Hsp25 and -90 Immunoreactivity in the Normal Rat Eye

Abstract

purpose. The distributions of heat shock protein (Hsp)25 and -90 in various regions of the rat eye are described to provide a basis for understanding their roles in normal, damaged, and diseased ocular tissues. This work complements the earlier examination of Hsp70 and Hsc70 (the constitutive form).

methods. Eyes of adult male Sprague-Dawley rats were fixed in methacarn and embedded in paraffin. Sagittal sections (10 μm) through the optic nerve were stained with hematoxylin and eosin, or incubated with anti-Hsp90, anti-Hsp25, or control IgG. Bound antibody was visualized using an avidin-biotin–horseradish peroxidase detection system.

results. Hsp90 immunoreactivity was abundant in the retina, whereas only low levels of Hsp25 were detected there. In the optic nerve, the relative difference in immunoreactivity for the two Hsps was reversed, with Hsp25 being considerably greater than Hsp90. Both Hsps were detected at low levels in the retinal pigment epithelium (RPE), except for that portion within 250 μm of the optic disc, where Hsp25 and -90 immunoreactivities were increased. Similar to the optic nerve, the corneal epithelium showed greater staining for Hsp25 than for Hsp90, and basal cells contained the highest levels of immunoreactivity for both Hsps. In the ciliary body and iris, Hsp25 and -90 were abundant and similarly distributed in the epithelial and stromal layers.

conclusions. Each of the ocular tissues had distinctive patterns of Hsp25 and -90 immunostaining. These results suggest that the various structures of the eye have unique requirements for the particular chaperoning and supportive functions of these two Hsp families.

When living cells are exposed to metabolic stress, they respond by increasing transiently the expression of a small subset of proteins known as the heat shock proteins (Hsps). These proteins, so named because temperature elevation (heat shock) was the first-described inducer of the response, have been found in virtually all organisms and are highly conserved phylogenetically. In mammals, Hsps are induced by a wide range of noxious stimuli, such as hypoxia and ischemia, as well as by damaging chemical agents, such as metabolic poisons and heavy metals. The major Hsp families in mammals consist of sets of proteins of approximately 110 kDa, 90 kDa, 70 kDa, 60 kDa, and 15 to 30 kDa. The most studied is the 70-kDa family. ^{1 2}  

In recent years, one of the Hsp families that has attracted serious interest is the 90-kDa family (Hsp90 family; for review, see Ref. ³ ). In mammals, Hsp90, a cytosolic protein, can exist in one of two forms, known as Hsp90α (or Hsp84) and Hsp90β (or Hsp86). These two isoforms, which are encoded by separate genes, are 86% homologous at the amino acid level. ⁴ A functional distinction between Hsp90α and Hsp90β has not yet been elucidated. Hsp90 is one of the most abundant proteins found in eukaryotic cells and accounts for 1% to 2% of all cellular protein under normal physiological conditions. Furthermore, its expression can increase severalfold after stress. ⁵ Similar to other Hsps, it functions as a molecular chaperone, but it binds to a specific subset of cellular proteins, including transcription factors, such as steroid hormone receptors ^{3 5 6 7 8} ; several tyrosine and serine-threonine protein kinases, such as v-Src/c-Src, v-Raf/c-Raf, and cyclin-dependent protein kinase 4 (CDK4) ^{3 6 8 9} ; and components of the cytoskeleton, including microtubules and microfilaments. ^{3 10} Hsp90 appears to regulate the signaling capacity of many of these transcription factors and protein kinases ^{3 6 8 11} and to buffer potentially disruptive alterations in amino acid sequence of regulatory proteins caused by minor gene mutations. ¹²  

Another family of Hsps garnering increasing attention lately are the low-molecular-weight proteins of 15 to 30 kDa. The most studied mammalian small Hsps (sHsps), identified as the αB-crystallins and two closely related proteins, are collectively referred to as Hsp25/27. Hsp25 (rodent) and Hsp27 (human) are cytoplasmic proteins that share more than 80% identity at the amino acid level. ¹³ The similarity of these two proteins has led to some ambiguity in the literature, in that many reports refer to the rodent Hsp25 as Hsp27. 

One of the most characteristic features of sHsps is their ability to oligomerize to form aggregates in excess of 800 kDa. This property may be related to the phosphorylated state of the protein, which, in turn, is influenced by the physiological state of the cell. For instance, it appears that stress-associated phosphorylation of Hsp27 may be linked to the disintegration of large sHsp aggregates ¹⁴ and there is evidence that the assembly of either large or small oligomeric structures may be important for modifying the protein to perform different tasks within the cell. ^{14 15} For example, small oligomers of Hsp27 have also been linked to resistance against fragmentation of the actin cytoskeleton and protection against cell death by oxidative stress. ^{14 16}  

In the mammalian eye, little is known about the distribution and function of either Hsp25/27 or Hsp90. It has been reported that mRNA species for murine Hsp90α and Hsp90β are expressed strongly during early ocular development (embryonic days 11.5–14.5) and that during later embryogenesis and adulthood, levels of Hsp90α mRNA decline, whereas those of Hsp90β remain elevated. ¹⁷ In something of a contradiction, it has also been reported that rat Hsp90α mRNA is expressed from embryonic day 17 to adulthood, except for a short period around postnatal day 5. ¹⁸ The functional implications of these results cannot be assessed because, until the present report, Hsp90 protein levels in the eye have not been evaluated. In the case of Hsp25/27, it has been determined that Hsp25 mRNA is expressed intensely in the eyes of adult mice ¹⁹ and that this protein is prevalent in lens. ^{20 21} This tissue was singled out for study by several groups, because sHsps have considerable homology toα B-crystallin, the major structural protein of the vertebrate lens. Recently, a link was made between Hsp25/27 and the response of the retina to damage in a report that this Hsp was elevated in the retinal ganglion cell layer and optic nerve head of human glaucomatous eyes. ²²  

In a previous study, we examined the expression of the constitutive and stress-inducible forms of the 70-kDa Hsps, Hsc70 and Hsp70, respectively, in ocular structures from rats maintained in the absence of overt metabolic stress and showed that they have distinct distributions, implying cell- and structure-specific functions. ²³ Because there is a growing body of literature indicating cooperative interactions between many of the members of the Hsp family in the performance of their chaperoning functions, ^{9 24} it is important to characterize the retinal distributions of all the major Hsps to understand how they support retinal function and protect it from damage. In this article, we extend that descriptive approach to the distributions of Hsp25 and -90 immunoreactivity in ocular tissues. Together with our previous work, ²³ this study will permit researchers, for the first time, to compare and contrast the expression patterns of the main Hsps of the 90-kDa and sHsp families with those of the 70-kDa family. Such information will be important in the interpretation of Hsp changes after retinal injury and degenerative disease. In the current study, Hsp25 and -90 had significantly different distributions in several ocular tissues. These distributions were also distinct from those reported for Hsc70 and Hsp70. ²³ Consequently, our results support the concept that there are regional and cell-type–specific differences in the ocular expression of Hsps and that these differences are likely to be important indicators of normal cellular function in the eye and could help to explain the various responses of different parts of the eye to injury. 

Materials and Methods

Animal Handling and Euthanasia

Animals were handled in accordance with the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. Male Sprague-Dawley rats (200–300 g) were maintained between 21°C and 24°C on a 12-hour light–dark cycle with food and water made available ad libitum. The light intensity in the housing cages was 2 foot-candles (ft-c). For euthanasia, the rats were transferred to portable cages covered with a blanket and transported to the dissection room. The animals were allowed to settle for approximately 30 minutes with the blanket in place before they were killed by CO₂ inhalation. The eyes, or appropriate ocular tissues, were removed as quickly as possible. The dissection room was maintained between 22°C and 24°C with a light level of not more than 6 ft-c during the euthanasia procedure. All animals were killed between 1 and 3 PM. 

Paraffin Embedding, Sectioning, and Histology

Eyes were fixed in methacarn solution (60% methanol, 30% 1,1,1-trichloroethane [TCE] and 10% acetic acid) at room temperature for 18 to 20 hours and embedded in paraffin (Surgipath Medical Industries, Richmond, IL), as previously described. ²³ Ten-micrometer-thick sagittal sections through the optic nerve were cut from each eye and mounted onto microscope slides (ProbeOn Plus; Fisher Scientific, Fairlawn, NJ). Some tissue sections were stained with hematoxylin and eosin (H&E), as described. ²³  

Immunohistochemistry

Slide-mounted sections were deparaffinized, rehydrated, and blocked with 10% normal goat serum (NGS; Vector Laboratories, Burlingame, CA) in PBS-T (10 mM phosphate-buffered saline [pH 7.4]), and 0.1% Tween-20), followed by incubation, overnight, with one of the primary antibodies in PBS-T, plus 2% NGS. The primary antibodies were anti-Hsp25 (the antisera were diluted 1:750 or 1:5000, SPA-801; StressGen Biotechnologies Corp., Victoria, British Columbia, Canada) and anti-Hsp90 (1 μg/ml, SPA-835; StressGen) or, in place of those antibodies, rabbit IgG and rat IgG (both purchased from Sigma, St. Louis, MO). The nonimmune IgGs were included to monitor nonspecific immunoreaction product and were used at a dilution at which their protein content was estimated to be equivalent to that of the Hsp antibody raised in the same species. The next day, slides were incubated for 1 hour in PBS-T, containing biotinylated secondary antibody, followed by 1 hour in PBS-T containing peroxidase-conjugated streptavidin (1 μg/ml; Kirkegaard & Perry Laboratories [KPL], Gaithersburg, MD). The secondary antibodies were goat anti-rabbit (1μ g/ml, KPL), in the case of rabbit IgG and anti-Hsp25, and goat anti-rat (1 μg/ml, KPL), in the case of rat IgG and anti-Hsp90. Bound antibody was visualized using the 3,3′diaminobenzidine (DAB)-nickel detection system (Vector Laboratories). 

Western Blot Analysis

Before use, the antibodies were tested by Western blot analysis of samples of the purified proteins (obtained from StressGen Biotechnologies) and homogenates of ocular tissues to verify the specificities claimed by the manufacturer. The dilutions of the antibodies used for the Western blot analysis were greater than those applied to tissue sections because of the well-known fact that the antigenic epitopes are much more accessible on a Western blot then in a tissue section (see the Results section and Fig. 1 for details). 

Results

Specificity of Anti-Hsp25 and Anti-Hsp90

To study patterns of ocular staining, it is imperative that the antibodies recognize only the appropriate Hsps. Thus, we began by testing the antibodies, a rabbit polyclonal antibody against Hsp25 (SPA-801; StressGen) and a rat monoclonal antibody against Hsp90 (SPA-835; StressGen), for their ability to recognize Hsp25 and -90, respectively, by Western blot analysis. The antibody against Hsp25 is reported to be highly specific for the rodent form and does not cross-react with other Hsps, including primate Hsp27, to which it has 80% amino acid homology (StressGen, personal communication, 1999). In our hands, anti-Hsp25 did not react with samples of recombinant human Hsp27 (SPP-715; StressGen), or bovine lensα -crystallin (SPP-225; StressGen), which is composed predominantly of the sHsp αB-crystallin, but it identified a single band of the expected size in samples of recombinant mouse Hsp25 (SPP-710; StressGen) and homogenates of the rat retina (Fig. 1) . In contrast, anti-Hsp90, which is reported to specifically recognize free and complexed Hsp90 (StressGen), did not react with samples ofα -crystallin, Hsp25, or Hsp27, but it identified a single product with purified Hsp90 (SPP-770; StressGen) and the homogenate of the rat retina (Fig. 1) . Because the antibody against Hsp90 is equally reactive to both isoforms of the protein (StressGen), we believe that rat Hsp90α and rat Hsp90β are sufficiently close in size to be inseparable under the denaturing conditions used in our system. We did not test an antibody that discriminates between the two isoforms of Hsp90, because such an antibody was not available when this study was begun. 

Retina

The level of immunoreactive Hsp25 in a retinal tissue section was found to be low. Even though Hsp25 was detected by Western blot analysis of retinal samples using a 5000-fold dilution of the antibody (Fig. 1) , when that concentration was used on a tissue section, it was indistinguishable from the nonimmune IgG control (compare Fig. 2 with top right panel of Fig. 3 ). However, a 750-fold dilution of the anti-Hsp25 antibody yielded detectable Hsp25 immunoreactivity (Fig. 3 , bottom left panel). This immunoreactive Hsp25 was most prominent in the cytoplasm of the retinal ganglion cells (RGCs; Fig. 3 , white arrow with black outline), the inner plexiform layer (IPL), the inner nuclear layer (INL), the outer limiting membrane (OLM) and adjacent photoreceptor nuclei (Fig. 3 , white arrowheads), and in the inner segments of the photoreceptors (IS; Fig. 3 , just above white arrowheads). The walls of blood vessels also stood out, indicating that they had more Hsp25 immunoreactivity than the adjacent structures (Fig. 3 , black arrow). In addition, a lower level of Hsp25 immunoreactivity was seen at the tips of the outer segments (OS), which probably included processes of retinal pigment epithelium (RPE; Fig. 3 , black arrowheads). There was an unexpected dramatic increase in Hsp25 immunoreactivity as the RGC fibers coalesced to form the optic nerve head and optic nerve (Fig. 4) . This increase was equally apparent even in retinal sections probed with the 5000-fold dilution of the Hsp25 antibody (Fig. 5) . Furthermore, RPE cells within 100 μm of the optic nerve head had Hsp25 immunoreactivity in their cytoplasm and nuclei (Fig. 4) , as did cells in other retinal layers that were close to the optic nerve head. 

Hsp90 immunoreactivity also was observed in most layers of the retina (Fig. 3) . However, in contrast to Hsp25, it was very intense, even though the concentration of this antibody used was more dilute than that of Hsp25. The most prominent staining for Hsp90 was observed in the OLM (Fig. 3 ; white arrowheads). The next most darkly stained regions were the IS, IPL, and outer plexiform layer (OPL; Fig. 3 ). Staining in the remainder of the retina was detected in the perinuclear regions of cells of the RGC, INL, and ONL and was slightly above background at the tips of the OS, where they are interdigitated with processes from the RPE (Fig. 3) . No greater reactivity was noted in the walls of blood vessels, in contrast to observations of Hsp25. Another distinction between Hsp25 and -90 was seen close to the optic nerve head, where staining for Hsp90 in the RGC axons remained the same as the axons approached and entered the optic disc (Fig. 4) . However, near the retina–optic nerve junction, more intense staining was observed in the innermost row of ONL nuclei, adjacent to the OPL (Fig. 4) . Furthermore, RPE cells located less than 200 to 250 μm from the optic disc also contained higher levels of Hsp90 reactivity, particularly within the cell nuclei, similar to that observed with Hsp25 (Fig. 4) . There was no nonspecific staining when rat IgG was used as a negative control (not shown). 

Optic Nerve

Hsp25 reactivity was greater in the optic nerve fibers than in the transversely oriented glial cell processes. The pattern and intensity of staining was similar whether a 750-fold dilution or 5000-fold dilution of the antibody was used (compare Figs. 6 and 7 ). The staining of the nerve fibers appeared as wavy lines running in parallel to the long axis of the optic nerve (diagonally from the upper left to the lower right in Fig. 6 and horizontally in Fig. 7 ). Two examples of stained axonlike structures are indicated by the white arrows with black borders in Figure 6 . More lightly stained glial cell processes can also be seen in Figures 6 and 7 , running in a nearly vertical orientation. In addition, a perinuclear pattern of immunoreactivity was apparent in the glial cells (solid black arrows; Figs. 6 7 ). Nonspecific staining with rabbit IgG was negligible (Figs. 4 6) . 

Hsp90 immunoreactivity was also detected in the optic nerve, although the overall level of staining was considerably less than with anti-Hsp25 (Figs. 4 6) . In addition, the staining had a pronounced perinuclear and, occasionally, nuclear distribution in the glial cells, even more so than that noted with Hsp25 (Fig. 6 , black arrows, Hsp90). Figure 6 also showed that the Hsp90 immunoreactivity in the neuropil of the nerve was more uniformly distributed than was true for Hsp25. No staining was observed with rat IgG (not shown). 

Cornea

Hsp25 immunoreactivity was very intense in all layers of the corneal epithelium, regardless of which concentration of primary antibody was used (1:5000 dilution, Fig. 8 ; 1:750 dilution, Fig. 9 ), but because the darkness of the staining with the 1:750 dilution of anti-Hsp25 made details difficult to see (Fig. 9) , the following descriptions are based on the staining using the 1:5000 antibody dilution shown in Figure 8 . The immunostaining was especially concentrated in a perinuclear distribution in many basal cells (Fig. 8 ; Hsp25, black arrow) and within the nuclei of the superficial, squamous cells (Fig. 8 , Hsp 25 and inset, white arrows). In the stroma, Hsp25 immunoreactivity was detected in nuclei of superficial keratocytes (Fig. 8 ; Hsp25, black arrowhead). Hsp25 immunoreactivity was also present in the corneal endothelium, but at relatively low levels (Fig. 7) . 

In the case of Hsp90, the overall pattern of staining in the corneal epithelium was distinct from that of Hsp25. Not only was the intensity of reactivity considerably lower, but also it was restricted mainly to the basal and wing cell layers (Fig. 8) . Little immunoreactivity was detected in the more superficial squamous cell layer, with the exception of occasional, well-stained nuclei (Fig. 8 , Hsp90, black-bordered white arrows). However, there was a similarity between Hsp25 and -90 immunoreactivity in the basal cells, in that it had a perinuclear distribution in the cytoplasm (Fig. 8 ; Hsp90, black arrow and inset) and in the staining of occasional superficial keratocyte nuclei in stroma (Fig. 8 , Hsp90, black arrowhead). The immunoreaction product in the corneal endothelium in this case was nonspecific, in that it was also detected after treatment with the control nonimmune rat IgG (not shown). 

Ciliary Body

The pattern of staining for Hsp25 and -90 in this tissue was similar in one respect to that described previously for the constitutive form of the 70-kDa Hsp, Hsc70, ²³ in that it was distributed uniformly throughout the cytoplasm of the epithelial cells (Fig. 10) . However, in contrast to Hsc70, Hsp25 and -90 immunoreactivity was prominent in many ciliary epithelial cell nuclei (Fig. 10) . The Hsp25 immunoreactivity shown here was detected only when the higher concentration (750-fold dilution) of primary antibody was used. Negligible nonspecific immunoreactivity was observed throughout the ciliary body with rabbit, or rat IgG (not shown and Fig. 10 , respectively). 

Iris

All layers of the iris contained both Hsp25 and -90 immunoreactivity (Fig. 11) . However, the posterior epithelium (which would contain melanin granules were this not from an albino rat) was clearly more Hsp25-immunoreactive than the other layers. Hsp90, in contrast, was relatively uniformly distributed throughout the iris. Nonspecific immunoreactivity was negligible with either nonimmune rabbit IgG (Fig. 11) or rat IgG (not shown). 

Discussion

Specificity of Antibodies

The anti-Hsp25 and -90 antibodies used in this study have been shown to be specific in studies on other tissues when used at similar or even considerably higher concentrations. ^{25 26} Nonetheless, because they had not been used previously for immunohistochemistry of the rat retina and other ocular tissues examined in the present study, we tested their specificities using Western blot analysis on which were run several concentrations of the purified protein standards and retinal homogenates. Although we were satisfied that the antibodies recognized the correct proteins on Western blot analysis (Fig. 1) , because the 5000-fold dilution of Hsp25 antibody used for the blots did not reveal Hsp25 immunoreactivity in the retinal tissue sections (Fig. 2) , we still included with each immunohistochemical staining series additional control slides. These consisted of a set of tissue sections exposed to nonimmune IgGs from the same species in which the antibodies were raised and at the same concentrations as those antibodies. The need for more concentrated antibody to detect Hsp25 in the tissue sections compared with the Western blot analysis is one that is commonly observed for many antibodies. It is related to the differences in accessibility of the antigens in the two types of samples and can be found in two previous publications on Hsp25 and -90. ^{25 27} The specificity seen in the Western blot analysis together with the absence of immunoreaction product in tissue sections treated with nonimmune IgGs, as shown in most of the figures in this report, makes it unlikely that the immunoreactivity detected with the antibodies was nonspecific. Nonetheless, when the level of immunoreactivity is low, as for Hsp25 in the retina (Fig. 3) , there is a small possibility that some of it may be nonspecific. 

Retina

The retina is known to be susceptible to damage, even under normal physiological conditions. As far back as 1966, Noell et al. ²⁸ reported that exposure to light, even at low irradiance levels, could induce retinal cell death. Therefore, it is not unexpected that proteins, such as Hsps, which can help protect cells from damage, might be found in abundance in the retina under normal conditions. We reported previously that Hsc70 and Hsp70 were expressed in the normal retina, although their cellular distribution patterns were quite different. ²³ In the present study, we discovered that Hsp90 was present at easily detectable levels throughout most of the normal retina. However, in a somewhat unexpected finding, Hsp25 immunoreactivity was very low in the normal retina. It is clear that this difference between Hsp25 and -90 in the retina was not a consequence of differences in affinities for the antigens or epitope masking, because the former was found to be at higher levels in other parts of the eye. For example, Hsp25 was found at high levels in the optic nerve, cornea, and iris (Figs. 4 5 8 10 11) . Furthermore, it is known to be the primary constituent of the lens. ^{20 21} Perhaps, Hsp25 is more of a stress-inducible protein in the retina, analogous to Hsp70. ²⁹ The recent report of Tezel et al., ²² showing that Hsp27 immunoreactivity (synonymous with Hsp25 in this case) is elevated in RGCs of human glaucomatous retinas supports this speculation. In the future, it will be interesting to determine whether retinal Hsp25 levels can be elevated by inducers of the heat shock response and whether it has a role in the survival of the retina after acute injury. 

Many of the functions of Hsp90 are thought to involve the formation of an Hsp90-protein substrate heterocomplex (foldosome), which includes at least nine accessory proteins, as well as Hsc70. ^{3 6 8} We observed that all retinal layers, from RGC to IS, were stained for Hsp90 (Figs. 3 4) and that these layers were the same ones that showed reactivity for Hsc70. ²³ Thus, our results are consistent with the functions of Hsc70 and Hsp90 being intertwined, such as would be the situation if both proteins were part of a foldosome. The only discrepancy in the staining patterns was observed in the plexiform layers. The IPL and OPL were much more prominently stained for Hsp90 (white arrows, Fig. 3 ) than for Hsc70. ²³ However, because several factors may affect the intensity of the immunoreaction product deposited in the tissue section, including differences in antibody affinity for antigen and several technical aspects of the reactions used to detect the bound primary antibodies, other more quantitative analyses will be necessary to determine whether the immunostaining differences are related to differences in the amount of antigen in the retina. 

One functional distinction between Hsp90 and Hsc70 that may contribute to their different distributions in the retina is the role of the former in maintaining steroid hormone receptors in an active state. Because the retina is known to be sensitive to corticosteroids and thus to contain receptors, ^{30 31} the retinal distribution of Hsp90 may be related partly to the receptor distribution. In addition, the plexiform layers contain the cell processes and synaptic connections of the retinal cells, and therefore one possibility is that Hsp90 may plays in maintaining the structures involved in forming synaptic junctions. There is some evidence to suggest that Hsps can be localized to synaptic connections between retinal cells (Ref. ²³ and this study, discussed below) and that Hsps can maintain synaptic functions in some eukaryotes by modifying the properties of their synapses. ^{32 33}  

We have reported punctate staining for Hsc70 and Hsp70 in the OLM of the normal retina. ²³ In that report, we suggested that Hsc70 and Hsp70 may be focally associated with the gap and zonulae adherentes junctions between Müller cells. In this study, we observed intense reactivity in the OLM of the Hsp90-stained retina (Fig. 3 , Hsp90, arrowheads) as well as detectable staining in the OLM of the Hsp25-stained retina (Fig. 3 , Hsp25, arrowheads). Based on both reports, we now suggest that several Hsps, including Hsp70 and -90, may play an important part in the maintenance of Müller–Müller and Müller-photoreceptor cell junctional specializations located at the OLM. Karunanithi et al. ³² have reported that synaptic transmission at the Drosophila larval neuromuscular junction is protected by the induction of Hsps and in particular by Hsp70. His group also showed that the process seems to involve pre- and postsynaptic modifications. ³² We have no evidence to suggest how the process may occur in the rat retina, but because of the functional importance of structural and gap junctions between the Müller cells and between the Müller cells and the photoreceptors, it is reasonable to suspect that Hsps may be involved in their maintenance. 

Retinal Pigment Epithelium

These cells, which are sandwiched between the choroidal and photoreceptor layers of the retina, are indispensable to the survival of the vertebrate retina. They perform a myriad of different functions, including the daily phagocytosis of effete photoreceptor OS fragments; the absorption of stray light energy; the uptake, processing, transport, and release of vitamin A and some its visual cycle intermediates; and the maintenance of proper retinal levels of hydration and thus, optical clarity. ³⁴ In the past, we have reported that epithelial cells located less than 200 to 250 μm from the optic nerve head show intense reactivity for Hsc70, whereas RPE cells located more than 250 μm from the optic nerve head show little or no immunoreactivity for either Hsc70 or Hsp70. ²³ In the present study, we found that the staining pattern for Hsp90 was similar to that described for Hsc70 (Fig. 4) . Hsp25 immunoreactivity was also seen in some RPE cells close to the optic nerve head (Fig. 4) , but, in contrast to Hsp90 and Hsc70, it did not extend as far from the optic nerve. The reason for elevated Hsps in the RPE cells near the optic nerve head is unclear. It is possible that the region of the retina adjacent to the optic nerve is subject to greater metabolic stress than the rest of the retina, perhaps because of mechanical factors (tension produced during eyeball rotation), or it could be because the blood–retinal barrier is defective in this region (Dean Bok, personal communication, 1998). 

Optic Nerve

The optic nerve is suggested to have three distinct morphologic regions near the retina–optic nerve junction: the optic nerve head (ONH), the retina-optic nerve transition region (ROT) and the optic nerve proper (ONP). ^{35 36} We observed intense staining for Hsp25 in ganglion cell fibers located within the ONH, ROT, and ONP (not shown), but not in RGC axons located within the retina (Figs. 4 5) . Because the fibers entering the ONH and other optic nerve regions are surrounded by glial cell processes and, in particular, astroglial processes, whereas those axons in the retina are not, ^{35 36} the increase in Hsp25 expression may be related to the presence of the astroglial cells. Closer inspection of the optic nerve revealed that much of Hsp25 staining was localized to the perinuclear regions of the glial cells. Such a distribution of Hsp25 has been reported previously for cultured astrocytes. ³⁷ In addition, optic axons showed high levels of immunoreactive Hsp25 (Figs. 6 7) . This pattern was surprising, because the intraretinal portions of the axons of the retinal ganglion cells had much lower Hsp25 immunoreactivity (Figs. 4 5) and prompts the question of how the Hsp25 immunoreactivity could increase in the optic nerve portion of those axons, in that it is well know that axons have no machinery for protein synthesis. We suggest that the axons may acquire Hsp25 from the astroglia. There is some evidence that proteins such as Hsps can be transferred from glia to axons, particularly during times of stress, ^{38 39} and recent work has shown that retinal cells can take up Hsp70 injected into the vitreous in the rat eye. ⁴⁰ Oligodendrocytes are not found in the ONH and are only occasionally present in the ROT ^{35 36} ; thus, their distribution does not correlate with the Hsp25 staining pattern. Finally, Hsp25 expression was inconsistent with changes in axonal myelination. It has been observed that the ONH is composed of unmyelinated axons, the ROT is composed of a mixture of myelinated and unmyelinated axons, and the ONP is composed of predominantly myelinated axons. ^{35 36} Thus, if a relationship exists between myelination and Hsp expression, there should be a gradual change in staining intensity as the nerve fibers pass from ONH to ONP. We never observed such a transition (Figs. 4 5) . Consequently, there must be some as yet unrecognized functional distinction between RGC axons and glia in the nerve fiber layer of the retina, compared with those in both intra- and extraretinal portions of the optic nerve. 

The distribution of Hsp90 in the optic nerve was remarkably similar to that reported for Hsc70 ²³ and was clearly different from the pattern observed for Hsp25 (Figs. 4 6) . Most of the Hsp90 immunoreactivity was concentrated in the juxtanuclear cytoplasm of the glial cells and not in the nerve fibers (Fig. 6) . There was also no abrupt increase in Hsp90 reactivity as the RGC axons left the retina and entered the optic nerve head (Fig. 4) . Our results suggest that Hsp25 is the more abundant Hsp in the axons of the normal optic nerve, whereas Hsc70 and Hsp90 are localized predominantly in the glia of the optic nerve. The functional significance of these differences in distribution remains to be determined, but it supports the hypothesis that each of the Hsps play distinctive roles in the various cellular components of the retina and optic nerve. 

Cornea

Within this tissue, the most prominent Hsp staining was found in the epithelium. This structure, which is five to seven cells thick in the rat, is composed of three cell types: the basal cells (innermost), which undergo mitosis to produce daughter cells that move toward the surface of the cornea and replace the cells sloughed from the corneal surface; the wing cells (middle), which are in an intermediate state of differentiation; and the superficial cells (outermost), which are terminally differentiated and in the process of degenerating. We observed that immunoreactivity for Hsp25, Hsc70, Hsp70, and Hsp90 tended to be greater in many basal cells of the corneal epithelium than in the superficial cells (Ref. ²³ and present study, Fig. 8 ). This pattern may be related to the fact that the basal cells are more metabolically active than the superficial ones ^{41 42} and that activity may increase the need for the chaperoning functions of the Hsps. The perinuclear distribution of the Hsps seen in the basal cells is consistent with this interpretation, because the protein synthetic machinery of the cells also tends to be in a perinuclear distribution. 

We reported in our prior study that a significant number of epithelial wing cells stain for Hsp70, but that very few wing cells exhibit reactivity for Hsc70. ²³ This observation was surprising at the time, because Hsc70 was more prominent than Hsp70 in other ocular tissues. In the present study, we found that many wing cells also stain for Hsp25 and/or Hsp90 and that the proportion of cells that stain for Hsp90 is comparable to that of cells that stain for Hsp70, rather than for Hsc70 (not shown). Because the staining pattern for Hsp90 in most ocular tissues resembles that of Hsc70, it suggests that Hsp activities in the corneal epithelium may be different from those operating in other ocular regions. It also suggests that this tissue could be useful for exploring interactions between Hsp70 and Hsp90. 

An interesting observation in the corneal epithelial layer was that Hsp25 staining seemed to be distributed more uniformly than that of Hsp90 (Fig. 8) . The broad expression of Hsp25 within this structure may reflect the fact that it has a structural role to play in corneal epithelial cells, as well as its possible role as a chaperone. This would be analogous to the situation suggested for the sHsps and the structurally related α-crystallins in the lens. Perhaps its function in these two tissues relates to the fact that they must both maintain optical transparency. In addition, the abundance of Hsp25 immunoreactivity even in the superficial cells that are beginning to undergo apoptosis is similar to that observed in epidermis, in which the amount of Hsp27, the human analogue of Hsp25, increases during the differentiation of keratinocytes. ⁴³  

Keratocytes are responsible for maintaining the extracellular matrix and collagen fibrils of the corneal stroma. Therefore, they play a pivotal role in regulating the thickness and composition of the cornea. We observed staining of these cells using antibodies to Hsp25 and -90 (Figs. 8 9) only in those layers close to the epithelium, but not using antibodies to Hsc70 and Hsp70. ²³ The reason for this observation is unclear, and the absence of any 70-kDa Hsp staining is unusual. This distinctive staining pattern may be related to the observation that the superficial keratocytes seem to depend on factors from the epithelium for survival, because they die when the corneal epithelium is lost due to injury. ^{44 45}  

Ciliary Body and Iris

Both Hsp25 and -90 were found to be abundant throughout the epithelial and stromal layers of the ciliary body and iris (Figs. 10 11) . In the iris, the pigmented epithelial cell layer, which faces the lens, had noticeably greater immunoreactivity than the stromal layer, particularly for Hsp25 (Fig. 11 , insets). This epithelial versus stromal difference in Hsp25 immunoreactivity was less apparent in the ciliary epithelium (Fig. 10) , which is continuous with that of the iris. For Hsp90, there was less of a difference between epithelium and stroma in the ciliary body and iris (Figs. 10 11) . The abundance of both of these Hsps in these structures, as well as Hsc70, ²³ may be related to their roles in the secretion of the aqueous humor and the fact that they form the barrier between the blood and the aqueous and vitreal chambers of the eye. 

Summary

Hsps play a vital role in cell survival. Many of them, including Hsp25 and -90, are known to be molecular chaperones and to participate in the folding and transport of newly synthesized proteins under normal physiological conditions. Of course, after stress, the levels of all Hsps are significantly elevated, because a primary function of these proteins is to trap partially dissociated or damaged proteins until they can be repaired, or destroyed. A failure to realize this outcome would lead to a build-up of misfolded and aggregated nascent polypeptides and, eventually, to cell death. It could be argued, therefore, that endogenous levels of these Hsps reflect directly on the natural tolerance of a cell and its ability to survive an insult and the types of metabolic stress typically encountered by the tissue. A growing body of information indicates that each of the Hsps has distinctive functions in different tissues and cells types, but the details of these differences remain to be clearly delineated. There seems to be considerably more overlap in the functions of Hsc/Hsp70 and Hsp90 than between either of them and Hsp25/27. Both Hsc/Hsp70 and Hsp90 are prominent in neurons but are found in low levels in or are absent from glia, especially in the unstressed condition. ^{46 47} Thus, our observations on the distributions of the Hsps in the retina compared with the optic nerve agree with what has been reported for the rest of the CNS. The reasons for the similarity in distributions of Hsc/Hsp70 and Hsp90 are a consequence, in part, of their known cooperative interactions in modulating cytoplasmic protein folding. ⁴⁸ Nonetheless, there are functional distinctions between the two, in that Hsp90 is tied specifically to the folding of various signal transduction pathway proteins and steroid hormone receptors, whereas Hsc/Hsp70 is not. ⁸ In contrast to the 70- and 90-kDa Hsps, Hsp25/27 is most prominent in astrocytes and found only at low levels, if at all, in neurons, even after hyperthermic or ischemic stress. ^{33 49} This localization in the brain correlates with our observation of the greater prominence of Hsp25/27 in the optic nerve, where astrocytes are abundant, than in the retina. The presence of relatively high levels of Hsp25/27 in astrocytes appears to be related to its role in intermediate filament stabilization. ⁵⁰  

Among the other ocular tissues examined in the current study—the cornea, ciliary body, and iris—only the corneal epithelium showed a difference in distribution between Hsp25 and -90. That Hsp25 was prominent throughout the cornea, whereas Hsp90 was preferentially localized in the basal layer may relate to the cytoskeleton-stabilizing role of the former. ⁵⁰ The reduction of Hsp90 in superficial epithelial cells makes sense, because those cells are less active metabolically and should be less in need of the signal transduction proteins that are stabilized by Hsp90. ⁸  

In this article, we have studied the expression of Hsp25 and -90 in ocular structures from the rat in the absence of overt whole-body stress. Combining these data with those of our earlier observations of the constitutive and inducible 70-kDa Hsps, ²³ we can now begin to use this road map of normal Hsp expression to identify and interpret changes after ocular damage and/or aging. In addition, our observations will help to show the relationship between ocular tissues and the larger body of data on Hsps in other tissue types. That information will be useful in suggesting how the Hsps may be used for therapeutic intervention in the treatment of various eye diseases and trauma. 

 Presented in part at the Cold Spring Harbor Symposium on Molecular Chaperones and the Heat Shock Response, Cold Spring Harbor, New York, April 1998.

 Supported by grant EY07616 (MT) from the National Institutes of Health.

 Submitted for publication May 9, 2000; revised November 6, 2000, and June 6, 2001; accepted June 27, 2001.

 Commercial relationships policy: N.

 

Figure 1.

 

Specificity of antibodies in Western blot analysis. Both anti-Hsp25 (1:5000) and anti-Hsp90 (1:3000) stained single bands in samples of retinal protein (lane R: 4.0 and 0.15 μg, respectively) that ran at nearly the same molecular weights as the standards (lane S) of pure protein included on the same blot (100 and 25 ng, respectively). The slightly lower apparent molecular weight of the Hsp90 band in the retinal sample compared with the standard is a common occurrence when comparing a complex mixture of proteins with a purified standard, and we have noted that Hsp70 displays this same behavior (unpublished observations, 1988). No immunoreactive bands other than those shown were observed. The portions of the lanes included in this figure for each sample extend from at least 10 kDa below each immunoreactive band to approximately 50 kDa above.

Figure 2.

 

Retinal Hsp25 immunoreactivity was undetectable with a high dilution of antibody. In the process of optimizing detection of Hsp25 immunoreactivity in ocular tissues, the 1:5000-fold dilution was first tried because that dilution was sufficient to detect this protein in Western blot analysis of the retina (Fig. 1) . However, in tissue sections, this dilution did not yield immunoreaction product greater than that seen with nonimmune IgG in the retina (Fig. 3) , although it generated a strong immunoreaction in the optic nerve and corneal epithelium (Figs. 5 8) .

Figure 3.

 

Hsp25 and -90 immunoreactivity in the retina. Tissue sections were stained with hematoxylin and eosin (H&E; top left), rabbit IgG (top right), anti-Hsp25 (bottom left; 1:750-fold dilution), or anti-Hsp90 (bottom right). Images of the retina were photographed at a position approximately halfway between the optic nerve and the peripheral end of the retina. The pattern of staining at this location was typical of all areas of the retina, except the region adjacent to the optic nerve. Hsp25 and -90, black arrowheads: location of the OS-RPE interface; white arrowheads: OLM; black arrows: two examples of Hsp25-immunoreactive blood vessels; white arrows: plexiform layers, which were prominently stained with anti-Hsp90; black-bordered white arrow: ganglion cell with Hsp25 immunoreactivity in its cytoplasm.

Figure 4.

 

Hsp25 and -90 immunoreactivity at the optic nerve–retina junction. Tissue sections were processed as described in Figure 3 . H&E-stained section, arrows: RPE and the fibers (axons) of the RGC, which converge at the optic nerve head.

Figure 5.

 

Hsp25 immunoreactivity at optic nerve head–retina junction, with the high dilution of antibody. This specimen was treated identically with the anti-Hsp25–immunostained specimen shown in Figure 4 , except that a 1:5000-fold dilution of anti-Hsp25 was used. It shows that the shift from low to high levels of Hsp25 immunoreactivity that occurs as the optic axons leave the retina and converge is the same, regardless of the dilution of antibody used. Arrows: examples of perinuclear staining of glial cells, which can be seen more clearly in this enlarged image than in Figure 4 .

Figure 6.

 

Hsp25 and -90 immunoreactivity in the optic nerve. These images are approximately at the center of the optic nerve, approximately 50 to 100μ m away from the optic nerve head. Note the Hsp25 immunostaining in the optic nerve fibers (Hsp25, black-bordered white arrows) and the perinuclear immunostaining of glial cells for both Hsp25 and -90 (Hsp25 and -90, solid black arrows). In the case of Hsp90, occasional glial nuclei appear to be filled completely with immunoreactivity (top center, black arrow).

Figure 7.

 

Hsp25 immunoreactivity in the optic nerve using the high dilution of antibody. This specimen was prepared identically with the Hsp25-immunostained section shown in Figure 6 , except that a 1:5000-fold dilution of anti-Hsp25 was used instead of a 1:750-fold dilution. The distribution of immunoreactive Hsp25 revealed is nearly identical with that shown in Figure 6 , with prominent staining of nerve fibers (appearing as discontinuous horizontal wavy lines), lighter staining of glial processes (appearing as occasional short vertical lines) and perinuclear staining of glial cell nuclei (black arrows).

Figure 8.

 

Hsp25 and -90 immunoreactivity in the cornea. Tissue sections were processed and photographed as described earlier. Images were captured from the central region of the cornea. The following three layers in the corneal epithelium were identifiable in these micrographs: the basal cell layer (adjacent to the stroma), the wing cell layer (middle three to four layers of flattened nuclei), and the superficial cell layer (outermost one to two layers of nuclei). The most darkly stained structures within the H&E-stained epithelium are nuclei. Comparison of their size, shape and location in this section with Hsp immunoreactivity in the Hsp25 and -90 sections indicated that staining in the wing and superficial cells (for example, see black-bordered white arrows) was largely nuclear. The lighter gray areas surrounding the cell nucleus in the H&E-stained specimen reflect eosin-stained cytoplasm. Comparison of this cytoplasmic staining with the distribution of Hsp25 and -90 reactivity in the basal cells (solid black arrows) indicated that it was localized to the cytoplasm around the nucleus. The insets in the Hsp25 and -90 sections show approximately twofold magnified images of the corneal epithelium to make it easier to see the cellular distribution of Hsp. Black arrowheads in the corneal stroma layers indicate examples of keratocytes (spindle-shaped fibroblasts), which stained for Hsp25 and -90.

Figure 9.

 

Hsp25 immunostaining of the cornea using the low dilution of antibody. This specimen was prepared identically with the Hsp25-immunostained specimen shown in Figure 8 , except that the 1:750-fold dilution of anti-Hsp25 was used instead of the 1:5000-fold dilution. Even though the immunoreaction product in the epithelium was so abundant that it obscured some of the details of the epithelial cells, the distribution was essentially the same as that shown in Figure 8 . Furthermore, the absence of immunoreactivity in the stroma, with the exception of several keratocyte nuclei seen next to the epithelium, showed that the more concentrated anti-Hsp25 antibody did not cause an increase in nonspecific staining of Hsp25-negative stromal collagen fibers.

Figure 10.

 

Hsp25 and -90 immunoreactivity in the ciliary body. These low-magnification images show that Hsp25 and -90 staining is found mainly in the epithelial cells. The higher magnification insets show that the immunostaining for both Hsps was found throughout the cytoplasm. Insets are approximately four times the magnification of the low-magnification views.

Figure 11.

 

Hsp25 and -90 immunoreactivity in the iris. The main regions of the iris are labeled in the H&E-stained section. Hsp25 and -90 immunostaining was present in the stromal and posterior epithelial layers of the iris, with the latter layer showing greater immunoreactivity, especially for Hsp25. Insets are approximately three times the magnification of the low-magnification views.

The authors thank Carol R. Kent for her contributions to the immunohistochemical techniques used in this study.