We examined neurofilament staining in the normal and visually deprived lateral geniculate nucleus (LGN), using the SMI-32 antibody. This antibody preferentially stains LGN cells that display the morphological characteristics of Y-cells. The soma sizes of SMI-32-stained cells were consistent with those of the overall population of Y-cells, and the Golgi-like staining of their dendrites revealed a radial distribution that often crossed laminar boundaries. Labeled cells were distributed within the A laminae (primarily near laminar borders), the magnocellular portion of the C laminae, and the medial intralaminar nucleus, but they were absent in the parvocellular C laminae. Electron microscopic examination of SMI-32-stained tissue revealed that staining was confined to somata, dendrites, and large myelinated axons. Retinal synapses on SMI-32-labeled dendrites were primarily simple axodendritic contacts; few triadic arrangements were observed. In the LGN of cats reared with monocular lid suture, SMI-32 staining was decreased significantly in the A laminae that received input from the deprived eye. Dephosphorylation of the tissue did not alter the cellular SMI-32 staining patterns. Analysis of staining patterns in the C laminae and monocular zone of the A laminae suggests that changes in the cytoskeleton after lid suture reflect cell class and not binocular competition. Taken together, the results from normal and lid-sutured animals suggest that the cat LGN offers a unique model system in which the cytoskeleton of one class of cells can be manipulated by altering neuronal activity.
The SMI-32 antibody stains the nonphosphorylated form of the high-molecular-weight neurofilament protein. It has been used extensively to document cytoskeletal changes in a number of neurological disorders (Sternberger et al., 1985; Troncoso et al., 1986; Hof and Morrison, 1990; Hof et al., 1990;Vickers et al., 1992; Duong and Gallagher, 1994; Gai et al., 1994;Smith et al., 1995; Su et al., 1996). Decreases in neurofilaments, or abnormal phosphorylation of neurofilaments, appear to be frequent consequences of neuronal disease or damage, but the sequence of events leading to these changes is unknown. An understanding of the normal functions of the cytoskeleton, and the potential transformations that occur during neuronal disease, requires an animal model. The lateral geniculate nucleus (LGN) of the cat is an excellent candidate for such a model. Its anatomy and physiology are well characterized, both in the normal condition as well as after experimental manipulations that produce subtle, yet reproducible, changes in neuronal morphology.
The pathway from retina through the LGN is composed of at least two (X and Y), perhaps three (W), morphologically and physiologically distinct neuronal streams. Each is designed to analyze different aspects of the visual scene (Sherman, 1985). These pathways develop at different rates and also respond differently to abnormal visual input (Friedlander et al., 1982; Sherman and Spear, 1982; Sur et al., 1982, 1984; Garraghty et al., 1986, 1988). For example, Y-cells seem particularly susceptible to monocular lid suture (MS). After early periods of MS, Y-cells in the LGN develop abnormal receptive field properties and show a reduced soma size and anomalous dendritic morphology. The abnormally thin and tangled dendrites of Y-cells in deprived laminae suggest an alteration in cytoskeletal support. Thus, MS may be a useful model to study how modifications in neuronal activity can alter the organization of neurofilaments.
The aim of the present study was to examine the staining pattern of the SMI-32 antibody in the LGN of normal cats and cats raised with MS. This antibody recently has been shown to stain preferentially the magnocellular layers of the monkey LGN (Gutierrez et al., 1995;Chaudhuri et al., 1996). However, SMI-32 staining has yet to be explored in the cat LGN, where the Y-cell pathway is well characterized and can be manipulated readily by altering visual input.
MATERIALS AND METHODS
A total of five cats were used for these experiments. Two were normal adult cats. Three other adult cats were reared with monocular lid suture that was performed before eye opening and was maintained until death (for details on lid suture, see Friedlander et al., 1982). The cats were given an overdose of sodium pentobarbital and perfused through the heart with saline, followed by a fixative solution of 4% paraformaldehyde (one normal and two MS) or 4% paraformaldehyde and 0.5% glutaraldehyde (one normal) or 2% paraformaldehyde and 0.1% glutaraldehyde (one MS) in 0.1 m sodium phosphate buffer (PB), pH 7.4. Previous studies that used the SMI-32 antibody excluded the use of glutaraldehyde fixation (Campbell and Morrison, 1989). However, we found that, although the low percentage of glutaraldehyde slightly decreased the staining intensity, the overall staining pattern was similar with each fixation protocol.
After fixation, the brains were removed and cut in the coronal or sagittal plane into 50-μm-thick sections with a vibratome. Series of sections from all five cats were stained with the SMI-32 antibody. Series of sections from two cats (one normal and one MS) were mounted on slides and stained for Nissl substance. Additional sections from two cats (one normal and one MS) were incubated in alkaline phosphatase (400 μg/ml; Sigma type VII-L, Sigma, St. Louis, MO) in 0.1m Tris buffer, pH 8, for 2 hr at 37°C, rinsed in Tris buffer, and stained with the SMI-32 antibody as described below.
For immunocytochemistry, sections through the LGN were incubated in 10% normal goat serum (NGS) in PBS (0.01 m; 0.9% NaCl) for 30 min. Then the sections were transferred to a solution of the SMI-32 antibody (monoclonal, made in mouse, Sternberger Monoclonal, Jarrettsville, MD) diluted 1:5000, 1:10,000, or 1:20,000 in 1% NGS in PBS with 0.5% Triton X-100 and incubated overnight with agitation at 4°C. The next day the sections were rinsed three times (10 min each) in PB and incubated for 1 hr in a 1:100 dilution of biotinylated goat anti-mouse antibody (Vector Laboratories, Burlingame, CA) in 1% NGS in PBS. Then the sections were rinsed three times (10 min each) in PB and incubated for 1 hr in a 1:100 dilution of a complex of avidin and biotinylated horseradish peroxidase (ABC; Vector Laboratories) in 1% NGS in PBS. The sections were rinsed three times (10 min each) and reacted with nickel-enhanced diaminobenzidine solution (Adams, 1981) for 5 min. The sections were rinsed and mounted on slides or prepared for electron microscopy as described below.
For electron microscopy the sections were post-fixed in osmium (2% in PB) for 1 hr, rinsed, dehydrated in an ethanol series, and embedded in Durcupan resin (Ted Pella, Redding, CA). Sections were cut from the resin blocks on an ultramicrotome, collected on Formvar-coated slot grids, stained with uranyl acetate (10% in methanol) for 30 min, and examined with a transmission electron microscope. SMI-32-labeled dendrites postsynaptic to retinal terminals were photographed.
Drawings of soma sizes from Nissl- and SMI-32-stained tissue were made with a camera lucida. The areas of these drawings subsequently were measured by a digitizing tablet and SigmaScan computer software (Jandel Scientific, Corte Madera, CA). For statistical analysis of soma sizes, a two-tailed nonparametric Mann–Whitney U test was used to evaluate the differences between population means. To determine the underlying distributions of cell soma size, we pooled the soma area data from the A laminae. We assumed for this analysis that the underlying distributions of neuronal soma sizes were distributed normally. We fit the histograms derived from these pooled data with gaussian functions, using the PStat program (Axon Instruments, Foster City, CA). Using a Simplex fitting algorithm, we achieved the best fit (least mean error per data point) for the Nissl data with two gaussian terms and with one gaussian term for the SMI-32 data. We evaluated with a comparison of Z scores whether significant differences existed between the gaussian distributions derived from these fits.
Nature of SMI-32 label in the normal LGN
A number of observations suggest that SMI-32 staining in the normal LGN is restricted to the Y-cell population. First, the distribution of staining is similar to that of Y-cells (Sherman, 1985). As shown in Figure 1, cells are distributed in laminae A, A1, the magnocellular portion of C, and the medial intralaminar nucleus (MIN), but the cells are lacking in parvocellular C laminae (C1–C3). Within the A laminae, the majority of the SMI-32-stained cells is near the laminar boundaries.
In addition, the morphology of SMI-32-stained cells is similar to that of Y-cells. As shown in Figures 1 C and2, well stained cells exhibit class I morphology (Guillery, 1966; Friedlander et al., 1981; Wilson et al., 1984; Raczkowski and Sherman, 1985); they have large somata, with numerous radiating dendrites. SMI-32-stained dendrites frequently cross the interlaminar zone between the A laminae as well as the boundary between the A1 and C laminae (Figs. 1 C, 2 C).
Like Y-cells, SMI-32-stained cells are the largest in the LGN, as demonstrated by a comparison of the soma areas of cells stained for Nissl substance to those stained with the SMI-32 antibody (Fig.3). The size of SMI-32-stained cells is significantly different from the overall population of LGN cells (Fig.3 A; n = 400 Nissl and 400 SMI-32;p < 0.0001) as well as cells in either laminae A or A1 (Fig. 3 B,C; n = 100 Nissl and 100 SMI-32;p < 0.0001 for lamina A and A1). In the magnocellular C lamina there is a slight, but significant, difference in the size distribution of SMI-32-stained cells and Nissl-stained cells (Fig.3 D; n = 100 Nissl and 100 SMI-32;p < 0.003). In the interlaminar zone between lamina A and A1 there was no significant difference between the sizes of SMI-32-stained and Nissl-stained cells (Fig. 3 E;n = 100 Nissl and 100 SMI-32).
In addition to these comparisons, we also examined whether there were significant differences between the population of larger (presumably Y) cells and the SMI-32-stained neurons. The LGN A laminae contain three major types of cells: X- and Y-cells and interneurons. The soma size distributions of these populations overlap, which prompted our effort to fit gaussians to the data set in an attempt to derive the underlying distribution functions of these cells. However, we note that the best fits of the pooled A laminae Nissl data were achieved with two gaussian terms. This likely reflects that two of these types could not be discriminated with the sample size we used. Of the three, previous studies have shown the most overlap between X-cells and interneurons (Friedlander et al., 1981). Thus, the distribution of large Nissl-stained cells likely represents the distribution of Y-cells. As is apparent from the graphical comparison of gaussian distributions in Figure 4, the distribution of SMI-32-stained neurons was significantly different from that of the small Nissl-stained cells (p < 0.001), but not significantly different from the distribution of large Nissl-stained cells. Thus, the most reasonable interpretation of these data is that the antibody is labeling Y-cells.
Electron microscopic examination of SMI-32-stained tissue (lamina A) confirmed that label is found in somata and dendrites. Within somata the label displayed a patchy distribution that was not clearly associated with any particular organelle (Fig.5 A). The label was seen to coalesce as dendrites emerged from somata, and most dendritic staining was distributed evenly throughout the cytoplasm. In some cases, reaction product was denser toward synaptic contact zones or puncta adherentia (Fig. 5 B–E).
A sample of 52 synaptic contacts made by retinal terminals onto SMI-32-stained dendrites was examined. Retinal terminals can be identified on the basis of their morphology as largeterminals with round vesicles and palemitochondria (RLP profiles). Previous studies indicate that Y-cells receive retinal contacts on proximal dendritic shafts; the synaptic arrangements of RLP profiles on Y-cells are generally less complex than those made by RLP profiles that innervate the dendritic appendages of X-cells (Wilson et al., 1984; Hamos et al., 1986, 1987;Bickford et al., 1992). Consistent with the hypothesis that SMI-32 stains Y-cells, a majority of contacts between RLP profiles and SMI-32-stained dendrites were simple axodendritic contacts on dendritic shafts. Occasionally (13 of 52 or 25%), retinal terminals presynaptic to SMI-32-labeled dendrites also contacted dendritic terminals (F2 profiles; Fig. 5 B). However, only one triadic arrangement (frequently seen in the X-cell pathway; Wilson et al., 1984; Hamos et al., 1987) was encountered in the sample that was examined.
Axonal labeling with SMI-32 was also consistent with staining in Y-cells. Intensely stained axonal profiles were of large caliber and heavily myelinated (Friedlander et al., 1981; Humphrey et al., 1985a,b). These axons did not contain pale mitochondria and thus were not retinal axons. The distribution and size of these axons suggested that they were thalamocortical Y-cell axons, most likely originating from the SMI-32-stained cells. However, if this is the case, the staining did not extend far from the somata, because axonal labeling was not seen in the optic radiations.
SMI-32 labeling after early monocular lid suture
Because the SMI-32 antibody appears to stain Y-cells preferentially, it was of interest to examine staining with this antibody in cats raised with MS, a manipulation that affects the form and function of the Y-cell pathway. As illustrated in Figures6 and 7, SMI-32 staining is reduced dramatically in geniculate A laminae deprived of normal visual input. The number of well stained cells and the general neuropil staining are both decreased in the deprived laminae when compared with the nondeprived laminae. Within the nondeprived laminae, cells with class I morphology continue to stain with the SMI-32 antibody. Within the deprived laminae, occasional cells that display class I morphology are labeled, but the staining intensity is decreased when compared with cells stained in the nondeprived lamina.
In contrast to the A laminae, the SMI-32 staining in lamina C is intense on both sides of the LGN of monocularly deprived cats. As previously described (Wiesel and Hubel, 1963; Guillery and Stelzner, 1970; Hickey, 1980; Murakami and Wilson, 1983, 1987), the deprived lamina C is slightly thinner than the nondeprived lamina C (Figs.6-8). Nonetheless, both deprived and nondeprived lamina C cells are well stained with the SMI-32 antibody and display normal class I morphology. However, some differences were detected when the soma sizes of lamina C cells on either side of the LGN were compared. In each of the three cases the SMI-32-stained cells in deprived lamina C were slightly smaller (15%) than those in the nondeprived lamina C. The size differences of samples of 100 SMI-32-stained cells in the deprived and nondeprived C laminae were significant in each of the three cases (Fig.9; case 97-11, p < 0.001; case 97-16, p < 0.0001; case 97-17,p < 0.0001).
It is also apparent that, unlike other changes associated with MS, the reduction in SMI-32 staining occurs in both the binocular and monocular segments of lamina A. As previously reported for Nissl-stained sections (Guillery and Stelzner, 1970), cells in the deprived binocular segment are noticeably smaller than those in the monocular segment (Fig.10 A,B). However, in adjacent sections it is apparent that, although monocular segment cells have normal soma sizes, they do not retain their normal capacity to stain with the SMI-32 antibody (Fig. 10 C,D). In contrast, the deprived monocular segment of lamina C is stained with the SMI-32 antibody, indicated by a band of label beneath the monocular segment of lamina A.
Alkaline phosphate pretreatment
To begin to examine possible mechanisms that could account for the decrease in SMI-32 staining after monocular deprivation, we pretreated sections with alkaline phosphatase before immunocytochemical staining. This treatment has been used previously to reveal abnormal phosphorylation of neurofilaments (Su et al., 1996). As shown in Figure11 A, SMI-32 staining in normal tissue revealed few axons because axonal neurofilaments normally are phosphorylated, and the SMI-32 antibody recognizes only nonphosphorylated neurofilament proteins. As shown in Figure11 B, dephosphorylation of the tissue before SMI-32 staining resulted in a dramatic increase in axonal staining. However, the alkaline phosphatase pretreatment did not affect the overall pattern of cellular staining in the monocularly deprived LGN. The nondeprived laminae served as an internal control for any effects of the alkaline phosphatase treatment on the intensity of cellular SMI-32 staining. After alkaline phosphatase pretreatment the SMI-32 antibody stained cells in the nondeprived laminae. However, staining was still reduced in the deprived A laminae as compared with the nondeprived A laminae. Thus, it is unlikely that abnormal phosphorylation of neurofilaments accounts for the loss of SMI-32 staining in LGN cells deprived of normal visual input.
The SMI-32 antibody as a marker for Y-cells
This study shows that the SMI-32 antibody preferentially stains Y-cells in the cat LGN. The distribution, morphology, and synaptic arrangements of LGN cells stained with this antibody match the characteristics of Y-cells. This is consistent with previous reports that the SMI-32 antibody preferentially stains the magnocellular layers in the monkey LGN (Gutierrez et al., 1995; Chaudhuri et al., 1996). In other areas of the brain, specific subsets of neurons stain with the SMI-32 antibody. In general, cells stained by this antibody are large, with thick dendrites and/or large-diameter axons. For example, the SMI-32 antibody stains the stout apical dendrites of cortical pyramidal cells (Campbell and Morrison, 1989; Carmichael and Price, 1994; del Rio and Defelipe, 1994; Cusick et al., 1995; Hof and Morrison, 1995; Hof et al., 1995). Within the visual system large cells in other structures are labeled. For example, the SMI-32 antibody labels a subset of large neurons in the pulvinar nucleus of the monkey (Gutierrez et al., 1995), and, in the retina, a relation to the Y-cell pathway has been demonstrated; in a variety of species SMI-32 or other neurofilament antibodies have been shown to stain large α-ganglion cells (Drager et al., 1984; Gabriel and Straznicky, 1992; Straznicky et al., 1992).
Evidence that neurofilament changes are attributable to cell class competition
Two mechanisms have been proposed to account for the changes seen in the LGN after MS, namely, cell class and binocular competition. Cell class competition arises in laminae A and A1, where X and Y retinogeniculate arbors co-mingle and compete for terminal space on their geniculate cell counterparts. After MS, Y retinogeniculate arbors are reduced significantly, whereas the X retinogeniculate arbors maintain a broad terminal field (Sur et al., 1982; Garraghty et al., 1986). Thus, the Y-relay cells receive less, or abnormal, retinal input (Friedlander et al., 1982). Binocular competition arises in portions of the LGN where input from the two eyes innervate separate laminae. In this case, competitive mechanisms in the cortex result in reduced thalamocortical arbors arising from cells in the deprived laminae (Shatz and Stryker, 1978; Friedlander and Martin, 1991). This leads to a corresponding reduction in the soma sizes of cells in the binocular portions of the deprived laminae (Guillery and Stelzner, 1970;Guillery, 1972). It has been proposed that the abnormal properties of Y-cells seen in the binocular portions of the A laminae result from the combined effects of both cell class and binocular competition (Spear et al., 1989).
Our results suggest that the reduction in SMI-32 staining that accompanies MS is attributable to cell class competition and not to binocular competition. SMI-32 staining is reduced only in A-laminae, where X and Y retinogeniculate arbors compete for terminal space. In laminae C, where cells are subject to binocular competition but not cell class competition, cells retain their ability to stain with the SMI-32 antibody. In the monocular zone of lamina A, where cells are subject to cell class competition but not binocular competition, SMI-32 staining is lost (Sherman and Spear, 1982; Spear et al., 1989).
Indeed, these results also suggest that the consequences of MS on Y-cell morphology are attributable to different mechanisms. Changes in the cytoskeleton of Y-cells, which presumably account for the changes seen in dendritic morphology (Friedlander et al., 1982), appear to be attributable to cell class competition. In contrast, changes in soma size that appear to occur independent of the cytoskeletal changes are attributable mainly to binocular competition (Guillery and Stelzner, 1970; Guillery, 1972; Sherman et al., 1975). Thus, the nature of the input that Y-cells receive may have a major influence on the organization of their neurofilaments, whereas the extent of their axonal arbors influences their soma size.
Modifications of the neuronal cytoskeleton
Our results suggest that abnormal input to Y-cells induces a restructuring of their cytoskeleton. Other reported changes in neurofilaments, resulting from neurological disorders and after a variety of experimental conditions, indicate the dynamic nature of the neuronal cytoskeleton (Sternberger et al., 1985; Troncoso et al., 1986;Hof and Morrison, 1990; Hof et al., 1990; Vickers et al., 1992; Duong and Gallagher, 1994; Gai et al., 1994; Smith et al., 1995; Su et al., 1996). In the visual cortex of cats with monocular lid suture or monkeys with monocular tetrodotoxin injections, neurofilament staining becomes patchy (Eckert et al., 1997; Yoshioka, 1997), suggesting that cortical cells receiving input from the deprived eye also undergo cytoskeletal changes.
A well documented change in neurofilaments is abnormal phosphorylation. Generally, in adult tissue, neurofilaments in somata and dendrites are not phosphorylated, but those in axons are (Matus, 1988). The phosphorylation of neurofilaments in axons may allow the bundling of groups of neurofilaments, which confers compactness and stability to the axonal cytoskeleton. In Alzheimer’s disease the neurofilaments in somata become abnormally phosphorylated, and this abnormal phosphorylation appears to accompany the formation of neurofibrillary tangles (Sternberger et al., 1985; Duong and Gallagher, 1994). Additionally, it is the cortical neurons that normally stain with the SMI-32 antibody that are particularly susceptible to cytoskeletal changes during Alzheimer’s disease (Hof et al., 1990). One explanation for the vulnerability of SMI-32-positive neurons to cytoskeletal changes may be that the large C terminus of the NF-H protein has a high affinity for several kinases that phosphorylate the protein (Wible et al., 1989; Xiao and Montiero, 1994).
We ruled out abnormal phosphorylation of neurofilaments as a reason for the decreased SMI-32 staining of Y-cells in the deprived laminae, because removal of phosphate in sections pretreated with alkaline phosphatase did not alter the cellular staining pattern. A similar result was found in the substantia nigra of brains from patients with Parkinson’s disease (Gai et al., 1994). Compared with the substantia nigra of control brains, the percentage of SMI-32-stained cells in the diseased brains was reduced, and this pattern remained in tissue pretreated with alkaline phosphatase. These authors suggested that the reduced SMI-32 staining might mark an initial stage of neuronal degeneration in which neurofilaments are one of the first proteins to be degraded.
Because LGN cells in deprived laminae do not degenerate, it is unlikely that the decreased SMI-32 staining shown in this study represents a process of neuronal degeneration. Thus, it remains to be determined whether deprived LGN Y-cells lack neurofilaments or whether their neurofilaments are reorganized such that they are not recognized by the SMI-32 antibody. In either case, our results demonstrate that changes in synaptic input can influence the organization of the neuronal cytoskeleton and suggest that the cat LGN offers a novel model system to study the normal function and pathological reorganization of neurofilaments.
This work was supported by National Institute of Neurological Diseases and Stroke Grant R29NS35377 and National Science Foundation Grant 97628089 to M.B., National Science Foundation Grant 9396270 to W.G., and National Eye Institute Grant EY11695 to D.G. We thank Martin Boyce for his excellent technical assistance.
Correspondence should be addressed to Dr. Martha E. Bickford, Department of Anatomical Sciences and Neurobiology, 500 South Preston Street, University of Louisville, School of Medicine, Louisville, KY 40292.