(A) Isolated immature RGC (MYFP) in the Q01 background (MCFP). Time-lapse images (5 minute intervals) of this cell (grayscale images) show that the dendritic tips are highly exploratory, extensions (E) and retractions (R) are apparent (arrows; Supp. Movie M1). (B–E) Quantification of dendritic branch dynamics for the RGC in (A). (B) Motility rates of terminal branches (n = 109). Branch length at each successive time point (i.e. every 5 min) was measured from branch node to branch tip. (C) Numbers of branches eliminated or added per time interval. (D) Net change in total terminal branch length per time interval. (E) Analysis of branch tip motility. Plots indicate total number of branch tips remaining within 1 μm of the branch tip position in the previous time-point. The vast majority of branch tips moved outside of this volume within 5 min. (F) Plots of terminal branch lifetimes for RGCs analyzed at two ages (n = 3 cells per age). Lifetime measurements were limited to branches present at the initial time-point (n = 59). Note that a larger fraction of branches persisted at the older age. (G) Time-lapse series of RGC with a substantial dendritic arbor prior to lamination in the IPL. Higher magnification of the boxed area shows a new branch forming and then retracting (arrow). (H) Time-lapse series of RGC with a stratified arbor (Supp. Movie M2). The dendritic terminals of stratified cells were relatively more stable, compared to RGCs that had yet to stratify. dpf = days post-fertilization; E = extensions; R = retractions.

(A) Time-lapse images of a region of retina in a Brn3c:MGFP/Q01 double transgenic animal. RGCs express MGFP while the majority of all other retinal cells express MCFP. Each image is an orthogonally oriented 3D projection of a 3 μm thick confocal stack. Arrows indicate displaced amacrine cells (dark profiles) embedded within the IPL at early stages. IPL inner and outer boundaries are marked by dashed lines. (B) Grayscale images of the MGFP-positive RGCs in (A). (C) Averaged fluorescence intensities within the boxed regions shown. Arrows indicate peaks in MGFP fluorescence, i.e. strata; red arrows indicate emergence of new strata. Numbers indicate presumed temporal sequence in the appearance of each peak. Five strata are apparent by 16 dpf; stratum 5 appeared to form in-between strata 2 and 3. (D) Location of each dendritic stratum expressed as a percentage of IPL thickness (e.g. S¹⁰; stratum at 10^th percentile).

(A) Simultaneous imaging of amacrine cells (MCFP) and RGCs (MGFP) at different ages in Brn3c:MGFP/Q02 double transgenic animals. Each image is an orthogonally oriented 3D projection of a 3 μm thick confocal stack. (B, C) Grayscale images providing better detail of the RGC dendrites (B) and amacrine cell neurites (C) shown in (A). At the bottom right corner of each image are the averaged fluorescence intensities within a rectangular region, as in . By day 6, the RGC-derived S⁵⁵ stratum appeared in-between the S⁴⁰ and S⁷⁰ strata (arrow in A, B, C).

Time-lapse imaging of isolated RGCs (MYFP) in the Q01 background (MCFP). Shown are orthogonally oriented image stacks of the entire arbor. (A) RGC whose dendrites initially elaborated diffusely throughout the depth of the IPL. Later, the cell developed an asymmetrically oriented, multistratified dendritic tree. (B) RGC showing biased elaboration within the inner half of the IPL that later became monostratified, occupying a single stratum in the middle of the IPL. (C) The inner arbor (arrow) of this RGC was lost as the cell went on to develop a monostratified arbor in the OFF sublamina.

Time-lapse imaging of isolated RGCs (MYFP) in the Q01 background (MCFP). Shown are orthogonally oriented image stacks of the entire arbor. (A) Sequential addition of dendritic strata, from inner (ON, S¹⁰) to outer (OFF, S⁷⁰) sublaminae, is evident in this bistratified RGC. (B) Addition of an inner stratum (S⁵⁵, arrow) to this bistratified cell. (C) Summary of sequential stratification patterns observed for a total of 41 RGCs imaged at at least two ages. Green bars indicate stratification pattern observed at initial time point; red bars at final time point. Numbers in parenthesis below bars indicate total number of cells observed with this pattern of strata addition. All others are n = 1.

Time-lapse imaging of isolated RGCs (MYFP) in the Q01 background (MCFP). (A) Multistratified RGC with a primary dendrite (arrow) that initially stratifies in the ON subregion but later extended and stratified within the OFF subregion. (B) Bistratified RGC with a stratified arbor in the OFF subregion (arrow) extending a branch back into the ON subregion at day 8.

Simultaneous time-lapse imaging of co-stratifying RGCs and amacrine cells. Injection of either an islet3:MCherry plasmid construct into the 220 background (A, C–D) or brn3c:MYFP plasmid construct into the Q02 background (B) was used to express reporter proteins in individual RGCs. (A,B) Vertical extensions of stratified RGC arbors (arrows) appear to target previously established amacrine-derived strata. New RGC strata subsequently extended from these processes in both the ON (A) and OFF (B) subregions. (C,D) Cells with dendritic distributions initially biased towards the inner IPL subsequently restricted their dendrites to at least one or more labeled amacrine-derived strata. (E) Schematic summarizing dendritic growth patterns observed for RGCs that targeted labeled amacrine-derived strata. Figures A–D represent examples of each of the four types of RGC growth patterns observed.

Comparison of the lamination strategies in mammals and zebrafish. (A) In mammals, immature RGC dendrites are thought to occupy the entire depth of the IPL; dendrites occupying inappropriate regions are then selectively eliminated (“pruned”) to form discrete strata. For technical reasons, mammalian studies have been limited to static imaging and have thus far focused on monostratified RGC subtypes; how bistratified and multistratified RGCs achieve their adult lamination patterns in mammals is not clear (denoted by question mark). Using in vivo time-lapse imaging in zebrafish we followed individual RGCs from immature to mature states and thereby correlate early arborization morphologies with final stratification patterns. Our analysis included observations of the dendritic behavior of mono-, bi-, and multistratified RGCs. We found several growth patterns by which the dendrites of zebrafish RGCs attain laminated arbors: (i) A particular multistratified RGC subtype was observed to behave more like mammalian RGCs; initially the dendritic arbor occupied the entire depth of the IPL, and subsequently an asymmetric multistratified arbor emerged. (ii) Most monostratified zebrafish RGCs showed a dendritic elaboration that is biased to the subregion of the IPL within which they later stratify. None of the monostratified RGCs we observed ramified throughout the IPL at early stages. (iii) The majority of bi- and multistratified RGCs demonstrated sequential layering of dendritic strata; strata can be added from the inner to the outer IPL, vice-versa, or even in-between existing strata. (B) In mammals, cholinergic amacrines appear to form laminated plexuses prior to the stratification of bistratified RGCs (). However, morphological identification of immature bistratified RGCs is problematic, making it difficult to ascertain the dendritic growth pattern of this subtype of RGC. In zebrafish, time-lapse analyses demonstrate that amacrine cells form laminated plexuses prior to the stratification of RGC dendrites within these same laminae. RGC dendrites thus appear to target and subsequently co-stratify with pre-patterned afferent plexuses.