Andreas Schedl, PhD, Université de Nice-Sophia Antipolis, laboratoire de génétique du développement normal et pathologique, Centre de Biochimie, 06108 Nice Cedex 02, France. Tel: +00 33 4 92 07 6401 Fax: +00 33 4 92 07 6475 e-mail: email@example.com
Background: SOX9 is a transcription factor that fulfills multiple functions during development. In the hair follicle SOX9 is expressed in the outer layer of the epithelial sheath, and the hair stem cell compartment. Recent data suggest that Sox9 acts as a downstream target of the Sonic hedgehog (Shh) pathway. Activation of the Shh pathway is a major cause of cutaneous basal cell carcinoma (BCC). Here we test whether activation of SOX9 is a general feature of BCC, or whether it could be used as a biomarker to better define subtypes of these skin tumors. In addition we investigated SOX9 expression in other skin epidermal tumors.
Methods: Tumors sections were stained with hematoxylin & eosin (H&E). SOX9 activation was determined by immunofluorescence.
Results: SOX9 activation was observed in all subtypes of BCC tested. Staining was heterogeneous and could be detected among the basaloid cells of the palisading cell layer as well as in the tumour nest. SOX9 expression was detected in all adnexal tumors analyzed and absent in Bowen’s disease and Merkel tumor.
Conclusions: SOX9 expression is a general feature of BCC and adnexal skin neoplasms, suggesting a contribution of SOX9 to the pathogenesis of these tumors.
With an estimated two million new cases every year in the US, basal cell carcinoma (BCC) is the most common form of skin cancer. Mortality associated with BCC is low as tumors are slow growing, rarely metastasizing and easily treated by surgical excision. However, BCC frequently causes significant local destruction and disfigurement, if neglected or inadequately treated. Reasons underlying the increased frequency of BCC are complex, but exposure to ultra violet radiation, especially during childhood and adolescence, increases the probability of BCC development, with people of skin type 1 being at the highest risk.1,2,3 Other risk factors include exposure to ionizing radiation or ingestion of arsenic compounds. Moreover, several genetic conditions such as albinism, xeroderma pigmentosa, Bazex’s syndrome and the naevoid BCC syndrome (Gorlin’s syndrome) are associated with an increased risk of developing BCC.
Research over the last few years has showed an essential role for the signaling molecule Sonic hedgehog (SHH) and its receptors patched and smoothened (SMOH) in BCC formation. Upon binding of SHH, patched releases the SMOH protein, which in turn leads to the upregulation of the glioblastoma transcription factors GLI-1 and GLI-2.4 In sporadic BCC, the most frequent genetic alteration found corresponds to a loss of heterozygosity on chromosome 9q22 at the region of the PTCH locus.5 Other mutations that have been identified include those affecting the function of the different players of the Hedgehog (Hh) cascade (e.g. SMOH), thus leading to a constitutive activation of this pathway.6,7 Analysis in mutant mice has showed that Shh is required for hair follicle development and for the onset of anagen.8,9,10 Transgenic mice which show a constitutively activated Shh pathway, develop skin tumors such as trichoepithelioma, trichoblastoma, cylindroma and BCC.11,12,13 While members of the Shh pathway have been clearly implicated in BCC development the molecular events downstream of Gli remain largely unknown.
Sox9 belongs to the Sox (Sry box related) gene family, members of which are known to be involved in a large number of developmental processes.14 In the skin, Sox9 is expressed in the sebaceous and sweat glands, the outer root sheath of the hair follicles, as well as the hair stem cells located in the bulge compartment.15 Molecular analysis showed that Sox9 is activated in BCC in transgenic mice and in a limited number of human biopsies from BCC patients.15
Correlating cancer phenotypes with molecular characteristics is very important, as it helps to define the basis of tumor development and contributes to its diagnosis and clinical management. The aim of this study was to investigate whether Sox9 activation is a general feature of BCC in human patients or whether it is specifically associated with a certain subtype of BCC. In addition, we tested whether other follicular-related and adnexal neoplasms also show activation of SOX9 and compared these results with SOX9 immunoreactivity in other skin epithelial tumors, i.e. Bowen’s disease and Merkel tumor.
Materials and methods
Formalin-fixed, paraffin-embedded tissue blocks of 17 BCC specimens (nodular (n = 8), superficial (n = 4), infiltrative (n = 5) including three sclerodermiforms and one basosquamous variants and fibroepithelial tumor of Pinkus (n = 1)), trichoepithelioma (n = 4), trichilemmoma (n = 2), seborrheic keratosis (n = 2), poroma (n = 2), porocarcinoma (n = 2), syringoma (n = 1), sebaceous carcinoma (n = 1) as well as Bowen’s disease (n = 2) and Merkel tumor (n = 3) were selected from the file of the department of Pathology of the Henri Mondor hospital, Creteil (France).
Paraffin sections of surgical specimens were cut at 5-μm thickness, and either stained with hematoxylin & eosin (H&E) or subjected to immunostaining.
Tissue sections were dewaxed in xylene then rehydrated in successive baths of decreasing ethanol amounts. Antigens were retrieved by cooking for 2 min in a pressure cooker in a solution of 10 mM Na citrate (pH 6). Slides were rinsed in PBS and incubated for 45 min in a blocking solution (3% BSA, 10% donkey serum and 0.1% Triton) at room temperature. Sox9 antibodies (courtesy of Professor M. Wegner),16 were applied in a 1/1000 dilution (3% BSA, 3% donkey serum, 0.1% Triton) and left to incubate overnight at 4°C. After three washes in PBS, Cy3-conjugated anti-rabbit antibodies (Jackson Laboratories, Westgrove, PA, USA) were applied at a 1/150 dilution in PBS and incubated for 1 h at room temperature. After three washes in PBST (PBS with 0.1% Tween, Sigma, Sigma, Saint-Quentin Fallavier, France) slides were mounted in Vectashield™ with DAPI as mounting reagent (Vector Labs, Burlingame, CA, USA).
SOX9 positive cells are detected in all BCC subtypes
To test whether SOX9 activation is associated with all BCC subtypes we have investigated a panel of 16 BCC biopsies of various subtypes for the expression of SOX9. For every surgical specimen, and before immunostaining, corresponding paraffin sections were stained with H&E and classified by histopathological examination (Fig. 1A,C,E,G and I). In healthy skin, SOX9 expression is restricted to the follicular outer epithelial sheath, the stem cell compartment, the sebaceous and the sweat glands.15 In 1 out of 16 samples analysed no positive signal could be detected in the tumor nor in the surrounding tissue indicating that the SOX antigen had been lost in this sample. Consequently this specimen was not further analysed. All of the nodular BCC tested showed nuclear staining for Sox9 (Fig. 1B). Interestingly the staining intensity was heterogeneous with some nuclei showing a stronger signal compared to others. All of the superficial BCC analysed disclosed SOX9-positive cells (Fig. 1D). Staining was disparate with positive cells located at the peripheral palisade of basaloid cells as well as in the tumor nests. SOX9 was activated in the infiltrative BCC (Fig. 1F), as well as in all of the sclerodermiform tumors analysed (Fig. 1H). In some cases, SOX9-positive nuclei could be observed in the neighboring epidermal cells (data not shown and Fig. 1H). SOX9 was also activated in the fibroepithelioma of Pinkus, with a staining restricted to the anastomosing cords of basaloid cells as well as in the overlying epidermis (Fig. 1J). In all biopsies analysed, SOX9 was restricted to the epithelial tumor and completely absent in the surrounding stroma.
SOX9 is activated in other follicular-related neoplasia: trichoepithelioma, trichilemmoma
Trichoepithelioma and trichilemmoma are tumors believed to be derived from hair follicles. A total of six biopsies were examined. They were first stained with H&E then classified by histological examination (Fig. 2A,C). Sections were submitted to immunodetection of the SOX9 antigen. All samples analyzed contained tumor cells that were moderately to strongly positive for SOX9 (Fig. 2B,D). SOX9 signal was restricted to the lobules of epithelial cells. Four out of the six neoplasms analysed showed ectopic SOX9-expressing cells among the surrounding upper epidermis (data not shown).
Analysis of Sox9 in other skin neoplasms
Apart from its expression in the hair follicle, SOX9 expression can also be detected in sebaceous and sweat glands.15 To answer whether SOX9 may also be present in other adnexal and non-adnexal tumors, immunostaining was performed on seborrheic keratosis, Bowen’s disease, Merkel tumor, syringoma, poroma, porocarcinoma and sebaceous carcinoma. Syringomas and poromas are benign sporadic tumors that arise from eccrine sweat glands.17 Porocarcinoma and sebaceous carcinoma are aggressive tumors with potential for local recurrence and distant metastasis.18,19 Interestingly, all other adnexal tumors showed strong staining for SOX9 (Fig. 3A–D, and data not shown). The immunofluorescent signal was nuclear and mostly located in basaloid cells of the tumor nest. In poroma, palisading cells were readily positive for SOX9 (Fig. 3B) whereas only few positive cells were detected at the periphery of the epithelial lobules of the sebaceous carcinoma (Fig. 3D).
By contrast, only few scattered positive cells were detected in seborrheic keratoses, and, in Bowen’s disease and Merkel tumor; SOX9 immunoreactivity was virtually absent, with only very few positive cells located at the upper layer of the adjacent normal epidermis ectopically expressing SOX9 (Fig. 3F,H and data not shown).
Genes involved in developmental processes are often implicated in tumorigenesis and overexpression of SOX genes has been showed in a large variety of tumors.20 SOX9 is activated and required for the development of many different organs and it has been detected in cancers related to those organs, including primary prostate cancer, ovarian Sertoli cell tumors and brain tumors.21,22,23 In normal skin, SOX9 expression is restricted to adnexal structures, especially the pilosebaceous system.
BCC can be classified according to histological growth pattern. We have focused our analysis on four main subtypes of BCC: the nodular, the superficial, the infiltrative and the sclerodermiform. Nodular BCC are considered as low-risk tumors, whereas the superficial, the infiltrative and the sclerodermiform are characterized by an increased probability of subclinical extension and/or incomplete excision and/or aggressive local invasive behavior and/or local recurrence.24 We have also analysed a fibroepithelioma of Pinkus, a rare, low risk subtype of BCC with a distinct growth pattern.25 The immunofluorescent assays that we report here show that SOX9 is strongly expressed in all BCC subtypes independent of their aggressiveness thus, showing that SOX9 activation is a general feature of BCC development. These results further suggest that BCC’s differentiation is related to hair follicle differentiation. Interestingly the staining was heterogeneous in all cases, suggesting that these tumors are composed of cells from various origins and/or it might reflect the different status of the tumor cells.
Trichoepithelioma are benign epithelial-mesenchymal neoplasms with follicular differentiation, they share many morphologic and immunologic features with BCC.26,27 Trichilemmoma was first described as a cutaneous neoplasm with differentiation towards pilosebaceous follicular epithelium in 1962. The periphery of trichilemmoma resembles the lower portion of normal hair follicles, immunological analysis of specimens has reported the presence of cells expressing cytokeratin markers specific for the hair follicle; taken together these data suggest that trichilemmoma is derived from the outer root sheath or is differentiated towards the outer root sheath.28,29 SOX9 was detected in both, trichoepithelioma and trichilemmoma, a result in agreement with the pilosebaceous origin of these neoplasms.
Transgenic analysis has showed that overexpression of GLI-1 can induce BCC, trichoepithelioma, cylindroma and trichoblastoma.11 As GLI-1 is activated by SHH, the expression of SOX9 in trichoepithelioma and trichilemmoma further supports our hypothesis that SOX9 represents a general effector of the SHH signaling pathway. Interestingly, in all of the tumor samples analyzed, SOX9 staining was restricted to the basaloid cells, whereas the surrounding stroma was negative. In contrast, some epidermal cells surrounding the tumor site showed ectopic activation of SOX9. This ectopic expression was not detected in all surgical specimens tested, suggesting that SOX9 activation in the epidermis is likely to be a consequence rather than the cause of adnexal neoplasm development. Ectopic activation of GLI-1 in histologically normal basal cells surrounding the tumor site has been reported in other BCC analysis, thus SOX9 ectopic expression in normal epidermal cells might reflect the activation of the SHH pathway in the neighboring tumor tissue.30
All other adnexal neoplasms analyzed also showed SOX9 activation. Interestingly, the Hh signaling pathway is involved in sebaceous gland development with Indian Hedgehog being the most prominent ligand.31,32 Constitutive activation of the Hh pathway by overexpression of Gli-1, Gli-2 or a mutated active form of Gli-2 lead to sebaceous gland hyperplasia and an increased number of sebaceous glands. 31 Components of the Hh pathway are up-regulated in mouse and human sebaceous tumors suggesting that this pathway may be involved in the development of those pathologies.32 The activation of SOX9 we have detected in the sebaceous carcinoma might therefore reflect the activation of the Hh signaling pathway.
BCC can be difficult to distinguish from non-adnexal epithelial skin tumors. Here we have shown that seborrheic keratoses, Bowen’s disease and Merkel cell carcinoma33 lack SOX9 expression, which may therefore be used as a diagnostic marker to distinguish these tumors from BCC in difficult cases. By contrast, we have shown in the present study that SOX9 is widely expressed in skin adnexal tumors, either with hair follicle, sebaceous or sudoral differentiation, thus indicating that SOX9 cannot be used as a marker for the differential diagnosis with trichoepithelioma or adnexal carcinomas.
In conclusion, SOX9 activation is a general hallmark of BCC, whatever their histological subtype. Moreover, SOX9 activation is also found in adnexal skin tumors with either hair follicle, sebaceous or sudoral differentiation, whereas it is absent in non-adnexal epithelial skin tumors. This general expression of SOX9 raises the interesting question that this transcription factor may be required and/or sufficient for tumors to develop. Future mouse models expressing Sox9 ectopically will allow us to address this issue.
We would like to thank Michael Wegner (Erlangen, Germany) for the provision of Sox9 antibodies. This work was supported by an INSERM Avenir fellowship, the ARC (Association pour la recherche sur le cancer) grant 5198, and the FRM (Fondation pour la Recherche médicale).