Identification and Localization of Retinal Cystatin C

Abstract

purpose. Cystatin C is a mammalian cysteine protease inhibitor, synthesized in various amounts by many kinds of cells and appearing in most body fluids. There are reports that it may be synthesized in the mammalian retina and that a cysteine protease inhibitor may influence the degradation of photoreceptor outer segment proteins. In the current study cystatin C was identified, quantitated, and localized in mouse, rat, and human retinas.

methods. Enzyme-linked immunosorbent assay (ELISA), reverse transcription–polymerase chain reaction (RT-PCR), DNA sequencing, Western blot analysis, and immunohistochemistry have been used on mouse, rat, and human retinas (pigment epithelium included).

results. Cystatin C is present in high concentrations in the normal adult rat retina, as it is throughout its postnatal development. Its concentration increases to a peak at the time when rat pups open their eyes and then remains at a high level. It is mainly localized to the pigment epithelium, but also to some few neurons of varying types in the inner retina. Cystatin C is similarly expressed in normal mouse and human retinas.

conclusions. Cystatin C was identified and the localization described in the retinas of rat, mouse, and human using several techniques. Cystatin C is known to efficiently inactivate certain cysteine proteases. One of them, cathepsin S, is present in the retinal pigment epithelium and affects the proteolytic processing by cathepsin D of diurnally shed photoreceptor outer segments. Hypothetically, it appears possible that retinal cystatin C, given its localization to the pigment epithelium and its ability to inhibit cathepsin S, could be involved in the regulation of photoreceptor degradation.

Cystatins are a group of proteins, functioning throughout the animal and plant kingdoms as inhibitors of papain-like cysteine proteases belonging to enzyme family C1. Mammalian C1 cysteine proteases, such as cathepsins B, H, L, and S, have important functions in the catabolism of peptides and proteins. ^{1 2} Because proteases can cause substantial tissue damage unless carefully controlled by specific inhibitors, great attention has been paid to natural protease inhibitors over the past decade, even though they were first discovered as long ago as the late 1950s. ³  

Cystatins regulate the activity of family C1 cysteine proteases by reversible binding to their active-site clefts in competition with enzyme substrate. Known mammalian cystatins are all composed of at least one 100- to 120-amino-acid-residue domain with conserved sequence motifs. ⁴ Of the 11 human cystatins known, cystatin C is the most extensively studied, showing a wide-spectrum inhibition profile and high-affinity binding to virtually all family C1 cysteine proteases. ¹ It is the dominating inhibitor in most body fluids, present in amounts allowing it to control the extracellular activities of cathepsin B and other family C1 cysteine proteases. ⁵  

The concentration of human cystatin C is markedly higher in cerebrospinal fluid than in blood. ⁵ The highest amounts of cystatin C in tissue homogenates from human, mouse, and rat are found in brain samples, at levels of 30 to 300 ng/mg protein. ⁶ It has been shown that cystatin C is expressed in rat, monkey, and human cerebral neurons. ^{7 8 9}  

The retina is ontogenetically a part of the central nervous system (CNS) and cystatin C might therefore be expected in retinal neurons as well. It has been shown by in situ hybridization that cystatin C mRNA is present in the rat retina, ¹⁰ but the precise retinal distribution of the protein, to our knowledge, has not been studied so far. Further, it has been shown that the production of cystatin C is upregulated in the CNS after ischemia or axotomy, ^{11 12} suggesting that it may play a role in neurodegeneration, neuroprotection, or repair. When it was further shown that the retinal pigment epithelium produces factor(s) with cysteine protease inhibitory characteristics, which may be of importance in retinal development or normal maintenance, ¹³ we decided to analyze the content and distribution of cystatin C in the retina. 

Materials and Methods

Normal pigmented (PVG) rats and C57BL mice were used in the study. Tissue was also obtained from unaffected parts of three human eyes enucleated for malignant tumors. The patients were 52, 64, and 80 years old. The experiments and the animal care were undertaken according to the rules and guidelines of the Society for Neuroscience and the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. The experiments were approved by the Swedish Committee for Animal Experimentation Ethics. The human tissue was obtained and examined as specified by Swedish laws and ethic regulations. 

All dissection work was performed in freshly prepared Ames’ medium (Sigma-Aldrich Chemie, GmbH, Steinheim, Germany) ¹⁴ that had been bubbled with a mixture of 95% oxygen and 5% carbon dioxide. The rats were killed by carbon dioxide asphyxiation and the mice by decapitation. Tissues were fixed at 4°C for 4 hours in 4% formaldehyde in 0.1 M phosphate buffer at pH 7.4 and processed for immunohistochemistry. Developing rat eyes were obtained at embryonic day 15 and 18 (E15 and E18), day of birth (PN0) and postnatal day (PN)7, 14, 21, 28, and 35 and prepared for immunohistochemistry in the same way as for adult tissue. The age of the embryos were determined by measuring their crown–rump lengths. For protein and RNA analyses, the retinas (pigment epithelium included) were dissected and instantly frozen on dry ice and stored at −80°C awaiting further preparation. 

Nucleotide-Based Analyses

Total RNA was prepared from tissue samples as described. ¹⁵ RT-PCR was performed using commercial reagents (PE Applied Biosystems, Foster City, CA). For rat cystatin C, the primer pair used had the sequences 5′-AGG AGA AGA GAA CCA GGG GAC AGC3-′ (KH727) and 5′-AGT ACA ACA AGG GCA GCA ACG ATG-3′ (KH728). For amplification of a mouse cystatin C cDNA segment, the primers KH723, 5′-CCA TGA CCA GCC CCA TCT GAT-3′, and KH724, 5′-CAC AAG TAA GGA ACA GTC TGC-3′, were used. PCR was accomplished using 0.4 μM primers and Taq polymerase (AmpliTaq Gold; PE Applied Biosystems) in a thermocycler (2400; PE Applied Biosystems), with 35 cycles of denaturation at 94°C, 30 seconds; annealing at 58°C, 30 seconds; and extension at 72°C, 30 seconds, with a 5-minute preincubation time at 94°C before the temperature cycling. PCR products were analyzed by electrophoresis in 2% agarose gels. The DNA sequences of PCR products were determined on both strands by dye dideoxy sequencing, using reagents in the sequencing kit (BigDye Terminator Cycle Sequencing; PE Applied Biosystems); oligonucleotides KH723, 724, 727, or 728 as primers; and a sequencer (310; PE Applied Biosystems). The sequences were evaluated by computer (Sequencher; Gene Codes Corp., Ann Arbor, MI). 

Antibodies

The monospecific polyclonal rabbit antisera used for ELISA, immunoblot analysis, and immunohistochemistry were raised against cystatin C isolated from human urine, ⁵ recombinant mouse, and rat cystatin C. ⁶ The antibodies against human cystatin C cross-react with mouse and rat cystatin C. ⁶ Of the three cystatin C antisera, the one raised against human cystatin C elicited the best immunohistochemical labeling and was therefore predominantly used. The two others did not result in any deviating observations. 

Secondary antibodies conjugated to FITC or Texas red and raised against rabbit IgG were obtained from Southern Biotechnology Associates, Inc. (Birmingham, AL), and Jackson ImmunoResearch Laboratories, Inc. (West Grove, PA), respectively. 

Cystatin C ELISA

Cystatin C was assayed as previously described. ⁶ Recombinant mouse or rat cystatin C were used for appropriate calibration curves. The cystatin C concentration was related to total protein concentration in the homogenates, the latter measured by a dye-binding assay. ¹⁶  

Western Blot Analysis

Immunoblot analysis to detect cystatin C after electrophoretic separation of proteins in tissue homogenates was performed as described, ¹⁷ using 16.5% SDS-polyacrylamide gels. Reference samples included recombinant human, mouse and rat cystatin C. ^{6 18}  

Immunohistochemistry

Cryostat sectioning (12 μm) and immunolabeling were performed with standard procedures. Labeling control experiments included using different secondary antibodies, omitting the primary antibodies, and preabsorbing the primary antibodies with excess recombinant human cystatin C. 

Microscopy and Image Analysis

The specimens were examined using a epifluorescence microscope (Eclipse E800; Nikon) equipped with a digital acquisition system (DEI-750; Optronics Engineering, Goletta, GA) and with a confocal laser scanning microscope (MRC1024UV; Bio-Rad, Richmond, CA). Images were viewed and processed using image analysis software (Confocal Assistant; free software, copyright Todd Clark Brelje, University of Minnesota, Minneapolis; and Photoshop; Adobe Systems, Mountain View, CA). 

Autofluorescence can be disturbing in the human pigment epithelium. Confocal scanning images were therefore obtained in selected regions with low lipofuscin autofluorescence and at two different wavelengths, one at 488/529 nm and one at 568/598 nm (excitation/emission wavelengths). The FITC-labeled antibody is known to elicit only weak signal in the 568/598-nm channel, which was also checked in independent model preparations. Care was taken to keep all image information well within the dynamic range of the confocal microscope, allowing subtraction images to be made without aberration caused by out-of-range fluorescence intensities. Fluorescence in the 488/529-nm channel that did not appear in the 568/598-nm channel was taken to be specific FITC fluorescence. For control purposes, the autofluorescence was studied in nonlabeled specimens, and subtracted images were made from such images, to affirm that the autofluorescence was similar in both channels, allowing such subtraction images to be made on the labeled sections. 

Results

Identification of Cystatin C in Rat and Mouse Retina

Homogenates of retinas from adult rats were investigated for cystatin C immunoreactivity, by using antisera raised against isolated human, mouse, and rat cystatin C. Cross-reaction of these antibodies with the three species variants of cystatin C previously seen and assessed by ELISA ⁶ was verified also to be prominent in immunoblotting from SDS-polyacrylamide gels (Fig. 1) . A single main band was detected in retinal homogenate, corresponding to a size of approximately 17 kDa, slightly larger than those of recombinant rat, mouse, or human cystatin C (Fig. 1) . Rat cystatin C harbors a glycosylation site and the protein isolated from, for example, rat urine, is glycosylated, ¹⁹ whereas the Escherichia coli–produced recombinant proteins are not, and the size difference of approximately 2 kDa agrees well with a normal-sized single-carbohydrate chain. 

By ELISA, the concentration of cystatin C in five adult rat retina homogenates (retinal pigment epithelium included) was 98 ± 10 ng/mg protein (±SEM; n = 17), which is a fairly high level in comparison with other tissues. ⁶ Similarly, cystatin C was detected by the antibodies in homogenates of adult mouse retinas. The amount of the inhibitor in a pool of 10 retinas (retinal pigment epithelium included), quantified by an ELISA, using mouse cystatin C for the calibration curve, was 23 ± 1.1 ng/mg protein (±SEM; n = 11). Samples of rat and mouse retinas were subjected to RNA purification and analyzed by amplification after reverse transcription copying of mRNA to cDNA (RT-PCR). By using PCR primers corresponding to segments of the coding region in mouse and rat cystatin C cDNA (Fig. 2A) , correctly sized PCR products of 152 and 454 bp, respectively, were obtained from total RNA of mouse and rat retina (Fig. 2B) . The DNA sequences of these fragments were identical with the published cDNA sequences for mouse and rat cystatin C, respectively. ^{8 20}  

Retinal Localization of Cystatin C

The cystatin C antisera most prominently and quite consistently stained parts of the cytoplasm of all retinal pigment epithelium cells (Fig. 3) . The labeling was subjectively similar in central and peripheral parts of the retinas. 

In most of the examined rat retinas, ganglion cells were labeled for cystatin C, but in other animals of the same age, there were no or only a few labeled cells in the same layer. The labeling of the ganglion cells was always localized to the cell soma and the most proximal part of the axons (Fig. 3C) . No arborization of labeled ganglion cells was observed in dendrites in the inner nuclear layer. 

At times, and again with variations between different animals, a small number of neurons with the localization and morphology of amacrine cells, bipolar cells, horizontal cells, and cone photoreceptor cells were labeled by the cystatin C antibody. Typically, these cells were few and scattered (Fig. 3D) . Horizontal cells were more common than bipolar and amacrine cells, and only occasional cone cells could be found. The cells appeared to be labeled in their entirety, which allowed us to identify them as belonging to the different cell classes. 

Control experiments using different secondary antibodies resulted in consistent staining. In experiments in which the primary antibody had been omitted or preabsorbed by using recombinantly produced antigen, there was no immunoreactivity in the retina (Fig. 3E) . 

In mouse (Fig. 3F) and human retinas, the staining was similar to that in the rat retina. The subcellular localizaton of the human cystatin C was examined in composite green–red (Fig. 4A) pictures and in images in which the autofluorescence recorded in the 568/598-nm channel had been subtracted (Fig. 4B) . It was different in many places from that of the autofluorescent lipofuscin granules. The autofluorescence in nonlabeled specimens was very close to identical in the 488/529- and 568/598-nm channels (Figs. 4C 4D) . In composite green–red pictures from nonlabeled sections, such channel subtraction completely eliminated the staining in the RPE (not illustrated). 

Cystatin C in Developing Rat Eyes

From E15 and throughout the development, cystatin C was present in the retina (Fig. 3A) . At E15, the pigment epithelial cells appeared cuboidal and had not yet formed the smooth, even structure seen in adult animals. Cystatin C immunoreactive cells also appeared in the neuroblast layer. However, their morphology was too undeveloped at this stage to make it possible to classify them. Neurons immunolabeled for cystatin C remained in the neuroretina throughout its development. As the tissue matured, it became possible to classify the stained neurons morphologically. 

The distribution of the immunoreactivity was similar at all stages to the pattern observed in adult animals. Thus, primarily the pigment epithelial cells were labeled, but also a small number of other retinal neurons of different classes. 

Analysis of tissue homogenates showed a gradual increase in retinal cystatin C content throughout development, reaching a peak at approximately day 14, after which a slight decline may have occurred (Fig. 5) . 

Discussion

For analysis of cystatin C in the eye, we mainly relied on immunochemical and immunohistochemical techniques. The specificity of the antibodies used was therefore critical. We used polyclonal antisera raised against human, mouse, and rat cystatin C, which in Western blot analysis of rat retinal proteins showed a single stained band corresponding in size to that expected for cystatin C. In immunohistochemical work with the antiserum against human cystatin C, preabsorbing it with a stoichiometric excess of recombinantly produced human cystatin C abolished the staining. Based on these results, we conclude that this antiserum specifically labels cystatin C in the rat and mouse in addition to human cystatin C, which it was raised against. Furthermore, omission of the primary antibody resulted in complete elimination of all specific labeling and switching between different secondary antibodies did not affect the labeling. There was thus no confounding staining by the secondary antibodies. 

The detection at the protein level of cystatin C in homogenates of adult rat and mouse retinas (with the pigment epithelium included), using these antibodies, is corroborated by the identification of cystatin C mRNA in such homogenates. According to our immunoassays, the inhibitor concentrations were, in the two species, more than 90 and more than 20 ng/mg protein, which is significantly higher than in homogenates from most tissues (liver, kidney, spleen, and muscle) of both animals, and of the same order as in homogenates of whole rat or mouse brain. ⁶ We therefore conclude that relatively high amounts of cystatin C are synthesized in the retina and retinal pigment epithelium. 

At the cellular level, cystatin C was seen in all cells of the retinal pigment epithelium. Certain ganglion, amacrine, bipolar, and horizontal cells and cones occasionally also showed cystatin C immunoreactivity. The distribution of cystatin C immunoreactivity was consistent and comparable in retinas of rats, mice, and humans. Based on the results presented here, we propose that cystatin C is localized to the retinal pigment epithelium as well as to some few neurons of different types in the inner retina. To our knowledge, this is the first direct examination of the distribution of a cysteine protease inhibitor in the retina. 

The presence of retinal cystatin C was previously investigated with an indirect method assaying cystatin C mRNA by in situ hybridization, ¹⁰ thus showing putative production sites. Our immunolocalization of the actual protein confirms the presence of cystatin C in the retina, but it appeared in structures other than the ones noted with in situ hybridization. However, expressed-sequence tag (EST) analyses of mRNA from cultured human retinal epithelial cells show the presence of cystatin C transcripts in these cells, ^{21 22} indicating that the protein we detect at least partially is located at its site of synthesis. Furthermore, in this study, specific ELISA showed that in the rat retina, cystatin C was present in fetal tissue from stage E18 and was expressed throughout the development of the retina, as well as in the adult retina, reaching a maximum at approximately PN14. Immunohistochemistry has shown that the distribution of cystatin C remains essentially the same throughout development and that the pattern is the same as that in the adult retina. 

The increase in retinal cystatin C content during the early development of the retina suggests that it may have a specific function controlling the development of the tissue. Because there was no addition of new cystatin C immunoreactive cell types during development, it appears likely that the increase reflects an increased production of cystatin C in predominantly the pigment epithelial cells. 

Cystatin C is expressed by neurons and glial cells in the rat CNS. ⁷ In the brain, the expression of cystatin C is upregulated after ischemia ¹¹ or facial nerve axotomy. ¹² Considering the similarities between the brain and the retina, cystatin C may have similar protective or regulatory functions in the event of pathologic or traumatic lesions affecting the retina. 

Based on the results presented herein, we propose that cystatin C is continuously present at a fairly high level in the retinal pigment epithelium as well as occasionally in retinal neurons of most classes. The increase in the amount of cystatin C present in the retina coinciding with the time when the photoreceptor cells mature at approximately PN14 raises the hypothesis that it may participate in regulating the degradation of photoreceptor outer segment proteins, potentially by affecting the activity of cathepsin S. This hypothesis is based on the fact that cathepsin S is present in the retinal pigment epithelium, ²³ where it can regulate the activity of cathepsin D, ²⁴ which is the predominant protease involved in photoreceptor outer segment degradation. ^{25 26 27} Further, cystatin C is a strikingly good cathepsin S inhibitor in vitro¹ and is a likely physiological regulator of cathepsin S activity. For example, the cystatin C–cathepsin S balance appears to control antigen presentation by dendritic cells, ²⁸ and it is important for the maintenance of normal vessel walls. ²⁹ The cystatin C we have demonstrated in the retinal pigment epithelium may influence the (cathepsin S–cathepsin D) system. 

 Supported by The Foundation Fighting Blindness, the Knut and Alice Wallenberg Foundation, the A. Österlund Foundation, the Segerfalk Foundation, the Crafoord Foundation, Project Grants K99-04X, 09915, and 14X-2321 from the Swedish Medical Research Council, and the Faculty of Medicine at the University of Lund.

 Submitted for publication July 24, 2000; revised October 30 and December 15, 2000, and February 5, 2001; accepted February 16, 2001.

 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.

 

Immunoblots of retinal cystatin C. Lanes for mouse, rat, and human contain recombinant cystatin C from the three species, and lanes PN14 to PN25 contain rat retina extracts from different postnatal days; relevant sizes are indicated. The cystatin C extracted from the rat retina was glycosylated, and as expected, it was approximately 2 kDa larger than recombinant cystatin C.

Figure 2.

 

Detection of cystatin C mRNA in retina. (A) The approximate positions of oligonucleotides used as PCR primers (KH723, 724, 727, and 728). Arrowheads: positions of introns in the corresponding genomic DNA. Expected sizes of the PCR products were 152 bp and 454 bp. (B) Agarose gel (2%) electrophoresis of the actual PCR products. Lane M: size markers (HaeIII-digestedΦ X174 DNA) with the base pair lengths of certain relevant bands indicated.

Figure 3.

 

Immunohistochemical localization of cystatin C in the retina. (A) E15 PVG rat. Labeling was mainly localized to the pigment epithelium (arrows), which is cuboidal at this stage. In the young (B, PN7) and adult (C, D) animals, the pigment epithelium was strongly labeled (arrows), as well as some ganglion (C, arrowhead), horizontal (D, arrowhead 1), bipolar (D, arrowhead 2), and amacrine (D, arrowheads 3, 4) cells. (E) Control staining of an adult PVG rat retina in which the antiserum had been preabsorbed with an excess of recombinant human cystatin C. (F) Adult C57BL mouse retina. Labeling was localized to the pigment epithelium and to retinal neurons, such as the horizontal cell indicated by the arrowhead. (A– D, and F) Confocal micrographs; (E) fluorescence micrograph. Scale bars, 50 μm.

Figure 4.

 

Cystatin C in peripheral parts of a human retina, selected for comparatively low autofluorescence in the pigment epithelium. (A) Composite green–red image of two recording channels in the confocal microscope, one at 488/529 nm and one at 568/598 nm (excitation/emission wavelengths). The short-wavelength channel shows both autofluorescence and FITC fluorescence, whereas the long-wavelength channel shows mainly autofluorescence. (B) Result of subtracting the long-wavelengths channel from the short-wavelengths channel–that is, in the pigment epithelium, mostly cystatin C immunofluorescence. Note that the subcellular localization does not coincide with the autofluorescence (arrows). (C, D) Green and red confocal microscope channels in a control experiment on a nonlabeled tissue section. The two were very nearly identical, allowing subtraction images such as that in (B) to be made. Scale bars, 50 μm.

Figure 5.

 

Cystatin C levels in developing rat eyes. Retinal tissue (pigment epithelium included) was dissected from developing PVG rat eyes. Homogenates of pooled tissue samples from three to four animals obtained at 0, 7, 14, 21, 28, and 35 days after birth were analyzed by an ELISA based on antibodies against recombinant mouse cystatin C. A dilution series of isolated recombinant rat cystatin C was used for the calibration curve. The cystatin C concentrations in the homogenates were related to total protein concentration determined by a dye-binding assay. Each data point represents the mean value of 20 analyses, with the SEM shown by the error bars.