Development 143: doi:10.1242/dev.133314: Supplementary information
Table S1. Composition of terminated CMZ clones from dpf5-7 fish
displaced AC
1
1
1
HC
PRs
1
4
1
MU
2
2
2
1
1
2
2
2
1
2
2
1
2
2
Development • Supplementary information
RGC
Clone composition at 7dpf
AC
BC
1
1
2
2
Development 143: doi:10.1242/dev.133314: Supplementary information
Supplemental Materials and Methods
Computation of division angles and the relevant statistics
The image analysis of the 4D movies (3d+t) was performed in three stages: 1) Detection of
the position of the CMZ edge; 2) Identification of the divisions and the orientation of the
division planes; and 3) Computation of the relevant statistics for each division (angle of
division and approximate distance from the CMZ edge).
To compute the position of the CMZ edge, first the retina was segmented and the blood
vessel present in the image is identified based on intensity and texture features via
thresholding and morphological operations (the blood vessel appears as a high intensity,
compact, ring shaped region, while the membrane staining in the retina produces sparser,
often dimmer fluorescent patterns, Fig. A). The identified blood vessel is an indicator of the
edge of the CMZ. We then computed the parameters of the circle fitted to the detected blood
Computation of the CMZ edge. (A) Segmentation of the retina (blue)
and blood vessel (green). (B) Fitted circle to the detected blood vessel
pixels as an indicator of the edge of CMZ.
Development • Supplementary information
vessel pixels as well as the plane in which this circle lies (Fig. B):
Development 143: doi:10.1242/dev.133314: Supplementary information
In the second step, the pixels belonging to the divisions are detected by analysing the
difference between two consecutive or temporally close frames (depending on the time
resolution of the imaging) (Fig. C). New division candidates were detected as sufficiently
large contiguous volumes of statistically significant pixels in the difference image, followed
by a manual validation step. Alternatively, manual indication of a division followed by
detection of pixels belonging to the division was also performed on one of the datasets).
Identification of the divisions and orientations of division planes.
Procedure to identify cell divisions (C) and determine the angle of
division as well as the distance to the CMZ edge (D).
arrow) are computed (Fig. D). Finally, for each division, the average position of the detected
significant pixels and the fitted normal of the division plane were projected onto plane of the
CMZ edge (dashed light blue arrow and its origin). We computed the distance between the
projected average position and the closest point on the CMZ edge, as well as the angle
between the projected normal and the radius of the CMZ edge (α). This angle α is equal to the
angle between the tangent to the CMZ edge and perpendicular to the projected normal of the
division plane (β).
Development • Supplementary information
For each division candidate the best fitting plane (solid yellow) and its normal (light blue
Development 143: doi:10.1242/dev.133314: Supplementary information
We then grouped the division events (based on the projection of the division) according to
rings of 5.3µm width around the CMZ edge. The number of division events in each ring is
depicted in Fig. 4E (main text). From these counts we calculated the division rates for each
ring as the number of observed divisions divided by the mean number of cells in the ring and
the total time of the analysed movies (Table 1).
Table 1 Average division rates in ring around the CMZ edge for two datasets
Distance from
Average division Average division Average division
CMZ edge (µm) rate (per hour)
rate (Dataset 1) rate (Dataset 2)
2.65
0.0120
0.0101
0.0138
7.95
0.0240
0.0123
0.0358
13.25
0.0311
0.0263
0.0360
18.55
0.0300
0.0272
0.0329
23.85
0.0369
0.0357
0.0380
29.15
0.0223
0.0164
0.0282
34.45
0.0141
0.0159
0.0123
39.75
0.0068
0.0068
45.05
0.0049
0.0049
𝜌𝑅 = ∑ 𝜌𝑅𝑖 =
𝑖
1
1
𝑛 1
1
𝑛
∑ 𝑖 = ( ∑ 𝑖) =
,
𝑇
𝑁𝑅 𝑇 𝑛
𝑁𝑅
𝑇〈𝑁𝑅𝑖 〉
𝑖
𝑖
where the index 𝑖 denotes a division event, 𝑁𝑅𝑖 the number of cells in ring R during the time of
division i, and n the total number of divisions observed in ring R during the time of analysis.
For a given division event we estimated the total number of cells in the corresponding ring by
dividing the estimated area covered by this ring by the average cells size (𝜋 (5.3⁄2)2 µ𝑚2).
The area of the CMZ ring observed in the image stack depends on several fitted parameters
for the respective time point (radius of CMZ ring and normal of the CMZ plane, influencing
Development • Supplementary information
More specifically, the division rate in each ring 𝜌𝑅 is obtained as the sum of rates 𝜌𝑅𝑖
Development 143: doi:10.1242/dev.133314: Supplementary information
the fraction of the CMZ inside the field of view). The estimated numbers of cells per ring in
the two datasets are shown in Fig. E.
To calculate the combined estimated division rates for both datasets we made use of the
assumption that individual division events are statistically independent. We therefore merged
the datasets for the final estimates, while still taking the individual estimates for the number
of cells in each ring and at each time point correctly into account. To estimate uncertainty, we
calculated one-sigma Poisson confidence intervals for the division counts in each ring. As the
uncertainty associated with the average cells size and the area of each ring was small
compared to this, the final estimates of uncertainty could simply be obtained by dividing the
limits of the Poisson confidence intervals by 𝑇〈𝑁𝑅𝑖 〉.
Estimated number of cells in each ring (1 - closest, 9 - furthest ring
to/from the blood vessel). A correction for the part outside the field of
view was applied. The left and right panel show the estimated number
of two datasets respectively.
Development • Supplementary information
E
Fig. S1. Spatial characterization of all the clones observed in the long-term clonal analysis. The geometrical
metrics of all the cells in a clone is plotted against the time point of observation. Cell positions are color-coded
so that cells in the ring 1-2 are represented as filled circles in black and gray, and cells with ring 3 or above are
plotted as empty circles in dark green to light green from peripheral to central retina. Clonal termination point is
represented as a black vertical line in the terminated clones. All the maintained clones retain at least 1 cell in
rings 1 or 2, while the terminated clones get detached from the peripheral end before termination.
Development • Supplementary information
Development 143: doi:10.1242/dev.133314: Supplementary information
Development 143: doi:10.1242/dev.133314: Supplementary information
Development • Supplementary information
Fig. S2. Bimodal distribution of the clone attachment position (the position of the most peripheral cells at the
last observed day) of all the 5-7 dpf maintained single clones (n=55) and the corresponding value for terminated
clones before termination (n=25).
Fig. S3. Cell divisions in the CMZ occur at the apical surface and with almost no apical-basal polarity. (A)
Schematic diagram of a retina cross-section, illustrating the actual distance between the central point of division
and the CMZ (d3) and the distance between the same two point as a projection in the plane of the apical surface,
i.e. in 2D (d2) (blood vessel represented in red). The graph shows that the distances measured in 2D are very
similar to those measured in 3D, highlighting the fact that via the Pythagorean theorem the majority of divisions
are within a few microns of the plane of the apical surface. Note that as d2 increases, d3 increases slightly
faster, and this is largely because of the curve of the retina, as the apical surface itself bends away from the
plane on which d2 lies. (B) Histogram of the angle distribution of cell divisions, showing that there are no
events where the divisions occurred parallel to the apical surface. Grey boxes show examples of division angles
(white bar: apical surface; dotted line: division plane).
Development • Supplementary information
Development 143: doi:10.1242/dev.133314: Supplementary information
Development 143: doi:10.1242/dev.133314: Supplementary information
Development • Supplementary information
Fig. S4. Distribution of size and cell type composition of the IPC clones compared to central retina clones
originating from 24, 32 and 48hpf progenitor cells (from He et al. 2012), with (A) showing clone size, and (B)
showing clonal composition (RGC: retinal ganglion cell, AC: amacrine cell or displaced amacrine cell, BC:
bipolar cell, MC: Müller glia cell, PR: photoreceptor).
Development 143: doi:10.1242/dev.133314: Supplementary information
Development • Supplementary information
Movie 1. 4D imaging of cell divisions in the CMZ of a 5dpf Rx2:GFPcaax transgenic retina. Each frame
represents a maximum intensity projection (10 frames, z=7.5μm). Stacks were acquired every 2.5min. Elapse
time shown in hr:min:sec. (Original movie: 0.18*0.18μm, stack size: 55.5μm, z-slice: 0.75μm, 135 timepoints,
interval: 2.5min). Full datasets of both movies (approx. 36Gb) are available upon request to the corresponding
author.