Effect of Brain-Derived Neurotrophic Factor on Mouse Axotomized Retinal Ganglion Cells and Phagocytic Microglia

Abstract

Purpose.: To assess the effect of a single intravitreal injection of brain-derived neurotrophic factor (BDNF) on the survival of mouse retinal ganglion cells (RGCs) and on phagocytic microglia after intraorbital optic nerve transection (IONT).

Methods.: One week before IONT or processing, RGCs from pigmented C57/BL6 and albino Swiss mice were traced by applying hydroxystilbamidine methanesulfonate (OHSt) to both superior colliculi. Right afterward unilateral IONT, BDNF or vehicle were intravitreally administered. At increasing time intervals postlesion retinas were dissected as flat-mounts and subjected to BRN3A and Iba1 immunodetection. BRN3A⁺RGCs were automatically quantified in all retinas and their distribution was assessed using isodensity maps. In all retinas, the Iba1-positive and OHSt-filled microglial cells present in the ganglion cell layer were manually quantified. Their distribution was observed by neighbor maps.

Results.: When vehicle was administered, IONT-induced RGC death was significant at 3 days, while BDNF treatment delayed it to 5 days. At 14 days after BDNF or vehicle injection, 45% and 18% of RGCs had survived, respectively. There was a significant increase in OHSt-filled microglial cells in the right (contralateral) retinas after both treatments, without concurring with quantifiable RGC death. In the injured eye, the number of OHSt-filled microglial cells increased as the population of RGCs decreased and spread from central to peripheral areas.

Conclusions.: In axotomized mouse retinas, a single intravitreal injection of BDNF protects RGCs throughout the whole retina. There is a strong contralateral response that involves microglial activation and OHSt phagocytosis.

Introduction

Axonal damage to central nervous system (CNS) neurons leads to their degeneration and death causing an irreversible loss of function. The retina is probably the most extensively studied part of the CNS in mammals. As a result, the visual system of rodents is widely used in developmental studies ^1,2 as a model of traumatic injury to the CNS, ^3–6 regeneration, ^7,8 transient ischemia, ^9,10 glaucoma, ^11–13 and retinal degeneration. ^14,15  

Retinal ganglion cells (RGCs) are the only output neurons of the retina. Since their axons form the optic nerve, damage to the optic nerve primarily affects RGCs. These cells are found in the innermost layer of the retina, sharing a location with the equally numerous population of displaced amacrine cells. ^16–18 In order to investigate the response of RGCs to trauma or disease as well as elucidating the effect of neuroprotective treatments, it is essential to clearly identify them. RGCs can be either traced from their retinorecipient targets in the brain, which in mice and rats are mainly the superior colliculi ^16,19–22 or by immunodetection of RGC-specific proteins such as BRN3A, ^4,23 or by in situ hybridization of RGC-specific mRNAs such as γ-synuclein. ^24–26 Tracers will identify the RGCs with a competent axon upon tracing, while BRN3A or γ-synuclein detection will identify living RGCs. 

Microglial cells are the resident macrophages in the CNS that originate from hemopoietic cells. ²⁷ During neurodegeneration, activated microglial cells engulf cell carcasses and debris to eliminate them from the tissue. ²⁸ When the dead cells are labeled with exogenous compounds, these are incorporated into the microglial phagosomes, resulting in the transcellular labeling of the phagocytotic microglial cell. ^29–35 Therefore, when traced-RGCs are phagocytosed, the phagocytotic microglial cells in the retina can be identified by their amoeboid morphology, ^35,36 by the expression of specific markers such as Iba1 ³⁷ and by transcellular labeling. 

Since the late 1990s it has been known that after unilateral optic nerve injury in rats, there is a microglial response in the eye contralateral to the lesion. ³⁸ This response has also been reported in rodent models of glaucoma ^39,40 and transient ischemia. ⁴¹ In the injured retina, the microglial response increases as the RGCs die ^36,42 and, in rats, this response is attenuated by neuroprotective trophic factors. ³⁶  

The temporal course of RGC death following IONT in mice and rats has been extensively studied by our group. ^4,23,30,43 This acute insult causes a significant RGC loss at 3 days in both species. At 14 days only 12% to 15% of the original population of RGCs survive. Extensive work has been devoted to rescue axotomized RGCs. ⁴⁴ In rats, a single intravitreal injection of brain-derived neurotrophic factor (BDNF) protects the whole RGC population up to 7 days after IONT. ^30,45–47 However, to the best of our knowledge, there are no reports of the effect of BDNF on RGC survival after axotomy in mice or its effect on phagocytic microglia. 

In this study, two commonly used mouse strains–an albino (Swiss) and a pigmented (C57/BL6) strain–were employed to investigate the effect of a single intravitreal injection of BDNF on the survival of RGCs, and to assess the appearance of transcellularly labeled microglial cells (phagocytic) in the ganglion cell layer of the injured and uninjured contralateral eyes after unilateral IONT. 

Materials and Methods

Anesthesia and Analgesia

Adult female albino Swiss (30–35 g body weight) and pigmented C57/BL6 (20–25 g body weight) mice were obtained from the University of Murcia breeding colony. All animals were treated in accordance with the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. 

For anesthesia, a mixture of xylazine (10 mg/kg body weight) (Rompun; Bayer, Kiel, Germany) and ketamine (60 mg/kg body weight) (Ketolar; Pfizer, Alcobendas, Madrid, Spain) was used intraperitoneally (IP). After surgery, an ointment containing tobramycin (Tobrex; Alcon S.A., Barcelona, Spain) was applied on the cornea to prevent desiccation. Mice were given oral analgesia (buprenorphine 0.3 mg/mL) (Buprex; Schering-Plough, Madrid, Spain) at 0.8 mg/kg (prepared in strawberry-flavored gelatin) the day of the surgery and for the next 3 days. 

All animals were sacrificed with an IP injection of an overdose of pentobarbital (Dolethal; Vetoquinol, Especialidades Veterinarias, S.A., Alcobendas, Madrid, Spain). 

Surgery

Retinal Ganglion Cell Tracing.

Hydroxystilbamidine methanesulfonate (OHSt) (Molecular Probes, Leiden, The Netherlands) diluted at 10% in 0.9% NaCl and 10% dimethylsulfoxide was applied to both superior colliculi (SCi) 1 week before surgery or 10 days before processing (intact animals), as previously described. ^21,48,49 In brief, after exposing the midbrain, a small pledge of gel foam (Espongostan Film; Ferrosan A/S, Soeborg, Denmark) soaked in OHSt was applied over the entire surface of both SCi. 

Intraorbital Nerve Transection.

The left optic nerve was severed at 0.5 mm from the optic disc while sparing the blood supply, according to standard procedures in our laboratory. ⁴ After surgery, the eye fundus of each animal was checked to verify that retinal vessels were intact. 

Brain-Derived Neurotrophic Factor Delivery.

After the IONT procedure, a group of pigmented (n = 30) and albino (n = 14) mice received in the left eye an intravitreal injection of 2.5 μL of BDNF (Peprotech Laboratories, London, UK) diluted at 1 μg/μL in 1% bovine serum albumin-phosphate buffer saline. A second group of pigmented (n = 29) and albino (n = 13) mice received an intravitreal injection of 2.5 μL of vehicle. 

Animals were sacrificed at increasing time intervals postlesion (see Results). 

In both groups, the injured retinas were the left ones, while the right retinas were the contralateral ones. Since we observed a microglial response in the contralateral retinas, the RGC population and the appearance of phagocytic microglial cells were also analyzed in intact retinas dissected 10 days after retrograde tracing with OHSt (6 or 12 retinas from pigmented or albino strain, respectively). These intact retinas were used as controls. 

Immunohistofluorescence

After euthanasia, all animals were perfused transcardially with 4% paraformaldehyde in 0.1 M phosphate buffer after a saline rinse. 

Retinas were dissected as flattened whole-mounts by the method described in previous studies. ^4,21,22 Then, all retinas were permeated in PBS 0.5% Triton X100 by freezing them at −70°C for 15 minutes, rinsed in new PBS 0.5% Triton X100 and incubated overnight at 4°C with goat anti-BRN3A (C-20; Santa-Cruz Biotechnology, Heidelberg, Germany) diluted 1:500 in blocking buffer (PBS, 2% bovine serum albumin, 2% Triton X100). Afterward, retinas were washed three times in PBS and incubated at room temperature for 2 hours with donkey antigoat dye (Dylight 594; Jackson ImmunoResearch, Newmarket, Suffolk, UK) (1:500 dilution in PBS 0.5% Triton X100). In 4 retinas per group, Iba1 was detected as well using rabbit anti-Iba1 antibody (1:500) (Dako; Rafer, Zaragoza, Spain), which was developed with donkey antirabbit dye (1:500) (Dylight 468; Jackson ImmunoResearch). Finally, the retinas were thoroughly washed in PBS and mounted vitreal side up on subbed slides and covered with antifading solution. Unless otherwise stated, all reagents were from Sigma-Aldrich Química S.A. (Madrid, Spain). 

Image Acquisition

Whole mounted retinas were analyzed for OHSt, BRN3A, and Iba1 signals. To reconstruct retinal whole-mounts, retinal photographs were taken under an epifluorescence microscope (Axioscop 2 Plus; Zeiss Mikroskopie, Jena, Germany) equipped with a computer-driven motorized stage (ProScan H128 Series; Prior Scientific Instruments, Cambridge, UK) controlled by image analysis software (Image-Pro Plus, IPP 5.1 for Windows; Media Cybernetics, Silver Spring, MD) by the method described in previous studies. ^4,21 Reconstructed whole-mounts made up of 140 individual frames were further processed using a graphics editing program, when required (Adobe Photoshop CS 8.0.1; Adobe Systems, Inc., San Jose, CA). 

Automated Quantification of the Total RGC Population

OHSt⁺ and BRN3A⁺RGCs were automatically quantified by the method described in previous studies. ^4,21 Briefly, the individual fluorescent images taken of each retinal whole-mount were processed by a specific cell-counting subroutine using the IPP macro language. A sequence of filters and transformations was applied to each image to identify cell limits and separate individual cells for automated cell counting. Quantitative data were exported to a spreadsheet application (Microsoft Office Excel 2003; Microsoft Corporation, Redmond, WA) for further analyses. 

BRN3A⁺RGCs were automatically quantified in all the retinas, while OHSt⁺RGCs were automatically quantified only in the control intact eyes, because at 3 days post-IONT the appearance of OHSt-filled microglial cells in the retina impaired the automated routine. Thus, traced-RGCs were manually counted in 12 retinal locations and their total number was inferred as previously described. ⁴  

RGC Spatial Distribution

The detailed spatial distribution of BRN3A⁺RGCs in all retinas was obtained through quadrant analysis and demonstrated on isodensity maps constructed as previously described. ^4,21  

Quantification of OHSt-Filled Microglia

The OHSt-filled microglial cells (i.e., phagocytic) present in the ganglion cell layer of 6 (pigmented) and 12 (albino) intact retinas and 8 experimental retinas (4 right and 4 left) per strain, group, and time point, were identified by their morphology ^30,36,50 and their expression of Iba1 (see Fig. 5A). These cells were dotted on the retinal photomontage by an experienced researcher using a graphics editing program (Adobe Photoshop CS 8.0.1; Adobe Systems, Inc.). An automatic IPP routine was developed to quantify the total number of dots in each flat-mount: first, the user was asked to mark the ON as a reference point in the retina and draw the outline of the retina to measure its total area. Then, we calculated the coordinates of the center of mass, (x, y) of each dot. Finally, all data, including the spatial coordinates of the ON, the number of dots, and the coordinates of their center of mass, were displayed and exported to a spreadsheet by dynamic data exchange (Microsoft Office Excel 2000; Microsoft Corporation) for further analysis. 

Spatial Distribution of OHSt-Filled Microglial Cells: Neighborhood Maps

To study the spatial distribution of the OHSt-filled microglial cells a new Java application (Oracle Corporation, Redwood Shores, CA) was developed. This routine was based on the k-nearest neighbor algorithm and the fixed-radius method. In a first step, the user set the radius of the study to r = 0.1023 mm and imported the spreadsheet with all the data previously recorded. Then, we converted all cell coordinates into the ON system. Next, we calculated the number of neighbors of each cell by measuring their Euclidean distance to the rest of cells. The cells within the fixed radius were counted as neighbor cells. Finally, the location of each cell with respect to the ON and their number of neighbors were recorded in a plain text file. Spatial information was used to plot every cell in the retina. Each cell was colored according to the number of neighbor cells around it. The scale ranged from purple (0–2 neighbors) to red (≥21 neighbors). Therefore, the warmer the color the more microglial cells in a given retinal location. All plots were performed using a software package for scientific graphing and data analysis (SigmaPlot 9.0 for Windows; Systat Software, Inc., Richmond, CA). 

Statistical Analysis

To compare values for retinas of different groups, the number of OHSt or BRN3A positive RGCs and the number of OHSt-filled microglial cells we used a statistical software package (SigmaStat for Windows Version 3.11 program; Systat Software, Inc., Richmond, CA). Differences were considered significant when P < 0.05. Test results are detailed in the Results section. Data are presented as mean ± standard deviation. 

Results

Response of Mouse Retinal Ganglion Cells to Axotomy and Vehicle or BDNF Treatment

After optic nerve axotomy, RGCs undergo progressive degeneration (Figs. 1A–C). This loss is delayed by BDNF treatment (compare left versus right columns in Figs. 1B, 1C). 

Figure 1. 

 

RGC loss in albino and pigmented mice after IONT and vehicle or BDNF treatment. Magnifications from flat mounted retinas showing BRN3A⁺ and OHTs⁺ RGCs in intact (A) and injured retinas (B, C) at increasing times post-IONT. (B) Pigmented; (C) albino. Arrows point to OHSt-filled microglial cells, which do not express BRN3A. Scale bar: 100 μm.

The presence of OHSt-filled microglial cells in the injured retinas and the contralateral retinas (see below) hindered the automated counting routine of OHSt⁺RGCs. Consequently, traced-RGCs were manually quantified in the injured and contralateral retinas, and their number was inferred as previously published. ⁴ The number of OHSt⁺RGCs in intact retinas and of BRN3A⁺RGCs in all retinas was automatically quantified. 

The number of BRN3A⁺RGCs in the intact retinas of the pigmented mice amounted to 83.4% of the RGCs traced from the SCi (Table 1A). This value is slightly lower, but not significantly different than the 85.5% reported in our previous work. ⁴ In the albino strain, the population of RGCs was higher than that in the pigmented strain, which was in agreement with previous reports. ²¹ Interestingly, in the albino strain, BRN3A detects 92.6% of the traced RGCs (Table 1A). The numbers of OHSt⁺RGCs or BRN3A⁺RGCs in the contralateral retinas decreased with time postlesion although not significantly. Thus data from these retinas were averaged (Table 1B). However, because of this decrease, we used the number of RGCs in the intact retinas as the control for the RGC population. 

Table 1. 

 

Total Number and Density of Surviving RGCs in Albino and Pigmented Mice after IONT and Vehicle or BDNF Treatment

	Pigmented (C57/BL6), n = 6; Mean (SD)					Albino (Swiss), n = 12; Mean (SD)
A: Intact
BRN3A	33,769 (1,210)					47,211 (1,346)
OHSt	41,011 (2,174)					51,025 (1,425)
	Pigmented (C57/BL6); Mean (SD)					Albino (Swiss); Mean (SD)
	Contra Lateral	Injured				Contra Lateral	Injured
		3d	5d	7d	14d		3d	7d
B: Experimental
Total numbers
IONT+VEHI
BRN3A	31,889 (765)	25,725 (1,809)	14,413 (3,488)	9,404 (1,397)	5,588 (773)	45,900 (2,443)	40,903 (2,031)	15,073 (3,858)
OHSt	40,317 (2,656)	34,937 (2,589)	23,661 (1,874)	15,121 (3,255)	6,900 (1,158)	49,553 (1,730)	42,232 (3,250)	19,599 (6,066)
IONT+BDNF
BRN3A	32,019 (711)	29,813 (1,805)	24,864 (4,129)	19,750 (3,192)	14,084 (1,870)	44,414 (1,092)	46,697 (1,846)	26,097 (3,392)
OHSt	40,274 (2,477)	40,222 (1,160)	30,496 (3,001)	26,981 (1,543)	18,315 (1,226)	49,182 (982)	49,972 (1,511)	30,364 (3,687)
Area (mm2)
IONT+VEHI	14.5 (0.3)	13.6 (1.6)	13.9 (0.5)	14.5 (1.1)	15.0 (1.0)	12.7 (0.7)	12.6 (1.3)	13.9 (1.1)
IONT+BDNF	14.1 (0.3)	13.3 (0.5)	13.6 (0.4)	14.3 (0.9)	15.0 (0.8)	14.4 (0.8)	13.5 (0.6)	13.1 (1.0)
Density (RGCs/mm2)
IONT+VEHI
BRN3A	2,193 (63)	1,916 (213)	1,043 (267)	655 (134)	375 (58)	3,625 (396)	3,273 (436)	1,102 (326)
OHSt	2,797 (172)	2,620 (446)	1,708 (150)	1,038 (192)	585 (48)	3,913 (361)	3,418 (322)	1,128 (556)
IONT+BDNF
BRN3A	2,255 (81)	2,240 (203)	1,835 (349)	1,381 (229)	938 (131)	3,115 (82)	3,458 (211)	2,000 (331)
OHSt	2,838 (86)	2,594 (1146)	2,245 (249)	1,888 (164)	1,222 (124)	3,444 (114)	3,702 (242)	2,317 (304)
Number of analyzed retinas
IONT + VEHI	29	8	6	7	8	13	8	6
IONT + BDNF	30	7	7	7	8	14	7	7

 The whole population of BRN3A⁺RGCs was automatically quantified in all retinas (intact, contralateral, and experimental). OHSt-traced RGCs in the intact retinas were automatically quantified and their number in the contralateral and experimental retinas inferred from the manual counting (see Methods for further explanation).

 (A) Number of RGCs in intact retinas. (B) Number and density of RGCs in the experimental groups. Compared to intact retinas, RGC loss became significant at 3 dpl in the vehicle treated animals (Mann-Whitney test, P < 0.001), and at 5 dpl in the BDNF-group (Mann-Whitney test, P < 0.001). After IONT the number of RGCs in both strains with both markers at all time points was significantly higher in the BDNF than in the vehicle-treated group (Mann-Whitney test, P < 0.001). d, days post-IONT; SD, standard deviation.

The number of RGCs was significantly higher in the BDNF than in the vehicle-treated group at all post-IONT times, in both mice strains, and with both markers (Table 1B). In fact, the number of RGCs in the BDNF group did not differ from that of the control group at 3 days postlesion (dpl), while it was significantly lower in the vehicle group. 

Another observation was that, regardless of the treatment, there was a significant decrease in the RGC population from 3 to 5, 5 to 7, and 7 to 14 dpl (P < 0.001). However, the RGC loss was more dramatic in the vehicle-treated retinas, as compared to that in the BDNF-treated retinas (Figs. 2A, 2B). As shown in the figures, the slope of the regression line between the surviving RGCs and postlesion time is greater in absolute terms (with a minus sign as the RGC population decreases with time) in the vehicle than in the BDNF group. This is observed both in the OHSt and in the BRN3A quantification. Moreover, Figures 2A and 2B show that after IONT+vehicle treatment, there was a daily loss of 2545 OHSt⁺RGCs and 1907 BRN3A⁺RGCs, while after BDNF treatment this loss was lower amounting to 1697 OHSt⁺RGCs and 1352 BRN3A⁺RGCs. This means that, regardless of the method used to identify the RGCs, survival is 1.5 times higher after treatment with BDNF than after treatment with vehicle. 

Figure 2. 

 

BDNF delays axotomy-induced RGC death in mouse retinas. (A, B) The temporal RGC loss after IONT and vehicle or BDNF treatment adjusts to an order 1 regression line (95% CI) for OHSt (A) and BRN3A (B). The slope (m) and the correlation coefficient (R ²) of each straight line are shown. Data are taken from the experiment on pigmented mice. (C, D) Histograms showing the percentage of surviving OHSt-traced or BRN3A⁺ RGCs after IONT and BDNF or vehicle treatment in pigmented (C) and albino (D) mice. The percentages were calculated considering 100% of the number of OHSt⁺ or BRN3A⁺ RGCs in the intact retinas. Notice that the percentage of RGC survival is similar for both markers, except at 5 dpl after IONT+vehicle ([C], asterisk), when the percentage of BRN3A⁺RGCs is significantly smaller than that of OHSt⁺RGCs (ANOVA, Bonferoni t-test, P = 0.023). C, control retinas.

However, BRN3A⁺RGC loss was lower than OHSt⁺RGC loss since the total number of BRN3A⁺RGCs was lower than the total number of OHSt⁺RGCs. The survival rates after BDNF and vehicle application showed that the RGC loss with respect to the original population was similar for both markers at all times, except at 5 dpl after vehicle application, when the percentage of BRN3A⁺RGC loss was significantly greater than that of OHSt⁺RGC loss (Figs. 2C, 2D). Finally, the percentage of RGC loss was similar for both mice strains, indicating that eye pigmentation does not have any effect after this type of lesion and BDNF treatment. 

Spatial RGC Loss after IONT and Vehicle or BDNF Treatment

Figure 3 shows the isodensity maps illustrating the distribution of OHSt⁺RGCs and BRN3A⁺RGCs in intact retinas from both mice strains (Figs. 3A, 3D) and the distribution of the surviving BRN3A⁺RGCs after IONT and BDNF (Figs. 3B, 3E) or vehicle (Figs. 3C, 3F) treatment. In the injured retinas and irrespective of the treatment, the RGC loss was diffuse and affected the whole retina, as the color distribution changes homogenously from warm (yellow, orange, red: higher densities) to cold (purple, blue, green: lower densities). In agreement with the quantitative data obtained, the distribution of BRN3A⁺RGCs at 3 dpl after BDNF administration is similar to that found in intact retinas, while after vehicle administration there was a substantial decrease in these neurons (Figs. 3A, 3B, 3D, 3E). Although the protection provided to RGCs by BDNF is transitory, RGC density was higher at all postlesion times after the BDNF application, as compared to that after vehicle treatment (Fig. 3: compare maps in 3B with those in 3C, and maps in 3E with those in 3F). 

Figure 3. 

 

After IONT, a single injection of BDNF protects RGCs throughout the retina. (A, D) Flat-mounted intact retina from a pigmented (A) and an albino (D) mouse showing OHSt-traced (first image) and BRN3A⁺ (third image) RGCs and their corresponding isodensity maps (second and fourth images, respectively). (B, E) Isodensity maps from pigmented (B) and albino (E) mice retinas showing the distribution of the surviving BRN3A⁺RGCs after BDNF treatment at increasing time intervals post-IONT. (C, F) Isodensity maps from pigmented (C) and albino (F) mice retinas showing the distribution of the surviving BRN3A⁺RGCs after vehicle treatment at increasing interval times post-IONT. The number of BRN3A⁺RGCs counted in the retina from which the maps have been generated is shown at the bottom of each map. Isodensity maps are filled contour plots generated by assigning to each of the 25 subdivisions of each individual frame a color code according to its RGC density within a 28-step scale ranging from 0 to 500 (purple) to 4800 (pigmented) and 5700 (albino) or more (red) RGCs/mm² (scale in [A] bottom right). S, superior; N, nasal; T, temporal; I, inferior. Scale bar: 1 mm.

Number of OHSt-Filled Microglial Cells in Control, Injured, and Contralateral Retinas

After BDNF/vehicle treatment, OHSt-filled microglial cells were counted in the injured and in the contralateral retinas at different postlesion times in both mice strains (Table 2). In the contralateral retinas, the appearance of OHSt-filled microglial cells was observed as early as at 3 dpl (Table 2). The reason for such early response in the right retina might have been that the appearance of OHSt-filled microglial cells was a normal reaction to OHSt-tracing, rather than a response to the contralateral injury. To further explore this issue, we quantified the number of OHSt-filled microglial cells (Table 2) and RGCs (Table 1A) in retinas from intact eyes processed 10 days after OHSt-tracing (7 days of tracing + 3 days, to make this group comparable to the 3 days IONT group). These data show that in the intact eyes the number of OHSt-filled microglial cells was significantly smaller than in the contralateral retinas at any time point. This suggested that the appearance of OHSt-filled microglial cells in the fellow retinas of axotomized retinas occurred as a consequence of the contralateral injury. Their number remained quite constant in both treatments and both strains, and did not vary significantly with time, though at 7 dpl their population was higher (Table 2, Fig. 4A). 

Figure 4. 

 

As the population of RGCs decreases, the number of OHSt-filled microglial cells increases. (A) Percentage of OHSt-filled microglial cells in contralateral (right) and injured (left) retinas after vehicle or BDNF treatment with respect to intact untouched retinas, which were considered 100%. (B) The temporal increase in OHSt-filled microglial cells after IONT and vehicle or BDNF treatment in the injured retina adjusts to an order 1 regression line (95% CI). The slope (m) and the correlation coefficient (R ²) of each straight line are shown. (C, D) Correlation between the temporal loss of OHSt⁺RGCs (C) or BRN3A⁺RGCs (D) and the appearance of OHSt-filled microglial cells. Data are taken from the pigmented mice experiment. C, control intact retinas.

Figure 5. 

 

Retinal distribution of OHSt-filled microglial cells. (A) Magnifications from a flat-mounted retina analyzed 7 days after IONT+vehicle shown from left to right: OHSt, Iba1, and BRN3A signal; the rightmost image corresponds to the merge image. The arrows point to some OHSt and Iba1 positive microglial cells. Scale bar: 50 μm. (B) Neighbor map showing the distribution of OHSt-filled microglial cells in an intact retina. (C, D) Neighbor maps showing the distribution of OHSt-filled microglial cells in representative contralateral retinas from pigmented mice at 3 and 7 days post IONT+BDNF (C) or IONT+vehicle (D). (E, F) Neighbor maps showing the distribution of OHSt-filled microglial cells in the injured retinas of pigmented mice at increasing time intervals post- IONT+BDNF (E) or IONT+vehicle (F). (G, H) Neighbor maps showing the distribution of OHSt-filled microglial cells in the injured retinas of albino mice at 3 and 7 days post- IONT+BDNF (G) or IONT+vehicle (H). The number of OHSt-filled microglial cells counted in the retina from which the map has been generated is shown at the bottom-left of each map. Color scale ranges from 0 to 2 neighbors (purple) to 21 neighbors or more (red), so the warmer the color the more OHSt-filled microglial cells in a given location. S, superior; N, nasal; T, temporal; I, inferior. Scale bar: 1 mm (in [B] bottom right).

Table 2. 

 

Number of OHSt-Filled Microglial Cells in the Injured and Contralateral Retinas after Vehicle or BDNF Treatment

	Pigmented (C57/BL6); Mean (SD)					Albino (SWISS); Mean (SD)
	Intact	IONT+VEHI		IONT+BDNF		Intact	IONT+VEHI		IONT+BDNF
		Contralateral	Injured	Contralateral	Injured		Contralateral	Injured	Contralateral	Injured
Intact	39 (5)					69 (11)
3d		234 (51)	410 (71)	212 (96)	182 (14)		181 (34)	227 (45)	166 (64)	158 (20)
5d		248 (113)	1664 (168)	211 (53)	1001 (217)
7d		373 (157)	2275 (215)	369 (141)	1465 (149)		211 (99)	1491 (78)	270 (136)	1043 (326)
14d		218 (44)	2630 (196)	281 (113)	2334 (118)

 Data from intact retinas were obtained from animals analyzed 10 days after OHSt-tracing but otherwise untouched. Pigmented mice: the number of OHSt-filled microglial cells in the right contralateral retinas was significantly higher at all time points as compared to control intact retinas (ANOVA, Bonferoni t-test, P < 0.05). In these contralateral retinas there were no differences between vehicle- and BDNF- treated animals or between the different time points during each treatment. There was a significant increase in OHSt-filled microglial cells from 3 to 5 days and from 5 to 7 days in the injured retina in both groups (ANOVA, Bonferoni t-test, P < 0.001), and from 7 to 14 days in the BDNF-treated group (ANOVA, Bonferoni t-test, P = 0.007). The number of OHSt-filled microglial cells was significantly higher after vehicle than after BDNF treatment at all time points (P < 0.001). The same occurs in the albino mice, with the exception that there was no difference between both treatments in the injured eye at 3 dpl.

As expected, in the injured eyes there was an increase of OHSt-filled microglial cells as the survival interval increased and at all time points, the number of OHSt-filled microglial cells was significantly lower after BDNF treatment (Table 2, Fig. 4A). This is graphically observed in Figure 4B. In these regression plots of OHSt-filled microglial cells versus time postlesion is shown that the slope of the straight line was more pronounced after the vehicle than after the BDNF treatment. The slope also shows that, each day, 195 and 175 new OHSt-filled microglial cells appeared in the retina after vehicle and BDNF treatment, respectively. 

Furthermore, there was an inverse correlation between the number of OHSt⁺RGCs or BRN3A⁺RGCs and the number of OHSt-filled microglial cells (Figs. 4C, 4D). This correlation is slightly weaker in BRN3A⁺RGCs. One possible explanation for this finding is that BRN3A expression disappeared as soon as the RGC died, while the OHSt⁺RGC carcass only disappeared from the tissue when it was phagocytosed. In addition, at the same postlesion times the increase in OHSt-filled microglial cells and decrease in RGCs was more significant after vehicle than after BDNF treatment. In fact, BDNF-treated retinas showed the same pattern at 7 dpl than vehicle-treated retinas at 14 dpl. Finally, the slope of the straight lines in Figures 4C and 4D indicated a ratio of OHSt-filled microglial cell and OHSt⁺RGC of approximately 1:10 or 1:12 when BDNF and vehicle were administered, respectively. Again, while these values were slightly lower for BRN3A⁺RGCs, the differences between vehicle and BDNF treatment are maintained. 

Distribution of OHSt-Filled Microglial Cells

Figure 5A shows OHSt-filled and Iba1 positive microglial cells. In control intact retinas, there were a few OHSt-filled microglial cells scattered without apparent pattern (Fig. 5B). Figures 5C and 5D show the distribution of OHSt-Filled microglial cells in the contralateral retinas at 3 and 7 dpl. These two time points were chosen to show their early appearance (3d) and their peak (7d). In these retinas, OHSt-filled microglial cells were not clustered in any specific region, but they rather were distributed over the whole retinal area. The same distribution was observed in both strains, after both treatments, and at all the survival time intervals examined. The distribution of microglial cells in the injured retinas at 3 days postlesion was similar to that found in the contralateral retinas (Figs. 5E–H), and from 5 days onward the number of microglial cell increased from the center to the periphery. 

Discussion

The neuroprotective effect of BDNF alone or in combination with other trophic factors on axotomized RGCs has been thoroughly studied in rats ^4,30,51–59 and cats. ^60–62 In mice, the effect of BDNF has been mainly analyzed in models of photoreceptor degeneration, ^63–65 but its effect on RGC survival after optic nerve injury is unknown. We studied two mouse strains, a pigmented and an albino strain, and found that in both strains the response to IONT and BDNF or vehicle treatment was similar in terms of RGC survival and microglial activation. Thus, eye pigmentation had no effect on this kind of lesion, neither in the injured nor in the contralateral retina. 

RGC Survival

The number of RGCs traced from the superior colliculi in pigmented and albino mice has been previously assessed, ²² while the number of BRN3A⁺RGCs has only been reported in pigmented mice. ⁴ In this study, we examined the total number of BRN3A⁺RGCs in albino mice and found that the percentage of BRN3A⁺RGCs in this strain amounts to 92.6% of the RGCs traced from the superior colliculi. Because RGC tracing from the SCi in albino mice amounts to 98.4% of the total RGC population, ²² we concluded that BRN3A detected approximately 94% of RGCs in albino mice. 

After IONT+vehicle administration, RGC loss became significant at day 3. Afterward, there was a continuous and significant loss of RGCs similar to that induced by IONT alone. ⁴ This suggests that eye puncture and vehicle administration do not have a deleterious or a beneficial effect on RGC survival. Isodensity maps show that RGC loss is diffuse and affects the whole retina, which is consistent with previous reports. ^4,23,66  

After IONT+BDNF treatment, the onset of RGC death was delayed until day 5. Although from this time onward the RGC population decreased substantially, the survival of RGCs was significantly higher at every time interval as compared to that after vehicle administration. Furthermore, a single BDNF injection protected the whole retina, as it has been previously shown in rats. ^45,47 However, there is a difference between species: the effect of BDNF is longer in rats. Indeed, a single injection of this factor protects the whole RGC population up to 7 days after the injury. ^30,45–47 This might be related to the fact that, after axotomy, RGCs die faster in mice than in rats. ^{4,23,30,46,66,67} Thus, in relative terms, the neuroprotective effect of BDNF might be comparable in both species. 

Microglial Response

In mammalian retina, resident microglial cells are found in four retinal layers: the nerve fiber layer, the ganglion cell layer, the inner plexiform layer, and the outer plexiform layer. ³⁶ Microglial cells are activated after injury, ⁶⁸ and they clear RGC debris when these cells degenerate after IONT. ^31,34–36 While there is no agreement on the role of microglial cells after injury or in neurodegenerative disease, they are known to restore tissue homeostasis, and their overactivation is known to be harmful. ^27,28 Indeed, some studies have shown that the suppression of microglial cells improves axonal regeneration and RGC survival after optic nerve injury. ³⁵  

In this study, we only examined the microglial cells in the ganglion cell layer that were transcellularly labeled (i.e., microglial cells that had phagocytosed debris from OHSt-traced retinas). This means that, although optic nerve injury triggers a microglial response in other retinal layers, ³⁶ this response was not studied in the present experiments. 

In control naïve animals subjected to OHSt-tracing alone, there were few phagocytotic microglial cells (<70), which were distributed over the whole retina. The presence of microglial cells in untouched retinas poses several questions: Do microglial cells phagocytose alive and traced-RGCs? Do they prune traced dendrites or axons? Do they phagocytose OHSt excreted from RGCs? Unfortunately, the data presented in this study cannot provide an answer to these questions. 

In the injured retina, there is a negative association between the microglial response and the number of surviving RGCs, as previously shown in rats ³⁶ and more recently in mice. ⁴² Accordingly, the appearance of OHSt-filled microglial cells is more pronounced after vehicle than after BDNF treatment, an effect also observed in rat. ³⁶ The lower number of microglial cells after BDNF-treatment is probably due to a slower RGC death rather than to BDNF inhibiting their activation, because it has been shown that BDNF activates these glial cells. ⁶⁹  

The microglial distribution over the injured retina changes with time. Thus, they initially are scattered at 3 dpl and, as postlesion time increases and more RGCs die, they proliferate throughout the retina with an increasing gradient from the center to the periphery and a peak density around the optic nerve. 

In the contralateral to the injured retina, OHSt-filled microglial cells were distributed over the whole retina without any apparent pattern. A contralateral response has been observed in rat models of optic nerve injury ^36–38 and in mouse and rat models of glaucoma. ^39,40 Liu and colleagues ⁴² reported recently the microglial response in mice injured retinas after IONC and glaucoma, but they did not observe any microglial proliferation in the contralateral retinas. This apparent discrepancy might be explained by the methodologic approach because, in the work of Gallego and colleagues ³⁹ as well as in the present study, the whole retina was analyzed while Liu and colleagues ⁴² only analyzed four representative areas. 

Why is there a contralateral response? Two main theories have been considered. ^36,38–40 In the first theory, microglial cells phagocyte the few retino-retinally projecting RGCs that will be injured and induced to death by unilateral axotomy. In rats, these RGCs are approximately 130 in number and, since they project to the superior colliculus, ⁷⁰ they become traced by OHSt. Thus, their removal from the retina would explain the appearance of OHSt-filled microglial cells in the contralateral retina. The second theory, which might be complementary to the first, is based on the role of microglial cells as a surveillance system, which is activated throughout the CNS as a result of a local insult. ^35,36  

The number of OHSt-filled microglial cells in the contralateral retina was the same after vehicle than after BDNF treatment of the left injured eye, and their number did not change with time. RGCs or microglial cells labeled with fluorogold (analogous to OHSt) ^34,67 remain labeled up to 3 months. This indicates that OHSt-filled microglial cells in the right eye appeared 3 days after the lesion and stayed there up to at least 14 days. After quantifying the whole number of RGCs in the right eye, we did not observe a decrease in the RGC population. This does not imply that the retino-retinally projecting RGCs have not died but, rather, because in rats they represent less than 0.01% of the total RGC population (similar values are expected to be found in mice) their loss would not have any statistical significance. 

In conclusion, this is the first study in mice to demonstrate the neuroprotective effect of BDNF on the axotomized RGCs and to show that after IONT there was a strong and constant contra-lateral microglial response. Therefore, the contralateral retina should not be used as internal control since, while changes in RGC survival in the contralateral retina may be insignificant, they may obscure the results of glial response analyses. 

Footnotes

 Supported by Spanish Ministry of Economy and Competitiveness and ISCIII-FEDER “Una manera de hacer Europa”: PI10/00187, PI10/01496; Fundación Séneca 04446/GERM/07; Spanish Ministry of Education and Science SAF-2009-10385, SAF-2012-38328, and Red Temática de Investigación Cooperativa en Oftalmología RD07/0062/0001.

Footnotes

 Disclosure: C. Galindo-Romero, None; F.J. Valiente-Soriano, None; M. Jiménez-López, None; D. García-Ayuso, None; M.P. Villegas-Pérez, None; M. Vidal-Sanz, None; M. Agudo-Barriuso, None

Footnotes

³  These authors are joint first authors.

Footnotes

⁴  These authors are joint last authors.