Generation of Activated Sialoadhesin-Positive Microglia during Retinal Degeneration

Abstract

purpose. The retina contains a rich network of myeloid-derived cells (microglia) within the retinal parenchyma and surrounding vessels. Their response and behavior during inflammation and neurodegeneration remain largely undefined. In the present study, the behavior of microglia was closely examined during the onset of photoreceptor degeneration in the rds mouse, to assess their role in photoreceptor apoptosis. The results may have relevance to similar degeneration in humans (retinitis pigmentosa).

methods. Retinas from rds and wild-type CBA mice aged 8, 14, 16, 17, 19, 21, 30, and 40 days were examined immunohistochemically, with antibodies to macrophage cell surface markers, inducible nitric oxide synthase (iNOS), and proliferating cell nuclear antigen (PCNA), during the most active phase of the disease. TUNEL was used to assess photoreceptor apoptosis.

results. In the rds mouse, microglia proliferated in situ (PCNA), migrated to the subretinal space, and adopted an activated phenotype. Maximum microglial cells occurred at postnatal day (P)21, 5 days after the peak in photoreceptor apoptosis (P16). Microglia did not express iNOS, and nitrotyrosine was absent. Sialoadhesin was expressed on microglia from P14, and expression was greatest at P21.

conclusions. During retinal degeneration, microglia are activated and express sialoadhesin. The temporal relationship between photoreceptor apoptosis and microglial response suggests that microglia are not responsible for the initial wave of photoreceptor death, and this is corroborated by the absence of iNOS and nitrotyrosine. Expression of sialoadhesin may indicate blood-retinal barrier breakdown, which has immune implications for subretinal gene therapeutic strategies.

The normal mammalian central nervous system (CNS) and retina possess two intrinsic populations of macrophage: the perivascular macrophage or microglia and retinal parenchymal microglia. ¹ These cells enter the CNS in utero and derive from the same myeloid/monocyte precursors as macrophages of other tissues. ²  

The function of resting microglia is still largely unknown, but their activation and migration to a site of injury are phenomena recognized throughout the CNS, both in rodent models and in humans. ^{3 4 5} These cells, when activated, are capable of secreting a wide range of substances, including cytokines, chemokines, prostaglandins, growth factors, TNFα, nitric oxide (NO), and neurotoxins. In particular, there is growing evidence that NO and microglia-derived neurotoxins play an important role in a range of neuropathologic conditions ^{6 7 8} and that microglia, when deregulated and activated, are a potentially destructive presence in the CNS. 

The rds (retinal degeneration slow) mouse displays the typical features of a photoreceptor dystrophy. The mouse is homozygous for a null mutation in the Prph2 gene encoding the structural protein, peripherin/rds, an integral membrane glycoprotein essential for the formation and maintenance of normal photoreceptor outer segments. ^{9 10} As a result, there is complete failure to develop photoreceptor outer segments and progressive loss of the photoreceptors by apoptosis. ^{11 12 13} Peripherin/rds mutations in humans have been linked to a variety of progressive retinal degenerations, including autosomal dominant retinitis pigmentosa and macular dystrophies. ^{14 15}  

One important unknown in the analysis of retinal degenerations is the sequence of cellular events from genetic miscoding of a protein to photoreceptor apoptosis. It remains unclear why a mutation in a gene encoding a protein such as peripherin/rds, should lead to cell death rather than merely loss of function. Similarly, why should an abnormality in the nonessential biochemical process of phototransduction instigate cell death? 

During normal murine retinal development, there is tissue remodeling, and microglia are involved in the phagocytosis of degenerate neurons in the inner retinal layers but demonstrate a resting phenotype shortly after birth. ¹⁶ An observation made by Sanyal et al. ¹¹ in their initial descriptions of the rds mouse, also seen in the Royal College of Surgeons (RCS) rat, ^{17 18 19} was the presence of macrophages in the outer retinal layers and subretinal space during the degeneration process. 

The finding of a prodigious microglial reaction, the knowledge of potential microglial cytotoxicity, and a lack of adequate explanation for photoreceptor death have led us to undertake a study of microglia in the rds mouse model, to define more precisely the changes in microglial phenotype and activity and ascertain their potential to influence retinal degeneration. 

Methods

Animals

Rds mice and wild-type CBA mice were treated in accordance with the Animal License Act (UK) and the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. Mice were terminally anesthetized with pentobarbital and perfused with PLP fixative (2% paraformaldehyde and 0.05% glutaraldehyde) at postnatal days (P)8, P14, P16, P17, P19, P21, P30, and P40. Eyes were enucleated and fixed in PLP for 4 to 6 hours followed by overnight cryoprotection in 20% sucrose at 4°C. Eyes were then snap frozen in optimal cutting temperature (OCT) compound for frozen sectioning. Cryostat sections (12 μm) were taken and transferred to poly-l-lysine precoated glass microscope slides for storage at −70°C. 

Immunohistochemistry

Experiments were performed using appropriate positive and negative controls (including no primary and appropriate isotype antibody controls). Sections were defrosted and air dried for 1 hour before blocking of endogenous peroxidase activity with 1% hydrogen peroxide in methanol. Nonspecific binding sites were blocked with 1.5% normal serum from the host of the secondary antibody diluted in 2% bovine serum albumin for 30 minutes before incubation with the primary antibody (see Table 1 ) for 2 hours at room temperature or overnight at 4°C. After washes in phosphate-buffered saline, slides were then incubated with a biotinylated secondary antibody (Vector, Burlingame, CA) diluted in 0.1% PBSA, followed by avidin-biotin peroxidase complex and diaminobenzidine (DAB; brown stain) reagent (Vector). Slides were counterstained with hematoxylin, dehydrated with alcohols, and mounted in Histomount (National Diagnostics, Atlanta, GA). Two-color immunohistochemistry was used to demonstrate the coexpression of F4/80 and proliferating cell nuclear antigen (PCNA). For this 3-amino-9-ethylcarbazole (AEC; Vector) was used as the F4/80 cell surface chromogen after which an avidin and biotin block was performed. Sections were then treated with Triton X-100 (Sigma, Poole, UK) to permeabilize before PCNA staining of the nucleus using a kit (Mouse-on-Mouse; Vector) to avoid background staining. Nuclear staining was developed with nickel-enhanced DAB. Two-color immunofluorescence was used to demonstrate coexpression of microglial cell surface markers (F4/80 and sialoadhesin). Sections were prepared as above using an FITC-conjugated goat anti-rat secondary antibody (Stratech, Luton, UK) for sialoadhesin, followed by a biotinylated F4/80 (Serotec, Oxford, UK) detected with streptavidin tetramethylrhodamine isothiocyanate (TRITC; Serotec). Photoreceptor apoptosis was detected using a commercially available TUNEL kit (In Situ Cell Death Detection Kit; Roche Molecular Biochemicals, Lewes, UK) according to the manufacturer’s instructions. 

Microscopy and Quantification

Sections were examined by light microscopy and counted by blinded observers (ADD, CCM). Positively stained subretinal microglia (F4/80 and sialoadhesin) and apoptotic photoreceptors (TUNEL) were counted in a standardized length of retina (1.2 mm) centered on the optic nerve head/posterior pole, using a graticule at ×40 objective magnification. The marked increase in number of microglia and their ramified morphology precluded any attempts at counting cells within the retinal parenchyma. Because subretinal microglia were rounded, individual cell counts could be obtained and corresponded to the semiquantitatively assessed increase in microglia throughout the retina. Three eyes were analyzed for each of P8, P14, P30, and P40 and four eyes for the time points spanning peak disease activity: P16, P17, P19, and P21. At least two nonadjacent sections per eye were counted for each stain. 

For F4/80–PCNA double staining, the number of dual-positive cells in all retinal layers of an entire section was counted. The area of each retinal section was calculated with an image-analysis system (Quantimet; Leica Cambridge, Cambridge, UK) and the number of proliferating microglia calculated per square millimeter. 

Data are expressed as the mean number of cells ± SD. An independent-samples t-test was used to determine statistical significance, where P < 0.05 was considered significant. 

Results

Peaks of Photoreceptor Apoptosis

Using a terminal deoxyribonucleotidyl transferase (TdT)-mediated biotin-16-dUTP nick-end labeling (TUNEL assay), we documented the time course and extent of photoreceptor apoptosis to compare with microglial activity in the same eyes. This revealed a significant peak in apoptosis on P16 (Fig. 1) , confirming previous findings. ¹² Only scattered apoptotic photoreceptors were present at P8 and after P21, when there was a level of continued photoreceptor apoptosis not seen in the wild type. There was a significant difference between amount of apoptosis at P16 compared with P21 (24.2 ± 5.2 and 7.2 ± 1.4, respectively, P = 0.02). 

Increase in Retinal and Subretinal Microglia during Rds Degeneration

In wild-type animals, microglia were found only in the inner retinal layers (Fig. 2A) . Microglia were typically ramified and highly arborized. Although initially present in a slightly increased number, a steady state was reached at approximately P21. In contrast, even as early as P8, microglia were noted in the outer retinal layers, the subretinal space (7.1 ± 1.0), and, in a greater number, in the inner layers of rds retinas (Fig. 2B) . Within the subretinal space, microglia became more rounded and ameboid, thought to be associated with an activated and highly phagocytic state (Fig. 2C) . Occasional rod-shaped microglia were visible crossing the photoreceptor cell body layer (Fig. 2D) . The number of retinal microglia rose markedly, peaking at P21 (20.3 ± 2.7), when the outer retinal surface was often coated with a lining of cells adherent to the vestigial photoreceptor outer segments (Fig. 2E) . By P30 the number of parenchymal and subretinal microglia was reduced (Fig. 2F) , but an increase was maintained up to P40. The time course of increase and reduction of microglia and the temporal relationship to photoreceptor apoptosis is illustrated in Figure 1 . The number of subretinal microglia at P21 is significantly greater than at P16 (P < 0.01), P17 (P < 0.01), and P19 (P < 0.05). Figure 1 also shows that there was a distinct pattern of an apoptotic peak preceding the maximum for microglia by at least 5 days. 

Changes in Microglial Phenotype during Rds Degeneration

After the observation that there is a dramatic increase in microglia and, in particular, subretinal microglia with an activated morphology, we wanted to assess whether there was any change in phenotype that would indicate alterations in cellular activity. Macrophages display marked heterogeneity, adapting to local environment, which regulated their development, differentiation, proliferation, and activation. Monoclonal antibodies directed against cell surface antigens can highlight changes in macrophage function. 

Throughout the time course, microglia in both wild-type and rds retina expressed F4/80 and CD11b. CD11c, a marker found on highly activated microglia and on dendritic cells, was only weakly expressed by a very small number of subretinal cells at the later time points (P30 and P40; Fig. 3A ). A dendritic cell marker CD205 (DEC205) was not expressed at any stage in the time course. 

Microglia in the wild-type retina did not express sialoadhesin (CD169), in keeping with current views on distribution of sialoadhesin-positive macrophages. ^{20 21} In the rds retina, no sialoadhesin expression was found at P8, despite an increased number of microglia. Indeed, only at P16 can a significant number of sialoadhesin-positive macrophages be detected (2.0 ± 2.2), contributing approximately 20% of the microglial population. By P21, there was florid and widespread expression of sialoadhesin by microglia (Figs. 3B 4) which was approximately 40% of microglia (8.9 ± 4.3) as confirmed by two-color immunofluorescence (Figs. 3C 3D 3E) . By P40 most of the microglia did not express sialoadhesin. 

The production of nitric oxide (NO) and reactive oxygen species (ROS) is a major source of tissue damage after activation of macrophages and/or microglia. However, no microglia expressed inducible nitric oxide synthase (iNOS) at any point in the time course studied, including the period of greatest microglial activity. In addition to this, no nitrotyrosine (a marker of oxidative protein damage) was detectable in rds retina. Thymus from 14-day-old mice was used as a positive control. iNOS was expressed by thymic dendritic cells and nitrotyrosine in lymphocytes. 

Proliferation of Rds Microglia In Situ

Microglial may increase in number in response to pathologic conditions in the CNS through in situ proliferation ²² or by recruitment of monocytes from the circulation. ^{4 23} Double staining of microglia with F4/80 and PCNA showed that in both wild-type and rds mice, proliferating retinal microglia were present at P8 in the inner retinal layers. By P14, very few proliferating microglia were present in the wild type, and none were found after this time point, reflecting the general reduction in microglia. However there was sustained microglial proliferation in the rds retina at P14 and P21, when the greatest number of proliferating microglia were located in the outer plexiform layer and on the outer retinal surface (Fig. 3F 3G) . Proliferation of microglia in the rds model thereafter subsided and was minimal by P30 (Fig. 5) . 

Discussion

We have shown that during retinal degeneration microglia became activated, proliferated, and expressed sialoadhesin but were not “classically” activated and failed to express iNOS. That microglia have a phagocytic role after migrating toward a degenerating photoreceptor layer is not in doubt and has been demonstrated in the RCS rat ¹⁸ and albino mice exposed to intense light. ²⁴ However, microglia, by secretion of a plethora of potentially cytotoxic molecules, are also capable of causing neuronal injury and death, and this has been demonstrated in vitro. ^{8 25 26}  

It has been suggested that microglia, having migrated to the photoreceptor layer in response to photoreceptor dysfunction or to debris buildup in the subretinal space, may be the instigators of photoreceptor apoptosis in the RCS rat. ¹⁹ Studies to date have demonstrated greatest microglial activity in these models at approximately the same time as the most rapid loss of photoreceptors. These data can be interpreted either as a phagocytic microglial response to photoreceptor apoptosis, or as apoptosis caused by microglial activity. By closely scrutinizing the period of greatest disease activity in the rds mouse, we have demonstrated that the peak rate of photoreceptor apoptosis precedes the peak in the number of microglia by at least 5 days. These findings indicate that microglia are very unlikely to be the initiators of photoreceptor death, in which case one would expect that maximum microglial activity would exactly coincide with or even precede the maximum photoreceptor apoptosis. It is possible, however, that although innocent in the initial wave of photoreceptor cell death, an exuberant microglial response both potentiates and perpetuates the disease process. A self-perpetuating process of microglial cytotoxic activity may be one of the reasons that replacement of the missing peripherin gene by subretinal gene delivery has no beneficial effect on the long-term rate of photoreceptor loss in the rds mouse. ²⁷  

Nitric oxide is a candidate for microglia-induced neurotoxicity, and inhibitors of NO production have beneficial effects on neuron survival after inflammatory and ischemic CNS injury. ^{6 7} However, even at the summit of microglial activity, we found no expression of iNOS or production of nitrotyrosine, making microglial cytotoxicity through oxidative damage by NO and reactive oxygen species unlikely. 

Under normal conditions, parenchymal microglia are a stable population of cells with little or no turnover with bone-marrow-derived cells. ²⁸ The source of the increased pool of microglia seen in CNS disease has been disputed: proliferation in situ or recruitment from blood? PCNA demonstrated that microglia in the rds mouse proliferated in situ in a far greater number than in the wild type at P14 and P21. Indeed, no proliferating microglia were seen in the wild type after P14. Maximum proliferation in the rds retina occurs before the maximum microglia presence is reached, and thus this phenomenon cannot be explained merely by the greater number of cells. This does not exclude additional recruitment of monocytes from the blood, and the finding of sialoadhesin-positive microglia from P14 in the rds retina may infer blood-retinal barrier compromise, because expression of this marker is thought to require contact with serum. ²⁰ In the healthy CNS, microglia that were protected by the blood-brain barrier were sialoadhesin negative, and sialoadhesin was completely absent in wild-type retina. At earlier stages of rds retinal degeneration, when there was already significantly increased microglial activity, there was no sialoadhesin expression, but by P21 approximately 50% of microglia were sialoadhesin positive. 

Sialoadhesin is one of a group of macrophage-restricted cell surface sialic acid receptor proteins named siglecs (reviewed by Munday et al. ²⁹ ), which is highly conserved between rodents and humans. Although sialoadhesin is not thought to be a phagocytic receptor, its expression may facilitate other phagocytic receptors ²⁹ and increase cell-cell and cell-matrix adhesion. Breakdown of the blood-retinal barrier in retinal dystrophies has implications for future gene therapeutic strategies, in which evasion of systemic immune responses is of paramount importance. It is hoped that the anatomic and immunologic properties of the eye will be advantageous, yet exaggerated systemic immune responses can occur when substances are injected into the subretinal space of a dystrophic murine retina when compared with normal mice. ³⁰ Although only a few DEC205⁻CD11c⁺ cells were observed subretinally, further assessment of immune responses in draining lymph nodes is warranted, to assess whether antigen capture and traffic has the potential to lead to systemic priming, particularly after gene transfer. 

 Supported by grants from The Iris Fund (EHH) and The Marie Curie Fellowship Association (FCS).

 Submitted for publication August 14, 2002; revised October 17, 2002; accepted November 14, 2002.

 Disclosure: E.H. Hughes, None; F.C. Schlichtenbrede, None; C.C. Murphy, None; G.-M. Sarra, None; P. Luthert, None; R.R. Ali, None; A.D. Dick, 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.

 

Details of Primary Antibodies Used

Primary Antibody	Host	Manufacturer
F4/80 (monoclonal)	Rat	Serotec, Oxford, UK
CD11b (monoclonal)	Rat	Serotec
Sialoadhesin (monoclonal)	Rat	Serotec
CD205 (DEC205) (monoclonal)	Rat	Serotec
CD11c (monoclonal)	Hamster	Serotec
iNOS (NOS2) (polyclonal)	Rabbit	Serotec
Nitrotyrosine (polyclonal)	Rabbit	Upstate Biotech, Lake Placid, NY
PCNA (monoclonal)	Mouse	Dako, Glostrup, Denmark

Figure 1.

 

Temporal relationship between photoreceptor apoptosis (TUNEL) and subretinal microglial numbers (mean ± SD). TUNEL staining of photoreceptors peaked at P16, significantly higher than at P21 (#P = 0.020), whereas microglia (stained for F4/80) peaked at P21. Microglia at P21 were significantly more numerous than at P16, P17, and P19. *P < 0.05, **P < 0.01 compared with the number on P21.

Figure 2.

 

Immunohistochemical staining of retinal microglia (brown) expressing F4/80. Normal retina had a small number of ramified microglia, restricted to the inner layers (A, P21). Microglia in rds retina were more numerous in the subretinal space (arrows) as early as P8 (B). Subretinal microglia adopted an ameboid morphology (C), whereas those crossing the photoreceptor layer had a rod-shaped configuration (D). The peak in number of microglia was reached at the P21 (E). By P30, the number of microglia was reduced again (F). Counterstain: hematoxylin. Scale bar, 20 μm.

Figure 3.

 

Immunohistochemical and immunofluorescent staining of rds mouse retinal microglia. A very small number of subretinal cells weakly expressed CD11c at P30 and P40 (A). Sialoadhesin expression peaked at P21 (B). Immunofluorescence demonstrated microglial coexpression of F4/80 (red) and sialoadhesin (green) with 4′,6′-diamino-2-phenylindole (DAPI) nuclear counterstain (C–E, blue). Two-color immunohistochemistry for F4/80 (red) cell surface marker and PCNA (black) nuclear staining showed proliferation of microglia. (F) Ramified plexiform layer PCNA⁺ microglia and (G) three ameboid subretinal microglia: one PCNA⁺ (black nucleus, large arrow), the other two PCNA⁻ (blue nucleus, small arrows). Scale bar, 20 μm.

Figure 4.

 

Time course of sialoadhesin expression by subretinal rds microglia (mean ± SD). As with F4/80, peak expression occurred at P21, significantly greater than at P16 (*P < 0.05).

Figure 5.

 

Comparison of retinal microglial proliferation (PCNA positivity, mean ± SD) in wild-type and rds mice. Proliferation of microglia occurred equally at P8 in rds and wild type (predominantly in the ganglion cell layer), but although this was sustained in rds until P21 (predominantly in the outer retina and subretinal space), it was nearly absent by P14 in wild-type mice.