Proliferation of Microglia, but not Photoreceptors, in the Outer Nuclear Layer of the rd-1 Mouse

Abstract

purpose. To establish whether photoreceptor apoptosis in the rd-1 mouse is accompanied by cell cycle progression. Studies of cell cycle proteins in other models of neuronal death provide consistent evidence that a repertoire of proliferative markers accompanies apoptosis.

methods. The spatiotemporal progression of photoreceptor loss in rd-1 and control mice at postnatal days (PN)8, -10, -12, -15, and -18 was correlated with markers of G₁- and S-phase progression. Photoreceptor death was detected by using morphology and terminal dUTP transferase nick end labeling (TUNEL). Cell-cycle-associated markers consisted of bromodeoxyuridine (BrdU) uptake, and immunolabeling for proliferating cell nuclear antigen (PCNA), Ki-67, and cyclin-dependent kinases-2 and -4. The identity of proliferating cells in the outer nuclear layer was established by double immunolabeling with PCNA and either F4/80 or recoverin.

results. A population of proliferating cells in the outer nuclear layer accompanies photoreceptor death along a central to peripheral gradient in rd-1 retinas. Double immunolabeling for PCNA and F4/80 readily identified these as microglial cells originating from the inner retina. Cell cycle progression in photoreceptors could not be demonstrated.

conclusions. These findings confirm that in rd-1, a preexisting condition for cell cycle progression does not exist as it does in other neurodegenerative conditions. Therefore, in this model, evidence of photoreceptor cell cycle progression in retinas exposed to neurotrophic factors is likely to result from the therapy itself. In addition, the results confirmed that proliferating microglial cells are intimately associated with the degenerative process in rd-1.

A variety of neurotrophic factors, given intraocularly, will slow photoreceptor death in animal models of retinal degeneration. ¹ Protective factors include ciliary neurotrophic factor (CNTF), ^{2 3 4 5 6 7} brain-derived neurotrophic factor (BDNF), and neurotrophin (NT)-3, ^{1 8} glial cell line-derived neurotrophic factor (GDNF), ⁹ fibroblast growth factor (FGF-2,-5, and -18), ^{10 11} lens epithelium-derived growth factor (LEDGF), ¹² and platelet epithelium-derived factor (PEDF). ¹³ Despite success in maintaining photoreceptor structure, concomitant preservation of normal electrophysiology may be lacking. ^{2 6} In addition, Bok et al., ² recently described morphologic changes in rescued rods which suggest increased transcriptional activity in response to CNTF therapy. The nature of such transcriptional alteration is completely unknown. Similarly, it is unclear what the consequences of preserving mutant rods would be. 

Photoreceptor death in the rd-1 mouse and other murine models of retinal degeneration occurs by apoptosis. ^{14 15 16} In these cases, apoptosis is an appropriate biological response to mutations resulting in unsustainable disruption of rod homeostasis. Suppression of this mechanism raises the possibility that mutant photoreceptors supported beyond their natural lifespan may become unable to respond normally to environmental cues and potentially undergo neoplastic transformation. The study of cell cycle proteins in other models of neuronal death provides the unexpected but consistent finding that aberrant cell cycle reentry accompanies apoptosis. In vitro expression of cell cycle proteins accompanies neuronal death resulting from trophic deprivation, ¹⁷ excitotoxicity, ¹⁸ and DNA damage due to UV irradiation, toxins, and oxidants. ¹⁹ These findings are reiterated in animal models of developmental, ischemic, and excitotoxic neuronal death ^{20 21 22 23} as well as in human neurodegenerative conditions. ^{24 25 26 27 28 29}  

Whether similar mechanisms exist in naturally occurring retinal degenerations is unknown. Determining whether an a priori condition for cell cycle reentry in degenerating postmitotic photoreceptors is important because some neurotrophic agents ^{30 31 32} transduce their signal through mitogenic pathways. The purpose of this study was to determine whether photoreceptor cells in one model of retinal degeneration, the rd-1 mouse, undergo cell cycle progression during the course of the disease. This will allow us to assess whether mutant photoreceptors display a predilection for cell cycle reentry that may predispose them to oncogenic transformation after exposure to neurotrophic agents. In addition, should evaluation of cell cycle kinetics in treated photoreceptors be indicated during preclinical trials, this study will provide information on how much proliferative activity is due to the primary disease and how much results from its therapy. 

Methods

Animals and Tissues

All experiments were performed in accordance with the ARVO Statement for Use of Animals in Ophthalmic and Vision Research. All animals were husbanded in accordance with the guidelines of the Association for the Assessment and Accreditation of Laboratory Animal Care. C3H/FeJ mice homozygous for the rd-1 mutation were examined at five time points: postnatal days (PN)8, -10, -12, -15, and -18. Age-matched congenic C3H mice (C3.BliA^Pdeb-rd1) which are wild-type (WT) at the rd-1 locus were used as controls. Both genotypes were obtained from the Jackson Laboratories (Bar Harbor, ME). Euthanasia was performed by placing mice in a prefilled CO₂ chamber, followed by cervical decapitation. Mice were always killed at the same time of day (2 PM). Eyes were fixed in 0.5% glutaraldehyde/Na cacodylate overnight for morphologic examination and in Bouin’s solution for 3 days for TUNEL assay and immunohistochemical procedures. 

Morphology

After death, eyes were enucleated, incised at the limbus, and fixed in 0.5% glutaraldehyde/sodium cacodylate overnight for morphologic examination. After graded dehydration in increasing concentrations of alcohol, they were embedded in Epon resin. Sections (500 μm) from the optic nerve to the ora serrata were cut with a microtome (Ultramicrotome; Carl Zeiss Meditec, Dublin, CA), stained with toluidine blue, and examined with a light microscope (Axioscope; Carl Zeiss Meditec). For each time point in both control and rd-1 animals, four to six eyes were evaluated. To quantify the spatiotemporal progression of photoreceptor degeneration, the number of photoreceptor nuclei in three adjacent columns of the outer nuclear layer were counted in two locations of the retina 500 μm peripheral to the optic nerve, and 500 μm central to the ora serrata. Nuclei in three adjacent columns were counted, and the mean and standard deviations calculated by computer (Excel; Microsoft, Redmond, WA) to assess outer nuclear layer thickness in the central and peripheral retina. 

TUNEL Assay

For all immunohistochemical procedures, skulls were placed in Bouin’s solution for 2 to 3 days before being processed for routine paraffin embedding and sectioning at 5 μm. 

Photoreceptor death in rd-1 and control retinas was detected by terminal deoxynucleotide transferase nick end labeling (TUNEL; Roche Diagnostics, Nutley, NJ) according to the manufacturer’s instructions. To quantify photoreceptor death, a photograph of the entire superior retina was taken. A point midway between the optic nerve and ora serrata divided the retina into central and peripheral halves. TUNEL-positive nuclei for each half were counted to obtain counts in the central and peripheral retina. TUNEL-positive nuclei in inner and outer nuclear layers were counted separately. 

Immunohistochemical Evaluation of Cell Cycle Progression

Bromodeoxyuridine (BrdU) incorporation was used to assess DNA replication in the retina. rd-1 and control mice were given intraperitoneal injections of 5 mg/kg of BrdU (Sigma-Aldrich, St. Louis, MO) 2 hours before death. BrdU incorporation was detected with a commercial kit (Zymed, San Francisco, CA). The signal was amplified with a tyramide signal-amplification kit (Perkin Elmer, Boston, MA). BrdU-positive nuclei were counted in central and peripheral retina as described for the TUNEL assay. To rule out the possibility that BrdU incorporation in photoreceptors occurs as a nonspecific event associated with DNA damage during apoptosis, immunohistochemistry was repeated using primary antibodies against proliferating cell nuclear antigen (PCNA) and Ki-67 (DakoUSA, Carpinteria, CA). PCNA immunohistochemistry was performed at all stages of disease; Ki-67 immunohistochemistry was performed only at PN12. To evaluate G₁/S progression, primary antibodies against cdk2 (H-298) and cdk4 (H-303) were obtained from Santa Cruz Biotechnologies (Santa Cruz, CA). The signal was visualized using tyramide amplification (DakoUSA). Nuclei in cornea and lens served as positive internal controls. Identity, concentrations and sources of primary and secondary antibodies are given in Table 1 . 

Immunohistochemical Double Labeling to Identify Proliferating Cell Types

To identify retinal microglia, F4/80 was applied at a concentration of 1:5. F4/80 was a generous gift from Colin Barnstable, but can also be obtained commercially (see Table 1 ). After application of a biotinylated secondary rabbit anti-rat antibody, cytoplasmic immunostaining was visualized with 3,3′-diaminobenzidine (DAB; brown). After the tissues were photographed with Nomarski optics, the PCNA antibody (developed with the purple pigment VIP) was applied to visualize proliferating nuclei. A rabbit polyclonal anti-human recoverin antibody (a generous gift from Alexander M. Dizhoor) was used to label rod and cone cytoplasmic regions. Recoverin staining was visualized with VIP (purple), followed by PCNA staining visualized with nickel-enhanced DAB (black). Identity, concentrations, and sources of primary and secondary antibodies as well as enzyme substrates are given in Table 1 . 

Results

Photoreceptor Death and BrdU Incorporation in the Outer Nuclear Layer of rd-1 Mice

As previously reported, a central-to-peripheral gradient of photoreceptor loss was noted. ^{14 16} Retinas from rd-1 and control mice are morphologically similar by light microscopy until PN10. At this time, outer nuclear layer width in both the central and peripheral retina is comparable in rd-1 and control mice. Reduction in the width of the outer nuclear layer by approximately one third of the width in WT mice is seen in rd-1 mice by PN12 and continues as photoreceptors degenerate. During the period of active degeneration (PN12–18), the width of the peripheral retina exceeds that of the central retina until most photoreceptors are lost (Fig. 1) . 

Both BrdU- and TUNEL-positive nuclei are noted in the outer nuclear layer of rd-1 mice during the period of active photoreceptor degeneration (Figs. 1 2 3) . At PN8, in both rd-1 and control retinas, BrdU labeling is limited to peripheral neural progenitors at the ora serrata. After PN10, BrdU immunopositivity is absent from the ONL of WT mice. In contrast, in rd-1 mice, by PN10 BrdU-positive nuclei appear in the ONL of the central retina (Fig. 2F) . By PN12 and -15 in rd-1, cell death and DNA synthesis have progressed centrifugally to involve the ONL of both central and peripheral retina (Figs. 2A 2B) . Numbers of TUNEL- and BrdU-positive nuclei in the peripheral retina exceed those present centrally (Figs. 1 2A 2B) . 

TUNEL- and BrdU-positive nuclear staining differ in two ways. First, the number of TUNEL-positive nuclei exceed the number of BrdU-positive nuclei (Figs. 1B 1C 2 3) . Second, the relative location of these two labels also vary. TUNEL positive photoreceptor nuclei are detectable at all levels of the outer nuclear layer in rd-1 retinas, whereas BrdU-labeled nuclei tend to be closer to the external plexiform layer (Figs. 2 3) . 

Markers of Proliferation Expressed by rd-1 in the Outer Nuclear Layer

In rd-1 retinas, nuclear staining for PCNA, Ki-67, Cdk-4, and Cdk-2 was detected first at PN10 and persisted through PN15 (Fig. 3) . No positive nuclei were present in control retinas. Immunostaining for Cdk-4 and Cdk-2 paralleled nuclear BrdU positivity and correlated with photoreceptor loss as retinal degeneration progressed. As with TUNEL and BrdU staining, positivity for proliferative marker staining assumed a central to peripheral progression over time (data not shown). 

Microglia Proliferation in the Outer Nuclear Layer of rd-1 Retinas

F4/80 immunostaining identified microglial cells in the outer nuclear layer of rd-1 retinas at both time points examined (PN10 and -12). These assumed the same central-to-peripheral distribution demonstrated by photoreceptor death and BrdU incorporation (Figs. 4A 4B) . F4/80-positive cells were also noted in the ganglion cell layer, usually around vasculature (Fig. 4B , arrow), as well as in the inner nuclear layer, inner plexiform layer (Fig. 4B , star) and outer plexiform layer. In the inner retina, microglial processes were extensive (Figs. 4B 4C 4D) . These were retracted, and microglial cells assumed a more amoeboid appearance as they traveled through the ONL (Figs. 4E 4F) . Once they emerged into the interphotoreceptor space, pseudopodia could often be seen (Fig. 4F) . In contrast, WT retinas displayed no F4/80 immunoreactivity within the outer nuclear layer (data not shown), and rare, light immunoreactivity in the inner retinal layers. 

Presence of Photoreceptors in the Outer Nuclear Layer of rd-1 Retinas

Double immunostaining for PCNA and either F4/80 or recoverin indicated that all PCNA-positive nuclei belonged to microglial cells (Figs. 5A 5B 5C 5D) . Microglial cells with PCNA-positive nuclei were noted through all cell layers of the retina. Although faint F4/80 immunoreactivity was noted, particularly in perivascular regions of the ganglion cell layer, no proliferating microglia were noted in WT retinas (data not shown). These data suggest that microglial proliferation in response to photoreceptor death is initiated in the inner retina and is sustained as microglial cells invade the outer nuclear layer. In all fields examined, recoverin-positive cytoplasmic profiles and PCNA-positive nuclei failed to colocalize (Figs. 5E 5F) . PCNA-positive nuclei had morphologic features typical of microglial cells and were surrounded by a clear, unstained space. Photoreceptor nuclei are typically round, 4 to 5 μm in diameter with coarsely heterochromatic (rods) or euchromatic chromatin (cones). In contrast, PCNA-positive nuclei were slightly larger (5–8 μm in diameter), pleomorphic, round, or fusiform and usually euchromatic with one to two nucleoli (Figs. 5E 5F 5G) . They were also most commonly located in the inner portion of the ONL layer where one would not expect to find cone nuclei. 

Discussion

In this study, nuclei expressing multiple markers of cell cycle progression were present in the outer nuclear layer of rd-1 retinas, and these were nuclei of microglia. Resident microglial cells are known to migrate into the outer nuclear layer in the Royal College of Surgeons (RCS) rat, ^{33 34} and in murine models of light-induced retinal degeneration. ³⁵ These studies describe activation and subsequent migration of resident microglial cells from the inner retina, where they are often associated with blood vessels, into the outer nuclear layer from the earliest time of photoreceptor damage. Our results in the rd-1 mouse recapitulate these findings. In addition, we demonstrate that proliferation of microglia occurs at all levels of the retina, including the outer nuclear layer. A similar finding has been noted in vitro by de Kozak et al., ³⁶ who demonstrated that microglial cells isolated from RCS rats have a higher capacity to proliferate in culture than those obtained from normal control animals. The distribution of proliferating microglial cells in our study coincides with the spatiotemporal pattern of photoreceptor death, implying that microglial cells are intimately engaged with the degenerative process. Their relationship to dying photoreceptors is likely to be complex. On the one hand, they are a known source of neuronal cytotoxins such as tumor necrosis factor-α, reactive oxygen intermediates, reactive nitrogen oxides, and excitatory amino acids. ^{36 37} Roque et al., ³⁸ demonstrated that retina-derived microglial cells will kill photoreceptors by apoptosis in vitro, thus lending support to the hypothesis that microglia accelerate death of dystrophic photoreceptors. On the other hand, the protective effects of neurotrophic factors on photoreceptors may be mediated both directly (in the case of FGF2) or in the case of BDNF and CNTF (whose receptors are not present on photoreceptors), indirectly through activation of Müller cells and inner retinal neurons. ^{39 40 41}  

In vivo demonstration of cell cycle progression in human and murine neurodegenerative diseases has relied heavily on immunohistochemical demonstration of markers for G₁- and S-phases of the cell cycle. ^{20 21 22 23 26 27 28 29} Neuronal identity of cells expressing cell cycle markers is confirmed either by their obvious neuronal morphology ^{27 29} or by double labeling with a neuron-specific marker. ²³ These reports conclude that G₁-S progression precedes apoptosis in terminally differentiated neurons. However, a question remains whether appearance of cell cycle enzymes indicates true cell cycle reentry or is merely a reflection of dysregulated enzyme synthesis. Using fluorescent in situ hybridization (FISH), Yang et al. ²⁸ showed in Alzheimer’s disease autopsy tissue that predegenerate neurons will replicate their DNA and persist in a tetraploid state for some time before their demise. The mechanism(s) responsible for aberrant cell cycle progression in degenerating postmitotic neurons are unknown. Similarly, it is unclear whether cell cycle reentry will promote neuronal apoptosis or whether it is an incidental epiphenomenon. ^{42 43} Mitotic activation appears to be a consistent phenomenon in a cohort of etiologically distinct neurodegenerative diseases characterized by neurofibrillary tangle formation, suggesting that a broad array of insults may induce cell cycle reentry. ²⁷ In the retina, unscheduled DNA replication induced by SV40 Tag in postmitotic photoreceptors results in their death by apoptosis, ⁴⁴ suggesting that inappropriate cell cycle progression results in elimination of postmitotic neurons rather than division. 

There is no evidence of photoreceptor cell cycle reentry in rd-1 retinas. Apart from the work by Al-Ubaidi et al., ⁴⁴ little is known about cell cycle dysregulation and retinal degeneration. Klein et al. ²³ describe cell cycle reentry in retinal neurons of the Harlequin (Hq) mouse. In this model, a proviral insertion in the apoptosis-inducing factor (AIF) gene results in progressive retinal and cerebellar degeneration first noticeable at 3 months of age. In the retina, degeneration begins in the ganglion cell layer and proceeds to affect all cell layers. Cell cycle reentry (PCNA immunoreactivity, BrdU incorporation, and Cdc-47 immunoreactivity) and caspase-3 activation were noted in postmitotic ganglion, amacrine, and horizontal cells. These phenomena were not present in photoreceptor cells, despite photoreceptor loss. This implies that photoreceptors may use alternate pathways of cell death, and that these pathways may not intersect as intimately with the cell cycle machinery as in other neuronal types. 

The potential for cell cycle progression in mutant photoreceptors may be of concern when considering neuroprotective strategies which use trophic agents which themselves stimulate proliferative pathways. In this study, there was no evidence of a preexisting condition for cell cycle reentry in rd-1 as there is in other neurodegenerative conditions. Therefore, in this model, evidence of photoreceptor cell cycle progression in retinas exposed to neurotrophic factors is likely to result from the therapy itself, rather than from the primary disease. 

 Submitted for publication March 24, 2003; revised July 28, 2003; accepted August 16, 2003.

 Disclosure: C.J. Zeiss, None; E.A. Johnson, None

 The publication costs of this article were defrayed in part by page charge payment. This article must therefore be marked “advertisement” in accordance with 18 U.S.C. §1734 solely to indicate this fact.

 

Table 1.

 

Sources and Concentrations of Immunohistochemical Reagents

Primary Ab	Source	Secondary Ab	Source	Substrate	Source
Biotinylated mouse anti-BrdU monoclonal (applied as is)	Zymed, San Francisco, CA (cat no. 93-3943)	—	—	DAB	With kit
Mouse anti-human Ki-67 monoclonal (applied as is)	DakoUSA (cat no. 7187)	Goat-anti mouse 1:250	Vector, Burlingame, CA (cat no. BA9200)	DAB	Vector (cat no. SK4100)
Mouse anti-rat PCNA monoclonal (applied as is)	DakoUSA (cat no. M0879)	Goat-anti mouse 1:250	Vector (cat no. BA9200)	VIP or nickel-enhanced DAB	Vector (cat no. SK4600; 2200)
Rabbit anti-human Cdk-2 polyclonal 1:200	Santa Cruz Biotechnologies (cat no. H-298)	Goat anti-rabbit 1:400	Vector (cat no. BA 1000)	DAB	Vector (cat no. SK4100)
Rabbit anti-human Cdk-4 polyclonal 1:200	Santa Cruz Biotechnologies (cat no. H-303)	Goat anti-rabbit 1:400	Vector (cat no. BA 1000)	DAB	Vector (cat no. SK4100)
Rat anti-mouse F4/80 monoclonal 1:5	Serotec, Raleigh, NC (cat no. MCAP497)	Rabbit-anti-rat 1:250	Vector (cat no. BA 4000)	DAB	Vector (cat no. SK4100)
Rabbit anti-human recoverin polyclonal 1:200	A. Dizhoor	Goat anti-rabbit 1:400	Vector (cat no. BA 1000)	VIP	Vector (cat no. SK4600)

Figure 1.

 

rd-1 retinas PN8 to -18. Comparison of outer nuclear layer width (A), BrdU uptake (B), and TUNEL staining (C) in the central and peripheral retina. X-axis: age in PN days; y-axis: number of nuclei constituting the width of the ONL (A) or number of positively BrdU- or TUNEL-stained nuclei in the ONL of each half of the retina (B, C). ONL width is expressed as the number of photoreceptor nuclei per column in the ONL and declines more rapidly in the central retina than in the periphery. The relative paucity of photoreceptors in the central retina compared with the periphery is reflected by correspondingly fewer BrdU- and TUNEL-positive cells in central versus peripheral retina. BrdU immunopositivity and TUNEL staining were maximum at PN12 and declined thereafter as the ONL diminished.

Figure 2.

 

(A, B) rd-1 mouse retina, PN12. TUNEL staining (A) and BrdU uptake (B) displayed the same central (less staining) to peripheral (more staining) gradient, corresponding to the spatial pattern of photoreceptor loss. Arrow: central retina. (C–H) rd-1 mouse retinas, PN10 to -15. Both TUNEL staining (C–E) and BrdU uptake (F–H) begin at PN10, peak at PN12 and decline as the ONL is lost. Horizontal arrow: outer plexiform layer BrdU-labeled nuclei generally occupied the inner region of the outer nuclear layer, whereas TUNEL-stained nuclei are present throughout its thickness. Bar: (A, B) 20 μm; (C–H) 30 μm.

Figure 3.

 

rd-1 (A–F) and WT (G–L) mouse retinas, TUNEL and immunostaining for G₁-S markers. (A–D) rd-1 mouse retinas at PN12 displayed positive staining of the ONL for TUNEL (A), and immunopositivity for PCNA, Ki-67, Cdk-4, and Cdk-2 (B–F). Staining of corresponding WT retinas was negative (G–L). Bar, 50 μm.

Figure 4.

 

rd-1 retinas. Distribution and morphology of microglia immunostained for F4/80. Microglial cells labeled with F4/80 were present to a greater extent in the outer nuclear layer of central (A) than in the peripheral (B) rd-1 retina at PN10. Microglial cells tended to localize in the deeper portions of the outer nuclear layer. Positive labeling was also noted near vessels of the ganglion cell layer (B, arrow) and in the inner plexiform layer (B, ), inner nuclear layer, and outer plexiform layer. Horizontal arrow: outer plexiform layer. (C–F) Nomarski images of F4/80 immunopositive glia in rd-1 retinas at PN10 (C, D) and PN12 (E, F). Immunopositive cells in ganglion cell (C) and inner plexiform (D) layers had ramified cell processes. These were withdrawn, and glia assumed a more amoeboid appearance as they traveled through the ONL (E, F). Pseudopodia were evident in those emerging in the interphotoreceptor space (F). Bar (A, B) 100 μm; (C–F) 15 μm.

Figure 5.

 

rd-1 retinas, PN 12. Nomarski images of F4/80 immunostaining (A, C) and double immunostaining for PCNA and F4/80 (B, D). Arrows: localization of microglial nuclei before (A, C) and after (B, D) PCNA immunostaining. Cytoplasmic for F4/80 was visualized with DAB and nuclear staining for PCNA with VIP. Recoverin (visualized with VIP) and PCNA (visualized with nickel DAB) double immunolabeling (E–G) demonstrated PCNA-positive nuclei within the ONL. These had morphologic features of microglial nuclei, surrounded by a clear space (white arrows) that failed to stain for recoverin. Bar: (A–E) 10 μm; (F, G) 4 μm.