Ultrastructural Localization of the HNK-1 Carbohydrate Epitope to Glial and Neuronal Cells of the Human Retina

Abstract

purpose. To localize the cell adhesion-related HNK-1 carbohydrate epitope in the human retina at cellular and subcellular levels.

methods. Retinas were obtained from seven normal human eyes at autopsy (age, 43–78 years). The specimens were embedded in medium-grade resin and studied by postembedding immunoelectron microscopy using the primary mouse mAb HNK-1 (Leu 7) to the HNK-1 epitope and secondary antibodies conjugated to 10-nm colloidal gold particles.

results. Prominent immunolabeling with mAb HNK-1 was observed on the outer surface of the entire plasma membrane of Müller radial glial cells, including their microvilli between the inner segments of rods and cones, on the plasma membranes of astrocytes in the ganglion cell layer, in bipolar cells in the inner nuclear layer, and in photoreceptor cells in the outer nuclear layer. Fewer gold particles were present on plasma membranes of other main types of retinal neurons, including ganglion cells. Only the outer segments of rods and cones and the endothelial cells of retinal capillaries were never labeled. In the ciliary epithelium, gold particles localized to the basement membrane of the nonpigmented and pigmented layers and to the cytoplasm of the pigmented epithelium.

conclusions. Unlike in many other species, the HNK-1 epitope in the human retina is found on both glial and neuronal cells, including photoreceptors. This epitope potentially contributes to neuron-to-neuron and glia-to-neuron adhesion of human retinal cells.

The HNK-1 epitope, defined as SO₄-3GlcAβ(1→3)Galβ(1→4)GlcNAc, ¹ is a prominent developmentally regulated carbohydrate sequence present in the retina and retinal pigment epithelium of vertebrates. ^{2 3 4 5 6} It is part of several plasma membrane and extracellular matrix glycoproteins, glycolipids, and proteoglycans involved in cell-to-cell and cell-to-substratum adhesion. ^{6 7 8 9}  

In the human eye, the HNK-1 epitope can also be found in subepithelial matrix cells of the inner connective tissue layer and pigmented and nonpigmented epithelium of the ciliary body, in the posterior pigmented epithelium of the iris, in the lens capsule and zonular lamella, in scleral fibroblasts, and in optic and ciliary nerves. ^{2 6 10 11 12 13 14} Immunoelectron microscopy has localized it to plasma membranes of the subepithelial matrix cells, microfibrillar bundles, and elastic fibers in the inner connective tissue layer, to the basement membranes of both layers of the ciliary epithelium, as well as to the cytoplasm of the pigmented ciliary epithelium. ^{12 13 14} In the lens, the HNK-1 epitope is present in the outer three fourths of the central anterior capsule, as visualized by immunoelectron microscopy. ^{12 13}  

Unlike in the anterior segment of the eye, the localization of the HNK-1 epitope in the vertebrate retina is not known in detail. Because of the extensive membrane-associated immunolabeling and the intimate approximation of retinal glial and neuronal cell membranes, light microscopy has allowed unquestionable localization of the HNK-1 epitope only to Müller radial glial cells in humans. ^{2 3 4 15} To gain a better understanding of the detailed distribution and subcellular compartmentalization in the human retina of the HNK-1 epitope, we determined by immunoelectron microscopy which additional cell types bear this carbohydrate sequence and to which cell organelles the immunoreaction localizes. 

Methods

Tissue Specimens and Tissue Processing

Retinal specimens from the posterior part of the eye were obtained from seven human cadaveric eyes that were used as donors eyes in corneal transplantation (age, 43–78 years; two men and five women). The donors had no known ocular disease and died of major injury, cerebral vascular accident, and cancer. Ciliary body and corneal specimens taken from the donor eyes were studied as positive and negative controls, respectively. 

For postembedding immunogold labeling, tissue specimens approximately 2 × 2 mm in size were fixed within 1 to 5 hours after death in freshly prepared 4% (wt/vol) paraformaldehyde and 0.1% (vol/vol) glutaraldehyde in 0.1 M cacodylate buffer (pH 7.4) for 2 hours at 4°C. After fixation, the specimens were rinsed in 0.1 M cacodylate buffer containing 3.7% (wt/vol) saccharose and sequentially dehydrated in 30% (vol/vol) ethanol at 4°C, 50% (vol/vol) ethanol at 0°C, and 70% (vol/vol) ethanol at −20°C. The specimens were then infiltrated with a 1:1 mixture of 70% ethanol and medium grade resin (LR White; Electron Microscopy Sciences, Fort Washington, PA) at 4°C and several changes of pure resin at 4°C and room temperature and finally polymerized in gelatin capsules for 24 hours at 45°C. Ultrathin sections were cut with a microtome (Ultracut; Leica, Bensheim, Germany) and mounted on uncoated nickel grids. Part of each specimen was also fixed in 2.5% glutaraldehyde in 0.1 M phosphate buffer (pH 7.4), postfixed in 2% osmium tetroxide in phosphate buffer, and routinely processed for embedding in epoxy resin (Epon; Serva, Heidelberg, Germany). 

The study was approved by the institutional review boards, and it adhered to the tenets of The Declaration of Helsinki. 

Immunogold Labeling

The mouse monoclonal antibody, mAb HNK-1 (Leu 7; IgM, diluted 1:100; Lot 50808; BD Bioscience, San Jose, CA) was used to detect the HNK-1 epitope. ¹⁶ This antibody has been found to react in immunoelectron microscopy. ^{12 13 14} An unrelated mouse mAb to cytokeratin 14, CK B1 (IgM, diluted 1:100; Lot 100H4800; Sigma, St. Louis, MO), was used as a control to exclude nonspecific binding of antibodies of the same isotype. Goat anti-mouse antiserum that reacts with mouse IgG and IgM, conjugated to 10-nm colloidal gold particles (diluted 1:30; Bio Cell, Cardiff, Wales, UK), was used as the secondary antibody. 

For postembedding immunogold labeling, ultrathin sections were incubated successively in drops of Tris-buffered saline (TBS, pH 7.4) for 5 minutes at room temperature; 0.05 M glycine in TBS for 30 minutes at room temperature, 0.5% (wt/vol) ovalbumin and 0.5% (vol/vol) teleost fish gelatin in TBS, both for 30 minutes at room temperature; and primary antibody diluted in TBS-ovalbumin overnight at 4°C. After 5 washing steps in TBS-ovalbumin-gelatin, the grids were incubated in 10-nm gold-conjugated secondary antibody diluted in TBS-ovalbumin-fish gelatin for 1 hour at room temperature. After being rinsed in TBS-ovalbumin-gelatin, TBS, and water (five times each), the sections were stained with 2.5% uranyl acetate for 3 minutes and examined with a transmission electron microscope (EM 906E; LEO, Oberkochen, Germany). 

Results

In the ciliary body, used as an internal positive control, gold particles localized to the periphery of subepithelial matrix cells, to microfibrillar bundles and elastic fibers of the inner connective tissue layer, and to basement membranes of the ciliary epithelium, as expected. ¹⁴ The basolateral infoldings of the nonpigmented ciliary epithelial cells were labeled only at crests of ciliary processes with no labeling in the valleys, and the cytoplasm of the pigmented layer was immunolabeled. No reaction was observed in the cornea, serving as a negative control. 

In the retina, all layers except the outer portion of the photoreceptor cell layer reacted with mAb HNK-1. Gold particles localized to the inner limiting membrane (Fig. 1A , ilm), formed by plasma membrane of Müller cells (mc), but vitreous collagen was not labeled. In the inner retina, gold particles were present along the inner processes of Müller cells (Fig. 1A) , on the cell membranes of astrocytes (Fig. 1B , ac), and along ganglion cell (gc) axons (Figs. 1B 1C 1D) , even when adjacent Müller cell processes were not identified (Fig. 1C) . In the inner plexiform layer, gold particles were also found along Müller cell processes and, to a lesser extent, on neuronal processes (Fig. 1E , nc). Gold particles were three to four times more abundant on glial than neuronal plasma membranes in these layers. 

In the middle third of the retina, immunostaining for the HNK-1 epitope was present on the plasma membranes of amacrine cell and horizontal cell bodies, in addition to Müller cell processes. Particles were prominent on plasma membranes of bipolar cells (Fig. 1F) , where they were as numerous as on the Müller cell bodies. In the outer plexiform layer, plasma membranes of both Müller cell and neuronal processes similarly were labeled with approximately equal numbers of gold particles (Figs. 1G 1H) . 

In the outer third of the retina, the membranes of perikarya and processes of rods and cones (Fig. 2A) and of the outer processes (Fig. 2A) and microvilli (Fig. 2B) of Müller cells were labeled with gold particles. The density of gold particles appeared equal on both cell types. Some gold particles were present on the inner segments of rods (Fig. 2B , ris) and cones (cis), where no Müller cell microvilli were identified. The outer segments of rods (Fig. 2C , ros) and cones (cos), as well as the mucopolysaccharide-rich interphotoreceptor matrix (ipm) remained unlabeled. 

No immunogold labeling was detected in endothelial cells of retinal vessels. No qualitative difference in immunostaining was evident between specimens from patients of different ages. 

Discussion

Our study confirms that the HNK-1 epitope, which is considered elementary for the function of diverse cell-adhesion molecules (for a review, see Ref. ⁶ ), is most prominently associated with Müller cells in the human retina. ^{2 3} This finding is consistent with the stabilizing role assigned to the radial glial cells. ¹⁷ We were also able to find this epitope on astrocytes and all main types of retinal neurons, especially bipolar cells, rods, and cones. During development, this epitope appears first in the neuroblastic layers. ¹¹ Consequently, in addition to glia-to-neuron, glia-to-glia, and neuron-to-extracellular matrix interactions, ^{7 8 9} the HNK-1 epitope may act in promoting neuron-to-neuron adhesion in the human retina. As judged by the number of gold particles, neuron-associated molecules that bear the HNK-1 epitope in the inner retina were more sparse than glia-associated molecules, whereas they were almost equally numerous in the middle and outer retinal layers. 

The human retina differs from most vertebrate retinas, in which the HNK-1 epitope is found mainly or exclusively on neurons and predominantly in the plexiform layers. In amphibians, reptiles, birds, and mammals with rod-dominated retinas, this epitope is observed in populations of ganglion, amacrine, bipolar, and horizontal cells, as in the human retina, but apparently not in glial and photoreceptor cells. ^{4 5 18 19 20 21 22 23} However, in these species immunoelectron microscopy has not been used, and low-grade expression by retinal glia consequently cannot be totally ruled out. 

In fish and primates, the HNK-1 epitope is also found on Müller radial glial cells, analogous to the human retina. ^{4 5} Similar immunoreactivity in tree squirrels has suggested a theory that expression of the HNK-1 epitope by mammalian radial glial cells may be linked with cone predominance. ⁵ The density of gold particles along rods and cones of humans in this study was shown to be similar by immunoelectron microscopy, indicating that both types of photoreceptor cells nevertheless participate in HNK-1 epitope-mediated interactions in the cone-dominant human retina, in which the Müller cell was a major cell type bearing this epitope. 

We did not detect any intracytoplasmic labeling for the HNK-1 epitope in the human retina, in contrast to the pigmented ciliary epithelium, in which intracytoplasmic gold particles were numerous. ^{12 14} Although the specimens were from cadaveric eyes, which may have had diminished immunoreactivity, it seems that no more than minor amounts of molecules bearing this epitope could have been in transit in the cytoplasm. The HNK-1 epitope-bearing molecules in the retina must therefore have a slow turnover rate, as would be fitting of macromolecules that supposedly have a mainly constitutive, structural role. Alternatively, the epitope is added at the cell surface when the corresponding macromolecules become incorporated into the plasma membrane or adjacent extracellular matrix. So far, we know that the HNK-1 epitope is added at least to myelin-associated glycoprotein intracellularly. ²⁴  

Although most gold particles were clearly associated with plasma membranes, it was difficult to exclude the possibility that some immunoreactivity may have been due to molecules secreted to the extracellular matrix. No immunoreactivity was observed in the interphotoreceptor matrix, however, secreted by the retinal pigment epithelial and Müller cells, ²⁵ which themselves are immunopositive for the HNK-1 epitope. ^{2 26}  

Given our observation that the HNK-1 epitope is present in neurons of the human retina, including photoreceptor and bipolar cells, it is of note that retinoblastoma cells do not react to this epitope. ^{2 3} Although considered to be mainly neuroblastic in nature, with differentiation toward photoreceptor cells and sometimes neurons of the inner nuclear layer, ^{27 28} they seem to be unable to express the HNK-1 epitope. It is of note that retinoblastoma is a noncohesive rather than a solid cancer, which typically seeds satellites to the vitreous and can even give rise to a pseudohypopyon of floating cells in the anterior chamber. The stable expression of the HNK-1 epitope in normal human retinal neurons throughout adult life, as shown in the current study, and its absence from noncohesive retinoblastoma cells ^{2 3} are observations consistent with the proposed structural function for the HNK-1 epitope in glial and neuronal adhesion. 

 Supported by grants from The Academy of Finland, The Finnish Eye Foundation, The Friends for the Blind, and The Mary and Georg C. Ehrnrooth Foundation.

 Submitted for publication May 22, 2002; revised September 23, 2002; accepted October 3, 2002.

 Commercial relationships policy: N.

 The publication costs of this article were defrayed in part by page charge payment. This article must therefore be marked “advertisement” in accordance with 18 U.S.C. §1734 solely to indicate this fact.

 

Figure 1.

 

Immunoelectron microscopic localization of the HNK-1 epitope in the inner human retina. (A) Gold particles concentrated in the inner limiting membrane (ilm) and plasma membrane of Müller cells (mc). Vitreous collagen (arrowhead) was unlabeled. (B) In the ganglion cell layer, numerous particles lined the plasma membrane (arrowhead) of astrocytes (ac). Some gold particles were present on cell membranes (open arrowhead) of ganglion cells (gc). (C) Gold particles localized to membranes (arrowhead) of ganglion cells (gc) even when adjacent Müller cell processes could not be identified. (D) In the nerve fiber layer, gold particles localized to plasma membranes of Müller cell (mc) processes, recognizable by dark cytoplasm among ganglion cells (gc). (E) In the inner plexiform layer, gold particles concentrated around processes of Müller cells (mc), but some were also associated with neuronal processes (nc). (F) Gold particles marked plasma membranes (arrowhead) of bipolar cells in the inner nuclear layer, but none were seen within their nuclei (bn) and cytoplasm (bc). In the outer plexiform layer (G, H) the particles label membranes of Müller cells (mc) and neurons (nc), highlighted at neuron-to-neuron interphases (arrowheads). Original magnification: (A) ×29,300; (B) ×69,900; (C) ×91,700; (D) ×58,700; (E) ×73,500; (F) ×47,300; (G) ×74,700 (H) ×62,400; scale bar, 500 nm.

Figure 2.

 

Immunoelectron microscopic localization of the HNK-1 epitope in the outer human retina. (A) In the outer nuclear layer, immunogold lined the plasma membranes of Müller cells (mc) and rods, as confirmed at rod-to-rod (arrowheads) and Müller-to-Müller cell interphases (open arrowhead). Nuclei (rn) and cytoplasm (rc) of rods did not bind immunogold. (B) Outside the outer limiting membrane (olm), gold particles localized to the microvilli of Müller cells (arrowheads) between the inner segments of rods (ris) and cones (cis). No immunogold was seen external to where the microvilli terminate (open arrowheads). (C) The outer segments of rods (ros) and cones (cos), and the interphotoreceptor matrix (ipm) between them, remained unlabeled. Original magnifications: (A) ×40,300; (B) ×41,900; (C) ×25,600; scale bar, 500 nm.