A method to quantify regional axonal transport blockade at the optic nerve head after short term intraocular pressure elevation in mice

Abstract

Axonal transport blockade is an initial step in retinal ganglion cell (RGC) degeneration in glaucoma and targeting maintenance of normal axonal transport could confer neuroprotection. We present an objective, quantitative method for assessing axonal transport blockade in mouse glaucoma models. Intraocular pressure (IOP) was elevated unilaterally in CD1 mice for 3 days using intracameral microbead injection. Longitudinal sections of optic nerve head (ONH) were immunofluorescently labeled for myelin basic protein (MBP) and amyloid precursor protein (APP), which is transported predominantly orthograde by neurons. The beginning of the myelin transition zone, visualized with the MBP label, was more posterior with elevated IOP, 288.8 ± 40.9 μm, compared to normotensive control eyes, 228.7 ± 32.7 μm (p = 0.030, N=6 pairs). Glaucomatous regional APP accumulations in retina, prelaminar ONH, unmyelinated ONH, and myelinated optic nerve were identified by objective qualification of pixels with fluorescent intensity greater than the 97.5^th percentile value of control eyes (suprathreshold pixels). This method segregated images with APP blockade from those with normal transport of APP. The fraction of suprathreshold pixels was significantly higher following IOP elevation than in normotensive controls in the unmyelinated ONH and myelinated nerve regions (paired analyses, p = 0.02 and 0.003, respectively, N=12), but not in retina or prelaminar ONH (p = 0.91 and 0.08, respectively). The mean intensity of suprathreshold pixels was also significantly greater in glaucoma than in normotensive controls in prelaminar ONH, unmyelinated ONH and myelinated optic nerve (p=0.01, 0.01, 0.002, respectively). Using this method, subconjunctival glyceraldehyde, which is known to worsen long-term RGC loss with IOP elevation, also produced greater APP blockade, but not statistically significant compared to glaucoma alone. Systemic losartan, which aids RGC axonal survival in glaucoma, reduced APP blockade, but not statistically significant compared to glaucoma alone. The method provides a short-term assessment of axonal injury for use in initial tests of neuroprotective therapies that may beneficially affect RGC transport in animal models of glaucoma.

1. Introduction

Orthograde axonal transport carries vesicles and mitochondria from retinal ganglion cell (RGC) somas in the retina along their axons through the optic nerve head and to brain target neurons (Anderson and Hendrickson, 1974). Retrograde transport carries material from central RGC axon terminals to the RGC somas (Edström and Hanson, 1973). Extremely elevated intraocular pressure (IOP) for 2–4 hours results in significant orthograde and retrograde axonal transport blockade in the monkey optic nerve head (ONH) (Anderson and Hendrickson, 1974; Minckler et al., 1977; Quigley and Anderson, 1976). Sustained IOP elevation is a major risk factor for transport obstruction that is associated with RGC death in animal models of glaucoma (Howell et al., 2007; Johnson et al., 2000) and human glaucoma eyes (Quigley et al., 1981). Among important transported materials are neurotrophins whose blockade at the ONH is linked to RGC injury and whose overproduction at the soma is protective in experimental glaucoma (Martin et al., 2003; Pease et al., 2009).

Laboratory methods to quantify axonal transport alteration in glaucoma have included autoradiography, scintillation counting, electron microscopy, and immunohistochemistry of transported materials (Bosco et al., 2008; Dapper et al., 2013; Minckler et al., 1977; Quigley et al., 1981, 1979; Quigley and Anderson, 1976). Most of these methods require injection of a precursor or tracer material into either the eye or the brain, such as horseradish peroxidase injection into the optic tract (Minckler et al., 1977), cholera toxin into the vitreous chamber (Angelucci et al., 1996) or FluoroGold particles into the brain (Wessendorf, 1991). Injection of material represents an additional variable in axonal transport blockade analysis. We report transport analysis using immunohistochemistry of endogenously transported amyloid precursor protein (APP), a ubiquitous neuronal transmembrane molecule. APP is synthesized at RGC somas, packaged into endosomes that move in fast orthograde transport through the ONH to the brain (Morin et al., 1993) and returns in retrograde transport (Brunholz et al., 2012). It is involved in axonal pruning of developing RGCs and cortical neurons (Marik et al., 2016) and may be a member of one pathogenic pathway in glaucomatous optic neuropathy (McKinnon et al., 2002). Since the earliest site of transport blockade is at the ONH (Howell et al., 2007; Minckler et al., 1977; Quigley et al., 1981, 1979; Quigley and Anderson, 1976), our method permits quantification of localized transport obstruction, compared to methods that report only total transport as recorded at the brain target site(s). After initial blockade with IOP elevation, axons begin degeneration within weeks in animal models (Quigley and Addicks, 1980; Quillen et al., n.d.); hence, we concentrated in this report on use of the method in rodents after 3 days of IOP elevation, prior to significant axonal degeneration. We developed a histogram analysis of immunofluorescent pixel intensity that was designed to identify and quantify foci of APP immunofluorescence hyperintensity that likely represent protein accumulations occurring in the context of transport blockade. This method was superior to assessment of mean image or region APP immunofluorescence intensity at capturing localized, initial transport block in the ONH. We further applied the technique to assess the effects of glyceraldehyde pre-treatment, which is known to exacerbate axonal loss (Kimball et al., 2014) and oral losartan treatment, which is neuroprotective by reduction in the severity of transport blockade (Quigley et al., 2015).

2. Materials and Methods

2.1. Animal Groups and Treatments

All animals were treated in accordance with the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research, using protocols approved and monitored by the Johns Hopkins University School of Medicine Animal Care and Use Committee. Female CD1 mice, purchased from Charles River (Strain 002, Wilmington, MA), were used at 2–4 months of age. Animals were purchased as two cohorts. A subset of animals from cohort #1 was pretreated unilaterally with glyceraldehyde, a collagen cross-linking agent (Kimball et al., 2014). We subconjunctivally injected 0.3mL of 0.5M glyceraldehyde (Millipore-Sigma, Burlington, MA) dissolved in 0.1M phosphate buffer (0.1M Na₃PO₄, pH=7.2) 3 times, once every 3–4 days under anesthesia by intraperitoneal injection of a mixture of ketamine (50 mg/kg), xylazine (10 mg/kg), and acepromazine (2mg/kg). Some glyceraldehyde-treated animals underwent subsequent experimental IOP elevation two weeks after the first glyceraldehyde injection (N=17), while others did not receive IOP elevation (N=12).

A subset of animals from cohort #2 was pretreated orally in drinking water with losartan (estimated daily dose of 40–60mg/kg/day, Cozaar, Merck, Whitehouse Station, NJ) starting twelve days prior to IOP elevation and continuing for the remainder of the experiment (Quigley et al., 2015). This group of animals underwent IOP elevation in one eye (N=9).

The animals from both cohorts that underwent experimental IOP elevation received intracameral microbead injection in one eye (Cone et al., 2012). Briefly, mice were anesthetized with intraperitoneal ketamine (50 mg/kg), xylazine (10 mg/kg), and acepromazine (2mg/kg). Next, one eye underwent intracameral injection of a 50:50 mixture of 6^m and 1μm microbeads (4μL, Polybead Microspheres®, Polysciences, Inc., Warrington, PA), followed by 1μL of viscoelastic compound (10mg/mL sodium hyaluronate, Healon, Advanced Medical Optics Inc., Santa Ana, CA). Animal procedures were performed on two independent cohorts (Table 1). In addition, animals which were not pretreated locally or systemically underwent experimental IOP elevation (cohort #1 N=13, cohort # 2 N=9).

Table 1:

Experimental Treatment Groups

	Group Title	N	Treatment
Cohort #1	Glaucoma	13	3 day IOP elevation
	Normotensive	12	Contralateral eye, no IOP elevation
	Glaucoma with Glyceraldehyde	17	Glyceraldehyde subconjunctival injection, 3 day IOP elevation
	Fellow eye	17	Contralateral eye, no IOP elevation
	Glyceraldehyde alone	12	Glyceraldehyde subconjunctival injection, no IOP elevation
	Fellow eye	12	Contralateral eye, no glyceraldehyde, no IOP elevation
Cohort #2	Glaucoma	9	3 day IOP elevation
	Normotensive	9	Contralateral eye, no IOP elevation
	Glaucoma with Losartan	9	Losartan in drinking water, 3 day IOP elevation
	Losartan contralateral control	8	Contralateral eye, Losartan in water, no IOP elevation

Sample sizes for APP fluorescence measurements

2.2. IOP Measurement

Mice were placed in the RC2-Rodent Circuit Controller (VetEquip, Inc., Pleasanton, CA) and anesthetized by inhalation of 2.5% of isoflurane in oxygen, 500cc/minute. IOP was measured with a TonoLab tonometer (TioLat, Inc., Helsinki, Finland) before and after any treatment (glyceraldehyde, losartan, or microbead injection) and additionally 1 and 3 days after induction of IOP elevation.

2.3. Sample Preparation

Three days after microbead injection, mice were euthanized by exsanguination under intraperitoneal anesthesia (ketamine (50 mg/kg), xylazine (10 mg/kg), and acepromazine (2mg/kg)), then were trans-cardially perfused with 4% paraformaldehyde in 0.1M phosphate buffer. Eyes were enucleated and placed into increasing concentrations of sucrose and cryopreserved in a sucrose-TissueTek ® O.C.T. compound solution (Sakura Finetek USA, Inc., Torrance, CA). The posterior pole of the eye was sectioned longitudinally to 8 ^m thickness for immunohistochemistry. Antibodies used were: rabbit anti-myelin basic protein (MBP) at 1:1000 (Ab218011, Abcam, Cambridge, MA), rabbit anti-P amyloid precursor protein (APP) at 1:125 (Cat # 51–2700, Invitrogen, Rockford, IL), goat anti-rabbit IgG Alexa 488 secondary antibody (Cat # A-11008, Invitrogen, Eugene, OR) at 1:200 against APP primary and 1:1000 against MBP primary antibodies. All sections were stained with DAPI at 1:2000 (5 μg/mL, Cat # 10–236-276–001, Roche Diagnostics, Indianapolis, IN) and mounted using DAKO mounting media (Cat # S3023, Dako North America, Carpinteria, CA).

All sections were reassigned random identification numbers by a second investigator in order to mask the investigators performing image acquisition, region identification, and image analysis. Images were unmasked and mapped to their experimental group only after image analysis results were extracted. Images with 0.208 μm per pixel resolution were obtained with a Plan-Apochromat 20x/0.8 M27 objective on a Zeiss confocal laser scanning microscope 710 (Carl Zeiss Microscopy, White Plains, NY), using independent two-channel acquisition with excitation at 488nm for Alexa 488 and at 405nm for DAPI. Images were stitched using ZEN 2.3 software (Carl Zeiss Microscopy, White Plains, NY) and converted to .tiff for image analysis. Animal treatment, IOP measurements, sample preparation, immunofluorescence staining, and imaging were performed concurrently per independent cohort (Table 1). Specifically, image acquisition for a unique cohort was performed in consecutive sessions with identical image acquisition settings, i.e. laser power, digital offset and gain, while images from both cohorts had identical scanning speed and image resolution.

2.4. Location of the Myelin Transition Zone

APP intensity measurements were extracted from specific optic nerve regions. Some regional boundaries were based on the position of myelinated optic nerve in tissue sections stained with MBP. ZEN 2.3 software was used to measure the distance between the midpoint of Bruch’s membrane opening (BMO) and the midpoint of the anterior boundary of the MBP signal. Where the anterior boundary of the MBP was not straight, the curved boundary was traced and its midpoint approximated as the midpoint of the chord bisecting the curve (Figure 4). Six eyes from each of the following groups: glaucoma of cohort #1, glaucoma of cohort #2, normotensive of cohort #1, normotensive of cohort #2, glyceraldehyde alone, glaucoma with glyceraldehyde were measured. There were five eyes each from the groups: glaucoma with losartan and losartan contralateral control that were used to mark the position of MBP label in Section 2.5. Statistical analysis was performed for each cohort independently, according to Section 2.7.

Figure 4. Myelin position moves posteriorly with glaucoma.

Longitudinal cryopreserved sections through the ONH and optic nerve immunolabeled with myelin basic protein (MBP, green) and DAPI (blue), from cohort #1. Distance (arrows) from center of Bruch’s membrane opening (BMO, top white line) to the anterior boundary of MBP immunoreactivity (curved white line with bisecting dashed line) was shortest in normotensive (A), longer in glaucoma(B), and longest in the glaucoma with glyceraldehyde group (C). Scale bar = 100 μm.

2.5. Demarcation of Regions in Nerve Head

Using the polygon selection tool in FIJI (Schindelin et al., 2012), we outlined the position of the retina, prelaminar ONH, unmyelinated ONH, and myelinated optic nerve on images of sections labeled for APP and DAPI, with one imaged section per eye analyzed (Figure 1). A line connecting the position of each side of the BMO separated the prelaminar ONH from the unmyelinated ONH. The width of the prelaminar ONH was arbitrarily set as the BMO + 25 ^m on each side. The retina and prelaminar ONH were separated by an arbitrary 25 μm space. Only the nerve fiber layer was included in retinal segments and the left and right retinal segments on each image were summed. The unmyelinated ONH was delineated as the tissue from BMO to a position 200 μm posterior to it (Figure 1). Since we found (Section 3.2) that the position of the myelin boundary moved significantly away from BMO with IOP elevation, the data for the myelinated optic nerve were taken from a position 350 ^m behind BMO to the end of the image.

Figure 1. Schema for demarcation of ONH regions.

Longitudinal cryopreserved section from normotensive group (cohort #1), immunolabeled with APP (green) and DAPI (blue). BMO = Bruch’s membrane opening labeled with yellow dots. ONH = optic nerve head. Scalebar is 100 μm. See text for detailed description.

2.6. Image Analysis Methods

APP images, i.e. the independently acquired channel excited at 488nm, were exported in grayscale format with intensity integer values from 0 (black) to 255 (brightest value). In MATLAB, the pixel intensity values for each outlined region in each sample were extracted and several metrics calculated, including the mean, median, standard deviation, and 97.5^th and 99^th percentile (%ile) intensity values. The pixel intensity values formed skewed distributions, weighted toward dim values (0) with a tail extending to brighter values. Therefore the 97.5^th%ile or 99^th%ile pixel intensity values for each region of each normotensive control nerve were calculated and then the means of these values computed and used as regional threshold cutoffs (T). Thus, four regional threshold cutoffs (T) based on the 97.5^th%ile of the normotensive control group were obtained, likewise four based on the 99^th%ile. Two additional parameters were calculated: the fraction of pixels within a region that fell above the threshold (T), or suprathreshold pixels, and the mean brightness intensity of the same suprathreshold pixels:

$\mathrm{Fraction}>{97.5}^{\text{th}}\%\mathrm{ile}\mathrm{or}{99}^{\text{th}}\%\mathrm{ile}=\frac{\text{\#}pixelswithintensityvalues>T}{total\#pixels}$, and mean>97.5^th%ile or $\mathrm{mean}>{97.5}^{\text{th}}\%\mathrm{ile}\mathrm{or}{99}^{\text{th}}\%\mathrm{ile}=\frac{\sum \text{(}intensityvalues>T\text{)}}{total\#pixels}$, where T = the threshold cutoff based on the 97.5^th or the 99^th%ile intensity value of each regional histogram of the normotensive control group and the denominator is the total number of pixels in the sample’s outlined region.

2.7. Statistical Methods

For the IOP analysis, measurements over time in a single mouse were assumed to have minimal correlation and two-way ANOVA with Sidak correction for multiple comparisons was used. For the two-way ANOVA, time since microbead injection for IOP elevation was always one of the factors. For all other comparisons, non-parametric methods were used due to the available sample sizes and distributions. For comparison between two groups, animals provided two eyes for analysis for the metrics acquired from immunofluorescence analysis; thus, we carried out a paired statistical approach. Two-tailed Wilcoxon Signed Rank tests were used for paired data and two-tailed Mann-Whitney U tests (Wilcoxon ranked sum test) for unpaired data. Multiple groups were compared using the Kruskal-Wallis test, followed by multiple pairwise comparisons between the groups if the test was significant. The tests use the rank of an observation when the groups are pooled to represent the observation. The mean rank for a group is the mean Wilcoxon score. The group with higher mean score will tend to have higher original measurements relative to a group with a lower mean score. A p ≤ 0.05 indicates a statistically significant difference among the groups. Analyses were performed using SAS 9.2 (SAS Institute Inc., Cary, NC, USA) or GraphPad Prism Version 8 (GraphPad Software Inc., La Jolla, California, USA).

2.8. Sample Size Estimation

The sample size analyses were performed for the unmyelinated ONH’s 97.5^th%ile, mean>97.5^th%ile, and fraction>97.5^th%ile values, using a significance level of 0.05 and a power of 80% using an online calculator for paired and unpaired t-tests (Dhand and Khatkar, 2014a, 2014b). The expected mean differences between two groups were either taken from the observed experimental data or hypothesized, and the expected standard deviation was the mean experimental standard deviation from the two groups to be assessed. First, paired sample size analyses were performed for a range of mean differences between the glaucoma and normotensive groups using group means and standard deviations from the first cohort. Next, primary unpaired sample size analyses were performed for a range of unpaired mean differences between a hypothesized experimental group and either the a) normotensive, or b) glaucoma groups. The expected standard deviation was the mean of the glaucoma with glyceraldehyde group and either a) or b) from the first cohort. Each sample size analysis also tested the observed mean difference between the glaucoma with glyceraldehyde and a) or b). In secondary analysis, the expected standard deviation pooled the glaucoma with losartan group and either a) or b) from the second cohort, and again included testing of observed mean differences in addition to hypothesized differences.

Subsets of the paired glaucoma and normotensive group values of the 97.5^th%ile, mean>97.5^th%ile, and fraction>97.5^th%ile from cohort #1 were randomly selected with decreasing sample size, from the 12 experimental pairs. Two-tailed Wilcoxon Signed Rank tests with significance level of 0.05 were repeated by reducing the sample size for totals from 11 down to 3 per group, with each random elimination repeated 4 times. An ideal size for the sample was taken as the subset with the fewest eyes per group that resulted in statistically significant differences in at least 3 of 4 repetitions.

3. Results

3.1. Glyceraldehyde and Losartan Treatment Do Not Prevent IOP Elevation

Pretreatment with losartan or glyceraldehyde did not change the baseline IOP (Figures 3, S1, and Table S1). Eyes of mice that received microbead injections had significantly elevated IOP 1 day after injection compared to their fellow eyes and their baseline IOP, regardless of additional treatment. The glaucoma and glaucoma with glyceraldehyde groups were also significantly elevated at 3 days, though the glaucoma with losartan eyes were higher but not statistically greater than baseline 3 days after microbead injection (Figures 3B, S1B, and Table S1).

Figure 3.

IOP differences between the microbead injected eye (glaucoma) and the fellow eye. A. Cohort #1, left to right: eyes from glaucoma (black bars, N=13), glaucoma with glyceraldehyde (green, N=17), and glyceraldehyde alone (blue, N=12). Both glaucoma with glyceraldehyde and glaucoma in cohort #1 had significantly increased IOP at days 1 and 3 compared to baseline, but did not differ from each other. Glyceraldehyde treatment alone did not alter IOP from baseline or normotensive control (Figure S1). B. Cohort #2, left to right: eyes from glaucoma (black bars, N=10) and glaucoma with losartan (orange, N=9). All glaucoma, otherwise untreated groups had significantly elevated IOP at either 1 or 3 days compared to normotensive eyes, see values in Table S1. Two-way ANOVA with Sidak correction for multiple comparisons, * = p < 0.05, ** = p < 0.001 compared to baseline IOP difference. Values are mean ± SEM.

3.2. Anterior myelination boundary position moves posteriorly with glaucoma

The mean distance from the BMO to the anterior boundary of the myelin zone in normotensive eyes was 228.7 ± 32.7 μm in cohort #1 and 217.1 ± 50.0 μm in cohort #2 (mean ± standard deviation, N=6 each, Figure 4). Therefore, we limited measurements of APP intensity in the unmyelinated ONH to 200 μm from the BMO. In glaucoma eyes compared to normotensive eyes, the position of the myelin zone was further from BMO, 288.8 ± 40.9 μm in cohort #1 (p = 0.030 Wilcoxon signed rank test, N=6 pairs) and 244.5 ± 28.8 in cohort #2 (p = 0.09, Mann-Whitney U test, N=6 per group). Losartan treatment in glaucoma eyes had longer distances from their controls and did not differ from the glaucoma eyes in their cohort, increasing to 252.1 ± 62.6 μm (p = 0.93 compared to glaucoma, Mann-Whitney U test, N=5). In the glyceraldehyde pretreated glaucoma eyes, the myelin zone was even further posterior than in non pre-treated glaucoma eyes, 337.4 ± 45.5 μm from BMO (N=6, p = 0.09; but p = 0.002 compared to normotensive eyes, cohort #1). In order to assure that the myelinated nerve region did not include any of the myelin transition zone, its anterior border was specified at 350 μm posterior to BMO (Figure 1). Treatment with glyceraldehyde or losartan without glaucoma did not lead to different distances to the myelinated nerve compared to the normotensive eyes from their respective cohorts (218.9 ± 37.2 μm, p = 0.70, N=6; and 212 ± 28.7 μm, p = 0.93, N=5, respectively).

3.3. IOP Elevation Causes Significant Increase in APP Intensity Data

Both the regional delineation of images and the extraction of image pixel intensity data were performed by two trained observers masked to the experimental group on 6 randomly-selected images of APP immunofluorescent sections spread across experimental treatment groups, one set analyzed twice without the observer’s knowledge. The intraobserver and interobserver comparisons indicated substantial consistency and agreement, with a mean difference <1 intensity value (out of 255 values). The differences were not statistically significant for overall mean, overall median, 97.5^th%ile, fraction>97.5^th, and mean>97.5^th in the retina, prelaminar ONH, and unmyelinated ONH (p > 0.05 using paired two-tailed f-test). In the myelinated optic nerve, there was a small interobserver difference in the mean APP intensity (5.49 vs. 5.46, p = 0.03), but not in the other parameters (Table S4).

The APP intensity data for each of the 4 regions was calculated using the following parameters: mean and standard deviation of pixel intensity, median pixel intensity, brightness intensity value at which the 97.5^th%ile and 99^th%ile pixel values occurred in normotensive and glaucoma groups, the mean>97.5^thile of control, the fraction>97.5^th%ile of control, and the same parameters for the 99^th%ile of normotensive control data. We examined the data parameters to determine which more clearly separated glaucoma eyes from normotensive controls. In all eyes, histograms of APP fluorescence intensity in retina, prelaminar ONH, unmyelinated ONH and myelinated optic nerve were skewed toward 0 intensity and terminated well before maximum intensity. The regional medians, means, 97.5^th%iles, and 99^th%iles differed between normotensive groups from cohort #1 and from cohort #2. The experimental procedures and image acquisition were carried out at different times for the two cohorts, indicating the need for concurrent controls in the method (Figure 5).

Figure 5. Normotensive prelaminar ONH region image intensity histograms from independent cohorts.

A: From cohort #1 (Table 1), N=12. The normotensive group mean 97.5^th%ile (red dashed line) is at intensity value 11. B: From cohort #2, N=9. The 97.5^th%ile is at intensity value 21. Group mean = solid line, ± standard deviation in gray shading, each data point is the sum of pixels within a bin, each bin spans 3 intensity values, e.g. 0–2, 3–5, etc. X axis is in 8-bit grayscale brightness values from 0 (dark).

Interestingly, the mean and median pixel intensities either did not significantly differ among regions between the glaucoma and normotensive groups or were occasionally inconsistent with the greater APP accumulation clearly visible in the immunofluorescence images (Figure 6; Table 2; and Table S2).

Figure 6.

APP image overall median intensity in each region, from independent cohorts (glaucoma and normotensive eyes from the same animal). Left to right, from cohort #1: glaucoma (black, N=12), normotensive (white, N=12) and from cohort #2: glaucoma (green horizontal dashes, N=8), and normotensive (brown diagonal dashes, N=8). No significant differences between glaucoma eye groups and their normotensive contralateral eyes using Wilcoxon signed rank test. Regional group mean and SEM. ONH = optic nerve head.

Table 2:

Median APP Pixel Intensity in Glaucoma Compared to Normotensive Eyes.

Median Intensity	Normotensive	Glaucoma	p
Retina	4.08	4.58	0.55
Prelaminar ONH	3.92	4.83	0.10
Unmyelinated ONH	5.20	6.17	0.66
Myelinated optic nerve	4.50	4.50	1.00

Glaucoma were compared to normotensive contralateral eyes, cohort #1, N=12 per group. Paired analysis performed with nonparametric two-tailed Wilcoxon signed rank test. For each group we show the regional mean of the 12 individual median values per group. ONH = optic nerve head.

In contrast, the pixel values for the 97.5^th%ile and 99^th%ile were significantly different between glaucoma versus normotensive eyes from cohort #1 for prelaminar ONH and unmyelinated ONH (p=0.006 and 0.004 for the 97.5^th%ile, respectively; p = 0.002 and 0.002 for the 99^th%ile). In comparing the differences between glaucoma and normotensive eyes in paired analysis, there were more significant differences in the 97.5^th%ile mean and fraction values than for the 99^th%ile values. Thus, these will be presented here and recommended for use with this method. Among the 4 regions, the mean>97.5^th%ile and fraction>97.5^th%ile were significantly higher in glaucoma than in normotensive eyes from cohort #1 in paired analysis in the prelaminar ONH, unmyelinated ONH, and in the myelinated optic nerve, but not in retina (Table 3; Figure 8). In the second cohort (performed in parallel with the losartan treated groups), glaucoma specimens were again significantly higher than normotensive specimens in the fraction>97.5^th%ile and mean>97.5^th%ile in unmyelinated ONH (p = 0.02 and 0.01, respectively, paired test, N=8). In addition, the glaucoma group was higher, but not significantly, than the normotensive group (both from cohort #2) in the prelaminar ONH fraction and mean>97.5^th%ile parameters (p = 0.15 and 0.15, respectively). In the comparison of glaucoma to normotensive eyes from cohort #2, the glaucoma eyes were not significantly different from normotensive eyes in median or mean overall APP intensity in any of the four regions (all regions p ≥ 0.11).

Table 3:

Glaucoma vs. Normotensive Group: Fraction and Mean APP Intensity >97.5%ile

	Fraction>97.5th%ile			Mean>97.5th%ile
	Normotensive	Glaucoma	p	Normotensive	Glaucoma	p
Retina	0.069	0.110	0.91	0.860	1.907	0.62
Prelaminar ONH	0.091	0.165	0.08	1.084	2.987	0.01*
Unmyelinated ONH	0.047	0.180	0.01*	0.549	3.335	0.01*
Myelinated optic nerve	0.005	0.035	<0.01*	0.057	0.467	<0.01*

Glaucoma eyes were compared to their normotensive eyes, from cohort #1, in the regional mean of the fraction>97.5^th%ile and mean>97.5^th%ile among 12 eyes per group. From Section 2.7, each normotensive sample region was analyzed separately; the 97.5^th%%ile brightness value for each region for every sample was calculated, and the group mean for each region calculated for the cutoff threshold, referred from then on as the control's 97.5^th%ile per region. Paired analysis by nonparametric two-tailed Wilcoxon signed rank test to determine statistical significance

^*= p < 0.05.

Figure 8. Fraction>97.5^th%ile and mean>97.5^th%ile in all regions of glaucoma and normotensive eyes.

Left to right, from the first cohort: glaucoma (black, N=12), normotensive (white, N=12) and from the second cohort: glaucoma (green horizontal dashes, N=8), and normotensive (brown diagonal dashes, N=8). Using two metrics related to the 97.5^th%ile of normotensive controls, glaucoma from both cohorts show significantly greater fraction>97.5^th%ile and mean>97.5^th%ile values in unmyelinated ONH, and additional regions show significant increases in glaucoma (paired Wilcoxon signed rank test, *p = < 0.05). Regional mean and SEM. ONH = optic nerve head.

Glaucoma eyes were compared to their normotensive eyes, from cohort #1, in the regional mean of the fraction>97.5^th%ile and mean>97.5^th%ile among 12 eyes per group. From Section 2.7, each normotensive sample region was analyzed separately; the 97.5^th%%ile brightness value for each region for every sample was calculated, and the group mean for each region calculated for the cutoff threshold, referred from then on as the control’s 97.5^th%ile per region. Paired analysis by nonparametric two-tailed Wilcoxon signed rank test to determine statistical significance * = p < 0.05.

3.4. Glyceraldehyde pretreatment exacerbates APP axonal transport blockade

To assess whether this APP quantification method was capable of detecting not only transport blockade and protein accumulation associated with glaucoma, but also the protective or detrimental effects of additional treatments in glaucoma eyes, we first evaluated the effect of subconjunctival glyceraldehyde pretreatment on APP immunofluorescence. Glyceraldehyde cross-links collagen in the sclera and increases scleral stiffness (Kimball et al., 2014), potentially in a similar manner as with age- related scleral stiffening (Danilov et al., 2008) and in this way may affect the load of elevated IOP on the unmyelinated ONH. In the unmyelinated ONH, there were significant increases in the fraction>97.5^th%ile and mean>97.5^th%ile in the glaucoma with glyceraldehyde group compared to the normotensive group (p = 0.009, 0.003, respectively, unpaired analyses). In the prelaminar ONH and unmyelinated ONH, the glaucoma with glyceraldehyde 97.5^th%ile, fraction>97.5^th%ile, and mean>97.5^th%ile were greater than the glaucoma eyes, but the difference was not statistically significant with the sample size in this initial study. In these regions and metrics, the comparisons of the glaucoma with glyceraldehyde eyes to their fellow eyes were more significant than the comparisons of the glaucoma to their normotensive eyes. For example, glaucoma with glyceraldehyde eyes’ fraction>97.5^th%ile was 416% of the value in their fellow eyes in the prelaminar ONH and 859% in the unmyelinated ONH (p = 0.001, < 0.001, respectively) while the glaucoma eyes’ fraction>97.5^th%ile was 181% of value of their fellow eyes’ in the prelaminar ONH and 383% in the unmyelinated ONH eyes (p = 0.08, p = 0.02, normotensive cohort #1). Both the glaucoma with glyceraldehyde and the glaucoma groups had dramatic increases in the brighter areas of APP fluorescence in a non-uniform distribution, exhibiting bright clumps of label (Figure 2D). Glyceraldehyde alone, without IOP elevation, did not alter quantitative APP labeling compared to fellow eyes (Table S3).

Figure 2. APP accumulation in optic nerve head in normotensive, glaucoma, and glaucoma with glyceraldehyde.

Longitudinal cryopreserved sections through the ONH and optic nerve immunolabeled with amyloid precursor protein (green) and DAPI (blue). A: normotensive. B: glaucoma. C: glaucoma with glyceraldehyde. D: Typical bright fluorescent clumps in magnified image from unmyelinated ONH of glaucoma nerve from (B). Outlined regions include retina, prelaminar ONH, unmyelinated ONH, and myelinated optic nerve. Scale bar = 100 μm.

3.5. Losartan treatment improves axonal transport blockade

The APP overall median pixel intensity in glaucoma eyes treated with losartan was not significantly different from normotensive eyes from cohort #2 (all regional p ≥ 0.31), except in the prelaminar ONH where it was somewhat lower (5.8 vs. 8.8, p = 0.01). Losartan treated eyes with experimental glaucoma (glaucoma with losartan) had mean intensity, fraction>97.5^th%ile, and mean>97.5^th%ile values that were either not significantly different from normotensive eyes or were closer to normotensive group’s values than glaucoma eyes of this cohort #2. For example, normotensive eyes’ fraction>97.5^th%ile (0.0007) was closer to the glaucoma with losartan group’s (0.0157, p = 0.01), than to the glaucoma eyes’ (0.0211, p = 0.002), in the unmyelinated ONH. Similar in the mean>97.5^th%ile, the normotensive eyes’ value (0.02) was smaller than the glaucoma with losartan group’s (0.46, p = 0.004), and even smaller than the glaucoma eyes’ (0.64, p = 0.001), in the unmyelinated ONH. In this initial series, only 9 glaucoma, 9 glaucoma with losartan, and 9 normotensive eyes were included. Post hoc sample size analysis (Section 2.8) indicated that 25–26 eyes per group would provide 80% power to find a statistically significant difference between the glaucoma with losartan group and the glaucoma group from this cohort #2 in the values of the 97.5^th%ile, mean>97.5^th%ile, and fraction>97.5^th%ile in the prelaminar ONH and unmyelinated ONH.

3.6. Sample Size Estimation

To judge the usefulness of the APP blockade assay to detect elevated IOP-related transport blockade, we performed sample size calculations based on parameter values in our two cohorts. The parameters were: 97.5^th%ile, mean>97.5^th%ile, and fraction>97.5^th%ile for the unmyelinated ONH region. The analyses determined the sample sizes needed for an 80% chance of finding a statistically significant difference with an expected true difference between two group means—a test group and a reference group.

In the first cohort’s unmyelinated ONH, the 97.5^th%ile, mean>97.5^th%ile, and fraction>97.5^th%ile values were higher in the glaucoma than the normotensive group, reaching 120%, 510%, and 280%, respectively, of the values of the normotensive group. The differences between the group means were statistically significant with 12 samples per group in all three parameters. Based on a sample size estimation, we found that the respective differences between the means of the glaucoma and normotensive groups in the 97.5^th%ile, mean>97.5^th%ile, and fraction>97.5^th%ile would remain statistically significant with only 7, 7, and 6, respectively, samples per group in paired analyses. The sample size of the first cohort was sufficient to determine statistical significance had the glaucoma 97.5^th%ile equaled 110% of the normotensive group value or greater.

With treatments hypothesized to either increase or decrease APP blockade in the unmyelinated ONH with IOP elevation, the expected mean differences for the sample size calculation can be informed by the measured differences and variabilities of the glaucoma, otherwise untreated (no injection and untreated water) groups and the glaucoma with glyceraldehyde or with losartan groups. If the expected mean differences are such that the treatment group’s value is 200% of the glaucoma value, only 7, 19, or 15 eyes per group are needed for such differences in the 97.5^th%ile, mean>97.5^th%ile, and fraction>97.5^th%ile, respectively. If the expected values are less than 200% of the glaucoma values, then more eyes per group are needed, for example for a treatment group with 97.5^th%ile value 150% of the glaucoma, then 27 nerves per group are needed.

Based on the values from the glaucoma with losartan group’s unmyelinated ONH, a value which is 50% of the glaucoma value in the 97.5^th%ile parameter would be identified with only 8 eyes per group, and a value equaling 75% of the glaucoma value would be identified with 29 eyes per group. The experimental glaucoma with losartan group was 82% of the glaucoma group which was not statistically significant with N=9. The comparable sample sizes for 82% of the untreated glaucoma for 97.5^th%ile are 57 eyes per group.

4. Discussion

We developed a novel method for quantifying axonal transport blockade in the optic nerve head in animal glaucoma models based on a histogram-based statistical analysis of APP immunofluorescence. The proposed method has some clear advantages over previously used methods. First, it does not require injection of exogenous materials into the eye or the brain. Secondly, it does not require lengthy post-processing, as does autoradiography. Thirdly, it can be more readily quantified than some past methods that used qualitative grading. Fourth, it provides detailed regional information about the location of transport obstruction. The method is reproducible in both intraobserver and interobserver comparisons, and replicable across independent animal cohorts (Figure 8 and Table S4).

In experimental monkey and human glaucoma axonal transport blockade with IOP elevation, some axons do not exhibit transport blockade, while others exhibit total blockade (Quigley et al., 1981; Quigley and Anderson, 1976). This poses a challenge in quantifying local axonal transport blockade in earlier time points, which has previously been addressed by qualitative grading (Minckler et al., 1977; Quigley et al., 1981, 1979; Quigley and Addicks, 1980). While the mean change in APP density might be modest, areas with axons swollen by blocked transported material appear as bright spots (Figure 2). Therefore, it was not surprising that the mean and median APP intensity were poor indicators of transport blockade (Figure 6, Tables 2 and S2). The use of metrics that assess the localized accumulation of APP is ideal for identifying the effect of short-term IOP elevation on axonal transport. The parameters based on exceeding the 97.5^th%ile brightness intensity value of the control group serve this purpose better than other metrics-mean and median—we evaluated. Our experience with the microbead mouse glaucoma model indicates that transport block in the mouse eye is sparse at one day after IOP elevation, but is significant by three days when examined by transmission electron microscopy (Quillen et al., n.d.). Thus, the three-day time point appears a practical choice for assessing early axonal transport blockade. The method can also be applied to later time points, with the caveat that at 1–2 weeks after IOP elevation, axonal degeneration has begun, producing a decrease in overall transport due to both loss of axons and axonal transport blockade in a higher number of axons (Quillen et al., n.d.). RGC soma death has not been observed at 3 days (Kimball et al., 2017), and thus axonal transport decrease is unlikely derive from a decrease in RGC synthesis of APP.

The method permits study of regional transport alteration from retina to myelinated optic nerve. Indeed, we found inconsistent effects in both retina and myelinated optic nerve, likely because the zone at which scleral hoop stress and translaminar pressure difference act most prominently is the prelaminar and particularly the unmyelinated ONH in mouse, as in human lamina cribrosa (Sigal et al., 2007). This site has been determined to be the location of transport obstruction in other mouse models of IOP elevation (Howell et al., 2007). The zones visible in our images were also smaller in retina, making assessment there less precise. Methods that assess only arrival of retrogradely moving material at the RGC soma, or the orthograde arrival of material such as cholera toxin at the brain centers cannot determine where, specifically, transport interruption has occurred. When attempting to confirm that a potential neuroprotective approach has acted by altering mechanical forces or remodeling at the ONH, localization as well as quantification can be derived from the reported method.

Axonal transport blockade following 3 days of IOP elevation using the proposed APP quantification parameters was significantly higher than normotensive eyes with a sample size of 8–12 pairs of eyes. However, if the intent is to determine a neuroprotective effect from therapies, the simulations given here allow planning for the appropriate sample sizes needed per group to identify a potential beneficial effect. The method was applied to two treatments, in both cases there were measurable effects in expected directions. We found greater localized APP accumulation with glyceraldehyde pretreatment prior to glaucoma. This is consistent with our previous qualitative grading of transport block in glyceraldehyde-pretreated glaucoma eyes that showed worsening of glaucoma damage (Kimball et al., 2014). Early increase in axonal transport blockade herein is associated with subsequently greater RGC degeneration. Glyceraldehyde pretreatment followed by 6 weeks of experimental glaucoma led to 43.6% axon loss, compared to 18.3% axon loss without pretreatment in glaucoma, using the same protocol and mouse strain (Kimball et al., 2014). Future studies will examine the mechanism by which glyceraldehyde pretreatment resulting in cross-linking of scleral collagen and increased scleral stiffness (Kimball et al., 2014) exacerbates IOP induced axonal transport blockade in the ONH. We also applied the APP blockade assay to test effects of the angiotensin receptor blocker, losartan, which reduced RGC loss in a 6 week glaucoma model from 25% loss to 6% (Quigley et al., 2015). Here, losartan treatment quantitatively reduced APP blockade at 3 days compared to untreated eyes, as suggested by our previous qualitative assessments. However, with the sample size utilized here, the losartan treatment in glaucoma did not significantly reduce axonal transport blockade to control levels in the unmyelinated ONH, but did so in the prelaminar ONH. Future studies will also examine the differences in blockade after longer or different levels of IOP elevation and quantification of the amount of axonal transport decrease due to axonal loss after longer IOP elevation.

The limitations of the assay are its need to label endogenous APP with antibodies. While there are available mouse strains with fluorescent APP, these did not have sufficient signal intensity in our hands to provide usable data in glaucoma models. We found that the degree of APP labeling in the normotensive eyes can vary from one dataset to another, due to differences in animals, antibody efficiency, image acquisition, and sample storage time. It may also be affected by systemic therapy. Therefore, it is advisable to use a contemporaneous control group in masked analysis. Another limitation is the large standard deviation of the image intensity in the glaucoma with losartan and the glaucoma with glyceraldehyde groups. Sample size analysis found large sizes (N=25–185) are needed to have at least an 80% chance of finding statistically significant differences between the glaucoma with losartan group and the glaucoma group in their 97.5^th%ile, mean<97.5^th%ile, and fraction>97.5^th%ile in the unmyelinated ONH. This variability has multiple potential contributors, including variability in the degree of IOP elevation and the posterior movement of the prelaminar ONH and unmyelinated ONH with IOP elevation. Another source of variability in the intensity signal is the inconsistent position of the vasculature in the longitudinal sections. This leads to a variability in the retinal surface of the ONH, so that portions of the pre-retinal vitreous are included in the prelaminar zone as performed here. If the prelaminar zone is to be included in further use of the method, specimen-specific demarcation of the retinal surface in each section may be preferred.

Only one strain of animals, CD1, was examined for axonal transport blockade herein. There may be differences in the effects of chronic IOP elevation on axonal transport and retinal ganglion cell loss among mouse strains (Cone et al., 2012; Schaub et al., 2017). Future work will examine axonal transport blockade in glaucoma induced in multiple mouse strains and also include male mice, as only female mice were included herein to exclude sex as a variable in establishing the methodology. Sample size analysis of section 3.6 will be used to examine sex and mouse genetic strain as variables. Additionally, serial longitudinal sectioning was not carried out to assess whether regional differences exist in transport block across the ONH, presumably indicating differential effects on RGC in various retinal locations.

We report a method for quantification of early axonal transport blockade in the rodent ONH glaucoma model. It does not require injection of exogenous material, it is user-independent, and detects changes in localized axonal transport blockade of APP in the ONH. The most effective outcome parameters assessed increases in local bright areas of immunofluorescence. Glyceraldehyde exacerbated and losartan reduced APP axonal transport blockade, consistent with their effects to worsen or improve ganglion cell loss at longer IOP elevation, respectively. The APP blockade assay can be used to assess the potential of a variety of neuroprotective therapies at early time points in animal glaucoma models.

Figure 7.

Pixel intensity histograms for glaucoma (red, N=13) and normotensive (black, N=12) by region from cohort #1: A. retina, B. prelaminar ONH, C. unmyelinated ONH, and D. myelinated optic nerve. Group mean frequencies (y-axis) for the respective pixel intensity values (x-axis) are solid lines with the range of ± one standard deviation shaded. Insets magnify the tail of each histogram, with x-axis beginning at the 97.5^th%ile intensity value for each regional normotensive (control) group (shown in larger graphs as vertical red dashed line). For each region, the glaucoma histogram tails exceed the normotensive control data beyond the threshold cutoff. ONH = optic nerve head.

• A method to quantify axonal transport blockade in mouse optic nerve is presented

• Localized accumulation in four zones of the optic nerve was quantified

• The method was applied to glaucoma pretreated with glyceraldehyde or losartan

• Method can assess neuroprotecation in glaucoma models at an early time point

Abbreviations

RGC

retinal ganglion cell

IOP

intraocular pressure

ONH

optic nerve head

APP

amyloid precursor protein

MBP

myelin basic protein

%ile

percentile

BMO

Bruch’s membrane opening