Abstract
purpose. To determine whether in glaucoma there is atrophy of relay neurons in
magnocellular and/or parvocellular lateral geniculate nucleus (LGN)
layers projecting to the visual cortex and to compare the degree of
neuronal atrophy in magnocellular layers with that in parvocellular
layers.
methods. Seven cynomolgus monkeys with unilateral experimentally induced
glaucoma and five control monkeys were studied. The left LGN neurons in
magnocellular layer 1 and parvocellular layers 4 and 6, connected to
the right glaucomatous eye were examined. Immunocytochemistry with
antibody to parvalbumin was used to specifically label relay neurons
connecting to the visual cortex. Neuronal cell body cross-sectional
area was estimated using unbiased point-counting methodology.
Experimental and control groups were compared using t-tests. Analysis of covariance (ANCOVA) tests were used
to compare the percentage of decrease in mean neuronal area between
layers 1, 4, and 6, as a function of percentage of optic nerve fiber
loss or mean IOP. There was significant correlation between percentage
of optic nerve fiber loss and mean IOP.
results. The mean cross-sectional area of relay neurons in magnocellular layer 1
and parvocellular layers 4 and 6 were significantly decreased in
glaucoma compared with controls by 28%, 37%, and 45%, respectively.
Neuronal area decreased in a linear fashion, with increasing optic
nerve fiber loss or increasing mean IOP for layers 1, 4, and 6. The
percentage of neuronal shrinkage in each of parvocellular layers 4 and
6, as a function of optic nerve fiber loss (P =
0.05; P = 0.001, respectively) or mean IOP
(P = 0.046; P = 0.0008,
respectively), was greater than that seen in magnocellular layer 1.
conclusions. Relay neurons in the LGN, which project to the visual cortex, undergo
significant shrinkage in glaucoma, and neurons in parvocellular layers
undergo significantly more shrinkage than neurons in magnocellular
layers.
Neurons in the central nervous system may undergo atrophy
or cell shrinkage in response to a given insult, and persistence of the
pathologic state may lead to neuronal death.
1 2 3 4 5 6 Decrease
in cell body size in atrophic neurons correlates with a decrease in
function
7 and is of particular interest, since it may be
reversible.
When disconnected from their major afferent pathways, neurons shrink
and may subsequently die.
6 Transneuronal degeneration is a
well-known process in the extension of neurodegenerative
diseases.
8 9 In glaucoma, severe optic nerve fiber loss
has been shown to be accompanied by loss of neurons in the lateral
geniculate nucleus (LGN),
10 in keeping with the concept of
transneuronal degeneration. Studies of the optic nerve in glaucoma have
shown evidence of preferential loss of larger
fibers.
11 12 13 In the central visual pathways, damage to both magnocellular and parvocellular pathways has been
shown in the experimental monkey model of
glaucoma
10 14 15 16 17 18 ; however, it is controversial
whether one of these pathways is preferentially affected. Whereas a
recent study showed preferential loss of neurons in the LGN in
magnocellular layers,
17 another study described evidence
of damage to both magno- and parvocellular LGN layers.
18
Within the LGN, there are two types of neurons: inhibitory interneurons
confined to the LGN and relay neurons that proceed to synapse in the
visual cortex.
19 20 Most of the degenerative and
compensatory changes in the LGN occur in the relay neurons rather than
interneurons after total deafferentation.
21 22 We have
previously shown loss of relay LGN neurons in both magno- and
parvocellular layers in glaucoma.
10 Using parvalbumin, a
selective marker for relay neurons of the LGN,
23 24 the
purpose of this study was to determine whether there is atrophy of
relay neurons in magnocellular and/or parvocellular LGN layers
projecting to the visual cortex and to compare the degree of atrophy in
magnocellular relay neurons with that in parvocellular relay neurons.
All studies were performed according to the guidelines of the
ARVO Statement for the Use of Animals in Ophthalmic and Visual
Research. Seven young adult cynomolgus monkeys (
Macaca
fascicularis) with experimental glaucoma induced in the right eye
by argon laser scarification of the trabecular meshwork (ALT) and five
normal control cynomolgus monkeys were studied. Experimental glaucoma
was induced in the right eye of each of seven monkeys according to the
protocol described by Gaasterland and Kupfer.
25 The
survival time after laser ALT was 14 months.
In experimental glaucoma and control monkeys, intraocular pressure
(IOP) measurements were performed under light sedation with a
pneumatonometer (Digilab, Norwell, MA) and an applanation tonometer
(Goldmann; Haag-Streit, Koniz, Switzerland) respectively as previously
described.
10 In the ALT-treated right eyes, the interval
between IOP measurements ranged from 5 to 39 days (mean, 26 days) over
a 14-month period after laser trabeculoplasty.
Table 1 summarizes the IOP values in the ALT-treated right eyes and in
control monkeys. The mean IOP in the ALT-treated right eyes was
significantly increased and ranged from 28.6 to 54.5 mm Hg. In all
ALT-treated eyes, IOP measurements were above 21 mm Hg for at least 13
months. In control monkeys, mean IOP ranged from 14 to 21 mm Hg.
The method by which optic nerve fibers were counted in the glaucomatous
right optic nerve and the control optic nerve in monkeys with
experimental glaucoma has been previously described.
26 Optic nerve fiber loss compared with the fellow control optic nerve
ranged from 0% to 100% as indicated in
Table 1 .
Under deep general anesthesia as previously
described,
10 perfusion through the heart in experimental
and control animals was performed with 4% paraformaldehyde, 0.1%
glutaraldehyde, and 4% paraformaldehyde, respectively, in 0.1 M
phosphate buffer (pH 7.4). For tissue processing and serial coronal
50-μm sectioning of the LGN, care was taken to ensure the same
procedures for all monkey brains.
10
The primary antibody was a monoclonal antibody against
parvalbumin (clones PA-235; Sigma, St. Louis, MO). Parvalbumin, a
calcium-binding protein, labels relay neurons in the LGN layers that
project axons to the visual cortex.
23 24 Tissue sections
were immunostained, according to a previously described
protocol.
10
The tissue was viewed using a bright-field microscope (Reichert
Jung, Vienna, Austria) with a color video camera (JVC, Yokohama,
Japan), video, and computer monitors. The six layers of the LGN were
easily identified on stained sections. The ventral layers 1 and 2 are
magnocellular layers, whereas the remaining dorsal layers three to six
are parvocellular layers. Layers 1, 4, and 6 of the left LGN are
connected to the right eye (glaucomatous, in this study), whereas
layers 2, 3, and 5 are connected to the left eye (nonglaucomatous). To
determine whether neurons in magnocellular and/or parvocellular LGN
layers connected to a glaucomatous eye are shrunken, the
cross-sectional area of neurons in the left LGN layers 1, 4, and 6 were
measured, and the measurements were compared with those from the left
LGN layers 1, 4, and 6 in control monkeys. Retinal ganglion cells
(RGCs) of the right nasal hemiretina and fovea project to the left LGN
layers 1, 4, and 6 and compose approximately 50% of the right eye
RGCs.
27 The difference in number of surviving nerve fibers
between the nasal and temporal quadrants of the right optic nerves was
not statistically significant in experimental glaucoma
(
P > 0.05) in the monkeys examined in this study;
therefore, changes observed in the left LGN layers 1, 4, and 6 are
representative of changes observed in target LGN neurons. Left LGN
layers 1, 4, and 6 of monkeys with a normal visual system were used as
controls rather than the left LGN layers adjacent to examined layers 2,
3, and 5 or the right LGN layers 1, 4, and 6 of the laser-treated
monkeys, because a significant decrease in cell size has been observed
in undeprived layers under monocular experimental
conditions.
28
Measurements were performed on three sections representative of the
anterior, middle, and posterior parts of each LGN. Cross-sectional area
measurements were made on parvalbumin-immunostained sections at high
power using an oil immersion objective (×100, numeric aperture, 1.32),
bright-field microscope, and color video camera. Immunostained neurons
were visualized on the computer and video monitors. Cell body
cross-sectional area measurements were made at locations determined by
a random and systematic sampling procedure using a superimposed grid
method.
29 Computer software (Neurozoom; Human Brain
Project, La Jolla, CA) enabled digital superposition of the sampling
grids on the tissue. Only sample locations within the LGN layers were
used for cross-sectional area measurements. The size of the sampling
grid was adjusted for each layer so that there were at least 65 samples
for that layer through the nucleus. To measure cross-sectional areas in
an unbiased fashion, optical dissector methodology was used to assess
the maximum cross-sectional area. A three-dimensional optical box was
composed of
x,
y, and
z axes of
50 × 50 × 10 μm, respectively. A 50 × 50-μm
counting frame was projected onto the video monitor. By measuring only
neurons completely within the frame and intersecting the upper or
righthand borders, sampling bias was minimized.
30 The
cross-sectional area of new neurons that came into focus as the
operator focused through the optical box was measured. The excursion
along the focusing axis (10 μm) and the thickness of the section were
measured with a microcator (MT12; Heidenhain, Traunreut, Germany)
mounted on the microscope stage. A point-counting grid generated by the
software was visualized on the computer monitor. Using the mouse, the
operator marked the points located on a cell body. Each point
corresponded to an area of 4 μm
2 for the grid
used for all layers. The cross-sectional area of the cell body was
calculated by multiplying the number of points on the cell body with
the area corresponding to a grid point by the software. Neuronal cell
body cross-sectional area per layer was estimated by calculating the
average cross-sectional area for at least 67 neurons. Neuronal radius
was calculated as the square root of the cross-sectional area divided
by π.
Student’s t-test was used to compare the mean
cross-sectional area means of neurons in the LGN of glaucomatous versus
control monkeys. To assess atrophic changes, neuronal area percentage
decrease (percentage of neuronal shrinkage) was used as the dependent
variable. Neuronal area percentage decrease for each layer of each
animal subject was calculated as the mean of neuronal area means in a
given layer of the control group minus the mean neuronal area divided
by the mean of neuronal area means for the control group. The
relationships between percentage of neuronal shrinkage and percentage
of optic nerve fiber loss and between percentage of neuronal shrinkage
and mean IOP were assessed separately, using linear regression models.
To compare percentage of neuronal shrinkage among magnocellular layer 1
and parvocellular layers 4 and 6, two analyses of covariance (ANCOVA)
were performed (GLM procedure of Statistical Analysis Software; SAS,
Cary, NC) one for percentage of optic nerve fiber loss and one for mean
IOP. The dependent variable was always percentage of neuronal
shrinkage, and the layer was the grouping factor. Using the interaction
term between percentage of neuronal shrinkage and layer, we determined
whether the slopes were significantly different. The model, without the
interaction term enabled assessment of whether the means differed among
the three layers, and of which combinations of means, if any, were
significantly different.
Light Microscopy.
Parvalbumin immunoreactivity was seen in the neuronal cell body and
processes in glaucoma and control magnocellular layer 1. Compared with
the control
(Fig. 1A) , in glaucoma, neuronal cell bodies were smaller, and cell body
shrinkage appeared to increase with increasing optic nerve fiber loss
of 29%, 61%, and 100% (
Figs. 1B 1C 1D , respectively).
Cross-sectional Area Measurements.
Morphometric studies were performed in left LGN magnocellular layer 1
connected to the right optic nerve in monkeys with experimental
glaucoma in the right eye (
n = 7) and in control monkeys
(
n = 5). The percentage frequency distribution of neuronal
radius in the glaucoma group shifted to the left, compared with the
control group
(Fig. 2) . An increase in the frequency of the smaller neurons was seen along
with a corresponding decrease in the frequency of the larger neurons.
Table 1 summarizes the cross-sectional area of neurons for
magnocellular layer 1 in control and glaucoma groups. The mean
cross-sectional area of layer 1 ranged from 135.1 ± 63.1 to
255.7 ± 70.6 μm
2 in the glaucoma group,
and from 227.3 ± 86.5 to 250.8 ± 91.3μ
m
2 in the control group. The mean of
cross-sectional mean areas (±SD) of relay neurons in magnocellular
layer 1 was significantly decreased by 28% in the glaucoma group
compared with the control group (171.3 ± 41.3 vs. 238.5 ±
12.3 μm
2;
P < 0.006).
Light Microscopy.
Parvalbumin immunoreactivity was seen in the neuronal cell body and
processes in glaucoma and control parvocellular layer 4. Compared with
the control
(Fig. 3A) , neuronal cell bodies were smaller in glaucoma, and cell body
shrinkage appeared to increase with increasing optic nerve fiber loss
of 29%, 61%, and 100% (
Figs. 3B 3C 3D , respectively).
Cross-sectional Area Measurements.
Morphometric studies were performed in left LGN parvocellular layer 4
connected to the right optic nerve in monkeys with experimental
glaucoma in the right eye (
n = 7) and in control monkeys
(
n = 5). The percentage frequency distribution of neuronal
radius in the glaucoma group shifted to the left, compared with the
control group
(Fig. 4) . An increase in the frequency of the smaller neurons was seen along
with a corresponding decrease in the frequency of the larger neurons.
Table 1 summarizes the cross-sectional area of neurons for
parvocellular layer 4 in control and glaucoma groups. The mean
cross-sectional area of layer 4 ranged from 94.9 ± 32.1 to
157.0 ± 39.0 μm
2 in the glaucoma group
and from 178.5 ± 46.0 to 230.8 ± 62.0μ
m
2 in the control group. The mean of
cross-sectional mean areas (±SD) in parvocellular layer 4 was
significantly decreased by 37% in the glaucoma group compared with the
control group (127.2 ± 24.9 vs. 202.6 ± 20.8μ
m
2;
P < 0.0005).
Light Microscopy.
Compared with the control
(Fig. 5A) , neuronal cell bodies were smaller in glaucoma, and cell body
shrinkage appeared to increase with increasing optic nerve fiber loss
of 29%, 61%, and 100% (
Figs. 5B 5C 5D , respectively).
Cross-sectional Area Measurements.
Morphometric studies were performed in left LGN parvocellular
layer 6 connected to the right optic nerve in monkeys with experimental
glaucoma in the right eye (
n = 7) and in control monkeys
(
n = 5). The percentage frequency distribution of neuronal
radius in the glaucoma group shifted to the left, compared with the
control group
(Fig. 6) . An increase in the frequency of the smaller neurons was seen, along
with a corresponding decrease in the frequency of larger neurons.
Table 1 summarizes the cross-sectional area of neurons for parvocellular
layer 6 in control and glaucoma groups. The mean cross-sectional area
of layer 6 ranged from 80.0 ± 23.1 to 151.3 ± 43.8μ
m
2 in the glaucoma group and from 176.9 ± 59.6 to 258.0 ± 62.4 μm
2 in the
control group. The mean of cross-sectional mean areas (± SD) in
parvocellular layer 6 was significantly decreased by 45% in the
glaucoma group compared with the control group (120.3 ± 31.8 vs.
219.7 ± 35.8 μm
2;
P <
0.001).
Neuronal Shrinkage as a Function of Optic Nerve Fiber Loss.
Figure 7A shows the plot of percentage decrease in mean neuronal area as a
function of percentage of optic nerve fiber loss for layers 1, 4, and
6. The figure shows that for each of the three layers, percentage
decrease in mean neuronal area increased with increasing optic nerve
fiber loss. The linear regression lines are superimposed on the plots.
The estimated regression equations for layers 1, 4, and 6 of percentage
decrease in mean neuronal area were 14.6 + (0.31 × percentage
optic nerve fiber loss), 26.6 + (0.24 × percentage optic nerve
fiber loss), and 31.1 + (0.32 × percentage optic nerve fiber
loss), respectively. These relationships were linear and statistically
significant; those for layers 1 and 4 were significant at
P < 0.05, whereas those for layer 6 were significant
at
P < 0.01. There was a significant correlation
between optic nerve fiber loss and IOP (
r = 0.94,
P = 0.002). The interaction term of the ANCOVA model
indicates that the difference in slopes was not statistically
significant (
P = 0.74). However, mean neuronal
shrinkage in parvocellular layer 6 was significantly greater than that
observed in magnocellular layer 1 (
P = 0.001). Mean
neuronal shrinkage in parvocellular layer 4 was significantly greater
than that observed in magnocellular layer 1 (
P = 0.05),
whereas the means of parvocellular layers 4 and 6 did not differ
significantly (
P = 0.08).
Neuronal Shrinkage as a Function of Mean IOP.
Figure 7B shows the plot of percentage decrease in mean neuronal area
as a function of mean IOP for layers 1, 4, and 6. The figure shows that
in all three layers neuronal shrinkage increased as mean IOP increased.
The linear regression lines are superimposed on the plots. The
estimated regression equations for percentage decrease in mean neuronal
area in layers 1, 4, and 6 were −32.0 + (1.46 × mean IOP),−
5.9 + (1.04 × mean IOP) and −12.1 + (1.39 × mean IOP),
respectively. These relationships are linear and statistically
significant:
P < 0.05 for layers 1 and 4 and
P < 0.01 for layer 6. There was a significant
correlation between optic nerve fiber loss and IOP (
r =
0.94,
P = 0.002). The interaction term of the ANCOVA
model indicates that the difference in slope was not statistically
significant (
P = 0.658). However, mean neuronal
shrinkage in parvocellular layer 6 was significantly greater than that
observed in magnocellular layer 1 (
P = 0.0008). Mean
neuronal shrinkage in parvocellular layer 4 was also significantly
greater than that observed in magnocellular layer 1 (
P = 0.046). The mean neuronal shrinkage in parvocellular layer 6 showed a
trend to be greater than that observed in parvocellular layer 4
(
P = 0.073).
Previous studies of normal monkey LGN have used Nissl stain that
labels all neurons including relay neurons and
interneurons.
4 31 32 In the present study, parvalbumin was
used for specific labeling of relay LGN neurons connecting to the
visual cortex.
23 24 Our cross-sectional neuronal area
measurements for magno- and parvocellular layers were similar to
measurements by Headon et al.
31
According to our results, relay neurons in magno- and
parvocellular layers undergo atrophy in experimental glaucoma. The mean
neuronal shrinkage observed in the two glaucomatous animals with
complete optic nerve fiber loss in this study was similar to that
observed by Matthews
6 12 months after ocular enucleation:
43% vs. 54% in magnocellular layer 1, 49% vs. 66% in parvocellular
layer 4, and 62% vs. 66% in parvocellular layer 6.
The present study demonstrates for the first time in the central
nervous system that cell size decrease relates to the degree of
deafferentation, previously suggested by findings in studies of the
olfactory system
33 and somatosensory sytem.
34 The linear relationship between the degree of optic nerve damage and
degree of atrophy in relay LGN neurons suggests a link between loss of
retinal ganglion cells and shrinkage of their target neurons.
Although significant input to the LGN from the cortex and several
subcortical structures has been described,
35 36 atrophic
changes observed in target relay neurons appears to be directly related
to the loss of connections with RGCs, the major afferent input to the
LGN. In addition, the linear relationship found in this study between
shrinkage of relay neurons and mean IOP is in keeping with the
correlation previously noted between atrophy of Nissl-stained neurons
and mean IOP.
17 We also observed significant neuronal
shrinkage in monkeys with ocular hypertension and no optic nerve fiber
loss, suggesting that neuronal atrophy in the LGN may be an early
event, at least partially related to elevated IOP. Statistical
determination of which of these parameters (i.e., optic nerve fiber
loss or IOP) is more important in atrophy was not possible in this
study, because of the large sample size needed for this type of
analysis.
Our results also suggest that atrophy of relay neurons in parvocellular
layers is greater than that observed in magnocellular layers in
glaucoma. In two monkeys with complete optic nerve fiber loss, neuronal
atrophy was noted in both magno- and parvocellular layers, as may be
expected. In addition, neuronal atrophy appeared to be more severe in
parvocellular layers than in magnocellular layers, in keeping with
previous studies showing that neuronal atrophy is more severe in parvo-
than in magnocellular layers after enucleation
1 6 and
similar to observations in human LGN after enucleation.
37 Investigators in a recent study were unable to detect a differential
cell size effect between magno- and parvocellular neurons in
experimental glaucoma.
17 In addition to overall shorter
survival times compared with that of the present study, the difference
may be due to the criterion used to identify neurons—namely, the
presence of a distinct nucleolus.
17 Because this organelle
is known to shrink during transneuronal degeneration, this criterion
may have introduced a bias in the selection of neurons that show less
atrophy for area measurement.
38 Finally, Nissl stain
labels relay neurons in addition to interneurons. The interneurons,
confined to the LGN, are relatively resistant to transneuronal
degeneration, and interneurons demonstrate less atrophic change
compared with relay neurons after enucleation.
21 22 This, combined with the fact that magnocellular layers are known to
have a greater percentage of interneurons than parvocellular
layers,
19 may also explain why no relative difference in
atrophy between magnocellular and parvocellular layers was detected in
the previous study.
17 In addition, the seven monkeys in
our study each had a longer survival time of 14 months compared with
survival times ranging from 0.5 to 6 months with similar mean
IOP.
17
Neuronal atrophy is believed to precede neuronal loss. Mathews et
al.
4 6 showed that whereas neuronal atrophy occurred
within 6 months after ocular enucleation, loss of neurons was not seen
until 12 months after ocular enucleation. In the present study, the
shrinkage of LGN neurons in experimental glaucoma was found at both
early and advanced stages of glaucomatous damage, and increased in a
linear fashion with optic nerve fiber loss. However, significant loss
of LGN magnocellular neurons (68% and 61%) and parvocellular neurons
(60% and 68%) was restricted to monkeys with optic nerve fiber loss
of 61% and 100%, respectively.
10 That neuronal atrophy
precedes neuronal loss is not supported by the results of a recent
study in which significant neuronal loss in the magnocellular layer was
reported as early as 2.5 weeks after IOP elevation and before the
detection of significant atrophy.
17 This discrepancy is
probably explained by the difficulty in detecting the nucleolus (the
criterion used by the investigators to identify neurons), particularly
in atrophic neurons with shrunken nucleolus.
38
In this study, optic nerve fiber loss ranged from 0% to 100%
and mean optic nerve fiber loss was 45%. Although this range of optic
nerve fiber loss may reflect the full spectrum of glaucomatous optic
nerve damage in humans, further studies are needed to assess the degree
of neuronal atrophy in LGN and its relationship to the degree of optic
nerve fiber loss in human glaucoma.
The presence of degeneration in LGN neurons has several implications
regarding progressive glaucomatous damage. The target neurons may
provide trophic support for RGCs and in fact, damage to LGN neurons has
been shown to cause RGC atrophy and degeneration.
39 The
degeneration of LGN neurons in experimental glaucoma may increase the
susceptibility of surviving RGCs to ongoing damage. Additionally, the
changes in relay LGN neurons that project to visual cortex, may explain
in part the metabolic and neurochemical changes seen in the primary
visual cortex in glaucoma.
15 18 40 41 We propose that in
addition to therapies to rescue RGCs directly, neuroprotective
strategies to rescue LGN neurons in glaucoma may further enhance RGC
survival. Indeed, a neurotrophic factor has been shown to prevent the
atrophy in LGN neurons induced by monocular visual
deprivation.
42 Although the mechanisms underlying neuronal
atrophy are not yet known, understanding the atrophic process in
glaucoma may provide insights into glaucomatous progression and its
prevention.
The authors thank Tamara Arenovich and Barbara Thomson,
Department of Statistics, University of Toronto, for assistance with
statistical analyses.