SOX10 Expression as Well as BRAF and GNAQ/11 Mutations Distinguish Pigmented Ciliary Epithelium Neoplasms From Uveal Melanomas

Abstract

Purpose: Adenocarcinomas or adenomas derived from pigmented ciliary epithelium (APCE) are exceptionally rare ocular tumors. These tumors have pigmented and epithelioid features, and some APCEs are negative for keratin markers and positive for melanocytic markers. It is especially difficult to distinguish APCEs from uveal melanoma (UM). Accordingly, we examined protein expression and genetic mutations associated with APCE to facilitate diagnosis.

Methods: Five APCE and 11 UM samples were obtained from patients during surgical resection at our institute. APCE and UM ocular structures were compared comprehensively. Protein expression and genetic alterations involved in malignant melanoma were evaluated.

Results: SOX10 was expressed diffusely in all 11 UMs and in surrounding uveal or choroidal melanocytes, but not in the APCEs or nontumorous pigmented epithelia. Additionally, the expression patterns of cytokeratins and melanocytic markers differed between UMs and APCEs. We identified BRAF V600E mutations in four of five APCE samples, but not in the 11 UM samples. Moreover, GNAQ or GNA11 mutations were found in 10 of the 11 UM samples, but not in APCE samples. NRAS mutations were not observed in either tumor group examined.

Conclusions: APCE is a separate entity distinguished from UM by the absence of SOX10 expression and presence of the BRAF V600E mutation. These results have implications for diagnosis, providing a means to distinguish between UM and APCE.

A pigmented epithelium is found in the iris, ciliary body, and retina. Destruction of the RPE causes several retinal diseases, including include age-related macular degeneration, which is the leading cause of irreversible blindness in the >55-year age group worldwide.¹ Recently, there have been remarkable developments in the field of regenerative medicine for retinal diseases. Embryonic stem cells and induced pluripotent stem cells have been used to generate RPE, and consequently, the pigmented epithelium has attracted considerable attention.² 

Adenocarcinoma or adenoma of the pigmented ciliary epithelium (APCE) is an exceptionally rare eye tumor arising from the pigmented epithelium. These tumors have pigmented, epithelioid, and glandular features, and often have PAS-positive basement membranes.³ In addition, some APCEs are negative for keratin markers and positive for melanocytic markers.^4–6 Recently, we reported a case of APCE with a diagnosis complicated by uveal melanoma (UM) markers.⁷ Therefore, we analyzed new markers and gene alterations in a series of APCE and UM tumor samples to identify a means to distinguish neoplasms of the pigmented ciliary epithelium from UMs. 

The SOX10 transcription factor is crucial for the specification and maturation of the neural crest. Previously, we reported that SOX10 is associated with salivary gland development and is a reliable marker of salivary gland tumors.⁸ SOX10 also is associated with melanocyte development and is a reliable marker of malignant skin melanomas.⁹ Unlike the pigmented epithelium, UMs arise from uveal melanocytes derived from the neural crest.^10,11 Therefore, based on their cells of origin, UMs and APCE may be distinguished by differential SOX10 expression. However, the roles of SOX10 expression in the human eye and eye tumors have not been investigated. Malignant melanomas of the skin frequently have BRAF or NRAS mutations,^12,13 while UMs do not.^14,15 Instead, UMs frequently have GNAQ or GNA11 mutations.^16,17 However, to our knowledge genetic alterations that influence APCEs have not been reported to date. To identify distinct differences between APCEs and UMs, we investigated SOX10 expression and the mutation status of BRAF, NRAS, GNAQ, and GNA11 in APCE and UM samples. 

Methods

Tissue Selection

Adenocarcinoma and adenoma samples of the pigmented ciliary epithelium and UM tissue samples surgically resected between 2000 and 2013 at the National Cancer Center Hospital, Tokyo, Japan (Table 1) were used. No tumor had extraocular lesions. Informed consent was obtained from all patients, and the Ethical Committee of the National Cancer Center Hospital, Tokyo, Japan approved the procedures (Approval #2010-077). 

Table 1

 

Clinicopathologic Characteristics of Cases of the Tumors

 

Animals

Wild-type C57BL/6J mice were purchased from SLC Japan (Hamamatsu, Japan). The Cre-expressing transgenic Wnt1-Cre strain¹⁸ was mated with an EGFP reporter CAG-CAT-EGFP strain¹⁹ to obtain Wnt1-Cre/CAG-CAT-EGFP double-transgenic mice. All animal experimental procedures were approved by the Institutional Animal Care and Use Committee of Keio University, and were performed in accordance with the Guidelines for the Use of Laboratory Animals from the National Institutes of Health (NIH; Bethesda, MD, USA) and the ARVO Statement for the Use of Animals in Ophthalmic and Vison Research. 

Immunohistochemistry

Formalin-fixed paraffin-embedded specimens were cut into 4-μm thick sections. Deparaffinized sections were subjected to hematoxylin–eosin, PAS, and immunohistochemical staining with the following primary antibodies: SOX10 (1:200, N-20; Santa Cruz Biotechnology, Santa Cruz, CA, USA), cytokeratin AE1/AE3 (1:100; Dako, Glostrup, Denmark), cytokeratin CAM5.2 (1:20; Becton Dickinson Immunocytometry Systems, San Jose, CA, USA), S100 (1:2000, rabbit polyclonal; Dako), HMB45 (1:10; Dako), Melan A (1:100, A103; Dako), and MITF (1:100, D5; Dako). Each section was exposed to 0.3% hydrogen peroxide for 15 minutes to block endogenous peroxidase activity. For staining, an automated stainer (Dako) was used according to the manufacturer's protocol. For SOX10; A103, ChemMate EnVision HRP/DAB Kits (Dako) were used for detection with melanin bleaching using warm 3% hydrogen peroxide in 0.05 M phosphate buffer (pH 7.4) at 55°C for 2 hours. In these conditions, the tissue morphology and antigenicity of various immunohistochemical markers were effectively preserved.^18–21 For detection of other antibodies, ChemMate Envision G2 System AP Kits (Dako) were used. Appropriate positive and negative controls were used for each antibody. Nuclear staining of tumor cells was considered positive for SOX10, S100, and MITF. For cytoplasmic staining, if no part of the tumor stained for cytokeratin AE1/AE3, cytokeratin CAM 5.2, HMB45, or Melan A, staining was considered negative. If less than half of the tumor was stained, the staining pattern was identified as single positive (1+). If more than half of the tumor was stained, the staining was considered double positive (2+). For cytokeratins AE1/AE3 and CAM5.2, intensity was categorized as strong or weak. For mouse sections, immunostaining analyses were performed as described previously.²⁰ The embryonic eyes of wild-type and Wnt1-Cre/CAG-CAT-EGFP double-transgenic mice were dissected with perfusion and post-fixed with 4% paraformaldehyde (PFA) in 100 mM PBS (pH 7.4) for 12 hours at 4°C, cryoprotected with 15% and 30% sucrose, and embedded into OCT compound (Tissue-Tech, Doral, FL, USA). The 20-nm thick frozen sections were prepared with a cryostat (Leica CM3050s, Wetzlar, Germany) and immunostained with primary antibodies for 12 hours at 4°C and with Alexa488 and Alexa555-conjugated specific secondary antibodies (Molecular Probes, Eugene, OR, USA) for 2 hours at 25°C. Primary antibodies used were rabbit polyclonal anti-GFP (1:500; MBL, Tokyo, Japan) and goat polyclonal anti-hSOX10 (1:200; R&D Systems, Minneapolis, MN, USA). The samples were examined with a laser scanning confocal microscope (LSM700; Carl Zeiss Meditec, Oberkochen, Germany). 

Transmission Electron Microscopy (TEM)

The detailed TEM procedure has been described previously.²² Briefly, the eyes from 3-week-old C57bl6j wild-type mice were dissected without perfusion and fixed with 2.5% glutaraldehyde in 60 mM HEPES buffer (pH 7.4) for 12 hours at 4°C and with 1% OsO₄ (Nissin EM) for 2 hours at 4°C. Samples then were dehydrated in diluted ethanol, acetone, and QY1, and embedded in Plain Resin (Nissin EM). After complete polymerization for 24 hours at 50°C and for 96 hours at 70°C, 70-nm thick ultrathin sections were prepared with an ultramicrotome (Leica UC7) and stained with uranyl acetate and lead citrate for 10 minutes each. The sections were observed under a TEM (JEOL 1400plus, Tokyo, Japan). 

Analysis of BRAF, NRAS, GNAQ, and GNA11 Mutations

Tumor specimen sections (10 μm thick) were deparaffinized and subjected to DNA extraction. The mutation status of BRAF, NRAS, GNAQ, and GNA11 was assessed by high-resolution melting analysis²⁰ and direct sequencing. Dissected samples were incubated in 50 μL DNA extraction buffer (50 mM Tris-HCl, pH 8.0, 1 mM ethylenediaminetetraacetic acid, 0.5% vol/vol, Tween 20, 200 μg/mL proteinase K) at 50°C overnight. Next, the samples were heated at 100°C for 10 minutes to inactivate proteinase K and then directly subjected to PCR using primer pairs encompassing exon 15 of BRAF, exons 1 and 2 of NRAS, exons 4 and 5 of GNAQ, and exons 4 and 5 of GNA11. The primers are listed in Supplementary Table S1. PCR products were separated on a 2% (wt/vol) agarose gel by electrophoresis and recovered using the QIAquick Gel Extraction Kit (Qiagen, Hilden, Germany). Isolated PCR products were sequenced using an Applied Biosystems 3130 Genetic Analyzer (Foster City, CA, USA). 

Results

Histochemical and Immunohistochemical Staining Characteristics of Pigmented Ocular Tumors

All five APCEs had epithelioid-glandular features, pigment granules, and obvious PAS-positive basement membranes (BMs). Three adenomas and one adenocarcinoma arose from the ciliary bodies/iris and one adenocarcinoma arose from the retina. Two adenocarcinomas showed obvious expansive invasion of the subepithelial tissue. All UMs had brown granules, at least in the focal areas. No UMs had PAS-positive BMs. All five APCEs were negative for SOX10 expression, while all UMs showed diffuse SOX10 expression in an immunohistochemistry analysis (Fig. 1; Table 2). All APCEs showed faint and/or rare focal expression of cytokeratin AE1/AE3. APCEs were positive for cytokeratin CAM5.2 (3/5 cases), S100 (4/5), Melan A (4/5), HMB-45 (4/5), and MITF (3/5). UMs were positive for S100 (11/11 cases), Melan A (11/11), HMB-45 (10/11), and MITF (7/11). One UM had distinct epithelioid features and was the only UM positive for cytokeratin AE1/AE3 and cytokeratin CAM5.2 (Figs. 1P, 1Q, 1R). 

Figure 1

 

SOX10 and PAS distinguish APCE from UMs. Histologic and immunohistochemical features of representative eye pigment tumors. PAS-positive BMs were observed in APCEs (B, E, H, K), but not in UMs. Additionally, epithelioid- and papillary-type UMs do not involve the BM (Q). SOX10 is not expressed in APCEs (C, F, I, L), but exhibits diffuse expression in UMs (O, R).

Table 2

 

Histologic and Immunohistochemical Staining Characteristics of APCE Cases and Mutation Status of Tumors Analyzed by Direct Sequencing

 

SOX10 Expression in Human Ocular Components

SOX10 expression was not observed in the pigmented or nonpigmented epithelium in nontumorous iris, ciliary body, or retina samples, even though melanogenesis occurs in the ocular pigment epithelium. However, the choroid and melanocytes in subepithelial tissues of the iris and ciliary body expressed SOX10 (Fig. 2A). To elucidate the morphologic differences between the normal RPE and choroid, TEM was performed. The RPE, just below the retina, was separated from the choroid by the capillary layer, and formed a bond with the BM. The RPE was a single layer of relatively large cells, while the choroid was flat and small, and overlapping was observed. In the RPE, melanin granules were relatively large and few in number; however, in the choroid, melanin granules were small and abundant (Fig. 2B). Next, we analyzed the expression of SOX10 in Wnt1-Cre/GFP transgenic mice on embryonic day 13. Wnt1-Cre/GFP transgenic mice were used widely to study the neural crest and its derivatives. The expression of SOX10 was observed only in the outer layer of the optic cup, which was derived from the neural crest at the completion of eye development (Fig. 2C). 

Figure 2

 

SOX 10 expression reflects the nontumor eye structure and development of neural crest-derived cells. (A) Nontumorous pigmented epithelium of the iris (a), ciliary body (b), and retina (c) do not express SOX10. Melanocytes of the subepithelial tissue of the iris (e), ciliary body (f), and choroid (h) express SOX10 (arrowheads). (B) Based on TEM, the layer of RPE on the BM and capillary vessel layer (CV) separate RPE from the choroid membrane (CM). (C) SOX10 is expressed in the outer layer of the optic cup region, but is not expressed in the optic cup region. The SOX10-positive cells were derived from the neural crest and were related to APCE and UM characteristics and their origin. DAPI (a), SOX10 (b), Wnt1-EGFP (c), and Merged (d).

Mutation Status of BRAF, NRAS, GNAQ, and GNA11 in APCEs and UMs

Using a high-resolution melting analysis and direct sequencing, we detected BRAF mutations in four of five APCEs (Table 3; Fig. 3). The four APCEs with BRAF mutations arose from ciliary bodies and/or irises. All of these mutations were heterozygous T-to-A transversions. These mutations resulted in a glutamic acid (p.V600E) substitution and have been detected in several human neoplasms, including skin melanoma. No BRAF mutation was detected in the single APCE that arose from the retina or in the 11 UMs. No NRAS mutations were found in any case. Mutually exclusive GNAQ or GNA11 mutations were detected in 10 of 11 UMs, including the tumor with epithelioid and papillary features. Five of six GNAQ mutations were heterozygous A-to-C transversions, predicted to result in a proline (p.Q209P) substitution. One GNAQ mutation was a heterozygous A-to-T transversion, predicted to result in a leucine (p.Q209L) substitution. All four GNA11 mutations were heterozygous A-to-T transversions, predicted to result in leucine (p.Q209L) substitution. GNAQ and GNA11 missense mutations were only detected in UMs. 

Table 3

 

Mutation Status of Tumors Analyzed by Direct Sequencing

 

Figure 3

 

Detection of BRAF, GNAQ, and GNA11 mutations in eye pigment tumors. Detection of heterozygous mutations by Sanger sequencing. BRAF V600E in APCE (A), and GNAQ (B, C) and GNA11 (D) in UMs.

Discussion

SOX10 expression was detected in all UM and uveal melanocytes, regardless of morphology, but not in any of the APCEs or nontumorous pigmented epithelia. Furthermore, the BRAF V600E mutation, but not the GNAQ or GNA11 mutations, was identified in APCEs. Thus, APCEs are distinct from UMs and can be distinguished by the absence of SOX10 expression and the presence of different genetic alterations. As an exception, a BRAF mutation was not observed in one APCE. This tumor originated from the retina and not the ciliary bodies or irises, like the other four APCEs, potentially explaining the different mutation profile. Although only a few reports of the epithelium of APCEs have shown extraocular extension or metastasis to other organs, the prognosis of the tumor remains unclear because it is very rare.²¹ In contrast, UM has high metastasis and mortality rates.²² The differential diagnosis of these two tumor types is important because they can be confused easily. However, as reported previously and shown here, with the exception of SOX10 expression, melanocytic markers yield variable results in APCEs and UMs.^4–6 All APCEs were positive for cytokeratin AE1/AE3, but expression was only weak and/or focal. Hence, immunostaining for keratin or melanocytic markers, including MelanA, HMB45, MITF, and S100, is not diagnostically useful. 

We used TEM to observe ocular melanocytic tissues and to elucidate detailed morphologic differences between the two tumor types. Despite similar pigmented tissues, the structure of the RPE and choroid were completely different. The Wnt1-Cre transgenic mouse line is used extensively in studies of the development of the neural crest, its derivatives, and the midbrain. Wnt1-Cre transgenic mice exhibit a substantial elevation of Wnt1 expression in the outer cap of the ocular site, and this increased expression was correlated with SOX10 expression. 

We encountered prominent epithelioid and papillary features in one particular ocular melanocytic tumor. Initially, this tumor was difficult to identify accurately (Fig. 1, case 15) as it expressed keratin and melanocytic markers, expressed SOX10, and did not have a PAS-positive BM. The diagnosis of UM was supported by the presence and absence of GNAQ and BRAF mutations, respectively. 

SOX10 remains a useful melanocytic marker in ocular lesions and has a valuable adjunctive role in the diagnosis of APCEs. A few reports exist of APCEs without PAS-positive BMs, some of which are called “papillary adenocarcinoma.”³ Indeed, based on our results, some of these tumors might have been misdiagnosed UMs. 

In conclusion, an adenocarcinoma or APCE can be distinguished from UM by the lack of SOX10 expression. This expression difference reflects differences in the origin of the tumor and underlying genetic alterations, such as a high frequency of BRAF mutations and a lack of GNAQ or GNA11 mutations. Our results strongly suggested that SOX10-positive tumors originate from the choroid or melanocytes in subepithelial tissues, and that SOX10-negative tumors originate from the pigmented epithelium of the iris and/or ciliary body. Accordingly, SOX10 is a valuable immunohistochemical marker to differentiate APCE from UM. These results present a means for the precise diagnosis of two easily confused eye pigment tumors. 

Acknowledgments

The authors thank Sachiko Miura, Toshiko Sakaguchi, and Chizu Kina for their skillful technical assistance. 

Disclosure: T. Mori, None; A. Sukeda, None; S. Sekine, None; S. Shibata, None; E. Ryo, None; H. Okano, None; S. Suzuki, None; N. Hiraoka, None