Light-Induced Rod and Cone Cell Death Andregeneration in the Adult Albino Zebrafish (Daniorerio) Retina
Computers and Technology
Submitted By yanym
Light-Induced Rod and Cone Cell Death and
Regeneration in the Adult albino Zebraﬁsh (Danio rerio) Retina
Thomas S. Vihtelic and David R. Hyde
Department of Biological Sciences, University of Notre Dame, Notre Dame, Indiana 46556-0369
Received 24 January 2000; accepted 25 April 2000
Light-induced photoreceptor cell degeneration has been studied in several species, but not extensively in the teleost ﬁsh. Furthermore, the continual production of rods and cones throughout the teleost’s life may result in regeneration of lost rods and cones. We exposed adult albino zebraﬁsh to 7 days of constant darkness, followed by 7 days of constant 8000 lux light, followed by 28 days of recovery in a 14-h light:10-h dark cycle. We characterized the resulting photoreceptor layer cell death and subsequent regeneration using immunohistochemistry and light microscopy. Within the ﬁrst 24 h of constant light, the zebraﬁsh retina exhibited widespread rod and cone cell apoptosis.
High levels of cell proliferation within the inner nuclear layer (INL) were observed within the ﬁrst 3 days of
Light has been an effective and popular environmental method to experimentally induce photoreceptor cell degeneration. The light treatment can irreversibly cause photoreceptor cell loss through apoptosis (Shahinfar et al., 1991; Li et al., 1996). Factors such as light intensity and exposure duration, length of dark adaptation before light exposure, and ocular pigmentation inﬂuence the retinal damage in mice and rats
(for reviews see Organisciak and Winkler 1994;
Rapp, 1995). In addition, the extent of light damage is affected by genetic background, pharmacological agents, growth factors, and transgene expression (La-
Correspondence to: D. R. Hyde (email@example.com).
Contract grant sponsor: University of Notre Dame Graduate
School and College of Science.
© 2000 John Wiley & Sons, Inc.
constant light, as assessed by immunodetection of proliferating cell nuclear antigen and BrdU labeling. The proliferating cells within the INL were closely associated with the radial processes of Muller glia, similar to the
pluripotent retinal stem cells observed during embryonic development. Using antibodies generated against the individual zebraﬁsh opsins, we determined that rods and the green, blue, and ultraviolet cone cells were replaced within the 28 day recovery period. While both rods and cones were replaced, the well-ordered cone cell mosaic was not reestablished. © 2000 John Wiley & Sons, Inc.
J Neurobiol 44: 289 –307, 2000
Keywords: light-induced degeneration; zebraﬁsh; retinal regeneration; retina; photoreceptor
Vail et al., 1987, 1992; Safars et al., 1997; Wang et al., 1997).
The cone-rich retina of diurnal ﬁsh (Teleostei) may be a useful model for the light-damaged human retina.
Although light treatment disrupted the retina of albino rodents, constant light treatment also damaged the retinas of both albino and pigmented teleost ﬁsh
(Marotte et al., 1979; Penn, 1985; Raymond et al.,
1988). However, the teleost retina was more resistant to light damage than the rodent retina. Albino rainbow trout (Oncorhynchus mykiss) exposed to 10,000 lux daylight at 11°C exhibited rod outer segment (ROS) truncation, but no signiﬁcant loss of rod cells (Allen and Hallows, 1997). In contrast, albino oscars (Astronotus oscellatus) exposed to 3000 lux constant light at 28°C exhibited both ROS truncation and rod cell loss (Allen et al., 1999).
Unlike mammals, the teleost retina grows through289
Vihtelic and Hyde
out its life (Johns and Easter, 1977; Fernald, 1985) and serves as a model to study mechanisms of neural cell differentiation and neuronal regeneration (Fernald, 1989; Raymond and Hitchcock, 1997; Stenkamp et al., 1997). During embryonic retinal growth, cells are added from the circumferential germinal zone
(CGZ) and from regions adjacent to the choroid ﬁssure (Johns, 1982). As the animal ages, a large proportion of the increased retinal size is due to retinal stretching and continued cell addition from the CGZ
(Lyall, 1957; Johns, 1977; Kock, 1982). Although growth-related retinal stretching decreases the density of most neurons, the density of rods is maintained by continuously generating new rods from precursor cells located within the outer nuclear layer (ONL;
Johns and Fernald, 1981; Johns, 1982). The initial population of rod precursor cells is likely deposited during early retinal development from cells proliferating in the inner nuclear layer (INL; Johns, 1982;
Raymond, 1985, 1991; Hagedorn and Fernald, 1992).
These proliferating cells in the INL were also identiﬁed in the retinas of trout (2–24 months posthatch; 2 g–3 kg body weight), where they migrate to the ONL and likely add to the population of rod precursor cells during normal retinal growth (Julian et al., 1998).
Injury to the adult goldﬁsh (Carassius auratus) retina can deplete photoreceptor cells and stimulate neural cell regeneration (Raymond and Hitchcock,
1997). For example, surgical removal of a small patch of the retina or widespread cellular destruction caused by the metabolic toxin ouabain leads to retinal cell regeneration in goldﬁsh (Maier and Wolburg, 1979).
Either of these methods of retinal cell destruction produces clusters of proliferating cells that span multiple retinal layers (Negishi et al., 1988; Raymond et al., 1988; Hitchcock et al., 1992). This is similar to the organization of proliferating cells in the INL of the growing trout retina (Julian et al., 1998). Thus, the replacement of rods and cones may simply be a stimulation of the normal retinal growth process. Both rods and cones were regenerated when goldﬁsh photoreceptors were selectively ablated by intraocular injection of the glycosylation-inhibitor tunicamycin or by directed thermal laser injury (Negishi et al., 1991;
Braisted et al., 1994). New cone cells were also generated when the adult green sunﬁsh (Lepomis cyanellus) retina was surgically lesioned (Cameron and Easter, 1995). Regeneration of goldﬁsh and sunﬁsh cone cells is signiﬁcant because this neuronal cell class is not normally produced during growth in these central regions of the adult retina. These results are consistent with the presence of a stimulus-activated retinal stem cell population and perhaps retinal cell dedifferentiation to replenish the damaged photoreceptor popula-
tion (Braisted et al., 1994; Cameron and Easter, 1995;
Reh and Levine, 1998). Recently, several genes expressed in the regenerating ﬁsh retina and peptide growth factors that affect cell proliferation in the CGZ were identiﬁed (Levine et al., 1994, 1997; Hitchcock et al., 1996; Passini et al., 1997; Boucher and Hitchcock, 1998).
We examined the effects of raising albino zebraﬁsh for 7 days in constant darkness, followed by 7 days in constant intense light, followed by 28 days of recovery in 14-h light:10-h darkness. The zebraﬁsh retinas experienced photoreceptor apoptosis that resulted in widespread rod and cone cell loss during constant light. After characterizing this inducible form of photoreceptor cell degeneration, we determined that both rods and cones were regenerated during the recovery period. We identiﬁed proliferating cells within the INL that migrated to the ONL and replaced the lost rods and cones.
MATERIALS AND METHODS
Zebraﬁsh were raised at 28.5°C in a 14-h light:10-h dark cycle under an average illuminance of 200 lux using standard husbandry techniques (Westerﬁeld, 1993). All experiments were performed using the albino mutant (University of Oregon, Eugene, OR) at 6 –12 months of age. Prior to enucleation, ﬁsh were anesthetized using 2-phenoxyethanol.
Euthanasia was performed by decapitation while ﬁsh were under deep anesthesia.
Light Treatment Protocols
Prior to each light treatment experiment, ﬁsh were placed in constant darkness for 7 days. A tungsten halogen lamp (500
W; Sylvania, Danvers, MA) was placed 50 cm from the side of a 2.5-gal glass aquarium to produce intense white light conditions. A hand-held illuminometer (Sekonic Model
246, Elmsford, NY) measured a light intensity of 20,000 lux
(no lamp correction factor) at the aquarium glass surface facing the light and 11,000 lux after light exited the aquarium (after light passage through the water and two layers of aquarium glass). The lamp was moved an additional 30 cm from the aquarium after 2 h (8000 lux at the aquarium glass surface and 5000 lux after exiting the aquarium) to prevent the water temperature from rising above 32.5°C. The light treatment was carried out in a large, well-ventilated room to aid heat dissipation from the halogen lamp. After 7 days of constant light, ﬁsh were returned to the standard facility light and temperature conditions. In the retinal terminal deoxynucleotidyl transferase (TdT)-mediated nick end labeling (TUNEL) and bromodeoxyuridine (BrdU) labeling experiments, eyes were examined after shorter periods of
Photoreceptor Death and Regeneration
Sample Collection Intervals and
Every 24 h during light treatment and after 28 days of recovery in LD
3, 6, 12, and 24 h after beginning constant light treatment
Every 24 h during constant light treatment and after 7, 14, and
28 days of recovery in LD
3, 6, 12, and 24 h after BrdU injection Sampling Methods
Measurements of PL and ONL width, and ONL nuclei density were made on central retina sections.
TUNEL(ϩ) cells were counted across entire retinal section(s).
PCNA(ϩ) cells were counted across entire retinal section(s).
BrdU(ϩ) cells in the ONL and INL were counted across entire retinal section(s). DD, constant dark treatment; LL, constant intense light treatment; LD, 14-h light:10-h darkness; PL, photoreceptor layer; ONL, outer nuclear layer; IP, intraperitoneal.
* albino zebraﬁsh were used in 3 independent experiments for retinal morphometric measurements, 1 TUNEL experiment, 2 independent
PCNA-labeling experiments, and 1 BrdU-labeling experiment.
BrdU was injected IP after 72 h of light treatment.
constant light treatment (see below). In all experiments, the preconstant light control eyes were taken from the ﬁsh raised for 7 days in constant darkness.
Histology and Retinal Morphometric
For histological analysis, eyes were removed, ﬁxed in 2% formaldehyde/2.5% glutaraldehyde/100 mM sodium cacodylate (pH 7.4) overnight at 4°C, washed in 100 mM cacodylate, and postﬁxed in 2% osmium tetroxide/100 mM cacodylate for
60 –90 min at room temperature. After washing in water, tissues were dehydrated through an ethanol series, placed in
1:1 xylene/ethanol and 100% xylene for 30 min each. The tissues were inﬁltrated in 1:1 xylene/Polybed 812 for 60 min, overnight in 1:2 xylene/Polybed 812, and ﬁnally placed into fresh Polybed 812. The resin was polymerized at 60°C. Sections of 5 m were cut using a glass knife on a JB4 microtome and stained with either toluidine blue or a combination of 1% methylene blue/1% azure II. Alternatively, the eyes were ﬁxed as described above, embedded in parafﬁn, and 7-m sections were stained with hematoxylin and eosin.
Changes in retinal morphology during constant light were quantiﬁed from three independent experiments. In each experiment, eyes were collected from single ﬁsh at 24-h intervals during constant light and after 28 days of recovery (Table 1).
Measurements were made from a single retina of each ﬁsh using four or ﬁve different sections from the central retina
(Table 2), which included the optic nerve or were immediately adjacent to the optic nerve. All the measurements were taken at a point two-thirds the distance from the CGZ to the optic nerve exit point. The photoreceptor layer thickness was measured between the external limiting membrane and the most distal
(scleral) tip of photoreceptor outer segments that were visible in that section. After 3 days of constant light, ROS were no
longer visible, and either the remaining cone outer segments or cone ellipsoid structures were used for the scleral-most component of the measurement. The same sections and retinal locations also were used to measure the ONL thickness and to count the number of nuclei in the ONL over a distance of 100
m. Thus, with the exception of the 4 days of constant light time point (where only two ﬁsh were examined), the retinal morphometric value at each time point represents the average of 14 or 15 different measurements taken from the retinas of three different ﬁsh (Table 2).
DNA strand breaks were detected in frozen retinal sections by TdT-mediated incorporation of ﬂuorescein-labeled dUTP (In Situ Cell Death Detection; Roche Molecular
Biochemicals, Indianapolis, IN). To quantify the change in the number of TUNEL-positive (ϩ) nuclei during the ﬁrst
24 h of constant light, the eyes of single ﬁsh were enucleated at 3, 6, 12, and 24 h after initiating constant light and immediately frozen in OCT (Table 1). We generated 16-m serial sections completely through each eye. For each sectioned eye, ﬁve sections from the peripheral retina and ﬁve sections from the central retina were chosen for TUNEL.
The retinal sections were ﬁxed with 4% paraformaldehyde/5% sucrose/PBS (pH 7.4) for 20 min at room temperature, washed for 30 min with PBS, and permeabilized with
0.1% Triton X-100/0.1% sodium citrate according to the manufacturer’s protocol. The labeling reaction was carried out at 37°C for 60 min in the dark. For each section examined, all the TUNEL(ϩ) nuclei across the entire retinal section were counted and an average number of TUNEL(ϩ) nuclei per retinal section determined for the 10 retinal sections at each time point (Table 3). Positive and negative
Vihtelic and Hyde
Retinal Morphometry and PCNA Cell Quantiﬁcation
No. of Retinas/No. of Total Measurements
(per time point)*
Duration of Treatment
(labeled cells/ retinal section)
Before light treatment
During light treatment
Recovery after light treatment 7
Retinal Morphometry Mean Values†
PL, photoreceptor layer; ONL, outer nuclear layer; NE, not examined.
* Morphometric measurements and PCNA cell counts were from one retina per ﬁsh; morphometric measurements were taken from 5 different sections per retina that included or were adjacent to the optic nerve; total number of PCNA measurements (each measurement was from an individual section) represent the combined total from two retinas (2 different ﬁsh from 2 independent experiments).
Numbers in parentheses represent S.D. from the mean.
controls included adding 0.5 mg/mL DNase I and labeling in the absence of TdT, respectively.
PCNA Immunocytochemistry and
Proliferating Cell Quantiﬁcation
Retinal cells that were actively dividing in the constant light-treated retinas were identiﬁed by the presence of proliferating cell nuclear antigen (PCNA). Whole eyes were ﬁxed overnight at 4°C in 9:1 ethanolic formaldehyde (95% ethanol:37% formaldehyde; Julian et al., 1998), rinsed with water, and cryoprotected in 30% sucrose/PBS (pH 7.4) for
4 h at room temperature or overnight at 4°C. Tissues were transferred to 1:1 30% sucrose/PBS:OCT overnight at 4°C, and frozen embedded in 100% OCT. Cryosections of 16 m were mounted onto glass slides and air-dried for several hours at room temperature before storing at Ϫ20°C. Stored sections were warmed and rehydrated in TBS (pH 7.4), blocked with TBS/5% normal goat serum for 60 min at room temperature, and incubated overnight at room temperature with monoclonal anti-PCNA (clone PC10; Sigma
Chemical, St. Louis, MO) diluted 1:3000 in TBS/5% normal goat serum/0.3% Triton X-100/1% dimethylsulfoxide
(DMSO). Sections were washed with TBS/0.05% Tween-20
(TBS-Tw) and incubated with either Cy3- or FITC-conjugated secondary antibody (Jackson ImmunoResearch, West
Grove, PA). Double-label experiments combined the PCNA antibody with rabbit polyclonal glial ﬁbrillary acidic protein
(GFAP) antiserum (Sigma) in the initial incubation.
PCNA(ϩ) cells were quantitated during the constant light treatment and the 28-day recovery period in two independent experiments. In each experiment, eyes from single ﬁsh were harvested at 24-h intervals during constant light and after 7, 14, and 28 days of recovery (Table 1). One light-treated eye from each time point was serially sectioned. Multiple sections were analyzed from each eye to ensure that the entire eye was represented (Table 2). The total number of PCNA(ϩ) cells (labeled CGZ cells were excluded) was determined in each of the frozen sections at each time point to obtain an average total number of
PCNA(ϩ) cells per section for each time point (Table 2).
Opsin Immunocytochemistry on Frozen
Sections and Whole-Mount Retinas
Rhodopsin and the different cone opsins were immunolocalized in frozen retinal sections from ﬁsh raised in constant light using rabbit polyclonal opsin-speciﬁc antisera, except for the polyclonal mouse anti–red opsin serum (Vihtelic et al., 1999). To visualize the rod photoreceptors, anti-rhodopsin serum was combined with the monoclonal antibody zs-4, which labels the rod inner segments (University of Oregon
Monoclonal Antibody Facility; 1:10). The frozen retinal sections for opsin immunostaining were ﬁxed and processed as described above for PCNA immunocytochemistry. For double labeling of opsin and BrdU, the opsin antibodies were applied to tissues that were processed for BrdU labeling (see below).
Photoreceptor Death and Regeneration
TUNEL and BrdU Cell Quantiﬁcation
No. of Sections Examined
(1 retina/time point)
TUNEL(ϩ) Cell Counts (total for each retinal section)
No. of Sections Examined
(1 retina/time point)
(mean per section)†
0, 0, 0, 3, 0, 3, 0, 0
0, 5, 4, 6, 5, 2, 0, 0, 8, 5
48, 66, 53, 63, 48, 31, 68
30, 79, 43
1199, 1983, 1502, 1545, 1573
1265, 810, 1131, 940, 758
1510, 1875, 1165, 1260, 1225
1668, 1334, 1092, 1036, 698
BrdU Cell Quantiﬁcation
BrdU(ϩ) Cell Counts
0/7, 1/8, 1/6, 12/56, 9/62, 12/74
58/249, 57/290, 37/192, 12/87
19/109, 11/96, 3/8, 3/27, 3/24
99/209, 104/231, 127/264, 100/228
122/270, 107/245, 54/214, 93/224
125/241, 95/265, 77/167, 76/151
63/148, 94/193, 120/268, 145/330
194/394, 210/415, 112/279,
104/299, 45/68, 69/101, 42/69
24/40, 25/37, 70/150, 108/232
Mean % ONL
INL, inner nuclear layer; ONL, outer nuclear layer.
* BrdU(ϩ) cell count ratios are shown for each retinal section examined; eyes were completely serial sectioned, and section values shown are in the order peripheral-central-peripheral (reading left to right in sequence).
Numbers in parentheses represent S.D. from the mean.
For whole-mount retina labeling, dissected retinas were ﬁxed in 4% paraformaldehyde/PBS/5% sucrose for 4 h at room temperature, washed with TBS and water (5 min each), and frozen in acetone (Ϫ20°C for 5 min). Following washes in water and TBS (5 min each), retinas were incubated in
TBS/1% BSA/1% DMSO/2% normal goat serum/0.1% Triton
X-100 for 2–12 h at room temperature. The primary antibody dilutions in blocking buffer were rabbit anti-blue opsin, 1:200; rabbit anti-green, 1:200; mouse anti-red, 1:500; monoclonal antibody zpr-1, 1:200; rabbit anti-UV, 1:333. Tissues were washed with four to six changes of TBS-Tw over 4 h and incubated in the appropriate Cy3- or ﬂuorescein isothiocyanate
(FITC)-conjugated secondary antibodies (Jackson ImmunoResearch, West Grove, PA) overnight and washed with TBSTw. Retinal cups were released by scissor cuts to enable ﬂattening and mounted using Vectashield (Vector Laboratories, Burlingame, CA).
To assess migration of dividing cells between the retinal layers during constant light, retinas were labeled by intraperitoneal injection of 200 g 5Ј-bromo-2Ј-deoxyuri-
dine (BrdU; Sigma) in PBS (pH 7.4) 3 days after initiating constant light. Enucleated whole eyes were ﬁxed overnight (4°C) in 3.7% formaldehyde/PBS, rinsed in water, cryoprotected in 30% sucrose/PBS, and frozen embedded in OCT. Sections of 16 m were rehydrated in
TBS, incubated in 2N HCl/TBS for 30 min, rinsed in TBS
(2ϫ 15 min), and blocked with TBS/5% normal goat serum/0.3% Triton X-100. Monoclonal anti-BrdU (clone
BU33; Sigma) was diluted 1:100 in blocking buffer and incubated overnight at room temperature. Sections were washed as described above and anti-BrdU was detected using FITC-conjugated secondary antibody (Jackson ImmunoResearch).
BrdU(ϩ) cells were counted in single retinas that were harvested at 3, 6, 12, and 24 h after BrdU injection (Table
1). The retinas for two time points (3 and 12 h postinjection) were from one ﬁsh, whereas the 6- and 24-h post-BrdU injection time points were from the retinas of two different ﬁsh. Similar to the PCNA protocol, BrdU(ϩ) cells were counted on selected sections from serially sectioned eyes to ensure that all retinal regions were examined. A total of 15,
12, 11, and 6 sections were evaluated for the 3-, 6-, 12-, and
24-h time points, respectively (Table 3). The total number
Vihtelic and Hyde
of BrdU(ϩ) cells was counted in both the INL and ONL of all sections examined at each time point, and the percentage of BrdU(ϩ) cells in the ONL was calculated (Table 3).
Tissues processed for BrdU labeling were also used for
BrdU/GFAP (rabbit anti-cow GFAP, 1:750; Dako, Santa
Barbara, CA) and BrdU/opsin double-label experiments. In both cases, primary antibodies were applied simultaneously, as were the secondary antibodies.
Light-Induced Loss of Rods and Cones in the Zebraﬁsh Retina
We designed a light treatment protocol to maximize photoreceptor cell damage in the zebraﬁsh retina. The albino (alb) zebraﬁsh were maintained in constant darkness for 7 days and then shifted to constant intense light for up to 7 days, which resulted in histological signs of severe damage (Fig. 1). Within 7 days of constant light, ROS were absent throughout the retina, and the ONL was thinner [cf. Fig. 1(B,D) and
(A,C), respectively]. The cone cells also exhibited light-induced damage, such as disorganization and swelling of the ellipsoids and loss of the cone outer segments (COS). Retinal sections with the most severe changes lacked any structurally normal cone cells [Fig. 1(D)]. We examined the time course of photoreceptor damage in alb ﬁsh by quantitating changes in retinal morphology at 24-h intervals during constant light. To maintain consistency, we conﬁned our measurements of photoreceptor layer thickness,
ONL and INL thickness, and ONL nuclei density to the central retina (see Materials and Methods). In the
preconstant light control retinas, the average photoreceptor layer measurement was nearly 100 m, but was only about 24 m in retinas treated for 7 days with constant light [Fig. 1(E), left].
Light microscopy revealed that the remaining photoreceptor layer structure was primarily due to some remaining cone ellipsoids, which were sparse and appeared aberrant. However, we measured the cone ellipsoid region in preconstant light control retinas at
Ͼ40 m. Therefore, 7 days of constant light results in loss of ROS, COS, and a reduction in the cone ellipsoid region. The greatest reduction in the photoreceptor layer occurred during the ﬁrst through third days of constant light [Fig. 1(E), left]. After 3 days of constant light, the ROS were nearly nonexistent [Fig.
2(B)]. Similarly, the greatest reduction in the ONL nuclei density also occurred in the ﬁrst through third days of constant light [Fig. 1(E), right]. The reduction in ONL nuclei density demonstrated an absolute loss of photoreceptor cells, rather than simply a shortening of the ROS as in albino rainbow trout (Allen and
Hallows, 1997). In contrast, the INL thickness did not change (data not shown).
We examined the changes in speciﬁc cell types during constant light treatment by immunolabeling with opsin-speciﬁc antisera (Vihtelic et al., 1999).
Rod photoreceptors that were double-labeled with the anti-rhodopsin serum and monoclonal antibody zs-4 revealed severe disorganization and truncation of
ROS after 3 days of constant light relative to preconstant light [Fig. 2, (B) and (A), respectively]. After 7 days of constant light, the only remaining immunodetectable rod structures were small amounts of disorganized rod inner segment membranes [Fig. 2(C)].
Figure 1 Constant intense light treatment results in retinal degeneration. The eyes from adult albino zebraﬁsh were isolated either before constant light treatment or after 7 days of constant light.
(A,B) Hematoxylin and eosin–stained, 7-m parafﬁn sections from a control retina isolated before constant light and a retina isolated after 7 days of constant light, respectively. (C,D) Plasticembedded retina sections, 5 m, that were stained with methylene blue and azure II (preconstant light control and after 7 days of constant light, respectively). Both control retinas possessed long abundant ROS, an ONL that is several nuclei thick and the expected three layers of cone photoreceptors (dc, lsc, ssc). After 7 days of constant light, the ONL was thinner, and nearly all rod photoreceptors and many of the cone cells were absent. (D) The large, dark bodies that appear adjacent to the degenerated photoreceptor layer may be phago-lysosomes. Changes in the photoreceptor layer during the course of light treatment were quantitated by retinal morphometry. The photoreceptor layer thickness (PL thickness; left graph in E) represents the distance between the external limiting membrane and the scleral-most tip of visible photoreceptor outer segments. (E) The right graph shows the reduction in the number of ONL nuclei over a retinal section distance of 100
m from eyes isolated during the 7 days of constant light. Error bars, S.D. (n values, see Table 2).
Scale bar ϭ (A,B) 25 m; (C,D) 12 m. CE, cone ellipsoid region; ONL, outer nuclear layer; OPL, outer plexiform layer; ROS, rod outer segments; dc, double cones; lsc, long single cones; ssc, short single cones.
Photoreceptor Death and Regeneration
The cone opsin antisera also revealed a marked reduction in all cone cell types after 7 days of constant light compared to preconstant light controls. The polyclonal antisera generated against the ultraviolet opsin, blue opsin, and the red and green opsins labeled the outer segments of the short single cones, the long single cones, and the double cones, respectively [Fig.
2(D), (E), and (F), respectively]. After 7 days of
constant light, the outer segments of the short single cones were truncated and had a lower level of labeling with the ultraviolet opsin antiserum relative to the control [Fig. 2, (G) and (D), respectively]. The remaining blue opsin– expressing long single cones were often characterized by extremely aberrant cell morphology and labeled opsin protein mislocalized toward the synaptic pedicle [Fig. 2(H)]. Similarly,
Vihtelic and Hyde
Figure 2 Rod and cone opsin immunolocalization during intense light treatment. (A–C) The rod inner segments in frozen retinal sections were labeled with monoclonal antibody zs-4 and detected with FITC-conjugated secondary antibody (green signal). The zebraﬁsh rhodopsin in the ROS was labeled with a rabbit polyclonal antibody and detected with a Cy3-conjugated secondary antibody
(red signal). Retinal sections from a preconstant light control retina (pre L; A), a retina from a ﬁsh isolated after 3 days of constant light (B), and a retina isolated after 7 days of constant light (C) are shown. (D–K) Sections were labeled with the different zebraﬁsh cone opsin polyclonal antibodies.
Preconstant light control retinas labeled with antiserum to the ultraviolet (UV) and blue opsin are shown in (D) and (E), respectively; (F) a preconstant light control section double-labeled with antibodies to the red and green opsins. Labeling frozen retinal sections from eyes isolated after 7 days of constant light with the UV (G), blue (H), or the red/green opsin (I) antibodies demonstrates a reduction of cone cells compared to the control sections. The remaining cones have shortened outer segments and exhibit severe ellipsoid swelling (asterisks in G and H). (I) Arrowheads point to shortened green cone outer segments, and the asterisk is near the swollen ellipsoid of a red opsin-expressing cone cell. The pathological reduction in the outer segments and mislocalization of the red (J) and blue opsin (K) proteins was evident as early as 24 h into the constant light treatment
(24 h L). The arrowheads in (J) and (K) denote the opsin-containing cone cell synaptic pedicles, and the arrows point to the shortened outer segments. A 50-m scale bar in (A) is the same for all panels.
GCL, ganglion cell layer; INL, inner nuclear layer; ONL, outer nuclear layer; ROS, rod outer segments; pre L, preconstant light.
many double cone cells totally lacked outer segments and/or possessed swollen ellipsoids that were outlined by detectable red and green opsin protein [Fig. 2(I)].
When these sections were examined by light microscopy, identiﬁable outer segment and ellipsoid structures were not detected. Similar to the rod cells, pathological changes were evident in the zebraﬁsh cone cells during the early stages of light treatment.
After 24 h of constant light, the COS were already truncated and the red and blue opsin proteins were mislocalized to the more proximal cell parts [Fig. 2,
(J) and (K), respectively]. However, the green opsin
protein was not detected proximal to the double cone outer segments [Fig. 2(J)]. This is in contrast to albino oscars, where intense light caused truncation of the
ROS and loss of rods, but did not affect the cone cells
(Allen et al., 1999).
Photoreceptor Loss from Intense Light
Is Due to Apoptosis
Light treatment of nonpigmented mammals causes apoptosis in photoreceptor cells (Shahinfar et al.,
Photoreceptor Death and Regeneration
ONL appeared within the ﬁrst 12 h of constant light
[Fig. 3(D)]. The nuclear morphological changes and the TUNEL(ϩ) nuclei suggested that apoptotic pathways were activated in the rods and also some cone cells during the ﬁrst 24 h of light treatment.
Light-Treated Zebraﬁsh Retinas
Regenerate Their Photoreceptor Cell
Figure 3 Photoreceptor apoptosis during the initial stages of light treatment. (A) Retinal section from a preconstant light control eye. After 24 h of constant light, large numbers of pyknotic nuclei are observed in the ONL (B; arrowheads) consistent with rod photoreceptor apoptosis. A darkly stained cone cell nucleus is also identiﬁed (arrow). The
TUNEL technique was used to detect DNA strand breaks in photoreceptor nuclei by ﬂuorescence microscopy (C; 24-h constant light treatment). Arrowheads point to a few of the positive nuclei in the ONL, whereas arrows mark labeled nuclei in the cone cell layers. The numbers of TUNEL(ϩ) nuclei per retinal frozen section were quantitated after 3, 6,
12, and 24 h of constant light (D). Error bars, S.D. (n values, see Table 3). Scale bar ϭ (A) 12 m; (B) 50 m. GCL, ganglion cell layer; INL, inner nuclear layer; ONL, outer nuclear layer; OPL, outer plexiform layer; ROS, rod outer segments; dc, double cones; lsc, long single cones; ssc, short single cones.
1991; Li et al., 1996). The loss of rod and cone cell structures and the reduced ONL thickness and nuclei density in the constant light-treated zebraﬁsh retina suggested that photoreceptor cell death was occurring.
Therefore, we examined if the retinas exhibited signs of apoptosis using histological and histochemical techniques (Fig. 3). After 24 h of constant light, nuclei in the ONL were condensed and darkly stained compared to the control [Fig. 3, (B) and (A), respectively].
The ONL also contained areas that lacked nuclei [Fig.
3(B)]. Consistent with these morphological changes, large numbers of nuclei in the ONL were labeled using TUNEL, which further suggested apoptosis
[Fig. 3(C)]. Although TUNEL-labeled cone nuclei were occasionally observed [Fig. 3, (B) and (C), respectively], they were rare compared to the large numbers of labeled rod nuclei within the ONL. A time course analysis revealed the TUNEL(ϩ) nuclei in the
Teleost retinal cell production occurs throughout the life of the ﬁsh and also subsequent to photoreceptor loss caused by chemical, laser, or surgical treatment
(Maier and Wolburg, 1979; Hitchcock et al., 1992;
Braisted et al., 1994). We assessed photoreceptor cell regeneration 28 days after terminating the 7 days of constant light in the alb retinas. Retinas that were isolated after 7 days of constant light completely lacked any histological evidence of rods, and the cone cell population was greatly depleted [Fig. 4(A)]. After
28 days of recovery, the retinas showed a regenerated rod and cone cell population [Fig. 4(B)] and an ONL
Figure 4 Photoreceptor regeneration in the light-treated zebraﬁsh retina. (A,B) Retinal sections from ﬁsh sacriﬁced either at the conclusion of 7 days constant light or after 28 days of recovery from constant light, respectively. (A) Absence of ROS, the extensive loss of nuclei in the ONL, and only a few remaining identiﬁable cone cells. (B) After 28 days of recovery from constant light, the ONL possesses a large number of nuclei, long ROS are present, and cone cells have regenerated. (C) Comparison of retinal measurements taken from retinas isolated before constant light (pre
L), at the conclusion of 7 days of constant light (7d constant
L), and after 28 days of recovery (28d post L). S.D.s are in parentheses (n values, see Table 2). Scale bar ϭ (A) 12 m.
CE, cone ellipsoid region; INL, inner nuclear layer; ONL, outer nuclear layer; OPL, outer plexiform layer; ROS, rod outer segments; dc, double cones; lsc, long single cones; ssc, short single cones.
Vihtelic and Hyde
thickness and density that were similar to the preconstant light retina [cf. Figs. 4(B) and 1(C)]. Although cone cell outer segments were visible, layered by type, and positioned proximal (vitread) to regenerated
ROS, they were irregularly spaced with some short single cones immediately adjacent to one another and long single cones occasionally in closely associated pairs [Fig. 4(B)]. Morphometric measurements revealed that the retinas returned to the preconstant light values for photoreceptor layer thickness, ONL thickness, ONL nuclei density, and PCNA(ϩ) cells [Fig.
To further assess cone cell regeneration, we immunohistochemically labeled retinal whole-mounts using the different cone opsin antisera (Fig. 5). After 7 days of constant light, the cone cell population was greatly reduced and the remaining photoreceptors exhibited severe abnormalities [Fig. 2(G,H,I)]. The adult zebraﬁsh cone cells are arranged in a stereotypical repeating array (Robinson et al., 1993), which is visualized by single and double antibody labeling of whole-mount retinas. The double cone photoreceptors consist of a paired green opsin– and red opsin– expressing cell (Vihtelic et al., 1999). Similarly, non– light-treated retinas double-labeled with the blue opsin antiserum and monoclonal antibody zpr-1, which labels the double cone plasma membrane (Larison and
Bremiller, 1990), revealed the repeating spatial arrangement of long single cones and double cone cells
[Fig. 5(C)]. Double labeling after 28 days of recovery revealed two abnormalities in the regenerated cone cell mosaic. First, many red and green opsin– expressing cone cells existed singly, and some retinal areas lacked one or the other [Fig. 5(B)]. Second, the apparent retinal recovery did not regenerate the regular spacing of the preconstant light mosaic pattern [cf.
Fig. 5, (B,D) and (A,C)]. Similarly, the ultraviolet
opsin-labeled short single cones were also irregularly arranged in the regenerated retina relative to the control [Fig. 5, (F) and (E), respectively].
Cell Proliferation in the INL
Accompanies Rod and Cone Cell Birth during the Zebraﬁsh Retinal
In goldﬁsh, cells within the INL likely play a role in retinal regeneration (Hitchcock et al., 1996). Furthermore, proliferative inner nuclear layer cells (PINCs) in trout participate in retinal growth (Julian et al.,
1998). To visualize the underlying cell proliferation that accompanied photoreceptor regeneration in zebraﬁsh, constant light-treated retinas were labeled with a monoclonal antibody that detects PCNA (Negishi et al., 1991a,b). In preconstant light control retinas, PCNA(ϩ) cells were primarily observed in the peripheral marginal zone or in the ONL, which likely represented dividing rod precursor cells (Marcus et al., 1999). On rare occasions, a single cell or small cluster of PCNA(ϩ) cells was observed in the
INL, which could be analogous to the PINCs in the adult trout [Fig. 6(A)]. After 24 h of constant light, an increase in the number of PCNA(ϩ) cells was observed in the inner aspect of the INL [Fig. 6(B)]. We also observed labeled cells in the ganglion cell layer adjacent to the iris–peripheral margin junction at this time (data not shown). This initial PCNA(ϩ) population was characterized by individual small, round cells. After 3 days of constant light, we detected an increased number of proliferating cells in both the
ONL and INL [Fig. 6(C,D)]. The PCNA(ϩ) cell morphology differed, depending on which retinal layer the cell was located. The PCNA(ϩ) cells in the ONL were round and compact, whereas the PCNA(ϩ) cells
Figure 5 Immunohistochemical analysis of cone opsin expression in whole-mount retinas. (A,B)
Retina whole-mounts that were double-labeled with the red and green opsin antisera and detected with Cy3- and FITC-labeled secondary antibodies, respectively. (A) A preconstant light control retina showing the regular repeating double cone spacing and the alternating red/green pair arrangement (labeled arrowheads). After 28 days of recovery from constant light, many of the red and green cones exist singly, and there are areas that lack either red or green opsin-expressing cells.
The arrow in (B) points to a patch of cone cells expressing red opsin surrounded by cone cells expressing green opsin. (C) Preconstant light control and (D) 28 days after constant light; retinal whole-mounts that were double-labeled with the anti-blue opsin serum (labels long single cone cell outer segments; Cy3-labeled secondary antibody; arrowheads) and monoclonal antibody zpr-1
(labels double cone cell plasma membrane; FITC-labeled secondary antibody). The long, single cones in the regenerated retina (D) lack the repeating pattern observed before the constant light treatment. Retinal whole-mounts labeled with anti-UV opsin serum show short, single cone disorder and small areas that lack short single cones after 28 days of recovery from constant light (F; arrows) compared to a preconstant light control retina (E). Scale bar ϭ 50 m in (A) applies to (B)–(F).
Photoreceptor Death and Regeneration
in the INL were organized in clusters of elongate, spindle-shaped cells that extended between the inner
(vitread) aspect of the INL and the ONL. These multicellular columns were often composed of PCNA(ϩ) cell clusters at each of their ends. In many cases, cell clusters were observed within the outer plexiform layer, suggesting that cells may be migrating from the
INL into the ONL. The elevated number of PCNA(ϩ) cells persisted for the duration of constant light, with the peak on the fourth day [Fig. 6(E)].
Proliferating Cells in the INL Migrate into the ONL to Regenerate Rods and
During teleost retinal development, neuroepithelial cells utilize the Muller glial cell scaffold for intrareti¨ nal migration (Raymond and Rivlin, 1987). Muller
cell nuclei were also identiﬁed in the ONL during laser-induced retinal regeneration in goldﬁsh
(Braisted et al., 1994). The columnar PCNA(ϩ) cells
Vihtelic and Hyde
that we observed in the INL [Fig. 6(C,D)] may be using the radial processes of Muller cells to migrate
from the INL into the ONL. This would be similar to the newly born cells observed crossing the outer plexiform layer of the trout retina (Julian et al., 1998). To examine cell– cell interactions and address whether cells migrated during the retinal regeneration in zebraﬁsh, we performed two types of experiments. First, we labeled the Muller cells in retinal frozen sections
using antibodies that detect GFAP [Bignami, 1984;
Fig. 7(A–C)] and performed GFAP/PCNA doublelabel immunohistochemistry experiments. In many cases, PCNA(ϩ) cell clusters were observed in direct association with GFAP(ϩ) processes [Fig. 7(B)]. Second, we labeled proliferating cells with the thymidine analog BrdU to examine changes in the retinal distribution of BrdU-labeled cells over time. The BrdUlabeled retinal cell pattern (24 h after intraperitoneal injection) resembled the PCNA(ϩ) cell pattern, with the BrdU(ϩ) cells in both the INL and ONL and associated with GFAP(ϩ) processes [Fig. 7(C)]. To examine the cell migration from the INL to the ONL, we quantitated the BrdU-labeled retinal cell population 3, 6, 12, and 24 h after BrdU injection and determined the percentages of labeled cells in the
ONL at each time point. The majority of labeled cells were located in the INL immediately after BrdU injection [Fig. 7(D,F)], which was followed by increasing numbers of BrdU(ϩ) cells in the ONL at longer intervals after injection [Fig. 7(E,F)]. This suggested that the proliferating cells were migrating from the
INL to the ONL.
Combining the BrdU cell labeling and opsin immunolocalization allowed us to determine some of the cell fates of the mitotically active cells (Fig. 8). To determine that the BrdU injections were successful, we sacriﬁced one ﬁsh 24 h after injection of BrdU and examined BrdU incorporation in its retina. As shown in Figure 8(A), clusters of BrdU(ϩ) cells were observed in both the INL and ONL. The remaining ﬁsh were sacriﬁced at 14 days post-BrdU injection and labeled nuclei were observed in both the photoreceptor layer and the INL [Fig. 8(B)]. Some small, round,
BrdU-labeled nuclei were clearly located in the ONL, which is consistent with the rod cell bodies. Other labeled nuclei were more elongate and located between the ROS and the ONL, which was consistent with cone cell nuclei. Double labeling with anti-BrdU and the different cone-speciﬁc opsin antisera conﬁrmed that at least the green, blue, and ultraviolet opsin-expressing cone cells were generated from the proliferating cell population [Fig. 8, (C), (D), and (E), respectively]. DISCUSSION
The major goal of these studies was to develop a vertebrate animal model suitable to study adult photoreceptor degeneration and permit a molecular genetic analysis of neural cell regenerative pathways.
Photoreceptor apoptosis was previously shown to occur during several genetic-based retinal degenerative disorders and also subsequent to light treatment of rodents (Wong, 1994; Papermaster and Windle 1995;
Abler et al., 1996). It was important to determine if intense constant light caused photoreceptor apoptosis because we were interested in regeneration of photoreceptor cells rather than the resynthesis of outer segments. For example, light damage in the albino trout retina was characterized by severe ROS truncation (“solar pruning”), but the absence of demonstrable cell death meant that the subsequent recovery did not require production of new rod cells (Allen and
Hallows, 1997). Histological and immunohistochemical characterization of the albino zebraﬁsh retinas indicated that 7 days of constant intense light resulted in widespread photoreceptor cell death that affected all rods and many cone cells. Although both rods and cones were regenerated within 28 days of recovery to nearly wild-type numbers, the wild-type mosaic cone pattern was not reconstituted. Similarly, the spatial patterns of regenerated inner retinal cells in the zebraﬁsh were also disrupted and were sometimes characterized by the closer than normal associations of like cells (Cameron and Carney, 2000). These results may suggest that the mechanisms used in regeneration are not identical to those utilized during retinal development, although the mosaic pattern may develop with additional time. Alternatively, as suggested by
Cameron and Carney (2000), because the regenerated retinas were largely similar in structure to the native tissue, retinal regeneration may be driven by the same mechanisms as during development but their operation occurs within an abnormal background of either absent or masked pattern-forming signals.
Our analysis of light-induced retinal degeneration in zebraﬁsh differed in several respects from previous studies. First, the major effect we observed during degeneration was the loss of both rods and cone cells from the photoreceptor layer. This photoreceptor cell loss was consistent with our observation that the ONL thickness and the nuclei density in the ONL decreased. This is in contrast to the ROS shortening in albino rainbow trout and the ROS shortening and rod cell loss in albino oscars (Allen and Hallows, 1997;
Allen et al., 1999). Additionally, neither report described loss of cone cells like we observed in ze-
Photoreceptor Death and Regeneration
Figure 6 Proliferating cell nuclear antigen (PCNA) immunolabeling during light treatment deﬁnes the initiation of the retinal regenerative response. (A–D) Retinal frozen sections labeled with a monoclonal antibody that detects PCNA, a marker of dividing cells. (A) Control section from a retina before exposure to constant light that, as expected, possesses very few PCNA(ϩ) cells
(arrowheads identiﬁes two cells in the ONL; arrow points to a PCNA(ϩ) cell in the INL). (B)
Retinal section after 24 h of constant light with an increased number of PCNA(ϩ) cells in the inner aspect of the INL (arrows). After 3 days of constant light, larger numbers of PCNA(ϩ) cells are present (C,D). (C) Dividing cells are located in both the INL and ONL. The asterisks identify clusters of proliferating cells in the INL that extend into the ONL. (D) A higher magniﬁcation of a confocal image that demonstrates the elongate morphology of the PCNA(ϩ) cells in the INL and the continuity of these cell columns between the INL and the ONL (asterisks). (E) Change in the number of PCNA(ϩ) cells per retinal frozen section during the 7 days of constant light treatment. Error bars,
S.D. (n values, see Table 2). Scale bars ϭ (A,D) 50 m; also applies to (B) and (C). GCL, ganglion cell layer; INL, inner nuclear layer; ONL, outer nuclear layer; OPL, outer plexiform layer; ROS, rod outer segments.
braﬁsh. Furthermore, some reports localized the initial retinal abnormality in light-induced degeneration to the INL. Penn (1985) reported that after 4 days of constant light treatment, the ﬁrst retinal layer to show damage in the golden shiner (Notemigonus crysoleucas) was the INL. Furthermore, at lower illuminance,
INL changes were detected in the absence of rod or
ONL changes (Penn, 1985). Similarly, Hollyﬁeld et al. (1980) reported a low level of INL pyknosis after
7 days of constant light in Rana pipiens. Contrary to these reports, we did not observe a change in the thickness of the INL during the constant light nor an increase in TUNEL(ϩ) cells in the INL during the ﬁrst 24 h of constant light. In fact, we detected increased numbers of PCNA(ϩ) cells in the zebraﬁsh
INL during the ﬁrst 3– 4 days of constant light, which
could actually increase the cell density in the INL.
Our ﬁndings are similar to the absence of inner retinal damage generally observed in the rodent models
(Rapp, 1995). Second, the photoreceptor cell damage was uniform throughout the retina. Previous studies with rodents revealed regions that exhibited different levels of sensitivity to light-induced degeneration. For example, the photoreceptor layer in the superior retinal hemisphere of the rat was more severely affected than the inferior hemisphere (Rapp et al., 1985). The zebraﬁsh alb retinas exhibited equally severe rod and cone cell changes in the inferior and superior hemispheres after 7 days of constant intense light.
Histopathology and internucleosomal DNA cleavage detection have both been used to assess apoptosis
(Wyllie et al., 1980; Compton, 1992; Gavrieli et al.,
Vihtelic and Hyde
Figure 7 GFAP, PCNA, and BrdU labeling of the Muller glial and proliferating cells in the
constant light-treated retina. Retinal Muller glial cells were identiﬁed by immunolabeling of glial
ﬁbrillary acidic protein (GFAP; A; arrowheads identify the radial processes of Muller cells in a
retina after 21 days of recovery from constant light). Double-label experiments of retinas after 3 days of constant light using anti-GFAP in combination with either PCNA immunolabeling (B) or immunodetection of incorporated BrdU (C) showed dividing cells in the INL and ONL that were associated with the radial processes of the Muller cells. In these confocal images, GFAP is detected
with an FITC-conjugated secondary antibody (green; A,C) or a Cy3-conjugated secondary antibody
(red; B). An arrow in (B) points to a PCNA-labeled cell crossing the OPL, whereas arrows in (C) identify clusters of BrdU-labeled cells associated with the radial processes of Muller glia. (D,E)
BrdU-labeled cells in frozen sections from retinas that were isolated 3 and 12 h post-BrdU injection, respectively. Initially, the majority of the BrdU-labeled cells were conﬁned to the INL. When
BrdU-labeled cells were counted in retinal sections obtained from ﬁsh at increasing time intervals after BrdU injection, an increase in labeled cells in the ONL was observed (F). Error bars, S.D. (n values, see Table 3). Scale bar ϭ (A) 50 m; also applies to (B)–(E). GCL, ganglion cell layer; INL, inner nuclear layer; ONL, outer nuclear layer; OPL, outer plexiform layer.
1992; Gerschenson and Rotello, 1992). Using light microscopy, we observed changes in the photoreceptor layer after 24 h of constant light that were consistent with apoptosis of the rod cells, such as widespread ONL nuclei densiﬁcation and disorganization.
Although we detected apoptotic rod and cone nuclei in situ by TUNEL, the apoptotic cone nuclei were not nearly as prevalent as the rod nuclei at 24 h of constant light treatment, although we did not quantitate this difference. This suggests that either the cone cells were more resistant to apoptosis or they possessed a slower time course of cell death than the rod cells.
The relatively rapid onset of light-induced apoptosis in zebraﬁsh is similar to the response latency observed in light-treated rats and mice. Our analysis demonstrated that the appearance of TUNEL(ϩ) cells in the
ONL began between 3 and 6 h into the constant light treatment and dramatically increased over the next
6 h. Similarly, TUNEL(ϩ) nuclei were observed in the mouse and rat ONL 6 h after light treatment began and increased dramatically by 12 h of light exposure
(Abler et al., 1996; Li et al., 1996; Hafezi et al., 1997).
Raymond and Hitchcock (1997) discussed three potential sources for the regenerative retinal cell population: (i) the rod precursor cells in the ONL, (ii) the
Muller glial cells, and (iii) stem cells in the INL. We
observed two different proliferating cell populations in the regenerating retina that differed in their temporal, physical, and spatial appearance. First, we observed small, round individual cells in the ganglion cell layer after 24 h of constant light. These cells likely represent the phagocytic microglial cells that were described in goldﬁsh retinal lesions (Negishi and
Shinagawa, 1993; Braisted et al., 1994). Although monoclonal antibody NN2 labels goldﬁsh microglial cells, it fails to identify zebraﬁsh microglia (P. Ray-
Photoreceptor Death and Regeneration
Figure 8 Double labeling of regenerated rod and cone photoreceptor cells with opsin and BrdU antibodies. (A) Cells labeled with anti-BrdU 24 h after BrdU was injected. Cell clusters in both the
INL and ONL were positive for the presence of BrdU. Fourteen days after BrdU injection, retinal sections were labeled with the zebraﬁsh rhodopsin, green, blue, or UV opsin antisera in combination with BrdU immunodetection. (B) Rhodopsin/BrdU double labeling. BrdU(ϩ) nuclei are identiﬁed in the ONL (arrowheads) as well as the cone photoreceptor layer (arrow). (C) Green cone outer segments (arrows) of cells that possess BrdU(ϩ) nuclei (arrowheads); (D,E) A similarly labeled long single cone (blue opsin) and short single cone (ultraviolet opsin; UV), respectively. Scale bar ϭ (A,B) 50 m; (C–E) 10 m. GCL, ganglion cell layer; INL, inner nuclear layer; ONL, outer nuclear layer; ROS, rod outer segments; grn opsin, green opsin; UV opsin, ultraviolet opsin.
mond, personal communication). Therefore, we were unable to conﬁrm that these proliferating cells in the ganglion cell layer were phagocytic microglia. Second, we observed another population of proliferating cells after 24 – 48 h of constant light. Unlike the small, round microglial cells, these cells were larger and elongate, similar to the neuroepithelial cells observed in developing retinas. Also, these cells were closely associated in groups that were initially conﬁned to the
INL and later extended into the ONL. On the basis of cell shape and size, cell– cell associations, and their
retinal location, these clustered PCNA(ϩ) cells may be equivalent to the proliferating cells identiﬁed in the
INL of the regenerating goldﬁsh retina (Hitchcock et al., 1996) and the proliferating cells in the INL of growing retinas from trout (Julian et al., 1998).
Previous studies suggest that the proliferating cells in the INL migrate through the outer plexiform layer to the ONL in the developing and regenerating goldﬁsh retina and the growing trout retina. Goldﬁsh retinas injected with ﬁbroblast growth factor (Negishi and Shinagawa, 1993) revealed a peak in PCNA(ϩ)
Vihtelic and Hyde
cell counts in the INL preceding the increase in
PCNA(ϩ) cells in the ONL, which is consistent with the cells moving from the INL to the ONL. When the growing retinas from trout (1–2 g body weight) were sequentially pulse-labeled with two different markers of cell division, Julian et al. (1998) demonstrated the likely movement of proliferating cells from the INL into the ONL. We also observed proliferating cell clusters within the outer plexiform layer. Taken with our observation that the percentage of BrdU(ϩ) retinal cells in the ONL increased over time, either the proliferating cells migrate from the INL to the ONL to repopulate the zebraﬁsh photoreceptor layer or the rate of cell divisions in the ONL increased during this time. Dividing cells in the developing INL of the goldﬁsh retina appeared to rely on the existing glial cell scaffold to seed the ONL with presumptive rod precursor cells (Raymond and Rivlin, 1987). Therefore, the morphology of the proliferating cells in the light-treated zebraﬁsh retina and their intimate association with Muller cells is very similar to the early
development of the goldﬁsh retina. Molecular analysis, to assess pax6 expression, for example, should conﬁrm the neuroepithelial identity of the dividing cells in the INL during the zebraﬁsh retinal regeneration response (Krauss et al., 1991; Hitchcock et al.,
One question remaining was if Muller cells or
closely associated stem cells were the source of the proliferative cell population in the INL. On the basis of GFAP immunoreactivity coupled with BrdU labeling, recently divided Muller cells were identiﬁed mi¨ grating proximal to the external limiting membrane in the ONL of the goldﬁsh retina that was regenerating in response to laser damage (Braisted et al., 1994). In mammalian retinas, Muller cell proliferation is only
observed during rapid or massive retinal injury
(Geller et al., 1995; Kono et al., 1995). Although our light-induced retinal damage in zebraﬁsh should be both acute and severe enough to overcome the Muller
cells’ high threshold against mitogenic stimuli
(Reichenbach et al., 1998), neither of our double-label experiments revealed any GFAP(ϩ) nuclei in the
ONL. Additionally, the numerous BrdU(ϩ) or
PCNA(ϩ) cells in the ONL that were in close physical association with GFAP-labeled processes exhibited no GFAP labeling themselves. Furthermore, the zebraﬁsh retinas contained clusters of BrdU(ϩ) or
PCNA(ϩ) cells in the INL that were associated with individual GFAP-labeled radial processes. This suggests that either the dividing cells were not Muller glia
or they were the progeny of recent Muller cell divi¨ sions that had not differentiated and begun expressing
GFAP. Because of their close association with pre-
sumed pluripotent cells in the teleost INL, the Muller
glial cells may serve as signaling conduits for activation of the stem cells in response to photoreceptor cell death. Alternatively, photoreceptor cell death may result in the loss of factors that normally maintain the
Muller cell differentiated state (Ikeda and Puro,
1995), which in turn leads to Muller cell proliferation,
followed by their migration and differentiation into neural cell types.
In summary, our histological and immunohistochemical analysis showed that light-induced retinal degeneration and the retinal regenerative response in zebraﬁsh could be divided into three distinct phases.
Phase one (cell death) is rapidly initiated during the ﬁrst 24 h of light treatment and is characterized by loss of nearly all rod photoreceptors and also many cone cells. Phase two (cell proliferation and migration) likely begins during the cell death phase, but becomes the predominant histological retinal response after 3– 4 days of constant light. This second phase is characterized by massive cell proliferation that begins in the INL and is followed by detection of large numbers of dividing cells in both the INL and ONL.
The proliferative response continues for several days, and many of the mitotically active cells are associated with the Muller glial cells and appear to migrate into
the ONL. The third and ﬁnal phase (cell differentiation) is characterized by differentiation of the newly generated cells into rods as well as green, blue, and ultraviolet opsin-expressing cone cells. Zebraﬁsh have recently been exploited as a genetic model system to identify a number of eye mutants that affect development, behavior, and degeneration (Brockerhoff et al., 1995; Malicki et al., 1996; Fadool et al.,
1997; Li and Dowling, 1997; Neuhauss et al., 1999;
Daly and Sandell, 2000). Additionally, new technologies are rapidly being developed for the zebraﬁsh system such as retroviral-mediated insertional mutagenesis and the transgenic expression of foreign genes (Higashijima et al., 1997; Amsterdam et al.,
1999). We hope to use these genetic and molecular techniques to further examine the cells and pathways involved in the neural cell regeneration response in the retina.
The authors thank Joseph O’Tousa for reading the manuscript, the anonymous reviewers whose suggestions greatly improved its quality, and Emily Cassidy, Debbie Bang, and the Freimann Life Science Center for their excellent care in maintaining the zebraﬁsh facility. Finally, the authors thank the University of Notre Dame’s Graduate School and College of Science for ﬁnancial support.
Photoreceptor Death and Regeneration
Abler AS, Chang C-J, Ful J, Tso MOM. 1996. Photic injury triggers apoptosis of photoreceptor cells. Res Commun
Mol Pathol Pharmacol 92:177–189.
Allen DM, Hallows TE. 1997. Solar pruning of retinal rods in albino rainbow trout. Vis Neurosci 14:589 – 600.
Allen DM, Pipes C, Deramus K, Hallows TE. 1999. A comparison of light-induced rod degeneration in two teleost models. In: Hollyﬁeld JG, Anderson RE, LaVail
MM, editors. Retinal degenerative diseases and experimental therapy. New York: Kluwer Academic/Plenum
Publishers. p 337–350.
Amsterdam A, Burgess S, Golling G, Chen W, Sun Z,
Townsend K, Farrington S, Haldi M, Hopkins N. 1999. A large-scale insertional mutagenesis screen in zebraﬁsh.
Genes Dev 13:2713–2724.
Bignami A. 1984. Glial ﬁbrillary acid (GFA) protein in
Muller glia. Immunoﬂuorescence study of the goldﬁsh retina. Brain Res 300:175–178.
Boucher S-EM, Hitchcock PF. 1998. Insulin-related growth factors stimulate proliferation of retinal progenitors in the goldﬁsh. J Comp Neurol 394:386 –394.
Braisted JE, Essman TF, Raymond PA. 1994. Selective regeneration of photoreceptors in goldﬁsh retina. Development 120:2409 –2419.
Brockerhoff SE, Hurley JB, Driever W, Neuhauss S, Dowling J. 1995. A behavioral screen for isolating zebraﬁsh mutants with visual system defects. Proc Natl Acad Sci
Cameron DA, Carney LH. 2000. Cell mosaic patterns in the native and regenerated inner retina of zebraﬁsh: implications for retinal assembly. J Comp Neurol 416:356 –367.
Cameron DA, Easter SS. 1995. Cone photoreceptor regeneration in adult ﬁsh retina: phenotypic determination and mosaic pattern formation. J Neurosci 15:2255–2271.
Compton MM. 1992. A biochemical hallmark of apoptosis: internucleosomal degradation of the genome. Cancer
Metast Rev 11:105–119.
Daly FJ, Sandell JH. 2000. Inherited retinal degeneration and apoptosis in mutant zebraﬁsh. The Anatom Rec 258:
Fadool JM, Brockerhoff SE, Hyatt GA, Dowling JE. 1997.
Mutations affecting eye morphology in the developing zebraﬁsh. Dev Genet 20:288 –295.
Fernald RD. 1985. Growth of the teleost eye: novel solutions to complex constraints. Environ Biol Fishes 13:
Fernald RD. 1989. Fish vision. Finlay BL, Sengelaub DR, editors. Development of the vertebrate retina. New York:
Plenum Press. p 247–266.
Gavrieli Y, Sheerman Y, Ben-Sasson SA. 1992. Identiﬁcation of programmed cell death in situ via speciﬁc labeling of nuclear DNA fragmentation. J Cell Biol 119:493–501.
Geller SF, Lewis GP, Anderson DH, Fisher SK. 1995. Use of the MIB-1 antibody for detecting proliferating cells in the retina. Invest Ophthalmol Vis Sci 36:737–744.
Gerschenson LE, Rotello RJ. 1992. Apoptosis: a different type of cell death. FASEB J 6:2450 –2455.
Hafezi F, Steinbach JP, Marti A, Munz K, Wang Z-Q,
Wagner EF, Aguzzi A, Reme CE. 1997. The absence of c-fos prevents light-induced apoptotic cell death of photoreceptors in retinal degeneration in vivo. Nat Med
Hagedorn M, Fernald RD. 1992. Retinal growth and cell addition during embryogenesis in the teleost, Haplochromis burtoni. J Comp Neurol 321:193–208.
Higashijima S, Okamoto H, Ueno N, Hotta Y, Eguchi G.
1997. High-frequency generation of transgenic zebraﬁsh which reliably express GFP in whole muscles or the whole body by using promoters of zebraﬁsh origin. Dev
Biol 192:289 –299.
Hitchcock PF, Lindsey KJ, Easter SS, Mangione-Smith R,
Jones DD. 1992. Local regeneration in the retina of the goldﬁsh. J Neurobiol 23:187–203.
Hitchcock PF, Macdonald RE, VanDeRyt JT, Wilson SW.
1996. Antibodies against pax6 immunostain amacrine and ganglion cells and neuronal progenitors, but not rod precursors, in the normal and regenerating retina of the goldﬁsh. J Neurobiol 29:399 – 413.
Hollyﬁeld JG, Rayborn ME, Medford D. 1980. Damaging effects of constant light and darkness on the retina of the frog. In: Williams TP, Baker BN, editors. The effects of constant light on visual processes. New York: Plenum
Press. p 401– 407.
Ikeda T, Puro DG. 1995. Regulation of retinal glial cell proliferation by antiproliferative molecules. Exp Eye Res
Johns PR. 1977. Growth of the adult goldﬁsh eye. III.
Source of the new retinal cells. J Comp Neurol 176:343–
Johns PR. 1982. Formation of photoreceptors in larval and adult goldﬁsh. J Neurosci 2:178 –198.
Johns PR, Easter SS. 1977. Growth of the adult goldﬁsh eye. II. Increase in retinal cell number. J Comp Neurol
Johns PR, Fernald RD. 1981. Genesis of rods in teleost ﬁsh retina. Nature 293:141–142.
Julian D, Ennis K, Korenbrot JI. 1998. Birth and fate of proliferative cells in the inner nuclear layer of the mature ﬁsh retina. J Comp Neurol 394:271–282.
Kock JH. 1982. Neuronal addition and retinal expansion during growth of the crucian carp eye. J Comp Neurol
Kono T, Kohno T, Inomata H. 1995. Epiretinal membrane formation. Light and electron microscopical study in an experimental rabbit model. Arch Ophthalmol 113:359 –
Krauss S, Johansen T, Korzh V, Moens U, Eriscson JU,
Fjose A. 1991. Zebraﬁsh pax[zf-a]: a paired box-containing gene expressed in the neural tube. EMBO J 10:3609 –
Larison KD, Bremiller R. 1990. Early onset of phenotype and cell patterning in the embryonic zebraﬁsh retina.
Vihtelic and Hyde
LaVail MM, Gorrin GM, Repaci MA, Thomas LA, Ginsberg HM. 1987. Genetic regulation of light damage to photoreceptors. Invest Ophthalmol Vis Sci 28:1043–1048.
LaVail MM, Unoki K, Yasumura D, Matthes MT, Yancopoulos GD, Steinberg RH. 1992. Multiple growth factors, cytokines, and neurotrophins rescue photoreceptors from the damaging effects of constant light. Proc Nat Acad Sci
USA 89:11249 –11253.
Levine EM, Hitchcock PF, Glasgow E, Schechter N. 1994.
Restricted expression of a new paired-class homeobox gene in normal and regenerating adult goldﬁsh retina.
J Comp Neurol 348:596 – 606.
Levine EM, Passini M, Hitchcock PF, Glasgow E,
Schechter N. 1997. Vsx-1 and Vsx-2: two Chx10-like homeobox genes expressed in overlapping domains in the adult goldﬁsh retina. J Comp Neurol 387:439 – 448.
Li L, Dowling JE. 1997. A dominant form of inherited retinal degeneration caused by a nonphotoreceptor cellspeciﬁc mutation. Proc Natl Acad Sci USA 94:11645–
Li S, Chang C-J, Abler AS, Fu J, Tso MOM, Lam T-T.
1996. A comparison of continuous versus intermittent light exposure on apoptosis. Curr Eye Res 15:914 –922.
Lyall AH. 1957. The growth of the trout retina. Quarterly J
Micros Sci 98:101–110.
Maier W, Wolburg H. 1979. Regeneration of the goldﬁsh retina after exposure to different doses of ouabain. Cell
Tissue Res 202:99 –118.
Malicki J, Neuhauss SCF, Schier AF, Solnica-Krezel L,
Stemple DL, Stainier DYR, Abdelilah S, Zwartkruis F,
Rangini Z, Driever W. 1996. Mutations affecting development of the zebraﬁsh retina. Development 123:263–
Marcus RC, Delaney CL, Easter SS Jr. 1999. Neurogenesis in the visual system of embryonic and adult zebraﬁsh
(Danio rerio). Vis Neurosci 16:417– 424.
Marotte LR, Wye-Dvorak J, Mark RF. 1979. Retinotectal reorganization in goldﬁsh-II. Effects of partial tectal ablation and constant light on the retina. Neuroscience
Negishi K, Shinagawa S. 1993. Fibroblast growth factor induces proliferating cell nuclear antigen-immunoreactive cells in goldﬁsh retina. Neurosci Res 18:143–156.
Negishi K, Stell WK, Teranishi T, Karkhanis A, OwusuYaw V, Takasaki Y. 1991a. Induction of proliferating cell nuclear antigen (PCNA)-immunoreactive cells in goldﬁsh retina following intravitreal injection with 6-hydroxydopamine. Cell Mol Neurobiol 11:639 – 659.
Negishi K, Sugawara K, Shinagawa S, Teranishi T, Kuo
CH, Takasaki Y. 1991b. Induction of immunoreactive proliferating cell nuclear antigen (PCNA) in goldﬁsh retina following intravitreal injection with tunicamycin.
Dev Brain Res 63:71– 83.
Negishi K, Teranishi T, Kato S, Nakamura Y. 1988. Immunohistochemical and autoradiographic studies on retinal regeneration in teleost ﬁsh. Neurosci Res Suppl 8:S43–
Neuhauss SCF, Biehlmaier O, Seeliger MW, Das T, Kohler
K, Harris WA, Baier H. 1999. Genetic disorders of vision revealed by a behavioral screen of 400 essential loci in zebraﬁsh. J Neurosci 19:8603– 8615.
Organisciak DT, Winkler BS. 1994. Retinal light damage: practical and theoretical considerations. Prog Retinal Eye
Papermaster DS, Windle J. 1995. Death at an early age.
Invest Ophthalmol Vis Sci 36:977–983.
Passini MA, Levine EM, Canger AK, Raymond PA,
Schechter N. 1997. Vsx-1 and Vsx-2: differential expression of two paired-like homeobox genes during zebraﬁsh and goldﬁsh retinogenesis. J Comp Neurol 388:495–505.
Penn JS. 1985. Effects of continuous light on the retina of a ﬁsh, Notemigonus crysoleucas. J Comp Neurol
Rapp LM. 1995. Retinal phototoxicity. In: Chang LW, Dyer
RS, editors. The handbook of neurotoxicology. New
York: Marcel Dekker. p 963–1003.
Rapp LM, Naash MI, Wiegand RD, Joel CD, Nielsen JC,
Anderson RE. 1985. Morphological and biochemical comparisons between retinal regions having differing susceptibility to photoreceptor degeneration. In: LaVail
MM, Hollyﬁeld JG, Anderson JG, editors. Retinal degeneration: experimental and clinical studies. New York:
Alan R. Liss. p 421– 437.
Raymond P. 1985. Cytodifferentiation of photoreceptors in larval goldﬁsh: delayed maturation of rods. J Comp Neurol 236:90 –105.
Raymond PA. 1991. Cell determination and positional cues in the teleost retina: development of photoreceptors and horizontal cells. In: Lam DM-K, Shatz CJ, editors. Development of the visual system. Vol. 3. Cambridge: MIT
Press. p 59 –78.
Raymond PA, Bassi CJ, Powers MK. 1988. Lighting conditions and retinal development in goldﬁsh: photoreceptor number and structure. Invest Ophthalmol Vis Sci
Raymond PA, Hitchcock PF. 1997. Retinal regeneration: common principles but a diversity of mechanisms. Seil
FJ, editor. Advances in Neurology. Vol. 72. Philadelphia:
Lippincott-Raven. p 171–184.
Raymond PA, Reiﬂer MJ, Rivlin PK. 1988. Regeneration of goldﬁsh retina: rod precursors are a likely source of regenerated cells. J Neurobiol 19:431– 463.
Raymond PA, Rivlin PK. 1987. Germinal cells in the goldﬁsh retina that produce rod photoreceptors. Dev Biol
Reh TA, Levine EM. 1998. Multipotential stem cells and progenitors in the vertebrate retina. J Neurobiol 206 –220.
Reichenbach A, Germer A, Bringmann A, Biedermann B,
Pannicke T, Francke M, Kuhrt H, Reichelt W, Mack A.
1998. Glio-neuronal interactions in retinal development.
In: Chalupa LM, Finlay BL, editors. Development and organization of the retina: from molecules to function.
Vol. 299. New York: Plenum Press. p 121–146.
Robinson J, Schmitt EA, Harosi FI, Reece RJ, Dowling JE.
1993. Zebraﬁsh ultraviolet visual pigment: absorption
Photoreceptor Death and Regeneration spectrum, sequence, and localization. Proc Natl Acad Sci
USA 90:6009 – 6012.
Safars M, Jeanny JC, Mosnier M, Courtois Y. 1997. Protection against light-induced retinal degeneration with
Naftidrofuryl (Praxilene) in the rat. J Ocular Pharmacol
Therapeutics 13:89 –94.
Shahinfar S, Edward DP, Tso MOM. 1991. A pathologic study of photoreceptor cell death in retinal photic injury.
Curr Eye Res 10:47–59.
Stenkamp DL, Barthel LK, Raymond PA. 1997. Spatiotemporal coordination of rod and cone photoreceptor differentiation in goldﬁsh retina. J Comp Neurol 382:
Vihtelic TS, Doro CJ, Hyde DR. 1999. Cloning and characterization of six zebraﬁsh photoreceptor opsin cDNAs and immunolocalization of their corresponding proteins.
Vis Neurosci 16:571–585.
Wang M, Lam TT, Tso MOM, Naash MI. 1997. Expression of a mutant opsin gene increases the susceptibility of the retina to light damage. Vis Neurosci 14:55– 62.
Westerﬁeld M. 1993. The zebraﬁsh book. Eugene, OR:
University of Oregon Press.
Wong P. 1994. Apoptosis, retinitis pigmentosa, and degeneration. Biochem Cell Biol 72:489 – 498.
Wyllie AH, Kerr JF, Currie AR. 1980. Cell death: the signiﬁcance of apoptosis. Int Rev Cytol 68:251–306.