected with pZNYX-GalVP16 and pUAS-MYFP and imaged at 2 hour intervals. Arrows point to two foci from which filopodia extended and retracted. ~B! Axon terminal imaged every 15 minutes froma3dpfnyx::MYFP transgenic fish. ~C! Cell expressing YFP at 5.5 dpf from a TG(Pax6DF4::mCFP)Q01 fish injected with pZNYX-GalVP16 and pUAS-MYFP and imaged at 2 hour intervals. Arrow indicates extension of a filopodia towards the outer half of the IPL. YFP, yellow; CFP blue.

1996!. Time-lapse imaging of populations of cells in the nyx::MYFP central retina as early as 7 dpf, but they were difficult to visualize transgenic performed at 3 and 7 dpf ~data not shown! shows that the with confocal microscopy because of the depth of tissue. To motility and exploratory behavior of the individual cells imaged is overcome this limitation we used two-photon microscopy, to im- typical of all labeled axon terminals. age central retina in vivo up to 14 dpf. At 14 dpf, mature bouton-like axonal terminals were readily apparent throughout most of the retina, although terminals with Axonal bouton-like structures emerge as filopodial activity irregular shapes and filopodial extensions were still present ~Fig. 6B!. diminishes To capture the developmental time course of bouton formation, we Axonal terminals at 3 to 7 dpf showed a very different structure to used time-lapse recordings of individual cells at 7 dpf. By imaging that of bipolar cells in adults. In mature animals, immunostaining at 15 min intervals, we observed the transition from a structure with anti-PKC revealed large bouton-like structures at the axon with many filopodia to the smoother mature terminal bouton terminals, which were localized to the inner half of the IPL ~Fig. 6C!. Thus, the foci along axonal shafts of immature bipolar ~Fig. 6A!. These large boutons are a characteristic feature of ON cells are likely to represent the sites where terminal boutons are bipolar cell axonal terminals in both zebrafish and goldfish ~Sherry established upon maturation. In addition, the boutons had a “hol & Yazulla, 1993; Connaughton et al., 2004!. Individual axon low” appearance, similar to that observed from PKC immunostain- terminals with distinct bouton-like endings were detectable in ing ~Fig. 6A!. This hollow appearance may indicate that the 840 E.H. Schroeter et al.

Fig. 6. ~A! PKC-immunolabeling of bipolar cells in a fixed section from a 28 dpf zebrafish showing the mature ball-like endings of the axons characteristic of this cell type. ~B! Two-photon image of bipolar axon terminals of a 14 dpf nyx::MYFP fish in vivo, demonstrating ~b! the presence of ball-like endings and ~f! terminals that have not yet attained this morphology. ~C! Time-lapse confocal imaging of an axon terminal ~7 dpf! in the nyx::MYFP fish acquired every 15 min from a 7 dpf axon terminal showing the transition of a terminal with many filopodia to the mature bouton like structure. ~D! Summary of morphological stages of ON bipolar cell axonal differentiation in the zebrafish retina.

cytoplasm is increasing in volume at the terminal with age, be- uted throughout the IPL, although there is a bias to the inner cause YFP is targeted to the plasma-membrane. half of the IPL ~Fig. 4!. With ensuing development, filopodia Fig. 6D illustrates the sequence of morphological changes become more restricted to discrete foci from which they extend that occur as ON bipolar cell axonal terminals develop in vivo. and retract, until eventually filopodia are lost and bouton-like Our recordings suggest that initially, axonal filopodia are distrib- structures are established. Zebrafish Bipolar Cell Development 841

Discussion terization of the bipolar cell types labeled in the nyx::MYFP transgenic fish will require combined physiological and anatomi- Nyctalopin promoter sequences drive expression in cal analyses. subsets of retinal bipolar cells In vivo maturation of ON bipolar cell axonal terminals Injection of two plasmids, pZNYX-GalVP16 and pUAS-MYFP into one-cell stage zebrafish embryos resulted in the generation of a Previous studies of the morphological development of bipolar cells stable transgenic line in which YFP is expressed specifically by a have relied on immunolabeling methods or dye-filling approaches subset of retinal bipolar interneurons. This result suggests that the to visualize cells ~Connaughton & Nelson, 2000; Connaughton two plasmids may have co-integrated into the same site in the et al., 2004!. However, common markers of bipolar cells often genome. To our knowledge, the nyx::MYFP fish is the first exam- label these interneurons late in development, when they have ple of a transgenic fish created by injecting two separate DNA already developed a stratified terminal arbor in the IPL. More elements to form a functional transgene. Although transgenic lines recently, genetic approaches have enabled bipolar cells in the expressing Gal4 have been established ~Scheer et al., 2001!, this mouse retina to be labeled and their development in retinal culture fish represents the first reported line using a Gal40VP16 fusion monitored prior to the appearance of a stratified axonal terminal protein as a transcriptional activator. Inclusion of Gal40VP16 in ~Morgan et al., 2006!. Here, we have created transgenic zebrafish this line, also allows it to be used to express any gene of interest that allowed us to perform the first in vivo visualization of bipolar in retinal bipolar cells by using a separate UAS promoter construct cell axonal development in a vertebrate retina. ~Koster & Fraser, 2001!. In addition, the pUAS-MYFP cassette Early in development axon terminals of ON bipolar cells gen- used in generating the transgenic may make nyx::MYFP useful as erate filopodial-like structures that transiently explore the entire a MYFP reporter line. depth of the IPL. However, our analyses indicate that at all ages The expression pattern of nyctalopin in the retina has been studied ~3–14 dpf!, the distribution of axonal filopodia is biased inferred from in situ hybridization studies and from the physio- towards the inner half of the IPL. Therefore, while ON bipolar cell logical defects associated with its absence in mice ~Gregg et al., axons do not immediately target their appropriate sublaminae, the 2003! and humans ~Bech-Hansen et al., 2000; Pusch et al., 2000!, bias in filopodia distribution indicates that there may be signals where it is thought to be involved in visual processing by ON important for their eventual targeting to the appropriate synaptic bipolar cells. In addition to expression in ON bipolar cells, expres- sublaminae and cellular partners. This finding is consistent with sion of nyctalopin has been described in ganglion cells and the recent observations of ON bipolar cell development in the Grm6- outer nuclear layers, as well as in several non-retinal tissues. GFP mouse retina ~Morgan et al., 2006!. In these transgenic mice However, our studies using the entire intergenic region upstream of the axonal terminals of ON bipolar cells are observed to originate the nyx gene plus intron 1 and part of exon 2 showed expression from epithelial-like processes that contact the internal limiting mem- restricted to ON bipolar cells in the retina. In addition, the trans- brane. Observations of Golgi stained retina also suggest that this genic line expresses MYFP in the pineal. The role of nyctalopin in interesting mechanism may also be occurring in rat ~Morest, 1970! this tissue is unknown, but it may be associated with ribbon and chick ~Quesada et al., 1981; Quesada & Genis-Galvez, 1985!. synapses as is thought to be the case at photoreceptor synapses. Whether bipolar cells in zebrafish show a similar morphological Our studies found that in the retina, the zebrafish nyx promoter pattern at early stages of development could not be determined since drove expression of MYFP only in morphologically classified ON YFP expression in the nyx::MYFP transgenic fish does not occur bipolar cells, even in transiently expressing embryos, consistent early enough to allow live imaging prior to the appearance of an with the possibility that nyctalopin is present in ON bipolar cells. axonal process. Thus, in zebrafish, it remains possible that ON bi- The lack of YFP expression in other retinal cells, however, does polar cells, like those of the mouse, may also undergo a stage of not preclude the possibility that nyctalopin could be expressed development whereby their primary axonal process extends beyond more extensively than in bipolar cells. the IPL. Whether Off bipolar cells follow a similar sequence of The mosaic labeling patterns when nyx::MYFP was transiently axonal differentiation remains to be elucidated. expressed allowed us to identify multiple types of ON bipolar The first visually-evoked responses in zebrafish are detected at cells. Comparison of the morphology of the YFP-expressing bi- 3 dpf ~Branchek, 1984; Easter & Nicola, 1996!, corresponding to polar cells with those described in previous reports using intracel- the formation of the first bipolar cell ribbon synapses ~Schmitt & lular fills ~Connaughton et al., 2004!, suggests that at least two Dowling, 1999!. Like dendritic filopodia ~Ziv & Smith, 1996; subclasses of bipolar cells with axonal terminals stratifying in Jontes & Smith, 2000; Wong & Wong, 2000!, the exploratory sublaminae S3 to S6 express YFP under the nyx promoter. Some behavior of zebrafish bipolar cell axonal filopodia at this age may YFP-positive bipolar cells were clearly monostratified whereas serve to increase the likelihood of contact with their postsynaptic others possessed at least two terminals. When imaged under con- targets. One of our more surprising observations was that bipolar ditions that allowed determination of position within the IPL, the cells continue to show considerable filopodial activity even at 14 outer terminal of bistratified YFP-labeled bipolar cells was located dpf, suggesting that synaptogenesis may not yet be completed even above the middle of the IPL where the glycinergic amacrine plexus though the overall structure of the retina appears adult-like. resides ~Connaughton et al., 1999!. This cell type most closely It is currently unclear what mechanisms are responsible for resembles the BON-s305 cell ~Connaughton et al., 2004!, which has organizing the stratification and morphological arrangement of the an ON type physiology with a glutamate activated chloride con- bipolar cell axonal terminals during development. Studies in mam- ductance, but is not sensitive to 2-amino-4-phosphonobutyric acid mals show that depletion of retinal ganglion cells as a consequence ~Connaughton & Nelson, 2000!. Bipolar cells with similar mor- of optic nerve section does not prevent the formation of ON and phology in the closely related species giant danio ~Danio aequipin- OFF bipolar axonal terminals ~Gunhan-Agar et al., 2000!. Like- natus! display ON, OFF or both ~ON0OFF! responses ~Wong wise, ablation of cholinergic amacrine cells in mammals does not et al., 2005; Wong & Dowling, 2005!. However, further charac- perturb bipolar axonal stratification ~Reese et al., 2001; Gunhan- 842 E.H. Schroeter et al.

Agar et al., 2002!. However, in the lakritz zebrafish mutant in cells within the zebrafish retina. Journal of Comparative Neurology which ganglion cells are absent, regions of perturbed amacrine cell 477, 371–385. Connaughton, V.P. & Nelson, R. ~2000!. Axonal stratification patterns lamination coincides with abnormalities in bipolar cell axonal and glutamate-gated conductance mechanisms in zebrafish retinal bi- stratification ~Kay et al., 2004!. It will be interesting in the future polar cells. Journal of Physiology 524, 135–146. to utilize the nyx::MYFP line for identifying other mutants in Cooper, M.S., Szeto, D.P., Sommers-Herivel, G., Topczewski, J., which bipolar axonal stratification is perturbed. Solnica-Krezel, L., Kang, H.C., Johnson, I. & Kimelman, D. Finally, in contrast to mammals, the axonal terminals of fish ~2005!. Visualizing morphogenesis in transgenic zebrafish embryos using BODIPY TR methyl ester dye as a vital counterstain for GFP. ON bipolar cells form large axonal boutons at maturity. Our in vivo Developmental Dynamics 232, 359–368. imaging observations suggest that this terminal structure in teleosts Crooks, J., Okada, M. & Hendrickson, A.E. ~1995!. Quantitative analy- arises only after a period of filopodial exploration located at sis of synaptogenesis in the inner plexiform layer of macaque monkey discrete levels within the IPL. Although our observations suggest fovea. Journal of Comparative Neurology 360, 349–362. Dubin, M.W. ~1970!. The inner plexiform layer of the vertebrate retina: a that axonal filopodia may play a developmental role, these thin quantitative and comparative electron microscopic analysis. Journal of processes may also have other functions. Specifically, filopodia Comparative Neurology 140, 479–505. extend from the giant terminals of goldfish ON bipolar axons in Easter, S.S., Jr. & Nicola, G.N. ~1996!. The development of vision in the response to dark adaptation ~Yazulla & Studholme, 1992; Behrens zebrafish ~Danio rerio!. International Journal of Developmental Biol- & Wagner, 1996; Job & Lagnado, 1998!. The nyx::MYFP trans- ogy 180, 646–663. Godinho, L., Mumm, J.S., Williams, P.R., Schroeter, E.H., Koerber, genic line will enable future investigations into how zebrafish A., Park, S.W., Leach, S.D. & Wong, R.O.L. ~2005!. Targeting of bipolar cell terminals respond dynamically to changes in illumi- amacrine cell neurites to appropriate synaptic laminae in the develop- nation. In addition, it will be possible to monitor in vivo how ing zebrafish retina. Development 132, 5069–5079. bipolar cell morphology is altered in response to perturbations Gregg, R.G., Mukhopadhyay, S., Candille, S.I., Ball, S.L., Pardue, M.T., McCall, M.A. & Peachey, N.S. ~2003!. Identification of the occurring during development. gene and the mutation responsible for the mouse nob phenotype. Investigative Ophthalmology and Visual Science 44, 378–384. Gunhan-Agar, E., Choudary, P.V., Landerholm, T.E. & Chalupa, Acknowledgments L.M. ~2002!. Depletion of cholinergic amacrine cells by a novel immunotoxin does not perturb the formation of segregated on and off We thank Amy Koerber for providing assistance with construction and cone bipolar cell projections. Journal of Neuroscience 22, 2265–2273. maintenance of fish lines and Josh Morgan for help with Monte Carlo Gunhan-Agar, E., Kahn, D. & Chalupa, L.M. ~2000!. Segregation of analysis. This work was supported by a Keck foundation fellowship to on and off bipolar cell axonal arbors in the absence of retinal ganglion E.H.S., by NIH grant EY14358 to R.O.W. and by NIH grant EY12354 to cells. Journal of Neuroscience 20, 306–314. R.G.G. Job, C. & Lagnado, L. ~1998!. Calcium and protein kinase C regulate the actin cytoskeleton in the synaptic terminal of retinal bipolar cells. Journal of Cell Biology 143, 1661–1672. References Jontes, J.D. & Smith, S.J. ~2000!. Filopodia, spines, and the generation of synaptic diversity. Neuron 27, 11–14. Bech-Hansen, N.T., Naylor, M.J., Maybaum, T.A., Sparkes, R.L., Kay, J.N., Roeser, T., Mumm, J.S., Godinho, L., Mrejeru, A., Wong, Koop, B., Birch, D.G., Bergen, A.A., Prinsen, C.F., Polomeno, R.O.L. & Baier, H. ~2004!. Transient requirement for ganglion cells R.C., Gal, A., Drack, A.V., Musarella, M.A., Jacobson, S.G., during assembly of retinal synaptic layers. Development 131, 1331–1342. Young, R.S. & Weleber, R.G. ~2000!. Mutations in NYX, encoding Koster, R.W. & Fraser, S.E. ~2001!. Tracing transgene expression in the leucine-rich proteoglycan nyctalopin, cause X-linked complete living zebrafish embryos. International Journal of Developmental Bi- congenital stationary night blindness. Nature Genetics 26, 319–323. ology 233, 329–346. Behrens, U.D. & Wagner, H.J. ~1996!. Adaptation-dependent changes of Lohmann, C., Mumm, J.S., Morgan, J., Godinho, L., Schroeter, E.H., bipolar cell terminals in fish retina: Effects on overall morphology and Stacy, R., Wong, W.T., Oakley, D. & Wong, R.O.L. ~2005!. Imaging spinule formation in Ma and Mb cells. Vision Research 36, 3901–3911. the developing retina. In Imaging in Neuroscience and Development. Bramblett, D.E., Pennesi, M.E., Wu, S.M. & Tsai, M.J. ~2004!.The eds. Yuste, R. & Konnerth, A., pp. 171–183. Cold Spring Harbor, NY: transcription factor Bhlhb4 is required for rod bipolar cell maturation. Cold Spring Harbor Laboratory Press. Neuron 43, 779–793. Miller, E.D., Tran, M.N., Wong, G.K., Oakley, D.M. & Wong, R.O. Branchek, T. ~1984!. The development of photoreceptors in the zebrafish, ~1999!. Morphological differentiation of bipolar cells in the ferret brachydanio rerio. II. Function. Journal of Comparative Neurology retina. Visual Neuroscience 16, 1133–1144. 224, 116–122. Morest, D.K. ~1970!. The pattern of neurogenesis in the retina of the rat. Burmeister, M., Novak, J., Liang, M.Y., Basu, S., Ploder, L., Hawes, Zeitschrift F ür Anatomie Und Entwicklungsgeschichte 131,45–67. N.L., Vidgen, D., Hoover, F., Goldman, D., Kalnins, V.I., Roder- Morgan, J., Huckfeldt, R. & Wong, R.O. ~2005!. Imaging techniques in ick, T.H., Taylor, B.A., Hankin, M.H. & McInnes, R.R. ~1996!. retinal research. Experimental Eye Research 80, 297–306. Ocular retardation mouse caused by Chx10 homeobox null allele: Morgan, J.L., Dhingra, A., Vardi, N. & Wong, R.O. ~2006!. Axons and Impaired retinal progenitor proliferation and bipolar cell differentia- dendrites originate from neuroepithelial-like processes of retinal bi- tion. Nature Genetics 12, 376–384. polar cells. Nature Neuroscience 9, 85–92. Cheng, C.W., Chow, R.L., Lebel, M., Sakuma, R., Cheung, H.O., Mumm, J.S., Godinho, L., Morgan, J.L., Oakley, D.M., Schroeter, Thanabalasingham, V., Zhang, X., Bruneau, B.G., Birch, D.G., E.H. & Wong, R.O. ~2005!. Laminar circuit formation in the vertebrate Hui, C.C., McInnes, R.R. & Cheng, S.H. ~2005!. The Iroquois retina. Progress in Brain Research 147, 155–169. homeobox gene, Irx5, is required for retinal cone bipolar cell develop- Nishimura, Y. & Rakic, P. ~1987!. Development of the rhesus monkey ment. Journal of Developmental Biology 287, 48–60. retina: II. A three-dimensional analysis of the sequences of synaptic Chow, R.L., Volgyi, B., Szilard, R.K., Ng, D., McKerlie, C., Bloom- combinations in the inner plexiform layer. Journal of Comparative field, S.A., Birch, D.G. & McInnes, R.R. ~2004!. Control of late Neurology 262, 290–313. off-center cone bipolar cell differentiation and visual signaling by the Olney, J.W. ~1968!. Centripetal sequence of appearance of receptor-bipolar homeobox gene Vsx1. Proceedings of the National Academy of Sci- synaptic structures in developing mouse retina. Nature 218, 281–282. ences 101, 1754–1759. Pusch, C.M., Zeitz, C., Brandau, O., Pesch, K., Achatz, H., Feil, S., Connaughton, V.P., Behar, T.N., Liu, W.L. & Massey, S.C. ~1999!. Scharfe, C., Maurer, J., Jacobi, F.K., Pinckers, A., Andreasson, Immunocytochemical localization of excitatory and inhibitory neuro- S., Hardcastle, A., Wissinger, B., Berger, W. & Meindl, A. ~2000!. transmitters in the zebrafish retina. Visual Neuroscience 16, 483–490. The complete form of X-linked congenital stationary night blindness is Connaughton, V.P., Graham, D. & Nelson, R. ~2004!. Identification caused by mutations in a gene encoding a leucine-rich repeat protein. and morphological classification of horizontal, bipolar, and amacrine Nature Genetics 26, 324–327. Zebrafish Bipolar Cell Development 843

Quesada, A. & Genis-Galvez, J.M. ~1985!. Morphological and structural Sherry, D.M. & Yazulla, S. ~1993!. Goldfish bipolar cells and axon study of Landolt’s club in the chick retina. Journal of Morphology 184, terminal patterns: A Golgi study. Journal of Comparative Neurology 205–214. 329, 188–200. Quesada, A., Prada, F., Armengol, J.A. & Genis-Galvez, J.M. ~1981!. Wassle, H. ~2004!. Parallel processing in the mammalian retina. Nature Early morphological differentiation of the bipolar neurons in the chick Reviews of Neuroscience 5, 747–757. retina. A Golgi analysis. Anatomia Histologia Embryologia 10, 328–341. Westerfield, M. ~2000!. The Zebrafish Book. Eugene, OR: University of Reese, B.E., Raven, M.A., Giannotti, K.A. & Johnson, P.T. ~2001!. Oregon Press. Development of cholinergic amacrine cell stratification in the ferret Wong, K.Y., Cohen, E.D. & Dowling, J.E. ~2005!. Retinal bipolar cell retina and the effects of early excitotoxic ablation. Visual Neuroscience input mechanisms in giant danio. II. Patch-clamp analysis of ON 18, 559–570. bipolar cells. Journal of Neurophysiology 93, 94–107. Ren, J.Q., McCarthy, W.R., Zhang, H., Adolph, A.R. & Li, L. ~2002!. Wong, K.Y. & Dowling, J.E. ~2005!. Retinal bipolar cell input mecha- Behavioral visual responses of wild-type and hypopigmented zebrafish. nisms in giant danio. III. ON-OFF bipolar cells and their color- Visional Research 42, 293–299. opponent mechanisms. Journal of Neurophysiology 94, 265–272. Rowan, S. & Cepko, C.L. ~2005!. A POU factor binding site upstream of Wong, W.T. & Wong, R.O. ~2000!. Rapid dendritic movements during the Chx10 homeobox gene is required for Chx10 expression in subsets synapse formation and rearrangement. Current Opinion in Neurobiol- of retinal progenitor cells and bipolar cells. Developmental Biology ogy 10, 118–124. 281, 240–255. Yazulla, S. & Studholme, K.M. ~1992!. Light-dependent plasticity of Scheer, N., Groth, A., Hans, S. & Campos-Ortega, J.A. ~2001!.An the synaptic terminals of Mb bipolar cells in goldfish retina. Journal of instructive function for Notch in promoting gliogenesis in the zebrafish Neurophysiology 320, 521–530. retina. Development 128, 1099–1107. Yazulla, S. & Studholme, K.M. ~2001!. Neurochemical anatomy of the Schmitt, E.A. & Dowling, J.E. ~1999!. Early retinal development in the zebrafish retina as determined by immunocytochemistry. Journal of zebrafish, Danio rerio: Light and electron microscopic analyses. Jour- Neurocytology 30, 551–592. nal of Comparative Neurology 404, 515–536. Ziv, N.E. & Smith, S.J. ~1996!. Evidence for a role of dendritic filopodia Sherry, D.M., Wang, M.M., Bates, J. & Frishman, L.J. ~2003!. Expres- in synaptogenesis and spine formation. Neuron 17, 91–102. sion of vesicular glutamate transporter 1 in the mouse retina reveals temporal ordering in development of rod vs. cone and ON vs. OFF circuits. Journal of Comparative Neurology 465, 480–498. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.