Expression pattern of Ccr2 and Cx3cr1 in inherited retinal degeneration

Though accumulating evidence suggests that microglia, resident macrophages in the retina, and bone marrow-derived macrophages can cause retinal inflammation which accelerates photoreceptor cell death, the details of how these cells are activated during retinal degeneration (RD) remain uncertain. Therefore, it is important to clarify which cells play a dominant role in fueling retinal inflammation. However, distinguishing between microglia and macrophages is difficult using conventional techniques such as cell markers (e.g., Iba-1). Recently, two mouse models for visualizing chemokine receptors were established, Cx3cr1GFP/GFP and Ccr2RFP/RFP mice. As Cx3cr1 is expressed in microglia and Ccr2 is reportedly expressed in activated macrophages, these mice have the potential to distinguish microglia and macrophages, yielding novel information about the activation of these inflammatory cells and their individual roles in retinal inflammation. In this study, c-mer proto-oncogene tyrosine kinase (Mertk)−/− mice, which show photoreceptor cell death due to defective retinal pigment epithelium phagocytosis, were employed as an animal model of RD. Mertk−/−Cx3cr1GFP/+Ccr2RFP/+ mice were established by breeding Mertk−/−, Cx3cr1GFP/GFP, and Ccr2RFP/RFP mice. The retinal morphology and pattern of inflammatory cell activation and invasion of Mertk−/−Cx3cr1GFP/+Ccr2RFP/+ mice were evaluated using retina and retinal pigment epithelium (RPE) flat mounts, retinal sections, and flow cytometry. Four-week-old Mertk−/−Cx3cr1GFP/+Ccr2RFP/+ mice showed Cx3cr1-GFP-positive microglia in the inner retina. Cx3cr1-GFP and Ccr2-RFP dual positive activated microglia were observed in the outer retina and subretinal space of 6- and 8-week-old animals. Ccr2-RFP single positive bone marrow-derived macrophages were observed to migrate into the retina of Mertk−/−Cx3cr1GFP/+Ccr2RFP/+ mice. These invading cells were still observed in the subretinal space in 18-week-old animals. Cx3cr1-GFP-positive microglia and Ccr2-RFP-positive macrophages were distinguishable in the retinas of Mertk−/−Cx3cr1GFP/+Ccr2RFP/+ mice. In addition, Ccr2 expression in Cx3cr1 positive microglia is a feature of microglial activation in RD. Mertk−/−Cx3cr1GFP/+Ccr2RFP/+ mice enabled observation of microglial activation over time during RD and may be useful for developing inflammation-targeted treatment strategies for RD in the future.


Background
Since photoreceptor cell death (PCD) is the proximal cause of blindness in retinal degenerative disorders such as retinitis pigmentosa and age-related macular degeneration, the mechanism of PCD should be clarified. Though traditionally PCD is thought to occur due to genetic predisposition, environmental risk factors, and old age [1], accumulating clinical and experimental evidence suggests that retinal inflammation can accelerate PCD [2][3][4][5]. Retinal inflammation is mediated by the retinal innate immune system, including microglia, which are resident macrophages in the central nervous system (CNS) and the complement system [1]. In the healthy retina, the retinal innate immune system plays a beneficial role in maintaining retinal homeostasis [1]. However, once a pro-inflammatory cascade is triggered, this retinal innate immune system can cause PCD [3]. The molecular details of the retinal inflammatory cascade, PCD, and the relationship between the two are not yet entirely understood. However, it is clear that not only microglia but also bone marrow-derived macrophages which invade to retina via a damaged blood-retina barrier play principal roles in this process [3].
We recently reported migration of microglia and macrophages into the subretinal space in during retinal degeneration (RD) [3]. As both of these cell types are stained by the Iba-1 Ab, it is unclear which is the dominant cell type in the subretinal space. To differentiate the contributions of microglia and macrophages in PCD, we administered the microglia suppressive drug "minocycline" or depleted macrophages systemically by injection of clodronate-liposomes in an RD animal model. However, interestingly, both of these treatments ameliorated PCD in light exposed Abca4 −/− Rdh8 −/− mice which show drastic RD [3,6], indicating that both microglia and macrophages play important roles in retinal inflammation and degeneration. Therefore, it is still uncertain which cell type initially triggers retinal inflammation and which plays a more dominant role in driving subsequent PCD.
In the current study, we employed two fluorescein protein knock in mouse models, namely Cx3cr1 GFP/GFP and Ccr2 RFP/RFP mice [7,8] to distinguish microglia and macrophage in the retina. Cx3cr1 is the sole receptor for Cx3cl1, also called fractalkine. Cx3cr1 is expressed by dendritic cells, natural killer cells, and macrophages [9]. Ccr2 is also the sole receptor for Ccl2. Ccr2 is required for macrophage infiltration to injure cites [10]. Furthermore, both Cx3cr1 and Ccr2 are upregulated in RD [3,11]. In a study of the brain, Cx3cr1 but not Ccr2 was expressed in microglia from embryonic development throughout adulthood [12]. However, whether this principle applies to retinal degeneration remains unknown. To shed light on microglia activation and to test whether microglia and macrophages are distinguishable in retinal degeneration, c-mer proto-oncogene tyrosine kinase (Mertk) −/− Cx3cr1 GFP/+ Ccr2 RFP/+ or Mertk +/+ Cx3cr1 GFP/+ Ccr2 RFP/+ mice were established by breeding Mertk −/− , Cx3cr1 GFP/GFP , and Ccr2 RFP/RFP mice. Mertk plays an essential role in retinal pigment epithelium (RPE) phagocytosis [13], and Mertk deficiency causes RD [14]. Furthermore, Mertk −/− mice show retinal inflammation associated with microglia and macrophage accumulation in the subretinal space [3,11]. The retinal morphology and expression pattern of Cx3cr1-GFP and Ccr2-RFP in Mertk −/− Cx3cr1 GFP/ + Ccr2 RFP/+ mice was examined by retinal sectioning, retina and RPE flat mounts, and flow cytometry.

Flat mount retina and RPE preparation
All procedures for retina and RPE flat mounts were described previously [3]. Images of flat mounts were captured by a confocal microscope (LSM, Carl Zeiss, Thornwood, NY, USA). For retina flat mount, the entire retina was captured at 5 μm intervals and all photographs were projected in one slice. For RPE flat mounts, the entire visible RPE was captured at 3 μm intervals and projected in one slice.

Histological analysis
All procedures to make sections for light microscopy were performed using a previously described method [15]. Rabbit anti-Iba-1 Ab (1:400, Wako, Osaka, Japan) was used for immunohistochemistry (IHC). Cell number was counted using ImageJ (National Institutes of Health, Bethesda, MD, USA). To observe microglia cell bodies, low melting point agarose-embedded (Sigma, St. Louis, MO, USA) thick sections (100 μm thickness) were prepared [16]. Images of IHC were captured on a confocal microscope (LSM, Carl Zeiss, Thornwood, NY, USA).

Flow cytometry
The neural retina was isolated from the eyecup with minimal inclusion of RPE cells, and the choroid plexus and ciliary body were carefully removed. The retina was then incubated in 0.25 % trypsin/PBS at 37°C for 15 min. After stopping the activity of trypsin by the addition of 10 % FBS/PBS with 0.1 % DNaseI (Invitrogen, Waitham, MA, USA), the cells were mechanically dissociated into a single-cell suspension by gentle pipetting. The dissociated cells were stained with 0.1 % propidium iodide (PI)/PBS, and 100,000 cells were analyzed by FACSCalibur (BD, Franklin Lakes, NJ, USA). Peripheral blood samples were collected from the orbital venous plexus. After the lysis of erythrocytes in an isotonic solution of ammonium chloride, white blood cells (WBC) were stained with PI and used for flow cytometric analysis. Data analysis was performed using FlowJo software.

Data analysis
Data represent the mean ± SD. At least three independent experiments were compared by the one-way analysis of variance test.

Results
First, the retinal phenotype of Ccr2 RFP/RFP and Cx3cr1 GFP/GFP mice was analyzed. No Ccr2-RFP-positive cells were detected in the retina of Ccr2 RFP/RFP mice, but Cx3cr1-GFP-positive microglia were ubiquitous in the inner retina (from the ganglion cell layer to the inner nuclear layer) of Cx3cr1 GFP/GFP mice. Ccr2-RFP-positive cells were found in Ccr2 RFP/RFP mice in the blood cell fraction corresponding to the monocyte population. No retinal degeneration was observed in either Ccr2 RFP/RFP or Cx3cr1 GFP/GFP mice (data not shown).
Retinal degeneration, microglia migration, and increase of Cx3cr1 and Ccr2 in Mertk −/− mice Next, the retinal phenotype of Mertk −/− and WT (Mertk +/+ ; littermate control of Mertk −/− ) mice was evaluated. Eight-week-old Mertk −/− mice showed a decrease in photoreceptor nuclei number in the outer nuclear layer (ONL) and disorganized inner segments (IS) and outer segments (OS) (Fig. 1a). Iba-1-positive cells were found to migrate to the IS, OS, and subretinal space in 8-week-old Mertk −/− mice (Fig. 1b) indicating microglia/macrophage migration [3]. Because Iba-1 is expressed in both microglia and macrophages, these cells were not distinguishable. qPCR was used to compare Cx3cr1 and Ccr2 mRNA levels in the retinas of 3and 8-week-old Mertk −/− and WT (Fig 1c). No significant differences in Cx3cr1 or Ccr2 expression were observed between 3-and 8-week-old WT mice. In contrast, for Mertk −/− mice, these mRNAs were increased in 8-week-old mice as compared to 3-week-old animals ( Fig. 1c, left). At both the 3-and 8-week time points, expression of Cx3cr1 was increased in Mertk −/− mice compared to WT mice, and Ccr2 levels were increased in 8-week-old Mertk −/− mice as compared to WT controls (Fig. 1c, right).

Migration of Cx3cr1-GFP-expressing microglia and invasion of Ccr2-RFP-positive monocyte-derived macrophage in degenerating retinas
To test the expression pattern of Ccr2 and Cx3cr1 in the degenerating retina and to determine whether microglia The insets are magnified images of the area within the broken rectangle. GCL ganglion cell layer, INL inner nuclear layer, ONL outer nuclear layer, IS inner segments, OS outer segments, RPE retinal pigment epithelium. b IHC was performed using rabbit anti-Iba-1-Ab. Eight-week-old Mertk −/− mice showed Iba-1-positive microglia/macrophages in the outer retina, though no Iba-1-positive cells were observed in WT mice. c Cx3cr1 and Ccr2 mRNA levels in the retina of Mertk −/− and WT were measured by qPCR. RNA samples were collected from 16 retinas at each time point. qPCR was performed 3-6 times (n = 3-6). Expression levels were compared between 3-and 8-week-old animals and between WT and Mertk −/− mice at each age. Error bars indicate the SD of the mean (n > 3). Asterisk indicates P < 0.05 vs 3-week-old Mertk −/− mice and macrophages are distinguishable in RD using Ccr2 RFP/ RFP and Cx3cr1 GFP/GFP mice (as previously reported in the brain [12]), Mertk −/− Cx3cr1 GFP/+ Ccr2 RFP/+ mice were established. Retinal sections of 4-, 6-, and 8-week-old Mertk −/− Cx3cr1 GFP/+ Ccr2 RFP/+ mice are shown in Fig. 2.

Discussion
Mutations in the MERTK gene cause retinal dystrophies in humans and in animal models [18]. MERTK belongs to a family of receptor tyrosine kinases that includes AXL and TYRO3 and plays an indispensable role in the clearance of photoreceptor debris by RPE phagocytosis [19]. Accumulation of photoreceptor debris in the Fig. 3 Retina and RPE flat mounts from Mertk −/− Cx3cr1 GFP/+ Ccr2 RFP/+ mice. Retina and RPE flat mounts from Mertk −/− Cx3cr1 GFP/+ Ccr2 RFP/+ mice were prepared to analyze Cx3cr1-GFP and Ccr2-RFP-positive cells in more detail. a Retina flat mounts from 4-(upper panels) and 6-week-old (middle panels) Mertk −/− Cx3cr1 GFP/+ Ccr2 RFP/+ mice are shown. Retina flat mounts from 6-week-old Cx3cr1 GFP/+ Ccr2 RFP/+ mice (lower panels) are shown as negative controls. Cx3cr1-GFP is shown in green, and Ccr2-RFP is shown in red. b RPE flat mounts from 4-(upper panels) and 6-week-old (middle panels) Mertk −/− Cx3cr1 GFP/+ Ccr2 RFP/+ mice are shown. RPE flat mounts from 6-week-old Cx3cr1 GFP/+ Ccr2 RFP/+ mice (lower panels) are shown as negative controls. Cx3cr1-GFP is shown in green, and Ccr2-RFP is shown in red subretinal space due to RPE phagocytosis deficiency is closely associated with the photoreceptor cell death seen in Royal College of Surgeons (RCS) rats (with disabled Mertk) and in Mertk −/− mice [14]. Recently, we reported that Mertk −/− mice show migration of microglia and retinal inflammation, which exacerbated retinal degeneration [3,11]. Chemokines and cytokines including Ccl2, Ccl3, Ccl12, and Il1b are increased in the degenerating retinas of Mertk −/− mice [3,11]. Furthermore, blockade of Ccl2 and Ccl3 can attenuate the retinal phenotype of Mertk −/− mice, clearly indicating that retinal inflammation contributes to PCD and RD [11].
To visualize inflammatory cells in a tissue, staining or immunohistochemistry procedures are required. However, tissue staining is limited for detecting inflammatory cells and immunohistochemistry requires clean and selective antibodies. Cx3cr1 and Ccr2, especially Ccr2, are difficult to detect by immunohistochemistry due to the lack of an appropriate antibody [7]. Newly developed fluorescent protein knock-in mouse models, including Cx3cr1 GFP/GFP and Ccr2 RFP/RFP mice have the potential to overcome the limitations of tissue staining and immunohistochemistry, and these mice will be instrumental for developing new treatment strategies, especially neuroinflammation-targeted therapy. From our current data, Cx3cr1-GFP and Ccr2-RFP dual positive cells were visualized not only by histology (Figs. 2 and 3) but also flow cytometry (Fig. 6).
To date, several studies used a combination of Cx3cr1 GFP/GFP and Ccr2 RFP/RFP mice in experimentally induced disease models such as experimental autoimmune encephalomyelitis (EAE), an animal model of multiple sclerosis [7,12]. However, these mouse models have yet to be employed in naturally occurring neurodegenerative disease models such as retinal degeneration or Alzheimer's disease. In experimentally induced disease models such as EAE, inflamed monocyte-derived macrophages play a dominant role in neuroinflammation and degeneration; hence, disease onset occurs outside the CNS. In contrast, in naturally occurring neurodegenerative diseases including retinal degeneration, microglia presumably play the dominant role at the onset of neuroinflammation and degeneration because the blood-retina barrier or blood-brain barrier is maintained at early disease stages [3,11,20]. This study provides evidence that microglia is the dominant inflammatory cell during the early stages of retinal degeneration (Figs. 2, 3, and 4). The precise mechanisms underlying microglial activation and morphological alteration during RD still remain unclear. However, evidence suggests that exposure to dead photoreceptor debris and subsequent phagocytosis is an important trigger for microglial activation in RD, as administration of photoreceptor OS proteins induced increased cytokine and chemokine production in microglia in vitro [3]. Currently, it remains unclear how Ccr2-RFP-positive macrophages infiltrate the retina. However, they likely invade via either the inner or outer bloodretina barrier. The inner blood-retina barrier is composed of tight junctions between neighboring capillary endothelial cells which rest on a basal lamina covered by the foot processes of astrocytes and Müller glia and tight junctions between RPE cells comprise the outer blood-retina barrier [21]. We previously reported disruption of the inner blood-retina barrier during RD [3], and the tight junctions between RPE cells are also reportedly damaged during this process [11,22].

Conclusions
Newly developed Mertk −/− Cx3cr1 GFP/+ Ccr2 RFP/+ mice were used to monitor the migration of Cx3cr1-GFPpositive microglia from the inner retina to the outer retina and subretinal space. Round-shaped Ccr2-RFPpositive monocyte-derived macrophages also invaded the retina. Activated microglia and macrophages that had migrated to the OS layer and subretinal space were Cx3cr1-GFP and Ccr2-RFP dual positive. Initiation of CCR2 expression in CX3CR1-positive microglia is a feature of microglial activation in RD. Currently, microglia suppressive approaches are being evaluated as new treatment strategies for retinal diseases including AMD, RP, and diabetic macular edema [3,24,25]. Mertk −/− Cx3cr1 GFP/+ Ccr2 RFP/+ mice will be a suitable model to assist in the development of future microgliatargeted treatment strategies.