Reactive Fibroblasts in Response to Optic Nerve Crush Injury

Abstract

Traumatic optic neuropathy leads to bidirectional degeneration of retinal ganglion cells and axons and results in optic nerve scaring, which inhibits the regeneration of damaged axons. Compared with its glial counterpart, the fibrotic response causing nerve scar tissue is poorly permissive to axonal regeneration. Using collagen1α1-GFP reporter mice, we characterize the development of fibrotic scar formation following optic nerve crush injury. We observe that perivascular collagen1α1 cells constitute a major cellular component of the fibrotic scar. We demonstrate that extracellular molecules and monocytes are key factors contributing to the pathogenesis of optic nerve fibrotic scar formation, with a previously unrecognized encapsulation of this scar. We also characterize the distribution of collagen1α1 cells in the retina after optic nerve crush injury based on in vivo and whole-mount retinal imaging. Our results identify collagen1α1 cells as a major component of fibrotic scarring following ONC and are a potential molecular target for promoting axonal regeneration after optic nerve injury.

Introduction

Traumatic optic neuropathy (TON) occurs in approximately 0.5-5% of head injuries and ultimately leads to severe visual impairment, including irreversible blindness[1]. Histopathologically, TON is associated with bidirectional (anterograde/retrograde) degeneration of retinal ganglion cells (RGC) and axons[2]. As a part of the central nervous system (CNS), injured RGCs, and their axonal nerve fibers typically do not regenerate[3]. No clinically effective evidence-based therapy currently exists to treat TON [4,5]. Discovering novel therapeutic targets to decelerate RGC degeneration or promote optic nerve regeneration after trauma is of paramount importance. RGC degeneration due to TON has been studied in the optic nerve crush (ONC) animal model, which has the basic pathological features and the molecular changes of traumatically injured optic nerve [6–8].

Following CNS trauma, scar formation at the site of trauma is a multi-factorial wound response and healing process[9]. CNS scar tissue not only protects the injured site from secondary damage but also can act as an extrinsic barrier for axonal regeneration and remodeling[10]. In spinal cord injury (SCI) model, spinal cord contusion site scar tissue can be categorized into glial and fibrotic components. The glial scar is made up of activated astrocytes, oligodendrocyte progenitor cells, and microglia that surround the lesion site, which has been studied for its role as a barrier to nerve regeneration[11]. However, far less is known about the fibrotic components which form the scar core. Studies have demonstrated fibroblasts express multiple molecules involved with axonal regeneration and transected axons end at the border of the fibrotic scar[12,13]. Inhibiting fibroblast migration after SCI significantly reduces fibrotic scarring and improves functional recovery in mice[14,15]. Therefore, the fibrotic component of the nerve scar tissue consisting of fibroblasts is less permissive to axonal regeneration than its astrocyte containing glial counterpart[16]. Elucidating the cellular components involved in preventing axonal re-growth and thereby allowing for functional restoration is necessary to gain an understanding of these inhibitory cues and for finding potential targets for therapeutic intervention that promote optic nerve axonal regeneration.

A reason for the gap in identifying fibrotic optic nerve scar cellular components is the lack of cell specific markers for identifying pericytes and other monocytic cells types. Recent studies demonstrate that perivascular collagen 1α1 cells are a major source of the cellular constituents of the fibrotic scar formed after contusive SCI in a transgenic spinal cord injury model[17]. Using this same SCI transgenic mouse model, in which GFP is expressed under the control of the collagen 1α1 promoter (Col1α1-GFP mice) [18] and the ONC model, we investigated the role of perivascular fibroblasts in optic nerve injury and characterize its cellular components. Our findings reveal detailed pathophysiologic features at the injured optic nerve injury site, highlight the molecular changes within the fibrotic scar that forms in response to traumatic injury, and provide novel insight into potential therapeutic targets following fibrotic scar development after ONC that may create a permissive environment for axonal regeneration after ONC injury.

Methods

Experimental animals

Col1α1-GFP transgenic mice were generated as described[18]. In brief, 3.2 kb of collagen1 (α1) promoter and enhancer (−8000 to −7000) sequences were cloned and positioned at 5’ of the open reading frame of enhanced GFP. GFP expressed when Col1α1 transcripts were generated. All experimental procedures were approved by the Animal Care and Use Committee of the University of Miami.

Optic Nerve Crush

Mice were subjected to optic nerve crush as previously described[19,20]. Col1α1-GFP mice at the age of two months were anesthetized with a Ketamine/Xylazine (90-100 mg/kg + 5-10 mg/kg) cocktail via intraperitoneal injection and eyes were topically anesthetized with 0.5% proparacaine hydrochloride. A conjunctival peritomy was performed above the eyeball and the eye muscles were gently retracted to expose the optic nerve. The optic nerve was crushed intra-orbitally with Dumont #5 forceps (FST) for 10 seconds approximately 0.5-1mm behind the globe without affecting the blood supply.

Immunohistochemistry

At 3, 7, 10, 14 days after optic nerve crush (n=5 mice per time point), mice were sacrificed and perfused transcardially with phosphate-buffered saline (PBS) followed by 4% paraformaldehyde (PFA) in PBS at 10ml/min. Murine optic nerves and retinas were harvested and fixed with 4% PFA in PBS overnight.

Ten micron thick sections of frozen optic nerve samples were prepared for immunofluorescence staining. Optic nerve sections were blocked for non-specific antibody binding using rodent blocker M (Biocare Medical, Concord, CA) for 1 hour at room temperature to minimize non-specific antibody binding. Primary antibodies were diluted in 0.5% Triton X-100/PBS and incubated overnight at 4 °C. Primary antibodies used are listed in Table 1. After PBS washes, tissue sections were incubated with the appropriate secondary antibody for 1 hour at room temperature and mounted using 4’-diamidino-2-phenylindole (DAPI, Vector Laboratories H-1200). Secondary antibodies used were: Alex Fluor 488 donkey anti-chicken (1:500, Jackson ImmunoResearch Laboratories, West Grove, PA), donkey anti-rabbit Cy3 (1:500, Jackson ImmunoResearch Laboratories, West Grove, PA), donkey anti-rat Cy5 (1:500, Jackson ImmunoResearch Laboratories, West Grove, PA), donkey anti-goat Cy5 (1:500, Jackson ImmunoResearch Laboratories, West Grove, PA). For whole-mount retinas, tissue was incubated in 0.5% Triton X-100/PBS with gentle rotating for 1 hour followed by incubating in rodent blocker M for 1 hour. Primary antibodies were diluted in 0.1% Triton X-100/PBS and incubated for 3 days at 4°C with gentle rotation.

Table 1.

Primary antibodies used in the study

Species	Antigen	Company	Dilution
Chicken	Green Fluorescent Protein (GFP)	Abcam, ab13970	1:200
Goat	Glial Fibrillary Acidic Protein (GFAP)	Abcam, ab53554	1:200
Rabbit	Glutamine synthetase (GS)	Abcam, ab73593	1:200
Rabbit	Vimentin	Abcam, ab92547	1:400
Rabbit	Platelet-derived Growth Factor Receptor beta (PDGFR-ß)	Abcam, ab32570	1:200
Rat	CD13	Abcam, ab33489	1:400
Mouse	Alpha-smooth muscle actin (αSMA)	Abcam, ab28052	1:200
Rabbit	Fibronectin	Abcam, ab2413	1:200
Rabbit	Laminin	Abcam, ab11575	1:200
Rat	CD68	Abcam, ab53444	1:200
Rat	F4/80	Abcam, ab6640	1:200
Goat	Ionized Calcium Binding Adaptor Molecule 1 (Iba-1)	Invitrogen, MA5-27726	1:200
Rat	CD11b	Invitrogen, 14-0112-82	1:200

Primary antibodies used for whole-mount retina include chicken-GFP (1:200, Abcam, ab13970), mouse-CD31(1:200, Abcam, ab24590), rabbit-RNA-binding protein with multiple splicing (RBPMS, 1:200, Abcam, ab194213), rat-CD13 (1:400, Abcam, ab33489) and rabbit- platelet-derived growth factor-beta (PDGFR-ß, 1:200, Abcam, ab32570). After PBS washes, retinas were incubated with the appropriate secondary antibody overnight at 4 °C and mounted using 4’-diamidino-2-phenylindole (DAPI, Vector Laboratories H-1200). Secondary antibodies used were: Alex Fluor 488 donkey anti-chicken (1:500, Jackson ImmunoResearch Laboratories, West Grove, PA), donkey anti-mouse Cy3 (1:500, Jackson ImmunoResearch Laboratories, West Grove, PA), donkey anti-rat Cy3 (1:500, Jackson ImmunoResearch Laboratories, West Grove, PA), and donkey anti-rabbit Cy5 (1:500, Jackson ImmunoResearch Laboratories, West Grove, PA). Immunostained tissue sections and whole-mount retinas were imaged by confocal microscopy (Leica TCS SP5 Confocal Microscope, Leica Microsystems, Buffalo Grove, IL).

Confocal scanning laser ophthalmoscopy (CSLO) imaging

In vivo CSLO images of the retina for Col1α1⁺-GFP cells was performed before ONC and at 3, 7, 10, and 14 days after ONC (n=6 mice per time point). Mice were anesthetized with injected intraperitoneally. Before imaging, both eyes were dilated with 2.5% phenylephrine hydrochloride eye drops (Akorn, Lake Forest, IL). Col1α1⁺-GFP cell number was determined by quantifying the percentage of bright GFP expressing cells along the retinal vessels and their signal strength within confocal retinal images. Mice were placed on an imaging platform with a heating pad at 37°C to maintain stable body temperature. Each eye was imaged with a customized Heidelberg CSLO HRA II (Heidelberg Engineering, Heidelberg, Germany). Images were acquired in red-free mode and at 488 nm for green fluorescence from a scan angle of 55 degrees with the angle adjusted to acquire peripheral retina images sweeps in all quadrants of the eye. The scan rate was 51 frames per second.

Quantification and Statistical Analysis

Cell density counts were performed by three masked observers using Image J software. Calculations were performed at the injury epicenter and in five adjoining sections of 5 mice at each time point by obtaining the average pixels of the Col1α1-GFP signals in selected 600×600 pixel figures. ECM cell density counts were performed at the injury site and in five adjoining sections by counting the average number of cells in the scar area in each tissue section. Co-localization was calculated by counting the cells stained by target antigen markers co-localized with Col1α1-GFP positive cells in each of the ten 1μm z-stack slices using Image J software. The quantification of Col1α1⁺-GFP cells in CSLO images was performed by three masked observers based on the cell density and signal intensity. The cell density was graded as follows: the percentage of Col1α1⁺-GFP cells <20%, 20%-50%, 51%-80%, >80% were assigned 1, 2, 3, 4 points, respectively. The signal intensity was graded as indiscernible, weak, moderate and strong, and assigned 1, 2, 3, 4 points, respectively. The two scores were multiplied together as the final result. All observers evaluated the same images and inter-observer variability was assessed. The intra-class correlation coefficient of three observers for cell density counts is 0.895. Differences in cell percentage were tested by one-way ANOVA with Tukey’s post-test. Statistical analysis for retinal Col1α1-GFP cells changes following ONC used longitudinal data analysis performing by SPSS (SPSS for Mac, v.23.0; IBM-SPSS, Chicago, Illinois, USA), p values less than 0.05 were considered statistically significant.

Results

Glial scar components surround the fibrotic ONC scar

To investigate the presence of reactive gliosis after ONC injury, we performed optic nerve crush on Col1α1-GFP mice and assessed the presence of glial fibrillary acidic protein (GFAP), glutamine synthetase (GS) and vimentin using immunostaining. The ONC lesion site was examined at different time points post-ONC injury. Concentrated GFAP expression is a hallmark of astrocyte reactivity after CNS injury. In the uninjured optic nerve, an even distribution of fibrillary patterned GFAP was present throughout the ON (Fig. 1a, c). Surprisingly, after ONC, we observed a demarcating border around the injury site with the absence of GFAP expression inside the formed scar (Fig. 1, marked with blue dotted line). At 3 days post-injury (dpi), the border of the glial scar was poorly formed with residual GFAP expression within the lesion area (Fig. 1d, f), while some Col1α1⁺-GFP cells began to appear within the lesion area. (Fig. 1e, marked with white dotted line). At 7 dpi, a distinct rim of the GFAP deficient scar area was observed around the ONC injury site (Fig. 1g, i), within which a visible mass of Col1α1⁺-GFP expressing cells was present (Fig.1h). At 10 dpi, GFAP-expressing cells formed a capsule-like structure (white arrow) surrounding the fibrotic component, with a faint expression of GFAP observed around the lesion site (Fig. 1j–l). Fourteen days after ONC, Col1α1⁺-GFP cells consolidated into a smaller region with a high density of GFAP expressing cells surrounded the scar but not within the scar (Fig. 1m–o).

Fig. 1.

Development of the glial scar and fibrotic scar after ONC. Time course of GFAP and Col1α1⁺cells distribution throughout the ON following ONC injury. GFAP negative area was marked with a blue dotted line, the fibrotic scar was marked with a white dotted line, a-c in the uninjured optic nerve, an even distribution of fibrillary patterned GFAP was present throughout the ON. d-f At 3 dpi, the border of the glial scar was poorly formed with residual GFAP expression within the lesion area, and a small number of Col1α1⁺-GFP cells began to appear within the lesion area. g-i At 7 dpi, a more distinct rim of the GFAP deficient scar area was observed around the ONC injury site, within which a visible mass of Col1α1⁺-GFP expressing cells was present. j-l By 10 dpi, GFAP-expressing cells formed a capsule-like structure (white arrow) surrounding the fibrotic component, with only a faint expression of GFAP observed around the lesion site. m-o At 14 dpi, Col1α1⁺-GFP cells consolidate into a smaller region with a higher density of GFAP expressing cells surrounded the scar but, most importantly, not within the ONC scar. Scale bar, 200 μm.

GS showed an even distribution pattern in the uninjured ON (Supplemental Fig. 1a, c). At 3 dpi and 7 dpi, a large area with the absence of GS expression was observed with a faint border (marked with blue dotted lines), meanwhile, Col1α1⁺-GFP expressing cells accumulated in the ONC crush site (Supplemental Fig. 1d–i). At 10 dpi and 14 dpi, a clearly demarcated GS deficient area presented at the ONC lesion site (Supplemental Fig. 1j–o) with fibrotic components expressed within the area (Supplemental Fig. 1k, n).

Vimentin is a key marker for reactive gliosis, but its expression pattern is different from GFAP and GS (Supplemental Fig. 2). In the uninjured ON, vimentin was expressed evenly in a fibrillary distribution pattern (Supplemental Fig. 2a, c). At 3 dpi, a faint border of vimentin deficient area was observed (marked with blue dotted line) with residual vimentin expression within the lesion area (Supplemental Fig. 2d, f), while Col1α1⁺-GFP positive cells began infiltrating the scar center (Supplemental Fig. 2e). At 7 dpi, vimentin-expressing cells together with Col1α1⁺-GFP cells were present at the center of the ONC lesion site, but vimentin and Col1α1⁺-GFP expressing cells were not entirely co-localized (Supplemental Fig. 2g–i). Ten days and 14 days post-ONC injury, as the Col1α1⁺-GFP cells formed a fibrotic scar (Supplemental Fig. 2j–o), a cluster of high-density vimentin-expressing cells were observed in the fibrotic scar (Supplemental Fig. 2l, o).

Our data show that following the ONC injury, both glial and fibrotic scar injury components are present at the ONC injury site, with the glial scar components surrounding a well-demarcated encapsulated fibrotic ONC scar.

Col1α1⁺ cells are predominantly pericytes forming the fibrotic optic nerve scar

To investigate the contributions of Col1α1⁺ cells to the development of the fibrotic scar after ONC, we analyzed the ONC lesion site at multiple time points after crush injury in Col1α1-GFP mice with immunostaining (Fig. 2). At 3 dpi Col1α1⁺ cells began migrating to the ONC lesion site (Fig. 2a). The number of Col1α1⁺ cells increased until a visible diffuse mass of Col1α1⁺ cells was observed by 7 dpi (Fig. 2b). The density of infiltrated Col1α1⁺ cells reached its peak between 10 dpi to 14 dpi with a clearly delineated sharp border with a spindle-shaped optic nerve scar (Fig. 2c, d). The fibrotic scar fdled the ONC site (GFAP-negative area), whereas the inner region contained a reticular distribution of Col1α1⁺ cells and the peripheral border adjacent to the glial scar contained a tight lining of Col1α1⁺ cells.

Fig. 2.

Distribution of Col1α1⁺cells during fibrotic scar formation (a-d). a At 3 dpi Col1α1⁺ cells were observed to have begun migrating to the ONC lesion site, b At 7dpi, a visible diffuse mass of Col1α1⁺ cells was observed at the crush site. c, d The density of infiltrated Col1α1⁺ cells reaches its peak between 10 dpi to 14 dpi with a spindle-shaped optic nerve scar with sharp borders clearly delineated. Time course of PDGFR-ß (red, e-h), CD13 (magenta, i-l) and αSMA(white, m-p) expression and their co-localized with Col1α1⁺-GFP cells. q Col1α1⁺-GFP cell quantification analysis of the epicenter of the encapsulated fibrotic scar, r-t Percentage of Col1α1⁺-GFP cells that express each cell type specific antigen at different time courses after ONC. Asterisk indicates significance in one-way ANOVA. * p<0.05, **p<0.01; NS, Non-significant by one-way ANOVA. Scale bar, 200 μm.

To characterize the phenotypic cellular identity of the ONC infiltrating Col1α1⁺ cells, we immunostained ONC histologic sections with pericytes-specific markers to detect the expression of PDGFR-ß, CD13 and α-smooth muscle actin (αSMA). In the mature fibrotic scar, Col1α1⁺ cells were the predominant PDGFR-ß-expressing cells at the fibrotic scar (Fig. 2e–h). Many CD13 positive cells were present within the fibrotic scar (Fig. 2i–l). However, only a few Col1α1⁺ cells co-expressed CD13, thereby suggesting other cell types which separately express CD13 also contribute to fibrotic scar formation. Some Col1α1⁺ cells co-expressed αSMA markers acutely after ONC (Fig. 2m–p). As αSMA is a marker for myofibroblasts, these results also suggest some Col1α1⁺ cells are predominantly pericytes which contribute to the formation of the fibrotic scar as part of a myofibroblastic scaffold.

Consecutive histological cross-sections of the ONC fibrotic scar area on 3, 7, 10, 14 dpi at 600 × 600 pixels were obtained for Col1α1⁺ cells’ density analysis. As shown in Fig. 2q, the percentage of Col1α1⁺ cells at the crush site increased from 13.3% on 3 dpi to 27.8% on 7 dpi and reached a peak level at 60.2% and 44.6% on 10 dpi and 14 dpi, respectively. The data showed no significant difference in GFP-positive cell density between 10 dpi and 14 dpi, thereby suggesting during the sub-acute phase of the ONC injury, the number of Col1α1⁺ cells reached its peak in the fibrotic scar at 10dpi with mature fibrotic scar formation. These results demonstrate a prominent role of Col1α1⁺ cells in the pathogenesis of the fibrotic scar after ONC.

To quantify the density of cellular components in the ONC scar, we calculated the percentage of PDGFR-ß, CD13, and αSMA that co-localize with Col1α1⁺-GFP positive cells during the time course of ONH scar formation. Only 33.0% of Col1α1⁺ cells expressed PDGFR-ß at the early stage of the fibrotic scar (3 dpi), and the percentage of co-localization was 85.8% at 14 dpi, suggesting that most of the Col1α1⁺ cells that formed the mature fibrotic scar were predominantly PDGFR-ß-expressing cells (Fig. 2r). At 3 dpi, 41.2% of Col1α1⁺ cells expressed CD13, while only 18.1% of Col1α1⁺ cells express CD13 at 14 dpi. This decreasing percentage indicates that CD13⁺ cells are recruited to the injury site at the early phase of ONC and these cells are not the predominant cells in the mature ONC fibrotic scar. As Fig. 2t shows, 45.0% of Col1α1⁺ cells were co-labeled with αSMA at 3 dpi. The distribution of αSMA-expressing Col1α1⁺ cells was relatively stable throughout assayed time points, the percent remaining at approximately 30%, suggesting that αSMA-expressing Col1α1⁺ cells reactive to the ONC injury in the acute phase, and have a stable structural role in the fibrotic scarring.

Extracellular matrix (ECM) molecules participate in organizing the ONC scar structure

Fibrotic scar formation after central nervous system injury is often characterized by the accumulation of perivascular fibroblasts and deposition of ECM. We assessed the presence of ECM in the formation of the fibrotic scar by immunostaining injured optic nerve sections to detect the expression of fibronectin and laminin after ONC (Fig. 3). At 3 dpi, fibronectin expressing cells were rare at the ONC site (Fig. 3a–b). At 7 dpi, a significant number of fibronectin-expressing cells accumulate at the ONC site and organize into an encapsulated fibrotic scar (Fig. 3c–d). At 10 dpi, after fibrotic scar formation, fibronectin expressing cells displayed a dense distribution pattern and are more organized into a cellular network highly co-localized with Col1α1⁺ cells at 14 dpi (Fig. 3e–h).

Fig. 3.

Extracellular molecules in fibrotic scar formation after ONC. Time course of fibronectin (red, a-h) and laminin (magenta, i-p) expression following ONC, and their co-localization with Col1α1⁺-GFP cells. a-b At 3 dpi, fibronectin-positive cells were scarce at the ONC site. c-d At 7 dpi, a significant number of fibronectin expressing cells are accumulated at the ONC site and organized into an encapsulated fibrotic scar. e-f By 10 dpi, after the fibrotic scar had formed, fibronectin expressing cells displayed a dense distribution pattern. g-h At 14 dpi, fibronectin became much more organized into a cellular network highly co-localized with Col1α1⁺ cells. i-j At 3 dpi, cord-like laminin is present at the ONC injury site. k-l At 7dpi, laminin expression is more organized than that of fibronectin. m-n By 10 dpi and 14 dpi, laminin expressing cells display a dense distribution pattern and co-localized highly with Col1α1⁺ cells (o-p). q-r Fibronectin cell and laminin cell count analysis at the injury site. s-t Percentage of fibronectin and laminin cells that co-labeled with Col1α1⁺-GFP cells at different courses after ONC. Asterisk indicates significance by one-way ANOVA analysis. * p<0.05. NS, Non-significant by one-way ANOVA. Scale bar, 200 μm.

Laminin is another key ECM molecule. At 3 dpi, cord-like laminin expressing cells migrated to the ONC injury site (Fig. 3i–j). At 7 dpi, laminin expression was more organized than that of fibronectin (Fig. 3k–l), indicating the scaffold of the scar formed by laminin has already been formed before Col1α1⁺ cell accumulation reached its peak (Fig. 3m–p). Laminin expressing cells, together with Col1α1⁺ cells, play an important role in the development of the ONC fibrotic scar.

To further quantify the ECM molecules at the ONC injury site, we counted the number of fibronectin and laminin-positive cells at the injury epicenter and calculated their percentage of co-localization with GFP- Col1α1⁺ cells. As shown in Fig. 3q, the number of fibronectin cells increased with the development of the fibrotic scar. However, no significant difference in cell number was observed between 7 dpi and 10 dpi, suggesting that the number of fibronectin expressing cells at the injury site did not change with scar structure becoming more organized by 10 dpi. After the mature fibrotic scar was formed, more fibronectin expressing cells were present at the ONC injury site by 14 dpi. The distribution of laminin was different from that of fibronectin (Fig. 3r). Laminin expressing cell numbers increased as the fibrotic scar matured and reached its peak density at 10 dpi and was stable through 14 dpi (p>0.05) suggesting that as the fibrotic scar grew, more laminin cells were recruited to form the ONC scar. Fibronectin and laminin expressing cells demonstrated 80% co-localization with Col1α1⁺-GFP cells at different time points as the scar formed, suggesting that ECM molecules play an important role in pericyte organization of the ONC scar structure (Fig. 3s–t).

Macrophages and microglia respond to optic injury and participate in ONC scar formation

We performed immunostaining on injured optic nerve sections using antibodies against CD68, ionized calcium-binding adaptor molecule 1 (Iba-1), CD11b and F4/80 at multiple timepoints after ONC to determine the role of macrophages and microglia in the formation of the ONC fibrotic scar. We compared the distribution of different macrophage and microglial cells at 10 dpi when the fibrotic scar was mature with fibrous encapsulation. CD68⁺cells accumulated at the ONC site acutely and reached a peak within the ONC fibrotic scar region at 10 dpi when the fibrotic scar was homogeneously filled with CD68⁺ cells (Fig. 4a–d). While some CD68⁺ cells were also present around the fibrotic scar, these cells did not co-localize with Col1α1⁺ cells. The expression of F4/80 was similar with CD68 (Supplemental Fig. 3). Iba-1⁺ cells were present around the ONC site and in the ON suggesting the presence of resident microglia. In contrast, Iba-1⁺ cells inside the fibrotic scar demonstrated a grid-like pattern and were not co-labeled with Col1α1⁺-GFP cells (Fig. 4e–h). A similar distribution pattern of CD11b expression was detected within and around the fibrotic scar without co-localizing with Col1α1⁺ cells (Fig.4 i–l). These results suggest that macrophages and microglia contribute to the formation of the optic nerve scar after ONC, probably from hematogenous circulation and migrate to the injury site, while some are the resident monocyte are localized at the ON. We hypothesize that both hematogenous and resident monocytes respond to the optic injury acutely and play a role in phagocytosis and fibrotic scar formation.

Fig. 4.

Distribution of monocytes in the mature fibrotic scar. a-d CD68⁺ cells accumulate at the ONC site acutely and reach a peak within the ONC fibrotic scar region by 10 dpi, some CD68⁺ cells are also expressed around the fibrotic scar, Col1α1⁺ cells do not co-express CD68. e-h Iba-1⁺ cells are detected around the ONC crush site and in the ON, Iba-1⁺ cells inside the fibrotic scar were detected as grid-like structures and were not co-labeled with Col1α1⁺-GFP cells, i-1 CD11b was detected within and around the fibrotic scar with no co-localizing presence of Col1α1⁺ cells. Scale bar, 200 μm.

Col1α1⁺ cells density decreases in the retina after ONC

To localize Col1α1⁺ cells in the retina, we performed CSLO with a green filter to detect GFP expression in the retina of Col1α1-GFP mice. CSLO retinal imaging demonstrates the presence of Col1α1⁺ cells along large diameter retinal blood vessels, whereas Col1α1⁺ cells were not detectable along microvessels (Fig. 5a). To further identify the cellular identity of Col1α1⁺ cells, we performed immunostaining on whole-mount retina using antibodies against PDGFR-ß, and CD13. GFP⁺ cells were detected along the large diameter retinal blood vessels and this expression pattern was consistent with CSLO retinal in vivo imaging for Col1α1⁺-GFP cells (Fig. 5a). Most GFP⁺ cells along the retinal vessels were co-labeled with anti-PDGFR-ß and anti-CD13 antibodies suggesting a perivascular origin (Fig. 5a).

Fig. 5.

Distribution of Col1α1⁺ cells in the retina and their change following ONC injury. a Col1α1⁺ cells were expressed mostly along the large-diameter retinal vessels based upon in vivo CSLO imaging and whole-mount retina laser confocal images. Most Col1α1⁺ cells along the retinal vessel were co-labeled with anti-PDGFR-ß and anti-CD13 antibodies. b Time course of Col1α1⁺ cell changes following ONC injury in vivo and whole-mount images. c Quantification of Col1α1⁺ cell density changes after ONC. Both veins and arteries demonstrate a significant decrease of Col1α1⁺ cells observed from early ONC scar formation (3 dpi and 7 dpi) to 10 dpi and 14 dpi. d Quantification of Col1α1⁺ cell density changes in the control eye. No difference was observed in the Col1α1⁺ retinal cell density.

We additionally observed the change of Col1α1⁺ cell retinal density in vivo using CSLO and calculated the portion of Col1α1⁺ cells over time accumulating after ONC injury. We performed ONC on one eye of the Col1α1-mice while using the contralateral eye as a control. In the control eye, no visible difference in Col1α1⁺ cells in the retina was observed over time after ONC injury. However, a decrease in Col1α1⁺ cell density and signal strength were observed in the ONC eye based on quantitative in vivo imaging. In the whole-mount retina, only a few GFP-expressing cells were detected at 10 dpi and 14 dpi compared with the baseline image and early time points after ONC (Fig. 5b). The expression of CD13 and PDGFR-ß didn’t show significant decrease after ONC injury (Supplemental Fig. 4). Quantitative analysis showed a significantly lower number of Col1α1⁺ cells along both veins and arteries after early ONC scar formation (3 dpi and 7 dpi) compared to 10 dpi and 14 dpi (Fig. 5c), while the number of Col1α1⁺did not change in the control eye (Fig. 5d). We hypothesize that a population of the Col1α1⁺-GFP cells which migrate to the ONC injury site may originate from retinal vessels.

Discussion

Glial and fibrotic cellular components form the scar present in CNS injury sites, inhibit axonal regrowth, and lead to severe and irreversible functional deficits. Compared with the glial scar, the fibrotic component of the CNS scar has received much less attention and the fibrotic scar in optic nerve crush has been observed but poorly characterized[21–23]. In the SCI model, studies have shown that perivascular collagen 1α1 cells are a major component of the cellular constituents of the fibrotic scar [17,24]. However, no study has focused on the molecular and cellular changes at the optic nerve injury site. Both the spinal cord and optic nerve are part of the CNS. We use a well characterized transgenic mouse model used in SCI studies, in which GFP is expressed under the control of the collagen 1α1 promoter, in an ONC model to investigate the role of perivascular fibroblasts in optic nerve injury and characterize the cellular components associated with optic nerve scar formation.

We firstly analyzed the formation of the glial scar at the lesion site following ONC using antibodies against GFAP, GS, and vimentin. As is shown in the present study and our previous study [19], when the mature ONC scar is formed, the glial scar encircles the fibrotic scar with no axons passing through the scar area. For the glial scar, a major rearrangement of its anatomic structure and cellular organization occurs after ONC injury. Resident astrocytes are activated with increased GFAP and vimentin expression along with the upregulation of GS. Both GFAP and vimentin are important intermediate filament proteins in reactive astrocytes, which are highly dynamic molecules that play key roles in cell signaling and cell migration[25,26]. Ten days after ONC injury, GFAP and vimentin-expressing cells form a glial scar that surrounds the fibrotic scar (Fig. 1, Supplemental Fig. 2). These results are in consistent with previous studies that glial scar formation was impaired after brain and SCI in mice lacking both GFAP and vimentin in astrocyte [27]. These data demonstrate the important role of GFAP and vimentin in the formation of nervous system scar tissue. In addition, vimentin also directly participates in fibroblast proliferation and collagen accumulation, and the loss of vimentin expression is associated with incomplete wound healing[28]. This may explain our observation that vimentin expressing cells are present at the center of the fibrotic scar. Our findings suggest that GFAP and vimentin play important roles in the development of the glial scar, while vimentin may also participate in the formation of the fibrotic scar. However, others have shown that GFAP^−/−Virri^−/− mice have reduced retinal degeneration [29] and improved recovery following SCI[30]. These findings suggest that in some pathophysiological processes or specific organs, the regenerative potential might correlate with the benefits of reactive gliosis, which provides a potential target for the treatment of CNS injury.

Oligodendrocytes play an important role in the metabolic support of axons in the CNS. Our data show an upregulation of GS in ON after ONC, which suggests a higher metabolic role from astrocytes. GS participates in the glutamate-glutamine cycle in the CNS to minimize glutamate neurotoxicity due to the excessive accumulation of excitotoxic glutamate[31]. Previous studies showed a significant decrease in brain GS immunoreactivity after fluid percussion injury in rats[32]. Others have shown that inhibition of GS activity leads to increased neuronal cell death[33]. The reasons for the difference in GS expression remain unclear but may involve variable metabolic status in different organs and disease processes. In the ONC model we observed increasing GS expression with glial scar formation, which provides indirect evidence that reactive gliosis and glial scar formation may be associated with metabolic environmental reactive change.

Our data suggest that following ONC, Col1α1⁺ cells play a critical role in forming the fibrotic scar at the ONC trauma site. Interestingly, Col1α1⁺ cells take approximately 10 days to migrate, accumulate and alter their distribution pattern at the ON crush site with the formation of an obvious encapsulated scar with a dense border present at the ONC site. The density of Col1α1⁺ cells reaches its peak at 10 dpi – a critical time point when the fibrotic scar forms a distinct boundary with the glial scar, which is earlier than that formed in the SCI model. Previous studies reported that Col1α1⁺ cells form the fibrotic scar in 2 weeks in SCI [17,24]. Possible reasons to explain this difference include different observation time points and differences between the spinal cord and the optic nerve. The richer blood supply and smaller cross-section of the ON make it easier for reactive cells to accumulate at the ONC lesion site. These findings suggest that although ON and spinal cord are parts of the CNS, the reactive time for injury response may be different based on anatomical differences.

In previous reports, researchers identified type A pericytes as the major source of PDGFR-ß and CD13 expressing fibroblasts following SCI[24]. We found the predominant accumulation of Col1α1⁺ cells were co-localized with PDGFR-ß around the ONC scar, and about one-third of Col1α1⁺ expressed CD13 and αSMA in the ONC site. These results suggest that different subsets of pericytes may contribute to the formation of fibrotic scar in ONC compared with SCI. Previous studies using the same transgenic Col1α1-GFP mouse line showed that GFP⁺pericytes expressed αSMA in subretinal fibrosis after photocoagulative damage[34], and similar results were observed in renal fibrosis[35]. In an SCI model, GFP⁺pericytes form the fibrotic scar without αSMA expression[17]. These findings suggest that Col1α1⁺ cell is the marker for different types of pericytes among different tissues, or the reactive program regarding the injury is distinct within different organs or tissues.

Similar to other organs, the fibrotic response following CNS injury has been described by the migration of fibroblasts to the lesion site and a deposition of ECM molecules that assemble into scar connective tissue[24]. Our results show fibronectin appears at the ONC crush site by 3 dpi - as Col1α1⁺ cells migrate to form the structural scaffold for fibrotic scarring. In the mature fibrotic scar, co-localization of Col1α1⁺ cells and fibronectin suggest Col1α1⁺ cells may produce fibronectin at the ONC site. Previous studies demonstrate Col1α1⁺ cells are a source of fibronectin [36,34]. Fibronectin participates in many cellular processes including cellular migration, proliferation, and differentiation [37,38]. These observations suggest that like pericytes, ECM molecules may play a vital role in fibrotic scar formation after ONC. In addition, removing or modifying the ECM molecules that accumulate at the injury site has the potential to promote axonal regeneration[39–41]. As fibronectin is a growth promoting substrate for axons, fibronectin may be a therapeutic target for promoting axonal re-growth. However, a large number of ECM proteins contribute to the fibrotic scar with complex functions which makes it difficult to distinguish among the key promoting or inhibitory factors [42].

In our study, a strong and prompt inflammatory response after ONC induced a high-density accumulation of CD68, Iba-1, and CD11b expressing cells at the ONC site. These findings are consistent with previous studies demonstrating reactive gliosis with an accumulation of inflammatory monocytes at the optic nerve injury site [43–45]. Migration of macrophages and microglia to the fibrotic scar suggest their participation in scar formation, while the accumulation of these cells over time after ONC suggests they could have a potential function in the maintenance of the ONC scar. Decreased the number of infiltrating inflammatory monocytes reduces chronic fibrotic scar formation in SCI and improves functional recovery[46,47] suggesting these monocytes may act as a molecular target for therapeutics that antagonize fibrotic scar formation.

We describe the distribution of Col1α1⁺ cells in the retina and the optic nerve after ONC injury. Previous study demonstrated that Col1α1⁺ cells may reside along both large-diameter blood vessels and microvessels throughout the spinal cord[17]. However, our data from both in vivo CSLO retinal imaging and immunostaining on whole-mount retina showed Col1α1⁺ cells were only visible along large-diameter retinal vessels, thereby suggesting a different distribution of Col1α1⁺ cells among organ structures. In addition, both our in vivo images and immunostaining of whole-mount retina show a significant decrease in Col1α1⁺ cells in the retina after ONC injury. We speculate that some of the Col1α1⁺-GFP cells migrating to the ONC injury site may originate from retinal vessels. However, further experiments are needed to trace the origin of Col1α1⁺ cells at the ONC site.

We show with Col1α1-GFP mice that Col1α1⁺ cells migrate acutely to the optic nerve injury site and are a major component of the fibrotic scar, which forms following ONC. We describe the upregulated expression of ECM components and a high percentage of infiltrating pericytes co-localizing with Col1α1⁺ cells at the injury site as major contributors to ECM in the ONC fibrotic scar, which we show is an encapsulated discrete structure within the optic nerve. Multiple monocytes are present at the ONC fibrotic scar, including hematogenous and resident macrophage/microglia responding to optic nerve injury acutely, and which play an important structural role in fibrotic scar formation. Col1α1⁺ cells in the retina concomitantly decrease in density following ONC, suggesting that some of the Col1α1⁺ cells at the fibrotic scar may have migrated from the retina to the site of optic nerve injury.

In conclusion, we demonstrate that Col1α1⁺ cells of pericyte origin migrate to the optic nerve injury site and aid in the formation of fibrotic scar. This fibrotic scar, different and uniquely characterized in this paper for the optic nerve compared to what is known for the spinal cord, is composed of multiple cell types which form an encapsulated scar previously not recognized. Reduction of specific extracellular matrix molecules or fibrosis forming cell types may be the potential therapeutic target for axonal regeneration after optic nerve injury since axons do not have the capacity to transit through this newly characterized fibrotic scar.

Supplementary Material

Supplemental Figure 1

Supplemental Fig. 1 Time course of GS distribution throughout the ON following ONC injury. GS negative area was marked with blue dotted line. The fibrotic scar was marked with yellow dotted line, d-i At 3 dpi and 7 dpi, a large area with absence of GS expression was observed with a faint border (marked with blue dotted lines), while accumulation of Col1α1⁺-GFP expressing cells were present in the ONC crush site, j-o By 10 dpi and 14 dpi, a clearly demarcated GS deficient area was observed at the ONC lesion site with fibrotic components expressed within the area. Scale bar, 200 μm.

Supplemental Figure 2

Supplemental Fig. 2 Time course of vimentin distribution throughout the ON following ONC injury. Lesion area stained by vimentin was marked with blue dotted line, the fibrotic scar was marked with white dotted line, a-c In the uninjured optic nerve, an even distribution of fibrillary patterned vimentin was present throughout the ON. d-f At 3 dpi, a faint border of vimentin deficient area was observed (marked with blue dotted line) with residual vimentin expression within the lesion area, while infiltrating Col1α1⁺-GFP positive cells began to appear in the scar center, g-i At 7 dpi, a number of vimentin expressing cells together with Col1α1⁺-GFP cells were observed at the center of the lesion site, but vimentin and Col1α1⁺-GFP expressing cells are entirely not co-localized. j-o By 10 dpi and 14 dpi, as the Col1α1⁺-GFP cells form an encapsulated fibrotic scar (k, n), a cluster of high density vimentin expressing cells are present at the fibrotic scar area. Scale bar, 200 μm.

Supplemental Figure 3

Supplemental Fig. 3 Distribution of F4/80 in the mature fibrotic scar. F4/80⁺ cells accumulate at the ONC site acutely and reach a peak within the ONC fibrotic scar region by 10 dpi, some F4/80⁺ cells are also expressed around the fibrotic scar, Col1α1⁺ cells do not co-express F4/80.

Supplemental Figure 4

Supplemental Fig. 4 Time course of Col1α1⁺-GFP (green), CD13 (red) and PDGFR-ß (magenta) expression in the whole-mount retina images, a-e Col1α1⁺-GFP expressing cells decreased significantly between 10dpi to 14dpi. CD13 (f-j) and PDGFR-ß expressing cells (k-o) did not show significant changes following ONC injury.