Expression of α-Crystallin in Retinoblastoma FREE

Crystallins, the major structural proteins of the eye lens, are primarily categorized into 3 distinct families: α, β, and γ. The two α-crystallins, αA and αB, are the principal members of the small heat-shock protein family, acting as molecular chaperones.¹ Although αA- and αB-crystallins have related amino acid sequences with similar structural properties, they vary significantly in their tissue distribution and have different functions; they protect different proteins and are active under different conditions.^{2 - 6} Oxidative stress in general is accompanied by upregulation of the hosts of heat-shock protein, including the α-crystallins.⁷ Rao et al⁸ demonstrated that photoreceptors selectively upregulate αA-crystallin to protect themselves against mitochondrial oxidative stress and stress-mediated apoptosis.

Agents that induce oxidative stress include hydrogen peroxide (H₂O₂), superoxide, hydroxyl radical, and others. The stress can cause cell death, aging, and development of malignancy.⁹ Hydrogen peroxide is a ubiquitous molecule that is able to diffuse freely into membranes owing to its nonpolar characteristics. It is recently reported¹⁰ that expression of α-crystallin varies in retinal pigment epithelial cells exposed to H₂O₂ oxidative stress.

Recent studies show that αB-crystallin is expressed in various malignant tumors.^{11 - 14} One of the major roles of αB-crystallin is to preserve the integrity of mitochondria and restrict the release of cytochrome c,^{5 ,13} subsequently resulting in tumor growth through escape from apoptosis.¹⁵ In addition, αB-crystallin has been suggested as a novel prognostic factor in malignant solid tumors, including breast,^{12 - 13} head, and neck cancers,¹⁴ because αB-crystallin expression has been correlated with poor patient prognosis.

Retinocytoma is a benign counterpart of retinoblastoma that shows photoreceptor differentiation. Singh et al¹⁶ reported that the proportion of retinocytoma in the population with retinoblastoma is 1.8%, suggesting that retinocytoma develops in rare instances rather than as the usual phenotype of retinoblastoma. From a genetic standpoint, retinocytoma is similar to retinoblastoma, with autosomal dominant inheritance involving a mutation in the RB1 gene.^{17 - 20} Moreover, malignant transformation of retinocytoma into retinoblastoma is rare, with a frequency of less than 5%.¹⁶ Pineda et al²¹ demonstrated that αB-crystallin was expressed in retinoblastoma cells; however, correlation with α-crystallin expression and apoptosis in retinocytoma and retinoblastoma has not been not elucidated.

The aim of this study was to examine the expression of α-crystallins A and B in retinocytoma and retinoblastoma using immunohistochemistry and to correlate crystallin expression with tumor cell apoptosis.

The institutional review board of the University of Southern California approved our use of human specimens obtained from the file of the Doheny Eye Institute Pathology Laboratory. All procedures conformed to the Declaration of Helsinki for research involving human subjects. We analyzed 1 histologically normal–appearing retina obtained from a patient with ciliary body melanoma, 1 enucleated globe with retinocytoma, and 12 enucleated eyes with retinoblastoma. None of patients had received chemotherapy and/or radiotherapy before enucleation of the tumor-containing eye. All eyeballs were fixed in 4% paraformaldehyde soon after enucleation.

The slides were dewaxed, rehydrated, and rinsed in phosphate-buffered saline twice for 10 minutes. As pretreatment, microwave-based antigen retrieval was performed in 10mM citrate buffer (pH, 6.0). These slides were incubated with 3% H₂O₂ for 10 minutes, then with normal goat serum for 30 minutes. Sections were incubated with antirabbit αA- and αB-crystallin polyclonal antibodies (1:100 dilution; Stressgen, Ann Arbor, Michigan) at room temperature for 2 hours. Binding of the primary antibody was localized with the fluorescein isothiocyanate–conjugated antirabbit secondary antibody (Jackson ImmunoResearch Laboratories, West Grove, Pennsylvania) for 30 minutes. Negative controls consisted of fluorescein isothiocyanate–conjugated mouse IgG incubated without treatment of the primary antibody. Slides were examined using a Zeiss LSM510 (Zeiss, Thornwood, New York) confocal microscope.

In evaluation of immunohistochemistry, necrotic areas of tumor tissues were excluded based on 4",6-diamidino-2-phenylindole (DAPI) nuclear staining. In the nonnecrotic viable tumor tissues, the number of immunopositive tumor cells included in the total tumor cells was evaluated using a microscope (objective lens, magnification ×20) to examine 3 to 4 fields in the same slide for each specimen. The positive rates counted in each field were then averaged, and the number of immunopositive cells in each case was shown as a percentage of viable tumor cells. Immunoreactivity for α-crystallin in tumor tissues was scored as strongly positive (>30% of cells), weakly positive (<30% of cells), or negative (background level staining only), according to the previous report.¹³

Serial sections 5 μm thick were cut for the terminal deoxynucleotidyl transferase-mediated dUTP nick end-labeling (TUNEL) assay to evaluate distribution of the TUNEL-positive reaction in the same part of immunoreaction with α-crystallin. An In Situ Cell Death Detection Fluorescein Kit (Roche, Indianapolis, Indiana) was used for the TUNEL assay. The slides were dewaxed, rehydrated, and rinsed in phosphate-buffered saline twice for 10 minutes. These slides were incubated with 3% H₂O₂ for 10 minutes, then permeabilized with 20 μg/mL of proteinase K at room temperature for 10 minutes. Texas red label with enzyme solution was added to each slide and incubated in a humidified chamber at 37°C for 1 hour. Deoxyribonuclease-pretreated slides were used as positive controls, and slides without added enzyme were used as negative controls. Apoptotic cells were revealed by confocal microscope (Figure 1). In the nonecrotic viable tumor tissues, at least 300 tumor cells were counted in 3 or 4 fields of the same slide for each specimen using a high-power field. Apoptotic cells were defined by the presence of perinuclear chromatin condensation and apoptotic bodies. The percentage of the apoptotic cells was considered to be the apoptotic index (AI) as previously described.²²

The H₂O₂ was obtained from Sigma Aldrich (St Louis, Missouri). The human retinoblastoma cell line Y79 was purchased from the American Type Culture Collection (Manassas, Virginia) and grown in RPMI-1640 medium (Invitrogen, Carlsbad, California) with 100 U/mL penicillin, 100 μg/mL streptomycin, and 20% fetal bovine serum. The Y79 cells were treated with different concentrations of H₂O₂ (range, 0μM to 400μM) for 24 hours. The cells were then harvested and protein was extracted.

Total tumor proteins were extracted from 2 retinoblastoma cases (cases 7 and 10) using formalin-fixed and paraffin-embedded tissue sections, as reported previously.²³ Briefly, 5-μm-thick paraffin sections were cut for hematoxylin-eosin staining and 30-μm–thick serial sections were cut for protein extraction and mounted on plain glass slides. To collect tumor tissues, the target areas were cut macroscopically with a razor blade referring to the microscopic observation of the morphology of serial hematoxylin-eosin staining. The dissected tissues were then placed in Eppendorf tubes. Three hundred microliters of ristocetin-induced platelet agglutination (RIPA) buffer (Cell Signaling, Danvers, Massachusets) that included a cocktail of proteinase inhibitors (Sigma, St Louis, Missouri) was added to each tube and the contents were incubated as follows: 0°C for 2 hours; 37°C for 2 hours; 60°C for 2 hours; and 100°C for 20 minutes, followed by incubation at 60°C for 2 hours. After incubation, the tissue lysates were centrifuged at 15 000g for 20 minutes at 4°C. The supernatants were collected to measure protein concentrations and for Western blot analysis.²³

The concentration of extracted proteins from Y79 cultured cells and paraffin-embedded tissues was measured by the Bradford protein assay (Bio-Rad, Richmond, California) with bovine serum albumin as the standard protein. A total of 25 μg of protein was separated by sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) (12.5% Ready Gel; Bio-Rad Laboratories, Hercules, California) at 110 V. The protein was then electrotransferred onto polyvinylidene difluoride blotting membrane (Millipore, Bedford, Massachusetts). The membranes were blocked in 5% milk and probed with primary antibodies against αA-crystallin (1:500 dilution; Stressgen) and glyceraldehyde 3-phosphate dehydrogenase (1:1000 dilution; Millipore, Billerica, Massachusetts) for 2 hours at room temperature. Membranes were washed and incubated with a peroxidase-conjugated secondary antibody (Vector Laboratories, Burlingame, California) for 1 hour at room temperature. Images were developed by adding enhanced chemiluminescence detection solution (GE Healthcare, Cleveland, Ohio).

The AI data are presented as mean (standard deviation). Statistical evaluations were performed using the t test. The accepted level of significance for all tests was P < .05.

In the normal-appearing retina, αA-crystallin immunoreactivity was predominantly detected in the cytoplasm of photoreceptors (Figure 2A-D). The negative control, in which the respective antibody was omitted, showed no staining in the retina (Figure 2E). Both crystallins were immunopositive in the cytoplasm of lens epithelial cells in all sections examined in this study. In addition, αB-crystallin was expressed in the corneal epithelium and endothelium and in the optic nerve head (data not shown).

Retinocytoma cells represented a relatively uniform nuclear shape without marked nuclear atypia. Mitotic figures were not observed (Figure 3A). Calcification was partially present in the tumor tissue. αA-crystallin was expressed in the cytoplasm of all tumor cells (Figure 3B). In contrast, αB-crystallin immunoreactivity was weakly detected in less than 30% of tumor cells (Figure 3C).

Retinoblastoma cells demonstrated a diffuse high cellular population with atypical cells containing prominent and hyperchromatic nuclei (Figure 4A). As summarized in Table 1, 6 retinoblastoma cases (50%) were strongly positive for αA-crystallin. The positive reaction was restricted to the cytoplasm of tumor cells (Figure 4A-D, green). Six cases (50%) were also strongly positive for αB-crystallin. In these tumors, αB-crystallin was homogeneously expressed in the cytoplasm of the neoplastic cells (Figure 4F-I, green).

Coomassie blue staining of the electrophoresed gel showed the extracted proteins, especially those ranging from 10 to 75 kDa, and these appeared well preserved (data not shown). Western blot analysis clearly revealed the expression of αA-crystallin in both cases. The expression was stronger in case 7 than in case 10 (Figure 5).

The αA-crystallin expressed in Y79 cells treated with 100μM H₂O₂ reached maximum expression at 150μM (Figure 6). Expression of αA-crystallin was low in Y79 cells exposed to a high concentration of H₂O₂, but was sustained up to 400μM addition (data not shown). In untreated Y79 cells, αA-crystallin protein expression was rarely detected.

Apoptotic cells were rarely detected in retinocytoma; the AI was 0.29% (Figure 3D). On the other hand, the mean AI in retinoblastomas was 5.53. The AIs in cases strongly and weakly positive for αA-crystallin were 3.55 (2.61) and 7.50 (2.61), respectively (Table 2). In tumors expressing αB-crystallin, the AIs for strongly positive and weakly negative cases were 5.94 (3.80) and 5.12 (2.92), respectively (Table 2). The AI was significantly higher in cases that were weakly positive for αA-crystallin than in those that were strongly positive (P < .05); no correlation was observed between αB-crystallin immunoreactivity and AI (Table 2). Indeed, as shown in Figure 4, no TUNEL-positive tumor cells were noted in the αA-crystallin–positive area (Figure 4E), but TUNEL-positive tumor cells were found in the αB-crystallin–positive area (Figure 4J).

αB-crystallin is frequently expressed in tumor tissues and is known both to preserve the integrity of mitochondria and to restrict the release of cytochrome c^{5 ,11 - 14} which, in turn, results in resistance to tumor cell apoptosis. In contrast, Rao et al⁸ have proposed that αA-crystallin, rather than αB-crystallin, has the more important role in protecting photoreceptor cells from apoptosis in some pathologic conditions. In the current study, all retinocytoma cells were positive for αA-crystallin where the number of apoptotic cells was low. Furthermore, immunoreactivity for αA-crystallin, but not for αB-crystallin, correlated with a low number of apoptotic retinoblastoma cells. These results suggest that αA-crystallin may protect retinal tumor cells from the apoptotic process.

The present study demonstrated that αA-crystallin was expressed in the photoreceptors of normal retinas and in retinocytoma cells that show clear photoreceptor differentiation. Rao et al⁸ recently demonstrated selective αA-crystallin upregulation in experimental uveitis retina. This crystallin was localized primarily in the photoreceptor inner segments, the site of mitochondrial oxidative stress. They also showed that upregulation of αA-crystallin was associated with photoreceptor protection against apoptosis resulting from oxidative stress. In the retina, αB-crystallin was not upregulated in this animal model of uveitis, suggesting that the retina may use αA-crystallin selectively for protection against stress.

It was previously reported that oxidative stress was high in retinoblastoma cells.²⁴ Such stress might contribute to the high expression of αA-crystallin in retinal tumors. This is supported in part by our current in vitro study showing that αA-crystallin was upregulated in cultured human retinoblastoma cells exposed to H₂O₂-related oxidative stress (Figure 6). Together with immunohistopathological results and oxidative stress known to induce apoptosis and to counteract such cell death, it appears that αA-crystallin is overexpressed.

Because it is likely that immunofluorescent signals from sections are highly variable depending on uncontrolled subtlety of specimen preparation and fixation, semiquantitative immunohistochemistry may be tricky. To support the data evaluated by immunohistochemistry, we extracted total protein from formalin-fixed paraffin-embedded tissue sections and performed Western blot analysis. As shown in Figure 5, αA-crystallin overexpression was clearly detected, supporting the findings of immunohistochemistry. Demonstrated absence of correlation between αB-crystallin levels and AI may complement and support the validity of the data noted with αA-crystallin.

Based on the inverse correlation with apoptosis in the current study, suppression of αA-crystallin expression, rather than αB-crystallin, may be useful in controlling tumor growth. This novel observation reveals the necessity of future studies that include upregulation of αA-crystallin in a retinoblastoma animal model as the next step. Antisense or nucleotide-based anti–αA-crystallin therapies may be tried to inhibit αA-crystallin expression in retinoblastoma, similar to the reported molecular approach of αB-crystallin suppression for treatment of various systemic cancers.²⁵

Correspondence: Narsing A. Rao, MD, Doheny Eye Institute, 1450 San Pablo St, DRVC 211, Los Angeles, CA 90033 (nrao@usc.edu).

Submitted for Publication: March 13, 2008; final revision received July 8, 2008; accepted August 18, 2008.

Financial Disclosure: None reported.

Funding/Support: This study was supported by grants EY015714 and EY03040 from the Institutes of Health and by an unrestricted grant from Research to Prevent Blindness.