Proliferative Response of Microglia and Macrophages in the Adult Mouse Eye after Optic Nerve Lesion

Abstract

Purpose.: The purpose of this in vivo study was to evaluate the proliferative response of immunologic cells during the acute phase after optic nerve (ON) lesion in the neural retina and the ciliary body (CB) in the adult mouse.

Methods.: The number of cells obtained 5 to 10 days after ON crush was compared with that counted after intraorbital ON transection. In addition, after ON crush, the time course of in situ proliferating Ki67⁺ microglia and macrophages was analyzed from 6 hours up to 10 days.

Results.: The number of BrdU⁺F4/80⁺ retinal microglia and ciliary macrophages increased over time, reaching the peak number 10 days after ON lesion. In the retina, both ON lesion types resulted in a similar number of BrdU⁺F4/80⁺ microglia. Approximately 85% of all BrdU⁺ cells were identified as F4/80⁺ microglia. However, this cell population represented only 30% of all F4/80⁺ microglia. The peak of microglial in situ proliferation was found 2 days after ON crush. In the CB, both ON lesion types induced a significant increase in the number of BrdU⁺F4/80⁺ macrophages. Of interest, the number of cells after ON transection further increased over time, whereas those after ON crush did not.

Conclusions.: ON lesion significantly increased proliferation of F4/80⁺ immunologic cells in both the retina and CB. Although no significant differences in cellular response were observed in the retina between both lesion types, ON transection had a more pronounced effect on ciliary macrophages than did ON crush. Therefore, both regions seem not to act in concert during the acute phase after ON lesion.

Cerebral and retinal injuries are accompanied by microglia/macrophage activation with similar characteristics: alterations in expression profiles and cellular distribution patterns, morphologic changes (decrease in ramification), and increased cell density as a consequence of proliferation. ^1–5 In contrast to numerous microglia studies in brain tissue, only a few reports are available on retinal microglia activation, in particular after ON transection ^6–8 or crush. ^9,10 These studies did not include a detailed evaluation of retinal microglia proliferation. 

Besides their presence in the neural retina, numerous immunologic cells reside in the vascular connective tissue of the CB. The CB, iris, and choroid belong to the uveal tract of the eye, an ocular structure known to consist of networks of macrophages and dendritic cells that maintain immunologic ocular homeostasis. ^11–13 As during development, these ciliary immunologic cells can enter the adult normal retina from the peripheral margin and populate the neural retina. ^1,14,15 This cell migration is increased after injury (e.g., retinal detachment or neurotoxic lesion). ¹⁶ Analyses concerning ciliary cellular migration or the proliferative response of these ciliary cells after occurrence of an ON lesion are lacking. 

In a previous study, we observed that in both areas, retina and CB, a substantial fraction of immunologic cells incorporated BrdU in response to ON injury, ¹⁷ indicating that immunologic cells proliferate and, moreover, suggesting that both ocular structures act in concert. Therefore, the purpose of the present study was to determine the number of proliferating retinal microglia and ciliary macrophages after both ON lesion types. ON crush and axotomy differentially induce RGC death. ^18–20 Thus, we suggested that this could result in differential immunologic responses, in particular in the neural retina, where the affected RGCs are located. Since comparative studies on immune cell activation in the eye after ON axotomy and crush have not been reported yet, we asked whether there were significant differences in the number, distribution, and morphology of retinal microglia. We also analyzed immunologic cells of the CB that may represent cells that immigrate after lesion. We focused on the fractions of BrdU⁺F4/80⁺ phagocytes labeled 5 to 10 days after both lesion types, at the time when most of the RGCs die, and analyzed their distribution patterns within the retinal layers and the CB. 

Cumulative BrdU labeling allows for tracking the fates of rarely dividing cells for a certain time but not for differentiating specific cells proliferating in situ. Therefore, an additional approach was performed by means of Ki67 labeling, to estimate the fractions and the time course of locally dividing cells. Since several studies reported on retinal microglia activation after ON axotomy, ^6–8 we analyzed this local proliferative response after ON crush and compared the number of cells with those after cumulative BrdU labeling. ON lesion increased the number of BrdU⁺ immunologic cells in the neural retina, the tissue where the lesioned RGCs are located, and in the CB, the structure known to be home to a pool of immunologic cells. Of interest, the proliferative response of retinal microglia did not differ between both lesion types; yet, in the CB, more ciliary macrophages were activated after ON axotomy. However, there was no obvious structural connectivity of both ocular regions until day 10 after ON lesion, as indicated by the lack of increase in cell density in the peripheral retina. 

Methods

Animals

All experiments were performed in adult female BALB/c mice (20–25 g body weight). The animals were kept in accordance with the European Convention for Animal Care and the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. All experiments were approved by and performed under supervision of the local Animal Care Committee. 

Optic Nerve Lesions

The animals were anesthetized by intraperitoneal (IP) injection of 5% chloral hydrate in phosphate-buffered saline (PBS; 400 mg/kg body weight). For intraorbital crush, the optic bulb was extruded carefully, and the ON was mechanically squeezed, directly behind the posterior eye pole, with fine, curved forceps, for 10 seconds. ²¹ For intraorbital transection (also referred to as axotomy), the ON was exposed, and with the area visualized by a binocular operating microscope (Carl Zeiss Meditec, Inc., Oberkochen, Germany), the dura was incised longitudinally, and the ON was transected ∼2 mm distal to the eye bulb. ²² Care was taken to prevent damaging the retinal blood supply. An intact retinal blood supply was verified funduscopically. Animals with permanent ischemia were excluded. Naïve, non–surgically altered, but anesthetized animals were used as control subjects. 

Injections

After surgery, 5-bromo-2′-deoxyuridine (BrdU, 50 mg/kg dissolved in sterile saline; Sigma, Munich, Germany) was injected intraperitoneally. Injections were made twice daily, starting after lesion and continued to 10 days after the lesion was imposed. Control animals for every time point were also treated, starting after anesthesia. 

Fixation and Sectioning

Animals were killed at the end of the treatment periods by an overdose of 30% chloral hydrate. The eyes were removed and fixed in 4% paraformaldehyde (PFA) for 20 minutes on ice. They were then washed in PBS, enucleated (removal of cornea, lens and iris), and incubated in 30% sucrose overnight at 4°C. After they were washed in PBS, the eye cups were frozen in an appropriate embedding medium (Tissue Tek; Sakura, Loeterwoude, The Netherlands), cryosectioned in 12-μm slices, and analyzed by immunofluorescence staining. For flatmounts, the eyes were removed and fixed in 4% PFA for 10 minutes. Others were enucleated in PBS, and the retina was isolated, further fixed for 30 minutes, washed again, and analyzed by free-floating immunofluorescence. 

Antibodies and Immunofluorescence Staining

For identification of microglia and macrophages, the F4/80 marker was used (Dianova, Hamburg, Germany). It is a commonly used marker for hematopoietic cells in the developing or adult mouse CNS, including the neural retina ^23–25 and nonneuronal tissues, such as the CB or choroid, that are part of the uveal tract. ^11,16,26 We defined microglia as ramified cells (highly branched or somewhat truncated, with thick processes) in the neural retina adjacent to blood vessels, and macrophages as ameboid or large globular cells in the nonneural tissue (nerve fiber layer [NFL], choroids, and CB), in accordance with previous studies. ^6,16 Microglia and macrophage identity was further confirmed by using a marker against the ionized calcium-binding adaptor molecule (Iba)-1. 

An established method of labeling dividing cells is based on the use and administration of the exogenous marker 5′-bromo-2-deoxyuridine (BrdU; Sigma-Aldrich), which is incorporated in the S-phase of the cell cycle (DNA synthesis). ²⁷ This method allows for tracking the fate of proliferating cells, and their progeny and is widely used, ^28–30 in particular for identification of cells with a low proliferation rate. ^29,31,32  

Since apoptotic cells also re-enter the cell cycle, ^33,34 terminal deoxynucleotidyl transferase-mediated dUTP nick-end labeling (TUNEL; Roche Applied Science, Mannheim, Germany) was performed to exclude apoptotic nuclear cell labeling and, moreover, to estimate the number and time course of apoptotic RGCs in acute phase after both lesion types. 

The in situ proliferation of retinal and ciliary F4/80⁺ immunologic cells was analyzed over time by using the Ki67 marker (Novacastra, Newcastle-upon-Tyne, UK). The Ki67 antigen is an endogenous nuclear protein that is expressed in all active phases of the cell cycle, including the late G₁, S, G₂, and mitosis phases, and is often used as a marker of mitosis. ^27,29  

For staining, retinal slices were dried, fixed with 4% PFA for 10 minutes, and washed twice in PBS at room temperature (RT). For BrdU and Ki67 co-staining, the tissue was pretreated with EDTA buffer [pH 8.0] for heat-induced epitope retrieval. For BrdU and F4/80 costaining, tissue was pretreated with 2 N HCl for 20 minutes at 37°C, followed by 0.1 M borate buffer (pH 8.5) usage for 10 minutes, and subsequent washing. Afterward, the slices were permeabilized and blocked with 1% bovine serum albumin (BSA) in PBS, supplemented with 0.3% TritonX-100 and 10% normal donkey serum (NDS) for 2 hours at RT. After nonspecific binding was blocked, the tissue was incubated with primary antisera in 2% NDS, overnight at 4°C. The following antibodies were used: mouse anti-BrdU (1:500; Serotec, Heidelberg, Germany), rat anti-BrdU (1:500; Serotec), rabbit anti-Ki67 (1:100; Novacastra), mouse anti-PCNA (1:100; Signet Laboratories, Dedham, MA), rat anti-F4/80 (1:100; Dianova), rabbit anti-Iba1 (1:500; Wako, Neuss, Germany), rat anti-CD68 (1:100; Serotec), rabbit anti–von Willebrand factor (1:250; Dako, Hamburg, Germany), and mouse anti-NeuN (1:500; Millipore, Eschborn, Germany). After the sections were washed with PBS, they were incubated with secondary antibodies in 10% NDS for 1 hour at RT. Secondary antibodies were donkey anti-rat IgG rhodamine (RhodamineX Red, 1:1000; Dianova); donkey anti-mouse IgG Alexa Fluor 488, donkey anti-rabbit IgG Alexa Fluor 488, and goat anti-rat IgG Alexa Fluor 488 (1:500; Molecular Probes, Leiden, The Netherlands); and goat anti-mouse IgG Cy3, donkey anti-mouse Cy5, and donkey anti-rabbit F(ab)′2 Cy5 (1:500; Dianova). As a test for specificity, some sections were incubated without primary antibodies. The nuclei were counterstained with DAPI (4,6-diamino-2-phenylindole; Sigma) for 5 minutes at RT. TUNEL labeling was performed with a cell-detection kit (Fluorescein In Situ Cell Detection Kit; Roche Applied Science) according to the manufacturer's instructions. For a positive control, some sections were treated with DNase I (1 μg/mL; Sigma) before the TUNEL reaction. 

Cell Counts and Statistical Analysis

The number of BrdU⁺F4/80⁺ microglia and macrophages was evaluated in the neural retina and CB, respectively, after two ON lesion types: crush and transection. Five to 10 days after ON crush, the period when RGC death primarily occurs, the number of BrdU⁺, TUNEL⁺, and F4/80⁺ cells was estimated and compared with the number after ON transection. In an additional approach, the time course of in situ proliferating F4/80⁺ retinal microglia and ciliary macrophages was analyzed with Ki67 and compared with BrdU labeling after ON crush. The number of BrdU⁺ and Ki67⁺ microglia and macrophages was estimated at 6 and 14 hours and 1, 2, 3, 5, 7, and 10 days after ON crush and in naïve control retinas. For cell counting, every fifth slice was completely evaluated in the neural retina and the CB. For the characterization of proliferating immunologic cells, a representative number of sections was analyzed for co-localization. All analyses were performed by confocal laser scanning microscopy (LSM 510 and 710 Meta; Carl Zeiss Meditec). All results are expressed as the mean ± SD. Each group consisted of at least five animals. Statistical significance was assessed by using the Mann-Whitney U test (*P < 0.05, **P < 0.01). Multiple testing was adjusted by means of the Holm-Bonferroni method. 

Results

Retinal Microglia and Ciliary Macrophages

Retinal microglia, which are highly ramified cells with small somata, build networks throughout the entire inner retina (Figs. 1A–C). Double-labeling with Iba1 (Figs. 1B, 1E, 1H, 1K) revealed that retinal F4/80⁺ microglia (Figs. 1A, 1D, 1G, 1J; arrowheads) and F4/80⁺ ciliary macrophages (Fig. 1J, arrows) were co-localized (Figs. 1C, 1F, 1I). However, retinal microglia displayed weaker F4/80 immunoreactivity (Figs. 1G, 1J; arrowheads) than did the immunologic cells of the choroid (Fig. 1G, arrows) or the CB (Fig. 1J, arrows), whereas Iba1 did not show any visible change (Figs. 1H, 1K). Therefore, F4/80 allowed for a better differentiation of retinal and ciliary cells. Stainings with the CD68 marker resulted in a smaller number of retinal (exclusively found in the ganglion cell layer/inner plexiform layer [GCL/IPL]) and ciliary immunologic cells (data not shown). Since F4/80 stains 90% to 95% of the total number of bone marrow–derived cells in the CB, ²⁶ CD68 seems to label only a fraction of retinal microglia and ciliary macrophages. Therefore, the F4/80 marker appeared to be the more appropriate marker for analyzing immunologic cells in this study. 

Figure 1.

 

F4/80⁺Iba1⁺ retinal microglia and ciliary macrophages in the adult mouse eye. Immunofluorescent staining with F4/80 (A, D, G, J; red) and Iba1 (B, E, H, K; green; merged in C, F, I) antisera on retinal flatmounts (A–F) and on horizontal sections of the retina (G–I) and the CB (J, K). Highly ramified F4/80⁺ microglia in the GCL and IPL of retinal flatmount preparations (A, D) were clearly co-localized with the Iba1 marker (B, E merged in C, F). Although F4/80 often labeled only the silhouettes of the nuclei (D), Iba1 strongly labeled the cell bodies (E, merged in F). (G–K) Ramified F4/80⁺ retinal microglia were observed in the inner retina and OPL (G, J; arrowheads) and co-expressed Iba1 (H, K; merged in I). F4/80 labeling within the neural retina (G, J; arrowheads) was weaker than that in the choroid (G, I; arrows) and the CB (J, arrows), whereas Iba1 staining showed no differences in labeling intensity within these ocular structures (H, K). PR, photoreceptor layer. Scale bar, 50 μm.

Comparison of Retinal Microglia after ON Crush and ON Transection

In the normal retina, ramified F4/80⁺ microglia were mainly found in the IPL and OPL (Figs. 2A/A′, 2B/B′). Some of these cells were also BrdU⁺, indicating continuous self-renewal under physiological conditions (Figs. 2A, 2B). 

Figure 2.

 

F4/80⁺ microglia in the naive and lesioned retina. (A–K) Immunofluorescent staining with F4/80 (green) and BrdU (red) antisera under physiological conditions (A, B) and after ON lesion (C–K). In naive control retinas, microglia were predominantly found in the IPL and OPL (A, B). In the IPL, at different time points, occasional BrdU⁺ cells were observed (A/A′, B/B′). (C–K) After ON lesion, the number of F4/80⁺ microglia increased, in particular in the inner retinal layers (C). BrdU⁺F4/80⁺ microglia were mainly found in the GCL (D/D′, E/E′), IPL (F/F′–H/H′), and OPL (I/I′). Most of the retinal microglia were ramified—either sparsely branched with thick processes (D/D′–F/F′) or highly ramified (G/G′–I/I′)—and a significant number had incorporated BrdU (C–K). In the NFL, but not within the retinal parenchyma, large ameboid macrophages were observed, in the same low number as in the control retinas (J, arrows). Ramified retinal microglia (J, K; arrowheads) were clearly distinguished from these macrophages in the NFL as well as from the ameboid cells located in the choroid (K, arrows). axo, axotomy. Scale bar: (A–B, J–K) 50 μm; (A/A′, B/B′, D/D′–I/I′) 25 μm.

After ON lesion, we observed an increase in F4/80 immunoreactivity of microglia in the GCL/IPL compared with that in the nonlesioned tissue (Fig. 2C). After both ON axotomy and crush, there was an increase in retinal microglia density, reaching the maximum number after 7 days (axotomy: 161 cells per section; crush: 152 cells per section; controls: 102 cells/section). Independent of the lesion type, 5 to 10 days after ON lesion, similar cell distribution and morphologies were observed (Figs. 2C, 2D/D′–I/I′). BrdU⁺F4/80⁺ microglia were found in the GCL (Figs. 2D/D′, 2E/E′), IPL (Figs. 2F/F′–H/H′), and OPL (Figs. 2I/I′). Nearly all (∼ 99%) parenchymal retinal BrdU⁺F4/80⁺ microglia displayed a ramified shape (Figs. 2D–I). In the GCL and IPL in particular, some cells were truncated with sparse, thick processes (Figs. 2D–F), whereas others remained highly ramified (Figs. 2G–2I). Ameboid, round, parenchymal microglia were rarely observed (<1% of the total retinal F4/80⁺ cell population) and were always BrdU⁻ (not shown). Large, globular F4/80⁺ macrophages were also seen but were restricted to the NFL and were not found in the retinal parenchyma (Fig. 2J, arrows). The large cells made only 3% (approximately two to five cells/section) of the total retinal F4/80⁺ population in naïve or lesioned tissue. The globular cells were almost exclusively BrdU⁻ and were clearly distinguishable from the ramified parenchymal microglia (Fig. 2J, arrowheads). In addition, at any time point analyzed and independent of lesion type, retinal ramified microglia were clearly distinguished from the ameboid cells from the choroid (Fig. 2K, arrows). 

Five to 10 days after ON crush, a threefold, significant increase in the number of BrdU⁺F4/80⁺ microglia was found (Fig. 3A) compared with that in the naïve control retinas (Fig. 3A), with the number of stained cells unchanged over time. The fraction of BrdU⁺ microglia in the total number of F4/80⁺ microglia amounted to ∼27%, which was significantly increased compared with the number in naïve control retinas (9%–11%; Table 1). The majority of proliferating cells in the lesioned retina (∼81% to 85%) were identified as F4/80⁺ microglia (controls 91%–96%; Table 1). 

Figure 3.

 

The number of BrdU⁺ microglia in the retina (A) and CB (B) after ON lesion. (A) In the neural retina, the number of BrdU⁺ microglia was significantly increased 5, 7, and 10 days after ON crush or axotomy compared with the corresponding control retinas and remained high. Values obtained for both lesion types were similar 5 to 10 days after ON lesion, with a threefold increase compared with control retinas. (B) In the CB, after ON crush, microglia proliferated at a low rate with no increase over time, but the number of cells was significantly increased compared with naïve controls at all three time points analyzed. After axotomy, the number of BrdU⁺ microglia further increased over time compared with that in naïve control retinas and were significantly higher than those after ON crush. Data are expressed as the mean ± SD; significant differences between the lesion and corresponding control groups as well as within the lesion groups over time are indicated: **P < 0.01, *P < 0.05; significant differences between the two lesion types are also indicated: ⁺ P < 0.05.

Table 1.

 

Percentages of Proliferating Immunologic Cells in the Retina and CB after ON Lesion

Condition	Retina		CB
	BrdU+ Microglia as Percentage of the Total Number of BrdU+ Cells	BrdU+ Microglia as Percentage of the Total Number of Microglia	BrdU+ Macrophages as Percentage of the Total Number of BrdU+ Cells	BrdU+ Macrophages as Percentage of the Total Number of Macrophages
5-d Crush	84.7 ± 3.7	25.8 ± 4.4*	79.6 ± 10.8	14.1 ± 5.3
5-d Axotomy	84.8 ± 8.6	26.9 ± 2.6**	72.0 ± 2.8*	24.0 ± 9.6*
5-d Naive	90.6 ± 5.9	9.5 ± 1.3	90.3 ± 6.5	8.5 ± 3.1
7-d Crush	80.5 ± 12.1	26.3 ± 4.3**	76.5 ± 7.8*	13.6 ± 5.5+
7-d Axotomy	81.3 ± 7.8	30.9 ± 5.4*	66.5 ± 13.6*	34.7 ± 11.7*,+
7-d Naive	94.6 ± 6.3	9.2 ± 2.9	91.5 ± 4.9	11.2 ± 1.7
10-d Crush	81.1 ± 6.5	27.1 ± 3.9*	68.3 ± 12.3*	12.0 ± 2.5+
10-d Axotomy	81.4 ± 4.2	29.7 ± 2.1*	68.2 ± 3.2*	56.5 ± 3.5**,+
10-d Naive	95.6 ± 8.2	10.6 ± 2.4	90.9 ± 6.6	10.7 ± 4.9

 Significant differences between lesion and corresponding control group are indicated: ** P < 0.01, * P < 0.05; significant differences between the two lesion types are also indicated: ⁺ P < 0.05.

Five to 10 days after axotomy, there was a significant three- to fourfold increase in BrdU⁺ microglial cells (Fig. 3A), with a similar number of cells and time course compared with those after the ON crush. The fraction of BrdU⁺F4/80⁺ cells constituted ∼81% to 85% of all BrdU⁺ cells; however, only 27% to 31% of all F4/80⁺ microglia were BrdU⁺ (Table 1). There was no significant difference in the absolute or relative number of cells between the two ON lesion types. 

Comparison of Ciliary Macrophages after ON Crush and ON Transection

In the naïve CB, numerous F4/80⁺ cells were found mainly at the surface of the vitreous but also within the stroma (Fig. 4A). Few of the stromal F4/80⁺ macrophages (approximately two to three cells/section), which were often associated with blood vessels, were BrdU⁺ (Figs. 4A, 4B; arrowheads). Thus, ciliary F4/80⁺ cells also proliferated under physiological conditions, but at a slow rate. 

Figure 4.

 

BrdU⁺ immunologic cells in the CB. Immunofluorescent staining with F4/80 (A–E; green) and BrdU (A, C–E; red) as well as von Willebrand factor (B; blue) antisera in the CB of naïve (A, B) and lesioned eyes (C–E). Under physiological conditions, F4/80⁺ cells were found at the surface of the vitreous (epithelial layers) as well as in the ciliary stroma, often adjacent to von Willebrand factor⁺ blood vessels (B; blue). Only occasional cells were BrdU⁺ (red), and these were mainly located in the stroma (A, B; arrowheads). After ON crush, an increase in F4/80⁺ cells was observed, and some of these cells were also BrdU⁺ (C, arrowheads). However, there were also nonhematopoietic BrdU⁺F4/80⁻ cells found (C, arrows). Compared with ON crush, the number of cells was even more increased after axotomy (D). BrdU⁺F4/80⁺ cells were mainly found in the ciliary stroma (arrowheads), but BrdU⁺F4/80⁻ cells (arrows) were also present. (E) The morphologic diversity of ciliary immune cells (sections 1–4). In the CB, small cells with fine-branched morphology were observed in a low number at the ciliary epithelium (section 1, *), which were BrdU⁺ (arrowhead) or remained BrdU⁻. Most ciliary cells displayed intense F4/80 staining and were large, ameboid cells (sections 2–4). Clusters of intensely labeled cells were seen at the surface of the vitreous (sections 2–3) or within the stroma (section 4). Some of these F4/80⁺ cells were BrdU⁺ (2,2′; 4,4′; arrowheads); BrdU⁺F4/80⁻ cells were also seen (4,4′, arrows) as well as intensely labeled BrdU⁻F4/80⁺ cells—in particular, in the pars plana, adjacent to the neural retina (section 3). axo, axotomy. Scale bar: (A–E) 50 μm; (F–I) 25 μm.

After ON crush, there was an increase in F4/80⁺ macrophages, especially in the ciliary stroma (Fig. 4C). After 5 to 10 days, a significant increase in BrdU⁺F4/80⁺ macrophages was found compared with that in the corresponding control retinas (Fig. 3B), with no further increase over time. After ON crush, 68% to 80% of all BrdU⁺ cells were identified as F4/80⁺ macrophages (control, 90%–92%; Table 1). Thus, the majority of ciliary BrdU⁺ cells had an immunologic identity (Fig. 4C, arrowheads) and ∼20% to 30% of the BrdU⁺ cells found in the ciliary stroma were nonhematopoietic (F4/80⁻; Fig. 4C, arrows). Of interest, only 12% to 14% of all ciliary F4/80⁺ cells were BrdU⁺, and the fractions of BrdU⁺F4/80⁺ cells did not increase over that in the control retinas (9%–11%; Table 1). 

After axotomy, a further increase in ciliary F4/80⁺ cells was found (Fig. 4D) compared with those after ON crush. BrdU⁺F4/80⁺ cells were observed at the surface of the CB; however, the majority of them were located in the stroma (Fig. 4D, arrowheads). The number of BrdU⁺F4/80⁺ cells further increased over time (Figs. 3B), reaching a fourfold increase in the number of cells after 10 days compared with the number after crush injury (Fig. 3B). The fraction of BrdU⁺F4/80⁺ cells did not markedly change at 5 to 10 days (∼68%–77% of all BrdU⁺ cells). However, the fraction of BrdU⁺F4/80⁺ cells of the total number of F4/80⁺ macrophages increased from 24% (day 5) to 35% (day 7) and further to 66% (day 10; Table 1). Similar to the observations after ON crush, approximately one fourth of the F4/80⁺ cells were BrdU⁻ and not every BrdU⁺ cell was F4/80⁺ (Fig. 4D, arrows). 

Besides the differences in the number of cells, independent of the lesion type, similar morphologies of ciliary cells were observed (Fig. 4E, sections 1–4). Large, ameboid, intensely labeled F4/80⁺ macrophages in the stroma or epithelium (Fig. 4E, sections 1, 2, 4); smaller cells in the cilioretinal junction (Fig. 4E, section 3); and fine-branched, lightly stained F4/80⁺ cells were found in the epithelial layers (Figs. 4E, section 1, asterisks). These cells were still located at the surface (epithelial layers) and in association with blood vessels, but were mainly diffusely scattered in the stroma of the CB. Independent of F4/80 expression level and morphology, some of these cells were BrdU⁺ (Fig. 4E, sections 2/2′; 4/4′; arrowheads). Not all BrdU⁺ cells were F4/80⁺ (Fig. 4E, section 4/4′; arrows). Despite the increase in macrophages within the CB, there was no cellular increase in the adjacent neural retina after ON crush (Fig. 4C) or axotomy (Fig. 4D) that indicated a cellular migration from the CB. 

Taken together, the results show that in the acute phase, both ON crush and transection resulted in a significant increase in the number of BrdU⁺ microglia and ciliary macrophages when compared with that in naive control retinas. The cell morphologies and distribution patterns of retinal microglia and ciliary macrophages were nearly identical after both injury types. All retinal microglia remained ramified, showing various levels of ramification, and the ciliary macrophages were round/ameboid. BrdU labeling demonstrated that most of the BrdU⁺ cells during the acute phase after ON lesion were F4/80⁺ retinal microglia and ciliary macrophages. In the retina, there were no significant differences in absolute and relative number of cells between the two lesion types. However, in the CB after axotomy, the absolute number of BrdU⁺F4/80⁺ cells as well as the percentage of BrdU⁺F4/80⁺ cells from the total number of F4/80⁺ cells further increased over time compared with those after ON crush. In contrast to the retina, ciliary cells showed a different proliferative response after both lesion types. Of interest, there was no obvious cellular migration from the CB into the retina, as evidenced by a lack of increase in cell density in the peripheral retina. 

Time Course of Retinal Microglia Proliferation during the Early Phase after ON Crush

To estimate the number and time course of local dividing cells after ON crush, we performed Ki67 labeling. Six to 14 hours after the nerve crush, no notable cell proliferation was observed. In the neural retina, up to two cells per section were found that entered the cell cycle and expressed the nuclear protein Ki67 or PCNA and were also BrdU⁺ (Figs. 5A–E, 6A). BrdU⁺ microglia found after 6 to 14 hours were in situ proliferating cells (100% BrdU/Ki67 co-localization). One day after the ON crush, the number of Ki67⁺ microglia was increased, and more Ki67⁺ than BrdU⁺ cells were observed (Figs. 5F, 6A), which were identified as mitotic, local dividing microglia (Fig. 5G). The peak in the number of Ki67⁺ cells was reached after 2 days. Thereafter, the number of cells continuously decreased over time, and at 10 days after ON lesion, in situ microglia proliferation was similar to that in the control retinas (Fig. 6A). By contrast, the number of BrdU⁺ microglia slowly increased in the first hours after crush and then further increased over time, reaching the maximum number after 7 to 10 days (Fig. 6A). Thus, the fraction of BrdU⁺Ki67⁺ microglia decreased over time (after 2 days: 54%, after 3 days: 21%), and consequently more BrdU⁺Ki67⁻ than BrdU⁺Ki67⁺ cells were detected (Fig. 5H). Five to 10 days after the ON crush, only ∼5% of BrdU⁺ microglia proliferated in situ. 

Figure 5.

 

In situ cell proliferation and TUNEL in the lesioned retina. Immunofluorescent staining of the retina with Ki67 (A, E, F, H; green), BrdU (D–F, H, I; red), and PCNA (C; green, E; blue) antisera and TUNEL (I, J; green) and DAPI (A–D, F–J; blue) nuclear staining. (A–E) Within the first 14 hours after ON crush, only a few cells underwent in situ cell proliferation and were positive for Ki67 (A, E; green) or PCNA (C; green, E; blue) and incorporated BrdU (D, E; red). (F) After 1 day, an increasing number of cells proliferated in situ (Ki67, green) and in most cases were BrdU⁺ (red). Mitotic cells were often BrdU⁻ (F, G; *). (H) Three days after ON crush, the number of BrdU⁺ cells (red) exceeded the number of in situ proliferating BrdU⁺Ki67⁺ cells (red/green, merged). Most BrdU⁺ cells were found in the GCL and plexiform layers (H, I). None of the cells were apoptotic (I, J; TUNEL, green); however, they were often located close to TUNEL⁺ cells (I). axo, axotomy. Scale bar: (A, F, H) 50 μm; (B–E) 25 μm; (G) 10 μm; (I, J) 100 μm.

Figure 6.

 

Time course of proliferating cells in the retina (A) and CB (B) after ON crush, with corresponding control retinas. (A) In the lesioned or naïve retina, within 14 hours, sparse cell proliferation was observed. After 1 day, the number of Ki67⁺ cells increased, reaching a maximum after 2 days, and decreased from 3 to 10 days after injury. In naïve control retinas, a low level of in situ cell proliferation was found during the period analyzed. After 1 day, the number of BrdU⁺ cells was somewhat lower than that of Ki67⁺ cells but further increased over time, reaching the maximum number 7 and 10 days after ON crush. The number of BrdU⁺ cells also slightly increased in the control retinas over time, reaching a plateau after 5 days. (B) In the CB, increases in BrdU⁺ and Ki67⁺ cells were observed after 1 day. Although the number of BrdU⁺ cells further increased over time up to day 10, the number of Ki67⁺ cells continuously decreased. In naive control retinas, moderate in situ proliferation was observed, with the number of BrdU⁺ cells slightly increasing over time.

In the naïve control retinas, sparse in situ proliferation was found. The number of BrdU⁺ cells also increased over time, with the number of cells unchanged after 3 days (Fig. 6A). At early time points (6 hours–1 day) all microglia were BrdU⁺Ki67⁺, but the percentage decreased steadily over time (after 3 days: 44%; after 10 days: 4%). 

TUNEL labeling confirmed that after ON lesion, NeuN⁺ RGCs underwent cell death and that no BrdU⁺ cells were apoptotic (Figs. 5I, 5J). Most of the TUNEL⁺NeuN⁺ RGCs in the GCL were found 5 days after ON lesion, with the number decreasing over time after both ON axotomy (5 days: 25.9 ± 0.94 cells/section; 7 days: 15.5 ± 3.32 cells/section; 10 days: 5.8 ± 1.49 cells/section) and ON crush (5 days: 26.6 ± 3.03 cells/section; 7 days: 13.3 ± 2.68 cells/section; 10 days: 5.4 ± 0.34 cells/section). 

The Time Course of Ciliary Macrophage Proliferation in the Early Phase after ON Crush

In the CB within the first 14 hours after ON crush, sparse cell proliferation occurred (Fig. 6B), similar to the observations in the retina. The number of Ki67⁺ macrophages was increased 1 day after ON crush and continuously decreased over time. Six to 24 hours after ON crush, every BrdU⁺ cell was identified as Ki67⁺ in situ proliferating macrophages (100% co-localization). Two to 10 days after crush, the number of BrdU⁺ cells further increased, but to a lesser extent than in the retina. The number of Ki67⁺ macrophages as well as the fraction of co-labeled BrdU⁺Ki67⁺ macrophages decreased over time (2 days: 40%; 5 days: 10%, 10 days: 1%). Similar kinetics were found in the control retinas. No TUNEL⁺ cells were detected in the CB. 

Taken together, the additional in situ proliferation assay demonstrated that in the retina and the CB, local cell division peaked on days 1 to 2 after ON crush and that BrdU⁺ cells were indeed dividing microglia and ciliary macrophages. Thus, in lesioned retinal tissue, the increase in microglia density was, to a great extent, due to local cell division. In the retina, one proliferative, intense event was detected between days 1 and 3 after ON crush, with a logarithmic increase of BrdU⁺ cells occurring from days 5 to 10. In the CB, a more moderate, but continuous division took place, leading to a linear increase over time. 

Discussion

In the rodent retina, microglial activation has been reported after ON injury, mainly with regard to phagocytic activity of dying RGCs. ^6,8,35 Of interest as well, immunologic cells of the CB can respond to ocular injury and migrate to affected areas. ^13,16 Characterization of retinal microglia proliferation and analyses of the proliferative capability of ciliary macrophages after ON injury have not yet been reported. In this study, we showed, for the first time, that an ON lesion results in proliferation of F4/80⁺ retinal microglia and ciliary macrophages, and that their number increases in these ocular tissues after ON lesion. 

Retinal Microglia under Physiological Conditions

In the naïve adult mouse retina, the increased number of cells and distribution of microglia are consistent with the results of other reports. ^7,9,24 Under physiological conditions, a continuous but very low rate of in situ proliferation from resident microglia was observed, and nearly all the proliferating cells were ramified microglia, consistent with results in sham-operation rats, ¹⁰ but contrary to a previous report on flatmounted naïve retinas, ¹⁵ possibly due to the different methods used. 

Retinal Microglia after ON lesion

After both ON lesion types, upregulation of F4/80 immunoreactivity was observed in the inner retinal layers within a few days, in accordance with previous reports. ^36,37 The observed increase in cell density and morphologic changes (larger somata and thicker processes accompanied by a reduction in ramification), indicated microglia activation. ^1,37,38 After imposition of both ON lesion types, we rarely observed ameboid F4/80⁺ microglia within the retinal parenchyma (<1%), in contrast to previous reports. ^7,8 Most microglia showed a ramified shape, in accordance with findings in another study, in which activated microglia remained ramified while phagocytizing cellular debris. ⁹ Both types of ON lesion resulted in a similar number of cells and distribution, despite reported differences in lesion severity. ^18,20,39 Therefore, similar mechanisms of microglia activation appear to be acting, independent of lesion type. Of note, in a previous study, we made similar observations concerning astrocyte and Müller cell responses after ON lesion. ¹⁷  

It was assumed that, after ON axotomy, the proliferation of microglia started a few days after RGCs began to die by apoptosis (day 5–7 after injury) ^18,19,21 ; however, this assumption was based on increased microglial density, peaking 12 days after ON transection. ^6,7 In the present study, the peak of RGC loss for both lesion types was found after 5 days (highest number of TUNEL⁺NeuN⁺ RGCs), consistent with another report. ⁹ That the maximum number of microglia was also observed a few days after the RGCs started to die (7 days after ON lesion), in agreement with previous observations, ⁷ led to the assumption that RGC death would precede microglia proliferation. However, in the present study, the peak of in situ proliferation was found in the neural retina 2 days after optic nerve ON crush, indicating that resident microglia rapidly respond to injury. Thus, there was one enhanced proliferative event between days 1 and 3, several days before the RGCs started dying and the ensuing histologic alterations became apparent. From days 5 to 10, the microglia in situ proliferation returned to physiological levels, leading to the conclusion that RGC death seemed not to be the main stimulus for microglial proliferation. The highest number of BrdU⁺ microglia was found up to day 7 after ON lesion. Since cumulative BrdU labeling allows the visualization of most of the proliferating cells for a certain time span, ^32,40 the number of BrdU⁺ microglia obtained between days 5 and 10 represent an accumulated view of all microglia that have undergone cell division between days 1 and 5, thus leading to the observed logarithmic increase in the number of BrdU⁺ microglia. Ki67⁺BrdU⁺ microglia were indeed proliferating cells, as was further confirmed by the presence of dividing cells, co-expression of PCNA, ²⁷ as well as the absence of TUNEL labeling. ^33,34 Thus, the lesion-induced increase in microglia density was to a great extent due to local cell proliferation, in accordance with conclusions reached in other reports. ^6,41.  

In previous work, we analyzed lesion-induced cell proliferation and found that approximately 18% of all retinal BrdU⁺ cells were also nestin⁺, identifying most of them as GFAP⁺-reactive astrocytes or Müller glia. A minority of cells were identified as endothelial cells and pericytes. ¹⁷ We suggested that the majority of the remaining unidentified cells represent microglia. ¹⁷ In the present study, we showed that approximately 80% to 85% of all BrdU⁺ cells were indeed proliferating microglia and ∼33% of all microglia did proliferate. These percentages are in good accordance with results found in the brain, after transection of the entorhino-dentate perforant path projection. ^42,43 Since this CNS lesion occurred without the junctional breakdown of the blood–brain barrier, the three- to fourfold increase in microglia density was proposed to occur primarily due to local proliferation. ^42,43 Nevertheless, a minor fraction of recruited bone marrow–derived cells from the circulating blood also contributed to the increased number found after a cerebral lesion. ^42,43  

ON transection occurs without damaging the BRB as well, ^6,8,44 and the entry of molecular particles or bone marrow–derived cells, such as macrophages, into the retinal parenchyma is prevented. ^6,45 Under physiological conditions, replenishment of retinal microglia from circulating bone marrow–derived cells has been reported in several studies. However, cellular migration was primarily not observed until 6 to 8 weeks after transplantation. ^14,15,46 In this study, independent of the lesion type, an unchanged small number (<3%) of globular macrophages, compared with the number in the control retinas, was exclusively found in the fiber layer and not in the retinal parenchyma. This result is in accordance with that in a previous study that demonstrated that microbead-labeled, circulating, bone marrow–derived macrophages entered the NFL, but no other retinal layer. ⁶  

Moreover, these cells were almost exclusively BrdU⁻, in agreement with the reported feature of fully differentiated macrophages of the central nervous system (CNS) which do not proliferate after maturation. ^25,43,47 Thus, after ON lesion, a replenishment of circulating bone marrow–derived cells may occur, but this event seems to be rare in the first weeks after the ON lesion. Since we analyzed microglia cell proliferation only up to day 10, we speculate that a substantial fraction of parenchymal microglia were generated through in situ proliferation, similar to those in the brain. ⁴²  

Ciliary Macrophages under Physiological Conditions

In the naive CB, we found numerous F4/80⁺ cells, mainly in the ciliary stroma but also close to the epithelial layers. These cells are reported to be bone marrow–derived macrophages ^11,12,26 and represent 90% to 95% of the total immunologic bone marrow–derived population in the CB. ²⁶ Previous studies reported a physiological replenishment of ciliary cells from monocyte-derived cells. ¹⁴ This replenishment is faster than that observed in the neural retina and starts as early as 2 weeks after transplantation, owing to different microenvironments. ¹⁴ The half-life of macrophages in the uveal tract is 3 days, ⁴⁸ but no data are available on the proliferative behavior of ciliary immunologic cells. In this study, we showed that under physiological conditions in the vascular connective tissue of the CB, a continuous local but very low proliferative turnover takes place. Thus, ciliary cells were not exclusively replenished by circulating bone marrow–derived cells. ¹⁴  

Ciliary Macrophages after ON Lesion

After crush injury, ciliary cells responded with increased in situ cell proliferation within 1 to 2 days, indicating that immunologic cells far away from the lesion site were also activated and indeed proliferated. However, this event was a more moderate one compared with that in the retina, leading to a linear increase in the number of BrdU⁺ cells. Nevertheless, the number of cells obtained for both lesion types after 5 to 10 days showed that an ON lesion led to an increased number of BrdU⁺ immunologic ciliary cells. Of note, independent of the ON lesion type, similar cell morphologies and distribution patterns were observed, indicating that the same cell types responded after ON injury. The only differences were found in the absolute number of F4/80⁺ cells and the corresponding fraction of BrdU⁺F4/80⁺ cells. Thus, differential inductive mechanisms seemed to prevail, contrary to the observations in the retina. Since most of the BrdU⁺F4/80⁺ cells were observed in the stroma, the area where ciliary macrophages and most of the blood vessels are mainly located, two possible explanations for the increased number of cells are proposed:

1.  
   In response to the ON lesion, it appears that ciliary macrophages can actually proliferate. This conclusion is in contrast to the accepted hypothesis that mature macrophages of the brain or neural retina do not proliferate. ^25,43,47 However, since the CB is a part of the uveal tract, the difference in the microenvironment could explain the presence of different macrophage phenotypes. ^14,43 Ciliary macrophages may also represent a more immature phenotype that still can re-enter the cell cycle. ¹¹
2.  
   An alternative explanation would be that these cells may have incorporated BrdU during division in the bone marrow and entered the CB as already-labeled cells, because previous studies have demonstrated that more than 50% of circulating monocytes leave the circulation within approximately 22 hours and enter different tissues. ^25,49

Since only moderate local cell proliferation rates in the CB were found and most of the BrdU⁺F4/80⁺ cells were located close to ciliary vessels, we assume that both mechanisms contribute to an increased number of BrdU⁺ macrophages. 

Activation of ciliary immunologic cells has already been reported after neurotoxic lesions and retinal detachment. ¹⁶ As early as 1 day after injury, increased and more rapid infiltration of ciliary cells at the peripheral retina was observed. ¹⁶ However, in the present work, after induction of ON lesion, at all time points analyzed, the number of F4/80⁺ cells in the adjacent peripheral retina remained unchanged and low. Therefore, despite an increase in F4/80⁺ immunologic cells in the CB, we observed no increase in the adjacent retina that could be due to the migration of ciliary cells. In addition, the lack of overall changes in the number of microglia in the neural retina argues against a migratory event. Thus, we suggest that putative cellular migration may be delayed, consistent with the notion that repopulation of the retina by bone marrow–derived cells does not occur in the acute phase after the lesion. Further work is necessary to confirm these hypotheses and to gain further insight into the possible mechanisms involved. 

Nevertheless, there was a significant number of BrdU⁺F4/80⁻ cells that increased further after ON axotomy. Therefore, (an)other cell type(s) may also respond with cell cycle re-entry (e.g., fibroblasts, endothelial cells, ^26,32 or retinal progenitors). ^32,50,51 In a recent study, proliferating progenitor-like cells were identified in the CB after ON lesion. ³²  

Conclusions

The results of the present study showed that an ON lesion increases the number of BrdU⁺ immunologic cells in both ocular structures, the neural retina and the CB, and that these cells are indeed proliferating in situ. The majority (70%–85%) of the lesion-induced proliferating cells in the retina and CB had an immunologic identity. Of note, within the neural retina, a nearly identical cellular response (i.e., similar number of cells, morphology, and laminar distribution), was observed in both lesion types, suggesting similar mechanisms of activation. In the CB, similar cell distribution and morphologies were observed. However, the number of macrophages after ON axotomy further increased compared with that after ON crush. Therefore, ON axotomy showed a more pronounced effect on the proliferative response of ciliary macrophages. However, there was no obvious migration of ciliary cells into the adjacent neural retina up to 10 days after injury. Thus, the CB did not function as a pool of immunologic cells in the acute phase after the ON lesion. These completely new findings, in particular concerning BrdU⁺ immunologic ciliary cells, may be of great importance in the assessment and interpretation of the lesion-induced response of ciliary and retinal cells. 

Footnotes

 Supported by the BMBF (Bundesministerium für Bildung und Forschung) and the IZKF (Interdisciplinary Centre for Clinical Research) Jena, and by a UWH (University of Witten/Herdecke) scholarship (SW).

Footnotes

 Disclosure: S.G. Wohl, None; C.W. Schmeer, None; O.W. Witte, None; S. Isenmann, None

The authors thank Josephine Walter for providing antibody probes and Iwa Antonow for helpful suggestions and comments on the manuscript.