Expression of CXCL8, CXCR1, and CXCR2 in Neurons and Glial Cells of the Human and Rabbit Retina

Abstract

purpose. Several eye diseases are accompanied by inflammatory processes. The authors examined the expression of the proinflammatory chemokine CXCL8 and the corresponding receptors in healthy human retinas, in cellular membranes from patients with proliferative vitreoretinopathy (PVR) or human glial cell cultures and in an animal model of PVR in rabbit eyes.

methods. The authors used immunohistochemical methods, Western blotting, RT-PCR, and real time RT-PCR to characterize the expression of CXCL8, CXCR1, and CXCR2 in human and rabbit retinas. Functionality of the receptors in cultured glial cells was tested by Ca²⁺ imaging.

results. Immunohistochemical examinations of normal human and rabbit retinas revealed a distinct expression of CXCR1 and CXCR2 in several neuronal cell types. CXCL8 mRNA was demonstrated only by RT-PCR in normal retinas, and receptor expression was confirmed by Western blotting and RT-PCR. The presence of CXCR1 and CXCR2, but not CXCL8, was detected by immunostaining in glial fibrillary acidic protein–positive glial cells of cellular PVR membranes. Immunoreactivity for CXCL8, CXCR1, and CXCR2 was observed in virtually all cultured glial cells and in the human Müller cell line MIO-M1. Müller cells responded to the application of CXCL8 with increased cytosolic Ca²⁺ concentrations. In PVR rabbit retinas, CXCR1 expression is increased in Müller cells, and CXCL8 and CXCR2 are strongly expressed in microglial cells.

conclusions. Expression of CXCL8 and CXCL8 receptors in glial cells of human PVR membranes and rabbit PVR retinas suggests an involvement in glial reactivity. Furthermore, the prominent expression of CXCR1 and CXCR2 in neurons of the healthy human and rabbit retina suggests additional physiological functions.

Interleukin-8 (IL-8, CXCL8), a proinflammatory chemokine, belongs to a family of structurally related peptides of 8 to 10 kDa and was originally identified as a neutrophil-activating protein. ¹ Many different cell types, such as macrophages, endothelial cells, epithelial cells, and fibroblasts, are able to secrete CXCL8 in response to inflammatory stimuli. ^{1 2} Additionally, many in vitro studies demonstrate CXCL8 expression in mostly stimulated glial cells from the human central nervous system (CNS), that is, in microglial cells ³ and astrocytes. ^{4 5} CXCL8 expression was also detected in astrocytomas, glioblastomas, and cell lines derived from malignant astrocytes. ^{5 6 7} However, little is known about the in vivo expression of CXCL8 in microglial and macroglial cells in human CNS under normal and pathologic conditions. ⁶ Several functions of CXCL8 were suggested for various cell types in different tissues. CXCL8 was shown to be involved in neutrophil recruitment, cell adhesion, homing of neutrophils and lymphocytes, tumor growth, angiogenesis, neuronal protection, and brain development. ^{8 9 10 11} Nevertheless, the main function of CXCL8 is the recruitment of neutrophils to inflammatory sites in response to injury or infection. CXCL8 induces three main responses in neutrophils: reshaping and directional migration, exocytosis of stored proteins, and respiratory burst, which is characteristic of stimulated phagocytes. ¹  

The action of CXCL8 in neutrophils is mediated by the activation of two distinct high-affinity receptors, CXCR1 and CXCR2. ¹² CXCR1 exclusively binds CXCL8 and GCP2 (CXCL6) with high affinity, whereas CXCR2 can be activated by additional ligands such as Gro-α, Gro-β, Gro-γ (CXCL1-CXCL3), ENA-78, GCP2, and NAP-2 (CXCL5-CXCL7) with an affinity similar to that of CXCL8. CXCR1/CXCR2 receptors belong to a subfamily of G-protein–coupled receptors. ^{1 2} Activation of the two CXCL8 receptors triggers the formation of distinct second messengers, which activate several signaling pathways through protein kinases and phospholipases. ² The main signal transduction pathways induce a transient mobilization of free Ca²⁺ from internal stores. 

In addition to their expression by neutrophils and monocytes/macrophages, CXCR1 and CXCR2 have been detected in stimulated basophils, T lymphocytes, and dendritic cells. ¹² CXCR2, but not CXCR1, has been identified on subsets of neurons in the human central nervous system by immunohistochemical methods. ¹³ Rat septal cholinergic neurons produce CXCR1 and CXCR2 mRNA, ¹⁴ and a recent study demonstrated that the CXCR1 and CXCR2 mRNA transcripts are widely expressed in glutamatergic, GABAergic, and cholinergic neurons in the rat brain. ¹⁵ CXCR2 expression was detected in cultured human microglial cells and astrocytes ^{16 17 18} and in microglial cells and astrocytes of the fetal and the adult human brain, respectively. ^{19 20} To our knowledge, only one group reported CXCR1 expression in cultured human microglial cells and astrocytes, ¹⁷ but there is no evidence of glial or neuronal expression of CXCR1 in the healthy or pathologic human brain thus far. The expression of CXC receptors in different cell types of the brain suggests their involvement in normal brain physiology and brain development and in various neurologic disorders, neuroinflammatory processes, and neuroimmune diseases. ^{10 11 14 21}  

In the neural retina, the Müller cells are the principal glial cells, spanning the entire thickness of the tissue and contacting every type of neuronal cell. Retinal astrocytes are located only in the nerve fiber layer, where they surround blood vessels and nerve fibers. Virtually every pathologic alteration in the retina is accompanied by reactive gliosis, that is, by distinct changes of the Müller cell properties. ²² For instance, during retinal detachment (RD) and proliferative vitreoretinopathy (PVR), major causes of severe visual loss, Müller cells become hypertrophic and express an intermediate filament, glial fibrillary acidic protein (GFAP), as an indicator of reactive gliosis. ^{23 24} Additionally, in patients with PVR, Müller cells transdifferentiate, migrate, proliferate, and form (together with other cell types such as astrocytes, retinal pigment epithelial cells, and leukocytes) deleterious cellular membranes. ^{25 26 27} Many retinal diseases are accompanied by inflammatory processes; in particular, increased levels of CXCL8 are detected in the vitreous of patients with PVR or proliferative diabetic retinopathy. ^{28 29 30} The exact intraocular cellular sources of the chemoattractants to recruit neutrophils, monocytes, and lymphocytes are unknown. 

The aim of our study was to evaluate the expression of CXCL8 and its high-affinity receptors in healthy human retinal tissue and PVR membranes. Additionally, we used a well-established animal model of PVR in rabbit eyes. ³¹ Contrary to the retinas of mice and rats, the rabbit retina seems to be particularly suitable for examining CXCL8 and CXCL8 receptor expression because of the high amino acid sequence similarities compared with human proteins (identities of rabbit compared with human amino acid sequences: CXCL8, 70%; CXCR1, 83%; CXCR2, 80%). 

Herein we show, for the first time, a prominent expression of CXCR1 in neurons of the human CNS, especially the expression of CXCR1/CXCR2 in several neuronal subtypes of the healthy human retina. Furthermore, we demonstrate the expression of CXCR1/CXCR2 in glial cells of surgically removed cellular membranes from patients with PVR. We detected that Müller glial cells in primary cultures from human retinas produce CXCL8 and express functional CXCR1 and CXCR2 receptors, indicating that they may participate in the development of intraocular inflammatory processes. Finally, we show neuronal expression of CXCR1/CXCR2 in healthy rabbit retinas similar to that observed in human retinas and demonstrate the upregulation of CXCR1 in Müller glial cells and CXCL8/CXCR2 in microglial cells in PVR retinas, respectively. Noteworthy, the prominent expression of CXCR1 and CXCR2 receptors in normal retinal neurons suggests additional functions of these receptors beyond their involvement in the inflammatory response. 

Materials and Methods

Primary Cultures from Human Retinas and Cultures of MIO-M1 Cells

All tissue was used in accordance with applicable laws and with the Declaration of Helsinki for research involving human tissue and was approved by the ethics committee of the Faculty of Medicine, University of Leipzig. Human retinas were obtained from seven organ donors (four females, three males) without known eye disease. Patient ages ranged from 28 to 75 years; donors died of gastrointestinal cancer or myocardial infarction. The postmortem time for tissue preparation varied between 12 and 24 hours. To prepare primary retinal cultures, retinal pieces were incubated in subtilisin A (0.1 mg/mL; ICN, Eschwege, Germany) containing Ca²⁺- and Mg²⁺-free phosphate-buffered saline (PBS) for 30 minutes at 37°C, followed by several washing steps with sterile PBS. After a short incubation in PBS supplemented with DNase I (200 U/mL), the tissue pieces were triturated by a wide-pore pipette to obtain suspensions of isolated cells. The cells were washed with sterile minimal essential medium (MEM) and were seeded on uncoated coverslips in Petri dishes and cultured at 37°C (5% CO₂ in air) for 2 weeks in MEM containing 10% fetal calf serum. 

The spontaneously immortalized human Müller cell line (MIO-M1) was a kind gift from Gloria A. Limb (Moorfields Eye Hospital, London, UK) and was characterized recently. ³² Cultures at passages 105 to 120 were used for the experiments. The cells were cultured in MEM containing L-GlutaMAX I (Gibco BRL, Invitrogen, Karlsruhe, Germany) at 37°C (5% CO₂ in air) for 7 to 10 days. Unfixed cultures were used for the calcium imaging experiments. In another series of experiments, the cultured cells were harvested for Western blotting and PCR analysis. All chemicals not indicated otherwise were purchased from Sigma-Aldrich (Taufkirchen, Germany). 

Surgically Induced PVR in Rabbit Eyes

All experiments were conducted in accordance with applicable German laws, with the European Communities Council Directive 86/609/EEC, and with the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. Ten adult pigmented rabbits (2.5–3.5 kg; both sexes) were used. The animals were held under 12-hour light/12-hour dark room conditions and had free access to food and water. According to a method described previously, ³¹ PVR was induced in one eye of each animal while the other eye remained untreated and served as the control. Anesthesia was induced by intramuscular ketamine (50 mg/kg) and xylazine (3 mg/kg; BayerVital, Leverkusen, Germany). Pupils were dilated with topical tropicamide (1%; Ursapharm, Saarbrücken, Germany) and phenylephrine (5%; Ankerpharm, Rudolstadt, Germany). After pars plana sclerotomy, circumscript vitrectomy was performed in the area of the future retinal detachment (in the ventro-nasal quadrant, below the medullary rays). A thin glass micropipette attached to a 250-μL Hamilton glass syringe was used to create a small local retinal detachment by injecting phosphate-buffered saline (pH 7.4) into the subretinal space. Another micropipette placed in the vitreous near the surface of the detached retina was used to inject 100 μL saline containing the proteolytic enzyme dispase I (0.5 U; Boehringer, Mannheim, Germany). Dispase I treatment (breakdown of blood-retinal barrier and extracellular matrix components) induced the development of moderate or massive PVR with broad retinal detachment and epiretinal membranes, as previously described in detail. ³¹ After injections, the sclerotomies and the conjunctiva were closed. Two weeks after surgery, the animals were anesthetized as described and were killed by intravenous T61 (3 mL; embutramide 0.2 g/mL, mebezonium iodide 0.05 g/mL, tetracaine hydrochloride 5 mg/mL; Hoechst, Unterschleissheim, Germany), and the eyes were removed. 

Immunohistochemistry

Surgically removed PVR membranes of six patients were obtained from the Eye Clinic of the University of Leipzig and were fixed in acetone on a glass slide. Isolated retinas from human donor and rabbit eyes were fixed in 4% paraformaldehyde (PFA) for 2 hours. After washing in PBS, the tissues were embedded in saline containing 3% agarose (wt/vol), and 90-μm–thick slices were cut with a vibratome. Retinal slices were incubated with PBS containing 10% normal goat serum (Dianova, Hamburg, Germany), 1% dimethyl-sulfoxide (DMSO), and 0.3% Triton X-100 for 2 hours. Subsequently, the slices were incubated with the primary antibodies in PBS/normal goat serum solution for 15 hours at 4°C. After washing in PBS, the secondary antibodies were applied for 2 hours at room temperature. Slices were washed several times with PBS, placed on glass slides, and covered with mounting medium (Fluoromount-G; Southern Biotechnology Associates, Birmingham, AL) and coverslips. 

Cultures were fixed with 4% PFA for 20 minutes and subsequently washed with PBS containing 1% DMSO and 0.3% Triton X-100. After blocking the unspecific binding sites with 10% normal goat serum, cultures were incubated with the primary antibody at 4°C for 12 hours. After washing in PBS, the secondary antibodies were applied for 2 hours at room temperature, and cells were counterstained with Hoechst 33258 (1:1000; Molecular Probes, Eugene, OR) for 30 minutes at room temperature to visualize cell nuclei. Cultures on the coverslips were washed several times with PBS, covered with mounting medium (Fluoromount-G; Southern Biotechnology Associates), and placed on glass slides. Immunolabeling was visualized by means of a confocal laser scanning microscope (LSM 510; Zeiss, Oberkochen, Germany). 

The following primary antibodies were used for human and rabbit tissue (dilutions 1:200–1000): mouse antibodies directed against CXCL8, CXCR1, CXCR2 (R&D Systems, Wiesbaden, Germany) and S100β (Sigma-Aldrich) and polyclonal rabbit antibodies directed against cellular retinaldehyde binding protein (CRALBP; a kind gift of John C. Saari, Seattle, WA), vimentin (Biomeda Corp., Foster City, CA), GFAP (DAKO, Glostrup, Denmark), and S 100 (Sigma-Aldrich). Secondary antibodies were carbocyanine (Cy3)-coupled goat anti-mouse IgG (1:750; Dianova) and Cy2-coupled goat anti-rabbit IgG (1:750–1000; Dianova). Cy3-tagged Griffonia simplicifolia agglutinin (GSA; isolectin B4) was used to label retinal microglial cells and blood-derived macrophages. 

Western Blotting

Retinal pieces were washed several times with cold PBS and solubilized in 300 μL lysis buffer (Tris 50 mM, pH 7.4, EDTA 2.5 mM) in the presence of protease inhibitors. Homogenates were centrifuged at 100,000g for 45 minutes at 4°C (Sorvall Ultraspeed Centrifuge; DuPont de Nemours, Wilmington, DE), and the resultant pellets were rehomogenized in 300 μL RIPA buffer. After incubation on a shaker for 60 minutes at 4°C, the homogenates were centrifuged at 140,000g for 45 minutes at 4°C, and supernatants were used for the determination of the protein concentration with the Bradford method. For preparation of lysates from Müller cell cultures (MIO-M1), the cells were washed twice with cold PBS, and the monolayer was scraped into 500 μL lysis buffer and 0.5% SDS (Mammalian Cell Lysis-1 Kit; Sigma-Aldrich). Total cell lysates were centrifuged at 10,000g for 10 minutes, and the supernatants were analyzed by immunoblotting. Equal amounts of protein (15 μg) were separated by 12% SDS-polyacrylamide gel electrophoresis. Immunoblots were probed with primary and secondary antibodies, and immunoreactive bands were visualized with diaminobenzidine (peroxidase substrate kit; Vector Laboratories, Burlingame, CA) or 5-bromo-4-chloro-3-indolyl phosphate/nitro blue tetrazolium (Sigma-Aldrich). 

PCR Analysis of Human Tissue

Total RNA from human donor retinas, whole blood, or MIO-M1 cells was extracted using a reagent (Trizol; Invitrogen) and treated with DNaseI (RNase-free; Stratagene, Amsterdam, Netherlands) to prevent contamination by genomic DNA. cDNA was synthesized from 5 μg total RNA by priming with oligo-dT using an RT-PCR system (Thermoscript; Invitrogen). The same approach without reverse transcriptase was performed as a negative control to visualize possible genomic DNA contamination. Primer pairs for the CXCL8 proof were generated based on the known human CXCL8 sequence (gi:28610153): forward, 5′-AGGGTTGCCAGATGCAATAC-3′; reverse, 5′-AAACCAAGGCACAGTGGAAC-3′. The expected size of the PCR product for CXCL8 was 420 bp. One microliter of the synthesized cDNA was amplified during 34 cycles of the PCR reaction in a thermocycler (GeneAmp 2400; PerkinElmer, Boston, MA) using Platinum Taq-Polymerase (Invitrogen). Each cycle of PCR consisted of 30 seconds at 94°C, 30 seconds at 56°C, and 1 minute at 68°C. To prove the expression of CXCR1 and CXCR2, primer pairs were generated based on the known human receptor sequences: CXCR1 (gi:559049), forward 5′-TTTGTTTGTCTTGGCTGC-3′ and reverse 5′-CCAAGAACTCCTTGCTGAC-3′; CXCR2 (gi:559053), forward 5′-ACATGGGCAACAATACAGCA-3′ and reverse 5′- CCTCCTCTGCTTCCTGTGAC-3′. The PCR cycle for these receptors consisted of 30 seconds at 94°C, 30 seconds at 58°C, and 1 minute at 68°C. The cDNA was amplified during 38 cycles, and the expected sizes of the PCR products were 532 bp (CXCR1) and 622 bp (CXCR2), respectively. Amplified samples were analyzed by agarose gel electrophoresis. 

PCR Analysis of Rabbit Tissue

Total RNA was extracted from control and treated rabbit retinas using a specific reagent (Trizol; Invitrogen), was then purified (RNeasy Mini Kit; Qiagen, Hilden, Germany) and treated with DNase I (Roche, Mannheim, Germany). cDNA was synthesized with 1 μg total RNA (RevertAid H Minus First-Strand cDNA Synthesis Kit; Fermentas, St. Leon-Roth, Germany), and semiquantitative real-time RT-PCR was performed (MyIQ-Single-Color Real-Time PCR Detection System; Bio-Rad, Munich, Germany). Primer pairs were selected according to the published rabbit cDNA sequences and homologue sequences of other mammals. The following primer pairs were used: glyceraldehyde 3-phosphate dehydrogenase (GAPDH; gi:126723532), forward 5′-AGGTCATCCACGACCACTTC-3′, reverse 5′-GTGAGTTTCCCGTTCAGCTC-3′; CXCL8 (gi:126723707), forward 5′-GAACTCCAAGCTGGCTGTGG-3′ and reverse 5′-TATGACTCTTGCTGCTCAGC-3′; CXCR1 (gi:165440), forward 5′-TTGTCCCTGCCCTTCTTC-3′ and reverse 5′-GAGGTGACCCGATGTCGT-3′; CXCR2 (gi:437661) forward 5′-ATGGGGAACAGCACTGCGAA-3′ and reverse 5′-TGAGCAGGCCATAGCGAAAC-3′. The PCR solution contained 1 μL cDNA, specific primer set (1 μM each), and 10 μL reagent (QuantiTect SYBR Green PCR Kit; (Qiagen) in a final volume of 20 μL. PCR parameters were initial denaturation and enzyme activation (one cycle at 95°C for 15 minutes); denaturation, annealing, amplification, and quantification, 45 cycles at 95°C for 30 seconds, 60°C for 30 seconds, and 72°C for 60 seconds; melting curve, 55°C with the temperature gradually increased (0.5°C) up to 95°C. mRNA expression was normalized to the levels of GAPDH mRNA, and the changes were calculated as described. ³³ After 45 cycles, the RT-PCR products were analyzed by agarose gel electrophoresis. 

Calcium Imaging

For fluorescence measurements, the MIO-M1 cells were allowed to rest in extracellular solution for 2 hours at 37°C. Then the cells were loaded with calcium-sensitive dye (Fura-2/AM; 10 μM; Molecular Probes, Eugene, OR) for 30 minutes at 37°C. Measurements were taken at room temperature with the use of a bath solution that contained 129 mM NaCl, 3 mM KCl, 1 mM CaCl₂, 0.2 mM MgCl₂, 20 mM glucose, and 10 mM HEPES (pH 7.4 adjusted with NaOH). A fluorescence measurement system (Fucal 5.12B; Till-Photonics, Munich, Germany) was used. Fluorescence was excited at 340 nm (F ₃₄₀) and at 380 nm (F ₃₈₀), and the ratio F = F ₃₄₀/F ₃₈₀ was calculated; images were recorded every 6 seconds. Cultures were continuously perfused for at least 20 minutes before application of the test substance. Recombinant human CXCL8 (300 ng/mL; R&D Systems) was administered by rapid switching of the perfusate. 

Results

CXCL8, CXCR1, and CXCR2 in the Normal Human Retina

We investigated normal retinal tissue from seven organ donors without known eye disease. Immunohistochemical examination revealed the expression of CXCR1 and CXCR2 proteins in several neuronal cell types (Figs. 1 and 2) . These neuronal cell types were unequivocally identified by their morphology and location in the well-structured retinal tissue. CXCR1 is expressed throughout the entire retina, that is, in ganglion cell somata and axons in the nerve fiber layer (NFL), in synaptic contacts in the inner plexiform layer (IPL) and outer plexiform layer (OPL), in various neuronal cell types of the inner nuclear layer (INL), and in the somata of photoreceptor cells in the outer nuclear layer (ONL). Cones were intensely labeled from their inner segments to the axon terminals (cone pedicles) in the OPL. Rod photoreceptor inner segments and somata were also immunoreactive for CXCR1. Immunostaining for CXCR2 revealed a slightly different expression pattern. Some ganglion cell somata were faintly labeled, but no labeled axons were found in the NFL. Amacrine cells located at the inner border of the INL and displaced amacrine cells in the IPL and ganglion cell layer (GCL) were intensely labeled. Neurons in the INL and synaptic contacts in the OPL were immunoreactive for CXCR2, which was also detected in the inner segments and somata of rods and cones. A horizontal band of CXCR2 immunoreactivity was observed at the outer limiting membrane. Western blot and RT-PCR confirmed the expression of both receptor subtypes in retinal tissue (Figs. 3A 3C 3D) . However, we detected CXCR1 mRNA in only five of seven donor retinas, but CXCR2 mRNA was detected in all examined specimens. RT-PCR analysis revealed identical PCR products for retinal tissue, blood-derived leukocytes, and the human Müller cell line MIO-M1. Additionally, CXCL8 expression was demonstrated by RT-PCR in four of seven examined donor retinas (one example is shown in Fig. 3B ). We were unable to confirm CXCL8 expression in specific cell types by means of immunocytochemistry (Fig. 1A) . 

In healthy human retinas, only astrocytes in the nerve fiber layer, but not Müller glial cells, are immunoreactive for GFAP. Under various pathologic conditions of the retina, Müller cells become GFAP positive as an indication of reactive gliosis. ²² In the examined donor retinas, only a few Müller cells were found to be GFAP positive (one example is indicated in Fig. 1B ); therefore, the retinas were considered healthy. We did not observe explicit double labeling of CXCR1 or CXCR2 with GFAP-positive cells. Surprisingly, CXCR1 and CXCR2 seem to be expressed exclusively in neurons of the normal healthy human retina. This raised the question whether activated glial cells might express CXCL8, or the corresponding receptor subtypes, during pathologic/inflammatory changes in the eye. We chose to investigate surgically removed cellular membranes from patients with proliferative vitreoretinopathy (PVR). 

CXCL8, CXCR1, and CXCR2 in Surgically Removed PVR Membranes

PVR is a common complication of retinal detachment that is associated with a proliferation of glial cells and the formation of periretinal cellular membranes. Surgical removal of these membranes is a common method to reduce deleterious tractional forces on the retina. We examined such removed PVR membranes of six patients by means of immunohistochemistry. Double labeling of slices revealed that GFAP-positive cells were also immunolabeled with antibodies against human CXCR1 or CXCR2 (Figs. 4D 4E 4F 4G 4H 4I) . However, GFAP-positive cells without labeling for one of the two CXCR receptors were also found. Some cells that were immunoreactive for CXCR1 or CXCR2 were not labeled with the anti-GFAP antibody. These results are not surprising because the PVR membranes consist of various cell types, including myofibroblasts, glial cells, retinal pigment epithelial cells, neutrophils, macrophages, microglial cells, and endothelial cells. We were unable to detect CXCL8-immunoreactive cells in the examined PVR membranes (Figs. 4A 4B 4C) . Furthermore, we examined parts of degenerated retina concomitantly removed with the PVR membranes through retinectomy (Figs. 4J 4K 4L 4M 4N 4O) . We observed massive GFAP immunoreactivity of Müller glial cells as a sign of reactive gliosis (Figs. 4K 4N) . Double-labeling experiments revealed that gliotic Müller cells in pathologically changed human retinas seemed to express CXCR1 but not CXCR2. Our observations in PVR membranes and tissue after retinectomy suggest that the expression of CXCL8 receptors might be a result of glial activation, dedifferentiation, or proliferation, characteristic of such cellular membranes and gliosis. 

Primary Cell Cultures from Human Retinas

To prove this assumption, we established primary cultures of human retinal glial cells because cultured cells dedifferentiate and proliferate and, therefore, may reflect characteristics of reactive glial cells during PVR. Primary cultures were prepared from retinas of healthy organ donors and were characterized by means of immunocytochemical markers (Figs. 5A 5B 5C 5D 5E 5F) . Nearly all the cultured cells were immunoreactive for vimentin (a mesenchymal marker but also characteristic for macroglial cells; not shown) and more than 98% of the cells were positive for the Ca²⁺-binding protein S100 and the S100β subunit (a glial marker in nervous tissue; Fig. 5 ). Immunoreactivity for GFAP (a specific glial marker) was observed in more than 96% of all cultured cells. Nearly all the S100β- and GFAP-positive cells were also immunoreactive for CRALBP (expressed by Müller cells and retinal pigment epithelium cells but not by astrocytes; Figs. 5A 5B 5C ). Because the combination of all the markers investigated is characteristic of Müller glial cells, the primary cultures were considered highly enriched Müller glial cell cultures. 

Double-labeling experiments revealed that all GFAP-positive cells were also immunolabeled with antibodies against human CXCL8, CXCR1, and CXCR2 (Figs. 5D 5E 5F) , suggesting the expression of CXCL8, and of both CXCL8-specific receptors, in cultured human retinal macroglial cells (primarily Müller cells). In the further course of experiments, we examined the immortalized Müller cell line MIO-M1. 

CXCL8, CXCR1, and CXCR2 in Müller Cell Line MIO-M1

The MIO-M1 cell line is a well-characterized human Müller cell line. ³² Cultures of MIO-M1 cells were intensively stained with antibodies against CXCL8, CXCR1, and CXCR2 (Figs. 5G 5H 5I) . The expression of CXCR1 and CXCR2 in MIO-M1 cells was confirmed by Western blotting and RT-PCR analysis (Figs. 6A 6B) . Additionally, CXCL8 expression in this cell type was confirmed by RT-PCR (Fig. 6B) . 

Ca²⁺-imaging experiments were carried out to test the functional expression of the CXCL8 receptors (Fig. 6C) . After the application of recombinant human CXCL8, 33% of the cultured Müller cells (18 of 54 cells in three different cultures) responded with an increase in the intracellular Ca²⁺ concentration. In a subpopulation of cells, we observed a sustained increase in the cytosolic Ca²⁺ concentration during CXCL8 application. Other responding cells displayed more oscillating intracellular Ca²⁺ alterations. Results indicated that Müller cells may express functional CXCL8 receptors. 

CXCL8, CXCR1, and CXCR2 in the Normal Rabbit Retina

We did not detect CXCL8 immunoreactivity in the healthy rabbit retina (Fig. 7A) , similar to the situation in healthy human retinas. Nonactivated microglial cells were faintly stained with GSA and were exclusively located in the NFL and IPL (Fig. 7B) . CXCR2 immunoreactivity was observed in ganglion cell somata, in a subpopulation of neurons in the INL and in the OPL (Fig. 7C) . Photoreceptor inner segments (rods and cones) and cone outer segments were intensely stained with anti-CXCR2 antibodies. With the exception of cone outer segment labeling in rabbit retinas, the CXCR2 immunoreactivity pattern was similar in human and rabbit retinas. CXCR1 immunoreactivity was prominent in the nerve fibers of ganglion cells (Figs. 7D 7F) , photoreceptor inner segments (rods and cones), and cone outer segments (Figs. 7D 7G) . Structures in both plexiform layers (IPL and OPL) were clearly CXCR1 immunoreactive but were stained less intensively (Figs. 7D 7E) . Furthermore, CXCR1-immunoreactive, vertically oriented cell processes in the GCL and INL (Fig. 7E)may indicate faint CXCR1 expression in Müller glial cells in the healthy rabbit retina. As observed for CXCR2 immunoreactivity, the labeling pattern for CXCR1 in human and rabbit retinas corresponded in general to each other. The exception was the prominent labeling of the cone terminals and cone pedicles in humans and the intense labeling of the cone outer segments in rabbit retinas, respectively. 

CXCL8, CXCR1, and CXCR2 in PVR Rabbit Retina

Treated rabbit eyes were examined by indirect ophthalmoscopy throughout the postoperative period. In 9 of 10 rabbits, we observed the development of focal epiretinal membranes and the formation of expanded tractional retinal detachments in one or two quadrants of the posterior retina (detachments sustained for the entire experimental period). This experimental procedure confirmed the observations and results described previously in detail. ³¹ The pathologically changed PVR retina was characterized by a massive loss of photoreceptor segments and somata in the ONL (Fig. 8)and microglia and macroglia activation. Microglial cells now displayed clear GSA labeling, changed their morphology, and migrated to the outer retinal layers. GSA-positive cells (microglial cells and probably blood-derived macrophages) in the nerve fiber layer and the outer retinal layers were strongly immunoreactive for CXCL8 (Figs. 8A 8B 8C 8D 8E 8F)and CXCR2 (Figs. 8G 8H 8I) . In PVR retinas, CXCR1 immunoreactivity was dramatically increased, particularly in ganglion cell somata, in both plexiform layers (IPL and OPL) and in Müller glial cells (Figs. 8J 8M 8N) . We did not observe CXCR1-immunopositive microglial cells in PVR retinas (Figs. 8K 8L) . 

We performed semiquantitative RT-PCR analysis to verify and quantify the changes in CXCL8, CXCR1, and CXCR2 mRNA expression (Fig. 9) . RT-PCR analysis revealed significantly increased mRNA expression for CXCL8 and the corresponding receptors (x-fold increase compared with control: CXCL8, 105.6 ± 26.9; CXCR1, 6.3 ± 1.9; CXCR2, 1.5 ± 0.06; mean ± SEM of seven separate animals). As for normal human retinas, we measured low levels of CXCL8 mRNA expression but did not detect CXCL8 immunoreactivity in healthy rabbit retinas. Microglial cells were responsible for the dramatic increase of CXCL8 expression and the moderate increase of CXCR2 expression, whereas neuronal expression seemed to decline because of neurodegenerative processes. The sixfold increase of CXCR1 mRNA expression was attributed to the massive upregulation synthesis in Müller glial cells in the pathologically changed retinas. 

Discussion

CXCL8 and CXCR1/CXCR2 Receptors in the Healthy Retina

We demonstrate here, for the first time, the expression of CXCR1 and CXCR2 receptors in distinct types of retinal neurons in the normal human retina. The prominent neuronal expression in the healthy retina suggests physiological functions of CXCR1 and CXCR2 receptors beyond their proinflammatory role, including the activation of neutrophils. Recent studies demonstrate that chemokines play important roles not only in brain pathology but also in the physiology of the normal brain. ^{34 35 36} Furthermore, the distinct distribution patterns of CXCR1 and CXCR2 suggest different functions of the two receptor subtypes. The prominent expression of CXCR1 in both plexiform layers and in cone pedicles may reflect a neuromodulatory role, as shown for cholinergic septal neurons in the rat brain ¹⁴ and for Purkinje neurons in the mouse cerebellum. ³⁷ By contrast, the pronounced CXCR2 expression in the inner segment of photoreceptors and in the somata of amacrine cells and other neurons of the inner nuclear layer may be indicative of CXCR2 involvement in neuroprotection or gene transcription. ^{36 38} However, the specific functions of retinal CXCL8 receptors and the exact expression pattern in transmitter-specific subtypes of neurons remain to be elucidated. A recent study demonstrated that the CXCR1 and CXCR2 transcripts are widely expressed in glutamatergic, GABAergic, and cholinergic neurons in the rat brain, ¹⁵ where they may modulate transmitter release or ion channel activities. ¹⁴ This idea is supported by the widespread expression of CXCL8 receptors in retinal neurons using distinct transmitter molecules (e.g., amacrine cells, ganglion cells, and bipolar cells). 

An open question remains: which ligand might activate the CXCR1/CXCR2 receptors in the normal healthy retina? Because of its potent proinflammatory properties, CXCL8 is tightly regulated, and its expression is low or undetectable in normal tissue. ⁷ We detected only the mRNA transcript of CXCL8 but did not detect the CXCL8 protein by means of immunohistochemistry, suggesting that the CXCL8 protein is only marginally expressed in retinal cells (i.e., at undetectable levels for immunohistochemistry) or that it may modulate the receptors in only very small, restricted (e.g., perisynaptic) areas. Another source of secreted CXCL8 might be retinal pigment epithelial cells. Indeed, cultured human pigment epithelial cells were shown to release CXCL8 under distinct conditions. ³⁹ Alternatively, other ligands, such as CXCL1–3 and CXCL5–7, might activate the receptors, particularly the CXCR2 receptor, in the normal retina. Whereas the physiological functions of the receptors remain to be elucidated, a pathophysiological role is easily conceivable. Increased intraocular levels of CXCL8 were detected in patients with retinal detachment, PVR, or proliferative diabetic retinopathy. ^{28 30 40} Hence, during pathologic changes, elevated levels of CXCL8 may activate neuronal CXCR1/CXCR2 receptors to exert neuroprotective effects. Conversely, deleterious consequences may be induced by uncontrolled cytosolic Ca²⁺ increases after neuronal CXCR1/CXCR2 receptor stimulation. 

CXCL8 and CXCR1/CXCR2 Receptors in PVR Membranes

To our knowledge, until now there has been no evidence of CXCR1 or CXCR2 expression in glial cells of the normal mammalian (especially human) retina; this lack of evidence is in accordance with our own findings. Although Danik et al. ¹⁵ demonstrate CXCR1 and CXCR2 mRNA expression in astrocytes of the normal rat brain, only one in vivo study suggests slight CXCR2 protein expression in astrocytes and microglial cells in the normal human brain. ²⁰ However, several groups demonstrate the expression of CXCR1/CXCR2 in human astrocytes or microglial cells in primary cultures or cell lines. ¹⁶ This finding corresponds to our present observations. ^{17 41} It remains to be proven whether this expression might reflect the activation and dedifferentiation of cultured cells, which is similar to the alterations observed under pathologic conditions in vivo. This assumption is supported by the observation that CXCR1 and CXCR2 are expressed in surgically removed cellular membranes of patients with PVR. 

The development of PVR is associated with inflammatory processes, cytokine release from various cell types, and intraocular accumulation of blood-derived cells such as neutrophils and macrophages. ^{40 42 43 44} PVR membranes consist of several cell types, such as blood-derived cells, retinal glial cells, fibroblasts, and retinal pigment epithelial cells ⁴⁵ ; the cellular composition changes with time. ⁴⁶ We observed CXCR1 and CXCR2 expression in GFAP-positive and GFAP-negative cellular structures, suggesting that these membranes consist of blood-derived cells (such as macrophages and neutrophils) and retinal glial cells, each of which can express CXCL8 receptors. In neutrophils, activation of CXCL8 receptors induces a variety of responses, including changes of cell shape, remodeling of the cytoskeletal proteins, directed migration, and respiratory burst coupled to phagocytosis. All these reactions have also been observed in gliotic Müller cells. Under pathologic conditions such as retinal detachment and PVR, Müller cells retract their main processes, migrate to the retinal surfaces, increase the expression of certain cytoskeletal proteins, and are able to perform active phagocytosis. ^{22 42 43 47} Therefore, the activation of glial CXCR1 and CXCR2 receptors might be involved in the induction and maintenance of gliotic processes in Müller cells and astrocytes of the retina. Cultured human astrocytes were shown to respond to tumor necrosis factor-α with enhanced CXCR1 expression, ¹⁷ and CXCL8 and CXCR2 upregulation is involved in a mechanism of self-defense against Fas-mediated cell death. ⁴⁸ The absence of CXCR1 and CXCR2 in glial cells within the normal human retinal tissue and their presence in glial cell structures of PVR membranes suggest an involvement of these receptors in proliferative gliosis. This assumption is further supported by our examinations of retinal pieces from human PVR eyes after retinectomy. Reactive Müller glial cells of PVR retinas seemed to express CXCR1, but not CXCR2, protein. Furthermore, examination of retinas from rabbit PVR eyes confirmed these observations in human retinas. In vivo data regarding a functional role of CXCL8 receptors in glial cells are rare. 

Increased intraocular levels of CXCL8 were detected in patients with retinal detachment, PVR, or proliferative diabetic retinopathy, ^{28 30 40} but the cellular source of CXCL8 is uncertain. There are several possible explanations for the observation that the ligand of both receptor subtypes, CXCL8, was apparently absent in all our examined human PVR membranes. First, the cellular composition of PVR membranes changes with time, ⁴⁶ and the cellular elements undergo a process of transdifferentiation and dedifferentiation. ^{42 45} Thus, we might have missed the adequate stage of this process. Second, other ligands (such as CXCL1–3 and CXCL5–7) produced by unidentified retinal cells might have interacted with the receptors. Third, CXCL8 may be secreted by cells of the ciliary body or from cells of retinal tissue (e.g., microglial cells) outside the epiretinal membranes not available for examination. However, primary cultures of glial Müller cells are able to produce and secrete CXCL8. ⁴⁹ Thus, it is conceivable that Müller cells may contribute to increased CXCL8 levels during certain pathologic changes in the eye. Furthermore, it is well known that cultured Müller glial cells dedifferentiate, undergo dramatic morphologic changes, and experience protein expression pattern changes similar to the alterations under pathologic conditions in vivo. ^{42 50} In particular, cultured retinal glial cells (and cultured pigment epithelial cells) strikingly resemble the cellular phenotypes found in epiretinal membranes during the course of PVR. ^{42 45} The glial expression of CXCL8 receptors in cellular membranes and cultured Müller cells supports the assumption that the expression of these receptors might be involved in phenomena such as cellular dedifferentiation, proliferation, and migration during gliosis. Furthermore, our results recommend such cultures as useful models to study CXCL8 signaling mechanisms and functions in glial Müller cells. 

CXCL8 and CXCR1/CXCR2 Receptors in Rabbit Retina

Our immunohistochemical examinations revealed strong similarities of the cellular distribution pattern of CXCR1 and CXCR2 receptors in healthy human and rabbit retinas. Only the prominent expression of CXCR1/CXCR2 in cone outer segments seemed to be a unique characteristic of the rabbit retina. Thus, the rabbit retina is appropriate for examining normal physiological functions of CXCL8 receptors in retinal tissue. During PVR, CXCL8 mRNA expression was dramatically upregulated and was exclusively produced by retinal microglial cells. The extreme increase in CXCL8 mRNA in retinal tissue was presumably a result of the very low corresponding expression level in the healthy retina. The massive upregulation and the immunohistochemical detection of CXCL8 in activated microglial cells in the rabbit retina confirmed observations that CXCL8 plays an important role in human retinas during retinal detachment, PVR, or proliferative diabetic retinopathy. ^{28 30 40} In addition to CXCL8, activated microglial cells express CXCR2 but not CXCR1. Conversely, gliotic Müller cells strongly express CXCR1 but not CXCR2. This suggests that CXCR1 and CXCR2 receptors exert different functions in different cell types and that other ligands, especially for the CXCR2 receptor, may be involved in the complex inflammatory process. 

Conclusions

The expression of CXCR1 and CXCR2 receptors in neurons of healthy human and rabbit retinas suggests physiological functions of the receptors beyond their well-known role in inflammatory processes. By contrast, receptor expression in glial cells of PVR retinas, PVR membranes, and cultured Müller cells, but not glial cells of the normal human and rabbit retinas, indicates an involvement of these receptors in the initiation or maintenance of reactive gliosis. Activated microglial cells contribute to increased intraocular levels of CXCL8 measured under pathologic conditions such as retinal detachment and PVR. Exclusive expression of CXCR1 or CXCR2 in different glial cell types suggests different functions of these receptors during inflammatory processes in the retina. 

 Supported by Project Grants C5 and C35 from the Interdisciplinary Centre for Clinical Research, Faculty of Medicine, University of Leipzig, and by Deutsche Forschungsgemeinschaft Grants RE 849/8, KO 1547/6-1, and GRK 1097/1.

 Submitted for publication February 19, 2008; revised March 28, May 3, and June 3, 2008; accepted August 15, 2008.

 Disclosure: I. Goczalik, None; E. Ulbricht, None; M. Hollborn, None; M. Raap, None; S. Uhlmann, None; M. Weick, None; T. Pannicke, None; P. Wiedemann, None; A. Bringmann, None; A. Reichenbach, None; M. Francke, 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.

 

Figure 1.

 

Double-labeling immunohistochemistry for CXCL8, CXCR1, CXCR2 (A, D, G), and GFAP (B, E, H) of a healthy human retina. Corresponding merged images (C, F, I). Whereas CXCL8 was not detectable (A), antibodies against CXCR1 (D) and CXCR2 (G) labeled several neuronal cell types in the human retina. All astrocytes in the nerve fiber layer, but only single Müller glial cells (B, arrows), were immunoreactive for GFAP. Clear double labeling was not obvious. Cell nuclei were counterstained with Hoechst 33258 (blue). Scale bar: (A–I) 20 μm. PRS, photoreceptor segments.

Figure 2.

 

CXCR1 and CXCR2 immunolabeling of the inner (A) and outer (B) healthy human retina. (A) Antibodies against human CXCR1 stained nerve fibers and somata of ganglion cells, IPL, OPL, and some neurons in the INL. Cone pedicles in the OPL were strongly CXCR1 immunoreactive. Antibodies against CXCR2 labeled only some ganglion cell somata and various inner neurons of the INL but not the nerve fibers or synaptic contacts in the IPL. (B) CXCR1 expression was detected in cones of the inner segments throughout the cone pedicles. Inner segments and somata of rods were also immunoreactive. The most prominent CXCR2 staining was observed at the inner segments of cones and rods. Scale bar: (A, B) 10 μm. PRiS, photoreceptor inner segments; PRoS, photoreceptor outer segments.

Figure 3.

 

Western blot and RT-PCR analyses of CXCL8, CXCR1, and CXCR2 expression in human retinal tissue. (A) Western blot analysis of human retinal tissue with antibodies specific for human CXCR1 and CXCR2. The most prominent bands indicate proteins of approximately 40 kDa. Bands at higher molecular weights may represent glycosylated receptor proteins. (B) Example of RT-PCR analysis of CXCL8 expression in human retinal tissue. The expected cDNA fragment of 420 bp was detectable in four of seven examined human retinas. (C) Example of RT-PCR analysis of CXCR1 (532 bp) and CXCR2 (622 bp) expression in human retinal tissue, human blood leukocytes, and Müller cell line MIO-M1. (D) Examples of CXCR1/CXCR2 cDNA detection of three additional donor retinas. (B–D, −) Control experiments with the omission of reverse transcriptase enzyme during the PCR procedure.

Figure 4.

 

Immunocytochemical examination of surgically removed human PVR membranes (A–I) and parts of degenerated PVR retinas after retinectomy (J–O). The examined membranes did not contain CXCL8-immunoreactive cells (A), but many of the membrane-forming cells were GFAP positive (B, E, H). Some of the GFAP-positive glial cells were labeled with antibodies against CXCR1 (D) or CXCR2 (G). Corresponding merged images (C, F, I). Specificity control (inset) was achieved by omitting the primary antibodies. Gliotic Müller cells, characterized by their massive GFAP immunoreactivity (K, N), seem to express CXCR1 (J, L) but not CXCR2 (M, O). Cell nuclei were counterstained with Hoechst 33258 (C, F, I, L, O, blue). Scale bars: (A–I) 50 μm; (J–L, M–O) 20 μm.

Figure 5.

 

Immunocytochemical characterization of primary retinal cultures from human organ donors (A–F). Nearly all cultured cells were immunolabeled with antibodies against GFAP (A), S-100β (B), and CRALBP (C), indicating a Müller cell-specific protein expression pattern. Additionally, all glial cells of primary retinal cultures were labeled with antibodies against CXCL8 (D), CXCR1 (E), and CXCR2 (F). Images were derived from double-labeling experiments (A/D, B/C). Additionally, CXCL8, CXCR1, and CXCR2 immunoreactivity was observed in all cultured cells of the immortalized human Müller cell line MIO-M1 (G–I). Scale bars: (A–F, G–I) 50 μm.

Figure 6.

 

Western blot and RT-PCR analyses of CXCL8, CXCR1, and CXCR2 expression in the human Müller glial cell line MIO-M1. (A) Western blot analysis of MIO-M1 cell culture with antibodies specific for human CXCR1 and CXCR2. The most prominent bands indicate proteins of approximately 40 kDa (arrow). Bands at higher molecular weights might represent glycosylated receptor proteins. (B) Example of RT-PCR analysis of CXCR1 (532 bp), CXCR2 (622 bp), and CXCL8 (420 bp) expression. (−) Control experiments with omission of the reverse transcriptase enzyme during the PCR procedure. (C) Examples of Ca²⁺-imaging experiments of cultured MIO-M1 cells during application of recombinant human CXCL8. CXCL8 evoked a sustained (upper trace) or an oscillating (middle trace) Ca²⁺ response in a subpopulation of cultured cells, whereas other Müller cells did not respond to applied hCXCL8 (bottom trace).

Figure 7.

 

Immunohistochemical examination of CXCL8, CXCR1, and CXCR2 expression in the healthy rabbit retina. We did not detect CXCL8 protein expression (A) in the healthy rabbit retina, and microglial cells were faintly labeled with GSA lectin (B). Prominent CXCR2 immunoreactivity was observed in photoreceptor inner segments and cone outer segments. Ganglion cell somata and subpopulations of neurons in the inner nuclear layer were also CXCR2 immunoreactive (C). CXCR1 protein was detected in nerve fibers of ganglion cells (D, F) and in all photoreceptor inner and cone outer segments, respectively (D, G). Structures in IPL and OPL were clearly CXCR1 immunoreactive but were stained less intensively (D, E). Furthermore, CXCR1-immunopositive vertical cell processes in the GCL and INL (E, arrows) may indicate faint CXCR1 expression in Müller glial cells in the healthy rabbit retina. (F, G) Views onto the most inner and outer retina of retinal flat mounts, respectively. Scale bars, 20 μm.

Figure 8.

 

Immunohistochemical examination of CXCL8, CXCR1, and CXCR2 expression in rabbit retina with experimentally induced PVR. We detected CXCL8 protein expression in microglial cells (GSA-positive cells in B, E) in the NFL (A, arrows) and the OPL (D, arrow). (C, F) Corresponding merged images, respectively. In addition to CXCR2 expression in ganglion cell somata (G, arrowheads), we detected CXCR2 expression in GSA-positive cells (G, arrows). A dramatic increase of CXCR1 immunoreactivity was observed in pathologically changed rabbit retinas (J). Plexiform layers and the nerve fibers of ganglion cells (K) were intensely stained, but microglial cells were immunonegative (K, L; view onto the most inner retina of a retinal flatmount). Müller cells in PVR retinas became strongly CXCR1 immunoreactive. Examples of isolated Müller cells (M, N). Cell nuclei were counterstained with Hoechst 33258 (C, F, I–N, blue). Scale bars: (A–M) 20 μm; (N) 10 μm.

Figure 9.

 

Quantitative RT-PCR analysis of CXCL8, CXCR1, and CXCR2 expression of rabbit retinas with experimentally induced PVR compared with control. (A) Dramatic increase of CXCL8 and CXCR1 mRNA expression and moderate increase of CXCR2 mRNA in PVR retinas. Mean (±SEM) relative mRNA levels from seven separate PVR retinas compared with control. (B) After 45 cycles, RT-PCR products were analyzed by agarose gel electrophoresis. Lane 1: control experiments with omission of cDNA during the PCR procedure; lane 2: control retina; lane 3: PVR retina. S, standard. ***P < 0.001, **P = 0.002 (Mann-Whitney rank sum test).

The authors thank Gloria A. Limb (Moorfields Eye Hospital, London, UK) for the spontaneously immortalized human Müller cell line MIO-M1 and John C. Saari (Seattle, WA) for the anti-CRALBP antibody.