| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
1From the Departments of Ophthalmology and 2Anatomy and Neurobiology, University of Maryland at Baltimore, School of Medicine, Baltimore, Maryland.
 |
Abstract |
|---|
 Top Abstract Materials and MethodsResultsDiscussionReferences |
|---|
METHODS. A double-transgenic mouse strain was used, containing a construct with cyan fluorescent protein (CFP) under Thy-1 promoter control, and a construct with ß-galactosidase (lacZ) linked to the stress gene c-fos promoter. Thy-1 in the retina is expressed predominantly in RGCs, enabling stereologic analysis of CFP(+) RGC numbers and loss post-rAION-using confocal microscopy. RGC loss was correlated with axonal counts using transmission electron microscopy (TEM). LacZ immunohistochemistry was used to evaluate retinal cell stress after rAION.
RESULTS. The 45,000 CFP(+) cells in the RGC layer of control animals compared with previous RGC quantitative estimates. rAION produced RGC stress, defined as lacZ expression, in patterns corresponding with later RGC loss. rAION-associated RGC loss correlated with regional nerve fiber layer loss. Axonal loss correlates with stereologically determined RGC loss estimates in transgenic mice retinas.
CONCLUSIONS. Post-ON infarct RGC stress patterns correlate with regional RGC loss. Cellular lacZ levels in most RGCs are low, suggesting rAION-affected RGCs express c-fos only transiently. CFP(+) cell loss correlates closely with quantitative axonal loss, suggesting that the Thy-1 (CFP) transgenic mouse strain is appropriate for RGC stereologic analyses.
We recently described a rodent model of NAION (rAION) in rats4 and mice.5 Intravenously administered rose bengal dye circulating through the anterior ON capillaries at the optic disc is photoactivated, using laser energy directed through the transparent ocular media, resulting in the generation of superoxide radicals that damage capillary vascular endothelium,6 producing a photoembolic blockade of anterior ON capillary blood flow and ischemic axonopathy. The lesion produces selective RGC loss, manifested by decreased RGC numbers, ON scarring, and ON-axonal dropout.4 RGC loss occurs in a regional fashion, similar to that seen in NAION.7
RGC quantitation is typically performed by either ON-axonal counts, using transmission electron microscopy (TEM),8 or direct RGC quantitation by serial step-cut sections,9 retrograde RGC-labeling,10 or RGC-specific immunostaining.7 11 12 The latter two methods can quantify RGC numbers, and regional RGC loss patterns occurring in NAION, but require additional tissue preparation and specimen manipulation. A more direct mechanism of RGC quantification would thus be advantageous.
The transgenic (Tg) Thy-1-CFP mouse strain possesses the Thy-1 gene promoter linked to cyan fluorescent protein (CFP).13 RGCs selectively express Thy-1 in the retina.12 This transgenic strain may thus enable direct ex vivo RGC identification by analyzing CFP expression in the retinal ganglion cell layer, without further processing. We evaluated the Tg-Thy-1-CFP line for consistency of RGC-based CFP expression and to determine its usefulness for analysis of the regional RGC loss patterns occurring after rAION. We used stereology, a statistically validated methodology for cellular quantification,14 to correlate CFP(+) cells in the retinal ganglion cell layer with previously published axonal numbers for this mouse strain.
We previously reported on a model of mouse rAION, using transgenic animals with the c-fos stress gene promoter fused to the lacZ reporter protein. Cells expressing lacZ turn blue with the appropriate reagents. We used this mouse line to identify RGCs and oligodendrocytes undergoing post-rAION stress.4 7 Our previous report suggested that rAION results in lacZ(+) RGCs and oligodendrocytes by 3 days after stroke, implying rAION produces both RGC and glial stress.5 However, the number of blue (i.e., lacZ positive) cells in the retinal (RGC) layer after rAION induction was disproportionately small in relation to ultimate RGC loss. We hypothesized that could be due in part to the sensitivity of the assay used to generate the blue cellular reaction.
To understand the reason for the difference between the number of RGCs that turn blue and the ultimate RGC loss, we generated c-fos-lacZ/Thy-1-CFP double-reporter transgenic animals. Incorporating both transgenes in a single animal model and using the more sensitive ß-galactosidase immunohistochemical localization technique, we were able to make a direct correlation of patterns of early rAION-induced RGC stress with later RGC loss.
|
Materials and Methods |
|---|
|
TopAbstractMaterials and MethodsResultsDiscussionReferences |
|---|
rAION Induction
rAION was induced in anesthetized animals as previously described.4 Briefly, rose bengal (2.5 mM in phosphate-buffered saline (PBS) 1 mL/kg) is administered intravenously, and the optic nerve of the treated eye is illuminated with a 535-nm wavelength, 300 µm laser spot (Iridex Corp., CA) for 12 seconds, using a fundus contact lens. One eye of each animal is left untreated as an internal control. Retinas and ON were photographed 3 days after induction.5
Tissue Preparation
Tissues were collected and prepared from euthanatized animals. Deeply anesthetized (100 mg pentobarbital/kg intraperitoneally [IP]) animals were perfused with 4%-paraformaldehyde-phosphate-buffered saline (PF-PBS). Eyes and ONs were isolated, and postfixed in either 4% PF-PBS (retina) or a mixture of 4% PF and glutaraldehyde (4FIG). Fixed retinas were flatmounted and examined with a 4-channel confocal microscope (Fluoview 400; Olympus, Lake Success, NY) excitation at 405 nm/visualization at 450 nm. Total retinal CFP expression was determined using composite photographs. Axonal distribution was analyzed using single 10x retinal photographs of the central retina.
Three days after rAION induction, animals used for c-fos-lacZ expression were euthanatized and their tissues evaluated. We found that global CFP expression was still apparent 6 days after rAION induction. Stereologic quantification of retinal CFP(+) cells was therefore determined from the control and experimental eyes of 7 animals, 21 days post-rAION induction.
ß-Galactosidase (lacZ) Analysis
We compared lacZ expression 3 days after induction by two methods: a primary chemical reaction,5 using the blue II reagent (5 mM potassium ferricyanide, 5 mM potassium ferrocyanide, 100 mM phosphate buffer, 100 mM NaCl, [pH 7.4], 2 mM MgCl2, 0.01% sodium deoxycholate, 0.02% NP-40, 1 mg/mL X-Gal). This chemical reaction generates a blue color in lacZ(+) cells. LacZ immunoreactivity was also evaluated using a mouse monoclonal antibody for ß-galactosidase (1:1000 dilution; Promega, Madison, WI). Immunolocalization was detected using Texas-red–labeled donkey-anti-mouse secondary antibody (1:500 dilution) (Jackson ImmunoResearch, West Chester, PA).
Stereologic Analysis
Stereologic analyses of the retinas of four control eyes included both RGC-regional density and total CFP(+) RGCs per retina. Stereologic analysis was performed on a system (Neuroleucida 6.0; MBC, Williston, VT) that utilizes a computer-driven counting stage to select a sufficient number of random sites within each defined region, with a sufficient number of cells to ensure valid statistical sampling. Estimated total CFP(+) cells were based on the total retinal area measured from each retina. For counting, a 10x objective was used with a 5.5x digital magnification, enabling analysis of approximately 100 cells per field. A minimum of 800 cells were counted per retina, which is greater than the number required by the Schmitz-Hof equation for statistical validity.14
We further defined three retinal subregions: central (<500 µm radius around the ON), midperipheral (500–1500 µm), and peripheral (1500–2500 µm). Calculation of regional density differences, were performed by counting three randomly selected fields within each region and using the standard exclusion–inclusion boundaries selected by the software (Neurolucida software ver. 6.0 package; MBC). ANOVA-statistical analysis of regional density was used to cross-compare RGC quantification in the central, midperipheral and peripheral regions among all four animals.
Quantitative Analysis of Post-rAION, CFP(+) RGCs
The number of RGCs in each rAION-induced retina and in the contralateral control retina was analyzed in each of seven animals. CFP(+) cells were counted 21 days after rAION induction. This time was chosen because CFP expression is still prominent 1 week after rAION induction, and preliminary results suggested ongoing RGC loss 2 weeks after induction (data not shown). The number of RGCs was quantified at a minimum of three sites within each retinal region, and not less than nine randomly selected locations were analyzed per retina. This approach yielded >800 cells/retina, which is greater than the number necessary for statistical validity.14 Differences in the number of RGCs between control and rAION-induced eyes were expressed as the percentage loss in each animal.
To correlate the number of RGCs with the number of ON-axons, we quantified ON axons by transmission electron microscopy. 4-FIG fixed ONs from control. eyes and eyes 21 days after induction were infiltrated with uranyl acetate, embedded in Araldite-Epon mixture, and sectioned at 200 nanometers. Sections were examined at 2500x. The relative number of axons was determined by using axonal counts from five evenly spaced sites in each of three different ON regions (total 15 counts per optic nerve). Results were statistically analyzed (mean ± SD) and compared with the control counts.
|
Results |
|---|
|
TopAbstractMaterials and MethodsResultsDiscussionReferences |
|---|
|
|
Three days after ON infarct, lacZ expression in double-transgenic animals was analyzed by the traditional blue cell reaction using blue II reagent, similar to that used for the c-fos-lacZ single transgenics in our earlier paper.5 Both the intraocular optic nerve and retina revealed ß-galactosidase-positive cells which appear dark in this figure (Fig. 2A) . Positive (dark) cells in the retina were concentrated in the RGC layer (Fig. 2B) . Closer examination revealed that LacZ(+) cells of different intensities were present in the retinal regions (Fig 2C) . The total number of dark (lacZ positive) cells did not correlate well with the estimated RGC loss in double-transgenic animals at 21 days (see Fig. 4G ). We wondered whether the Blue-II chemical reaction, which relies on sufficient product to generate a visible reduced iron product, was too insensitive to yield a true estimate of total RGC c-fos expression, which is rapidly turned on and off after rAION stress.4 16
|
|
|
Our analysis of individual post-ON infarct retinas revealed that lacZ expression was largely regional—that is, within the retina of one induced eye, we detected a few individual lacZ(+) cells in some regions (Fig. 2F , arrow), whereas lacZ expression was strong in all cells in other regions (Fig. 2H , arrows). The boundaries between strongly and minimally positive regions were also visible (Fig. 2H , arrowheads). Positive retinal regions bordered regions with relatively few positive cells (Fig. 2H ; asterisk). Cells overlaying the intraocular portion of the optic nerve were also positive (Fig. 2G ; ON). Thus, not only is immunohistology a more sensitive indicator of RGC lacZ expression than the blue II chemical reaction, but also reveals that after rAION induction, there is strong regional RGC-lacZ expression.
Confocal microscopy revealed that within the lacZ(+) regions, it was the CFP(+) cells in the RGC layer that expressed lacZ (Figs. 3A 3B 3C) . CFP(+) cells lined the radial retinal vessels in the RGC layer (Fig. 3A) . LacZ expression was present at the same cellular level in many retinal regions in rAION-induced eyes 3 days after induction (Fig. 3B) . Merged confocal images revealed that it was the CFP(+) cells that were lacZ positive (Fig. 3C) .
To reconfirm that the RGCs of double transgenic mice regionally express c-fos-lacZ post-ON infarct, we evaluated the retinas of these animals using an antibody for brain-expressed X-linked protein 1/2 (Bex-1/2), both in flatmounts and retinal cross-sections. Retinal Bex-1/2 expression was primarily found in RGCs.7 Retinal immunohistology for Bex-1/2 localized to both RGCs and their axons (Fig. 3D ; arrows and AxB). Retinal vessels were seen as negative images against the strong Bex-1/2 background (Fig. 3D , Vs). In contrast, lacZ expression in the same cell layer concentrated in the cell bodies (Fig. 3E , arrows) and was strong throughout individual regions (Fig. 3E) . LacZ expression was also strongest in RGCs and in the axons of the RGC nerve fiver layer (NFL; Figs. 3G 3H 3I ) and was minimal in deeper retinal cell layers (compare expression in Fig. 3H ). The merged confocal images revealed that it was the Bex-1/2(+) RGCs and their axons that expressed lacZ (Figs. 3F 3I ; arrows). Thus, post-rAION, affected RGCs of c-fos-lacZ/Thy-1 (CFP) animals expressed lacZ in a regional fashion that was limited to the RGC and NFL layers.
Regional rAION-Induced RGC Loss in Transgenic Animals
Quantification of RGC layer CFP(+) cells in these animals enabled precise analysis of rAION-associated RGC loss. These results are shown in Figure 4 . Cell loss was restricted to the RGC layer, consistent with previous studies.4 5 No change was seen in the average cell density of other retinal layers (data not shown). The RGC layer revealed dense packing of CFP(+) cells in control eyes (Fig. 4A) . CFP(+) axon bundles radially converge at the ON (Fig. 4D) . There was a loss of CFP(+) cells and axons 21 days after rAION induction (Figs. 4B 4C) . A variable degree of CFP(+) cell loss was present in different animals induced with similar induction parameters (compare Figs. 4B 4C , and their converging axons in Figs. 4E 4F ). Stereologic analysis of the number of RGCs after rAION revealed a loss of 25.5% to 72% CFP(+) cells (Fig. 4G ; white bars), with an average 53.3% ± 5.6% SEM loss (Fig. 4G ; black bar).
rAION-induced RGC loss (Fig. 4C) also resulted in RGC-axon bundle loss observable at the retinal (ON) center (Figs. 4E 4F) . Thus, axonal distribution changes mirrorred regional RGC loss. There were residual CFP(+) cells and axons present in stroke-affected regions (Fig. 4C) , suggesting that rAION produces regional RGC loss, but not all RGCs in affected areas may be affected.
TEM Correlation of RGC and Axonal Loss
Previous RGC quantification studies have used RGC labeling17 18 or TEM-axon counts.19 In addition to RGC counts, we evaluated relative ON-axonal loss using TEM. Twenty-one days after rAION induction, axonal loss was qualitatively apparent by toluidine staining of ON sections (data not shown).
TEM-based RGC axonal analysis is shown in Figure 5 . In control ONs, myelinated RGC axons of varying diameters were loosely arranged, peripherally near the ON sheath (Fig. 5A) and surrounded by pial septae. Central RGC axons (Ax) were more tightly packed (Fig. 5B ; arrow), and surrounded by thin pial septae (5B; S). After induction, axonal loss occurred in the ON periphery (Fig. 5C) . There was degeneration of some large-caliber axons (Fig. 5C ; arrow), surrounded by other intact, smaller-diameter fibers (Fig. 5C ; arrowheads). In the ON center, an almost complete loss of RGC axons occurred (Fig. 5D) , with a few intact remaining fibers (Fig. 5D ; arrow).
|
|
Discussion |
|---|
|
TopAbstractMaterials and MethodsResultsDiscussionReferences |
|---|
The results of the stereologic analysis of CFP(+) cells correlates with the results of the TEM-based axonal quantification method. There is no statistically significant difference between both RGC counting methods. The coefficients of variance of RGC counting (0.1) versus axon counting (0.07) are quite close, which suggests that both techniques are prone to the same types of error and that neither method is superior to the other with regard to variability. It also suggests that this transgenic mouse model is no different from the nontransgenic with regard to total RGCs.
Our data reveal that the pattern of RGC loss in rAION is likely to be identical with the pattern of RGC stress. Initial reports using transgenic c-fos-lacZ mice suggested that fewer RGCs visibly express lacZ than are ultimately lost.5 In contrast, there is good correlation of oligodendrocyte stress with blue cell numbers.5 Confocal analysis of poststroke transgenic animals reveals that nearly all RGCs in the retinal region affected by ON stroke express immunohistochemically detectable ß-galactosidase. However, only scattered RGCs in the retinal area affected by axonal stroke express sufficient ß-galactosidase to generate a positive colorimetric reaction.
Most, if not all RGCs in an affected region are likely to express ß-galactosidase. This suggests that rodents have retinal region-specific localization of ON fibers (retinotopy), similar to primates. Thus, rAION appears to affect individual ON regions, rather than diffusely. Most RGCs in a retinal region may be stressed, yet few cells accumulate enough ß-galactosidase product to be chemically detectable, because stressed neurons rapidly alter c-fos expression. Our current data correlate well with previous reports indicating rapid poststroke c-fos activation and deactivation,4 16 implying that most post-rAION–stressed RGCs probably rapidly reprogram themselves for ultimate RGC death and loss.
Unlike neurons, poststroke oligodendrocyte lacZ expression is strongly positive with colorimetric methods.5 This suggests that c-fos expression in stroke-affected oligodendrocytes is either more robust or more prolonged than in stroke-affected neurons. Differences in c-fos activation may be from stress pathway differences between neurons and glia. After axonal stroke, axon signal loss may initiate prolonged stress, ultimately leading to oligodendrocyte apoptosis.5 20 Conversely, increased oligodendrocyte c-fos levels may indicate a more robust survival response.15
Three weeks after stroke, ONs exhibited regional axonal loss by TEM, alongside intact axons. The average axonal loss was 55%, compared with the count in control ONs. TEM studies showed a loss of intact, myelinated axons in the ON. Visible changes included demyelination as well as axonal dropout. Our results correlate closely with the 54.8% loss derived from retinal stereology, which suggests that flatmount stereologic analysis of fixed CFP(+) cells may be sufficient for evaluation of RGC loss. The close statistical correlation of the estimated number of RGCs determined by stererological of CFP(+) cells, compared with published norms, implies that there is likely to be little quantitative difference in RGC layer cells expressing CFP and the actual number of axons.
Mice exhibited considerable variability in RGC loss after rAION induction, with an average RGC loss of 55%. With similar induction parameters, rAION in rats resulted in 
45% RGC loss.4 After rAION, RGC loss occurred at least partially by apoptosis (Slater et al., manuscript in preparation). Post-induced rat retinas revealed more consistency in pathologic features (data not shown), whereas identical rAION induction parameters in mice produced a 25% to 90% RGC loss range by CFP(+) cell counts, and 4% to 79% by axonal quantitation. Mouse induction variability may be due to several factors, including rose bengal intravenous concentration differences, increased difficulty in small animal handling, compared with rats, or ON vascularization differences between mice and rats.21 22 Unlike mice, rat ONs have extensive communication between the ON and the retinal vascular beds.21 The variability of mouse rAION response suggests the necessity to use larger numbers of mice than rats, to obtain statistically valid results. Thus, rAION is similar, but not identical in different rodent species. These differences must be taken into account when comparing results between species, and in comparative pharmacological and transgenic studies.
Because the ON is a central nervous system (CNS) white-matter tract, information obtained using the rAION model may be relevant to white matter strokes in other CNS regions. The c-fos-lacZ/Thy-1-(CFP) double-reporter transgenic strain improves the analysis of the early regional RGC stress associated with ON infarct, and later RGC loss, enabling direct ex vivo identification of affected RGC neurons. This gene combination may increase our ability to analyze ischemic disease impact on the retina in general and enhance our efforts to identify effective neuroprotective therapies.
|
Acknowledgements |
|---|
|
Footnotes |
|---|
Submitted for publication April 28, 2006; revised September 18 and October 27, 2006; accepted March 5, 2007.
Disclosure: S.L. Bernstein, P; Y. Guo, None; B.J. Slater, None; A. Puche, None; S.E. Kelman, None
The publication costs of this article were defrayed in part by page charge payment. This article must therefore be marked "advertisement" in accordance with 18 U.S.C. 
1734 solely to indicate this fact.
Corresponding author: Steven L. Bernstein, MSTF 5-00B; 10 S. Pine Street, Baltimore, MD 21201; slbernst{at}umaryland.edu.
|
References |
|---|
|
TopAbstractMaterials and MethodsResultsDiscussionReferences |
|---|
This article has been cited by other articles:
|
|
 |
|
  B. J. Slater, Z. Mehrabian, Y. Guo, A. Hunter, and S. L. Bernstein Rodent Anterior Ischemic Optic Neuropathy (rAION) Induces Regional Retinal Ganglion Cell Apoptosis with a Unique Temporal Pattern Invest. Ophthalmol. Vis. Sci., August 1, 2008; 49(8): 3671 - 3676. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
). Relative difference from the (100%) stereologically determined RGC number 
) Averaged loss of CFP(+) cells in all animals. rAION resulted in a 53.3% ± 5.6% (SEM) loss of RGCs, compared with control retina. Magnification: (A–C) x20. Scale bar: (A–C) 500 µm; (D–F) 100 µm.