Background
Ischemic events in the CNS cause traumatic tissue damage and irreversible loss of neurons present at the ischemic core and surrounding areas. The metabolic demands of the retina are among the highest of any tissue within the body, receiving a dual blood supply from both the choriocapillaris and the central retinal artery (CRA) [
1]. Thus, transient retinal ischemic attacks often cause permanent tissue damage resulting in irreversible vision loss. Following ischemic events in the CNS, there is a significant and detrimental upregulation of biological factors including excess ion influx, excitatory neurotransmitter release, free radical formation, and inflammation [
2‐
4]. Neuroinflammation represents a complex event in the CNS, often involving an activated cellular response from resident immune cells. Once activated, these glial cells secrete a host of pro-inflammatory proteins propagating a chronic cascade of apoptotic or phagocytic events [
4].
Two of the primary effector cells during neuroinflammation are the microglia and astrocytes. Both constitute two of the resident immune cells of visual system (the retina, optic nerve, and visual centers of the brain). Astrocytes maintain direct contact with the neurons, blood vessels, and other glial cells, providing metabolic support, modulating synaptic activity, and maintenance of the blood-brain barrier [
5,
6]. Microglia represent the resident myeloid cells of the CNS, derived from primitive macrophages of the yolk sac that colonize the neuroepithelium prior to the formation of the blood-brain barrier, separating microglia from circulating macrophages. In addition to their antigen-presenting and phagocytic capabilities, their processes are constantly in a mode of surveillance [
5,
7‐
10]. However, when glia are stimulated by disease or injury, they shift to an activation state leading to the production of inflammatory mediators such as chemokines, cytokines, and complement proteins [
11‐
13]. Activation of brain astrocytes causes a dramatic upregulation by at least four-fold of over 260 genes compared to that of quiescent astrocytes, with some genes showing 10 to 100-fold changes in expression [
14]. Accompanying these changes, reactive astrocytes undergo cell proliferation, somatic hypertrophy, overlapping of processes, and scar formation [
15]. Likewise, microglia when stimulated undergo a well-defined morphological transformation, proliferation and migration, as well as expression of adhesion molecules [
16]. These reactive microglia are often observed at sites of pathology in several neurodegenerative diseases including Alzheimer’s disease and glaucoma [
17]. Coupled with their secretion of pro-inflammatory factors such as TNF-α, IL-1β, and C1q, microglia are an ideal candidate as a primary mediator of damage.
C1q, the initiating protein of the classical Complement cascade, is a large complex comprised of six A, six B, and six C chains. Each chain contains a globular region at the carboxyl terminus and a collagen-like stem region at the amino-terminus [
18]. Traditionally, the classical pathway is the antibody-dependent activation of complement, as C1q is known to bind to the surface of foreign substances and antibodies [
19]. However, recent evidence demonstrates classical complement proteins contribute to various neurodegenerative and age-related diseases [
20]. In the DBA/2 J model of spontaneous glaucoma,
C1qa was among the earliest differentially regulated genes not only in the retina, but also at the optic nerve head (ONH) prior to onset of a glaucomatous phenotype [
21,
22]. More surprisingly, during development in the dorsal lateral geniculate nucleus (dLGN), C1q colocalized with either immature pre- or post-synaptic markers, and loss of C1q resulted in retention of overlapping inappropriate connections [
22,
23]. Furthermore, C1q expression in certain areas of the brain was found to increase more than 300-fold during normal aging in the mouse. Reduced levels of cognitive and memory decline were observed in aged
C1qa-knockout mice compared with age-matched WT mice [
24].
In this study, we demonstrate the impact of the neuroinflammatory response following an ischemic event in the retina. We identify time dependent increases in cell density and morphological changes of both astrocytes and microglia not only in the retina but also the superior colliculus, a primary termination site of retinal ganglion cell (RGC) axons exiting the retina. Furthermore, we observed significant increases in C1q expression correlating with reactive gliosis. Our goal was to determine if genetic deletion of
C1qa could morphologically and functionally protect the retina from pathological changes resulting from retinal I/R injury that we have previously characterized [
25], as well as attenuate the activation of glial cells in the visual system.
Discussion
Our studies have demonstrated a significant temporal increase in C1q expression that is accompanied by an increase in reactive microglial and astrocytic retinal cell density following retinal I/R injury. Further, this response is mirrored at RGC termination sites in the visual system through increased C1q expression observed in the SC. These findings agree with previous reports indicating the involvement of both the complement system and glial response to ischemic events not only in the retina, but also areas of the brain affected by I/R Injury [
4,
12,
28,
29]. The absence of C1q provided retinal neuroprotection through rescue against I/R-induced retinal thinning and neuronal loss. It should be noted that while
C1qa
+/−
mice do have measureable levels of C1q, following ischemic injury, these levels were significantly reduced in comparison to WT retina and SC (data not shown). Though present, this dramatic reduction likely explains the partially protective phenotype observed in these mice. Additionally, attenuation of the activated microglial response in the retina and SC was observed in
C1qa-deficient animals. To our knowledge, this is the first report to identify a time-dependent neuroinflammatory response in the visual system following retinal ischemia, and to establish C1q as a primary mediator of damage.
Given the wide variety of animal models of ischemic injury (e.g. global, cardiac, cerebral, or retinal), several methods for neuroprotection have met with mixed results [
30]. The retina provides a unique target for testing of neuroprotective compounds as investigators have the option of local [
31‐
33] or systemic [
34‐
36] delivery of therapeutic compounds or biologics. We have demonstrated significant protection against both functional and morphological deficits in the retina and SC subjected to I/R injury using the JNK inhibitor SP600125 (Kim et al., submitted for publication). Time dependent gene profiling following retinal ischemia identifies several members of the complement family to be among the most significant changes as early as 7 days post injury [
3,
37]. Previous studies identifying protection of RGCs in a
C3-deficient mouse following retinal I/R demonstrated a delayed loss of cells after 3 weeks; however, no difference in axonal damage was seen compared with WT mice [
3]. While our endpoints differed from Kuehn et al., we were able to show a significant rescue of RGCs and cells in the GCL. While there exists sufficient evidence of activation of the downstream complement cascade including C3 following retinal ischemia, our results suggest C1q may be acting in a cascade-independent role.
Activation of the complement system in the CNS during times of injury and pathogenesis has been well documented [
38,
39]. Increased expression of C1q has been specifically observed to colocalize with hallmark pathological changes observed in several neurodegenerative diseases including Alzheimer’s disease (AD), Parkinson’s disease (PD), and glaucoma [
20,
40]. Furthermore, C1q is known to directly bind to the membrane of neurons, which have a poor capacity to regulate activation of downstream complement cascade factors [
41]. Our studies demonstrate significant upregulation of C1q at both local and distal sites in the visual system following I/R injury. As such, C1q has been identified as a potential prime therapeutic target. Although not indicated for treatment of chronic neurodegenerative disease, previous strategies to modulate C1q have targeted C1r/C1s, which is necessary with C1q to form the C1 complex, using recombinant C1-INH as well as soluble CR1 [
42]. In animal models of brain ischemia, administration of C1-INH was observed to reduce infarct volume and neurological deficits [
28,
43]. Similar to our findings of cell rescue and retinal morphology preservation, deletion of C1q has been identified as a protective mechanism in various other disease models. Recent in vitro data suggested C1q promoted neurite progression by modulating expression of genes necessary for outgrowth [
44]. These findings were later supported, and extended in vivo
, demonstrating axon regeneration and improved guidance following spinal cord injury [
45]. In both studies however, downstream C3 and C5 had non-growth promoting effects, suggesting distinct mechanistic roles for different complement components. AD mice deficient for
C1q (APPQ
−/−) significantly preserved functional neurons and increased dendritic staining in the hippocampus compared to Aβ pathological (APP) mice [
46]. Accumulation of C1q during normal brain aging has been implicated in cognitive decline; age-matched
C1qa-deficient mice were observed to perform significantly better in a series of learning and memory behavioral tests compared to WT mice [
24]. Further, in the DBA/2 J spontaneous glaucoma model,
C1qa was identified to be differentially expressed in the retina and ONH preceding phenotypic glaucomatous damage. Deletion of
C1qa significantly delayed RGC axonal damage [
21,
22].
In our current study, we identified for the first time a temporal response pattern for microglia and astrocyte activation in the retina and SC after retinal I/R injury. Furthermore, we demonstrated the significant increase in microglial activation and cell density resulted from accumulation of C1q in the injured tissue. In the CNS, astrocytes, microglia, and neurons have been identified as the primary producers of complement proteins [
40,
47]. Our data support previous reports that disturbances to the retina from injury and inflammation results in a dynamic response of microglia, known to phagocytose dying neurons [
48‐
50]. Evidence suggests the increased cell density may result from infiltrating microglia through compromised blood-retinal barriers [
51]; however, our methodology was unable to differentiate resident and migratory microglia. Further studies utilizing fluorescently-labeled transplanted cells will be needed to identify whether our observed changes were due to infiltration, proliferation, or a combination of both. Given the implication of chronic reactive microglia in neurodegenerative disease as well as recently discovered developmental roles [
52,
53], several approaches have been made to abate microglial activation. Neuroprotective treatment strategies such as minocycline, irradiation, and TSPO have shown significant suppression of microglial activation and proliferation, while other methods such as modulation of fractalkine receptor CX3CR1 have led to mixed results [
29,
48,
54‐
57]. These studies, along with our current work, demonstrate the therapeutic potential of targeting microglia in several animal models of neurodegeneration. Furthermore, early clinical trials using oral minocycline resulted in moderately improved visual acuity and reduction in vascular permeability in patients with diabetic macular edema [
58].
Despite our
C1q-dependent modulation of activated microglia and observed morphological protection in the retina, we were unable to totally rescue I/R induced loss of retinal function as assessed through scotopic ERG. Coinciding with this finding, mice deficient in
C1qa displayed increased levels of GFAP similar to WT animals following retinal I/R. This supports previous results that microglia are the primary synthesizers of complement proteins, specifically C1q, leading to pathogenesis from prolonged inflammatory responses [
7,
56]. This suggests following ischemic episodes in the retina, prolonged astrocyte and Müller cell activation is stimulated by proinflammatory factors different from microglia. Reactive astrogliosis is known to be triggered by cytokines such as TNFα, CNTF, IL-1β, and IL-6 [
59]. Further, these macroglial cells facilitate a certain degree of neurodegeneration independent from microglia, which may explain our findings of visual deficits, as well as the second wave of C1q expression observed in the SC on day 28. Similar to microglia, astrocytes posses the machinery for synaptic engulfment; however, this appears to be a C1q-independent process [
60]. GFAP represents an essential intermediate filament of reactive astrocytes necessary for the many astroglial functions following injury [
61]. Genetic deletion of GFAP has provided mixed results of protection and exacerbation that appear to be dependent on the nature of the injury [
62‐
64]. Astrocytes, both neighboring and tissue-specific, have been identified as a highly heterogenous population in their gene expression patterns, morphology, and proliferative responses to injury, which are further complicated by their proximity to the trauma [
15]. Therefore, it is likely that retinal astrocytes would react much differently to retinal I/R injury compared to astrocytes in the SC. Therefore, we find it unsurprising that elimination of
C1q had no significant effect on controlling this gliotic response in the SC. Combinatorial therapies have had recent success in improving pathological endpoints in models of injury and disease [
65]; therefore, targeting multiple proinflammatory proteins may further suppress the glial response to I/R injury, thereby providing enhanced neuroprotection.
Methods
Animals
Male and female transgenic mice for C1qa (B6.129P2-C1qa < tmlmjw>/Sj) that were backcrossed 10 generations onto a pure C57BL/6 J background and (hereafter referred to as either C1qa
+/−
or C1qa
−/−
), were generously provided from the Simon John laboratory (Jackson Laboratory; Bar Harbor, ME). Female C57BL/6 J (Jackson Laboratory) and male and female C1qa mice (3–4 months of age) were used for transient retinal I/R studies. Animals were maintained in 12-h:12-h/light:dark cycle. All studies and animal care were performed as approved by the Institutional Animal Care and Use Committee at the University of North Texas Health Science Center and followed the Association for Research in Vision and Ophthalmology (ARVO) Statement for the Use of Animals in Ophthalmic and Vision Research.
Retinal I/R
Retinal I/R was induced as described previously [
25]. Briefly, mice were anesthetized with a ketamine/xylazine/acepromazine cocktail (100/10/3 mg/kg), and the left eyes were dilated (2.5 % Phenylephrine HCl; Paragon BioTeck, Inc.) followed by cannulation of the anterior chamber with a 30-gauge needle connected to a reservoir containing sterile PBS. The reservoir was elevated to generate an intraocular pressure of 120 mmHg for 1 h to induce retinal ischemia. Afterwards, the cannula was removed and blood was allowed to naturally reperfuse the retina. Body temperature was maintained on a digitally controlled heating pad for the duration of the procedure and recovery.
Immunofluorescence of the retina and SC
Following euthanasia of the mice, ischemic and contralateral control eyes were gently removed and placed in 4 % paraformaldehyde (PFA) for 4 h followed by a 20 % sucrose solution overnight. Whole brains were delicately excised and fixed in 4 % PFA for 6–8 h, after which a 3 mm region was cut to include the superior colliculus (SC) using a mouse brain block. The SC-containing brain section was also embedded in a 20 % sucrose solution overnight. Eyes and brains were placed in Tissue-Tek OCT compound (Sakura Finetek USA, Torrance, CA) and frozen over dry ice before being stored at −80 °C. Eyes and brains were frozen sectioned at 12 μm and placed on Superfrost Plus slides (VWR, Radnor PA). Slides were triple washed in PBS, blocked and permeabilized for 1 h in blocking buffer (PBS with 10 % fetal bovine serum and 0.15 % Triton X-100) before incubation with primary antibodies overnight at 4 °C. For retinal sections anti-C1q (A201, 1:1000) from Quidel (San Diego, CA) was used; however excessive non-specific labeling was observed with this antibody in SC sections, therefore anti-C1q (ab182451, 1:200) from Abcam (Cambridge, MA) developed by Stephan and colleagues was used [
24]. For glial cell analysis, antibodies to Iba1 (019-19741, 1:500) from Wako Chemicals (Richmond, VA) and GFAP (ab7260, 1:300) from Abcam were used. RGC counts were performed using the specific marker Rbpms (GTX118619; 1:200) from GeneTex (Irvine, CA). Tissues were incubated in secondary antibodies conjugated with Alexafluor 488 and Alexaflour 592 (1:250; Invitrogen/Molecular Probes, Carlsbad, CA) for 1 h at room temperature. Slides were mounted with cover slips using ProlongGold anti-fade reagent with DAPI (Molecular Probes, Life Technologies, Grand Island, NY). Images were taken using a Nikon Eclipse Ti inverted microscope (Nikon; Melville, NY) and CRi Nuance FX multispectral imaging system (Caliper Life Sciences; Hopkinton, MA). Autofluorescence was subtracted using Nuance 3.0 software (Caliper Life Sciences).
Cell density analysis
Density of astrocytes, Müller cells and microglia in the retina and SC was calculated using thresholding analysis in ImageJ (NIH; Bethesda, MD). Briefly, images were separated by color channel into an RGB stack, after which a region of interest (ROI) was drawn around the entire tissue. A threshold was set to positively select only Iba1+ or GFAP+ cells, and this threshold was maintained for all images analyzed. Measurements were limited to this threshold and the percent area occupied of the section was determined. A minimum of three images per section and three slides spanning the depth of the tissue were analyzed and averaged per animal. A similar methodology was used to assess fluorescence intensity of C1q in retina and SC sections.
Histological assessment of the retina
Whole globes were immersion fixed in 4 % PFA overnight at 4 °C, followed by paraffin processing. Eyes were sectioned at 5 μm and stained with hematoxylin and eosin (H&E). Entire retinas were imaged, ora serrata to ora serrata through the optic nerve head, and thickness was measured using calibrated calipers in ImageJ from the nerve fiber layer (NFL) through the outer nuclear layer (ONL) at two peripheral and two central locations of the retina. Three slides were selected per retina and the four cross sectional measurements from each retina were averaged together. Nuclei in the H&E ganglion cell layer (GCL), including RGCs and displaced amacrine cells, were counted from each retina and averaged.
Scotopic flash electroretinography (ERG)
All animals were dark adapted overnight; mice were anesthetized with isoflurane and connected to the HMsERG system (Ocuscience; Rolla, MO). Body temperature was maintained at 37 °C. A ground electrode was placed subcutaneously by the tail and reference electrodes inserted under each eye. Silver-thread electrodes were placed across the apex of the cornea and held in place with a Gonak (Akorn; Lake Forest, IL) coated contact lens. Eyes were exposed to a series of light flashes at increasing intensities (0.1, 0.3, 1, 3, 10, and 25 cd.s/m2). Amplitudes and implicit times of waveforms were measured and analyzed.
Statistics
Statistical analysis was performed using SigmaPlot 12 (Systat; San Jose, CA). Student’s paired t-test was used to compare experimental groups within animals, ischemic versus contralateral control. One-way ANOVA was used to compare among three or more groups, such as comparing between time points for experimental groups. Holm-Sidak post hoc analysis was used for multiple comparisons. All data are expressed as mean ± standard error mean (SEM), and p-values less 0.05 were considered statistically significant.
Competing interests
The authors declare that they have no competing interests.
Authors’ contributions
SMS performed in vivo procedures, immunoassays, statistical analysis, and prepared and edited the manuscript. BJK peformed in vivo procedures and revised the manuscript. JM performed immunoassays, data analysis, and helped revise the manuscript. RJW, GRH, and SMWJ participated in study design and manuscript revisions. AFC conceived of study, interpretation of results, and outlined and edited the manuscript. All authors read and approved the final manuscript.