Expression of the hair stem cell-specific keratin 15 in pilar tumors of the skin

ARTICLE

Keratins comprise a multigenic family of fibrous proteins that form the cell cytoskeleton; they are made of distinct polypeptides that have been classified, according to their MW and isoelectric point, as type I (or acidic or low MW) and type II (or basic or high MW) polypeptides. So far, about 30 different polypeptides have been recognized [1, 2]; their expression varies according to the state of differentiation of each individual epithelial cell. We and others [3] recently observed that the monoclonal antibody C8/144B, generated against the CD8 molecule of suppressor and cytotoxic T cells, also reacts with cells of the outer root sheath of hair follicles around the bulge area, but not with other parts of the hair follicle or the epidermis. Recent investigations (expression cloning and immunoprecipitation) showed that this antibody recognizes native (but not denatured) keratin 15 (K15) which seems to be specifically expressed, within human skin, by hair follicle stem cells located in the bulge [4]. So far K15 has not been studied extensively in normal and diseased human skin. We therefore investigated the expression of K15 in a group of cutaneous tumors of known or alleged follicular origin, in order to assess the usefulness of the C8/144B antibody in the diagnosis of pilar tumors of the skin.

Material and methods

Tissue samples

These included biopsy or excision specimens of various epithelial tumors of known or alleged pilar differentiation (Table I). The material had been collected in our dermatopathology laboratory, formalin-fixed and paraffin-embedded.

Immunohistochemistry

4 µm-thick sections placed on electrostatically-treated glass slides were deparaffinized and rehydrated, then immunolabelled according to a labelled streptavidin-biotin-peroxidase technique (kit LSAB Dako, Copenhagen, Denmark), including the following steps: a) antigen retrieval (heating the sections in citrate buffer with a microwave oven); b) inhibition of endogenous peroxidase with 1% H₂O₂ in PBS; c) incubation of the sections with blocking (non-immune) serum; d) incubation with the antibody C8/144B (Dako, Copenhagen, Denmark, diluted 1: 10) for 60 min in moist chambers at room temperature; e) incubation with biotin-conjugated antiserum to mouse immunoglobulins (10 min); f) incubation with peroxidase-conjugated streptavidin (10 min). The reaction was revealed with aminoethylcarbazole as chromogen. Negative controls were performed by omitting the first layer antibody and they proved consistently negative.

In most specimens normal hair follicles were present that showed cytoplasmic labelling of basal cells of the outer root sheath in the bulge area (Fig. 1). Several specimens (namely BCC) also contained a variable amount of lymphocytes that showed membrane staining (corresponding to the CD8 antigen). These structures served as built-in positive controls, and only specimens with such internal controls were considered. The remaining cutaneous structures (including basal epidermal keratinocytes and sweat glands) were unlabelled.

Among hair-follicle tumors, 5/8 trichoepitheliomas showed diffuse cytoplasmic staining in a proportion of cells varying from 10 to over 75% (Fig. 2). Both basal-cell nevi showed positive staining, but the percentage of immunoreactive cells was lower. One case each of three trichofolliculomas, six trichilemmal cysts and 11 trichilemmomas comprised some K15-positive cells. The remaining cases (including all 7 epidermoid cysts, 11 pilomatricomas and 17 BCC, most of which showed histologically signs of pilar differentiation) were unreactive (Fig. 3).

Discussion

The C8/144B monoclonal antibody, raised against the carboxy-terminal peptide of the CD8 antigen [5], was found to produce cytoplasmic immunostaining of keratinocytes of the hair follicle bulge. Recently it was shown that this clone recognizes native keratin 15 (K15) [4]. Although K15 shows molecular homology with K14, expression cloning and immunoprecipitation experiments have shown that the C8/144B antibody does not cross-react with K14 [4]. In order to further clarify this point, we carried out comparative studies on specimens of normal skin and some K15-positive tumors with an antibody to K14 (clone LL002) (data not shown). In keeping with previous results [6], we found K14 expression in basal and suprabasal cells of the hair follicle and the surface epidermis, whereas C8/144B never produced such labelling in paraffin-embedded tissue sections. Also in the tumors studied, the C8/144B and LL002 antibodies yielded different labelling patterns. These results further confirm the fact that the C8/144B antibody does not recognize K14.

K15 is a type I (acidic) keratin polypeptide of 50 kDa MW (pI 4.9) that has so far received little attention in immunohistochemical studies of normal and diseased skin. Biochemical studies of human [7] and animal (sheep and mouse) skin [8] have claimed that K15 is expressed at variable levels in the basal layer of the outer root sheath, in basal epidermal keratinocytes and sebaceous glands; however in our study and in that of Lyle et al. [4], K15 was not detected in basal keratinocytes outside the hair follicle. Within this structure, the distribution of K15 resembles that of K19, a 40 kDa type I (acidic) polypeptide, also expressed along the outer root sheath (including cells of the hair bulge) [6]; however K19 has a broader pattern of expression as compared with K15, since it is also expressed by more differentiated transient-amplifying cells located within the lower hair follicle and the epidermis [4, 6]. In keeping with this fact, K19 has been found to be expressed by a wider array of cutaneous (adnexal) tumors, including trichoblastomas and BCC [9, 10]. Thus, K15 appears as a more specific immunohistochemical marker of hair follicle bulge cells.

Our study shows that K15 has a characteristic expression pattern within pilar tumors. The tumor most frequently expressing K15 is trichoepithelioma (TE), both of the ordinary and the desmoplastic type, followed by basal cell nevi. Trichofolliculomas and trichilemmal tumors only rarely contain K15-positive cells. It seems likely that K15-positive cells correspond to stem cells of the bulge, suggesting that TE differentiate (at least partly) toward these cells. By contrast, pilomatricomas do not express K15, in keeping with their alleged origin from hair-matrix cells [11]. Epidermal cysts are also devoid of K15-positive cells, in keeping with their alleged origin from the upper part of the hair follicle (infundibulum). BCC (of the keratotic type) were also found consistently K15-negative, and this finding may have histogenetic implications. Indeed, despite the fact that BCC is the commonest human malignancy, its origin is not yet known with certainty. On the basis of immunohistochemical studies, namely using antibodies to various keratin polypeptides, BCC cells have been found to resemble either bulge [10] or matrix follicular cells [12]. Our results, showing no expression of K15 in BCC, suggest that this neoplasm is not related to hair follicle bulge cells.

From a practical point of view, the differential expression of K15 in TE and (keratotic) BCC may be useful for their differential diagnosis. Indeed, these tumors may be difficult to separate on histological examination since both are made of nodular growths of basophilic cells, surrounded by abundant fibrovascular stroma. The presence of CD34-positive dermal dendritic cells in the peritumoral stroma [13], the expression of the bcl-2 oncoprotein [14] and a continuous peritumor staining with the lectin PNA [15] have been claimed to be features favoring the diagnosis of TE over BCC, but these findings were subsequently disputed [9, 16, 17]. Our results suggest that K15 staining may be an additional useful adjunct for the differentiation between BCC and TE, since K15 expression (even at low levels) would strongly favor the diagnosis of TE vs BCC.

CONCLUSION

REFERENCES

1. Moll R, Franke W, Schiller D, Geiger B, Krepler R. The catalog of human cytokeratins: patterns of expession in normal epithelia, tumors and cultured cells. Cell 1982; 31: 11-24.

2. Heid H, Moll R, Franke W. Patterns of expression of trichocytic and epithelial cytokeratins in mammalian tissues. I. Human and bovine hair follicles. Differentiation 1988; 37: 137-57.

3. Lyle S, Elder D, Cotsarelis G. A cellular marker for the human hair follicle bulge identifies hair follicle stem cells (abstr). J Invest Dermatol 1998; 110: 491.

4. Lyle S, Christofidou-Solomidou M, Liu Y, Elder D, Cotsarelis G. The C8/144B monoclonal antibody recognizes cytokeratin 15 and defines the location of human hair follicle stem cells. J Cell Sci 1998; 111: 3179-88.

5. Mason D, Cordell J, Gaulard P, Tse A, Brown M. Immunohistological detection of human cytotoxic/suppressor T cells using antibodies to a CD8 peptide sequence. J Clin Pathol 1992; 45: 1084-8.

6. Schirren CG, Burgdorf W, Sander C, Plewig G. Fetal and adult hair follicle. An immmunohistochemical study of anticytokeratin antibodies in formalin-fixed and paraffin-embedded tissues. Am J Dermatopathol 1997; 19: 334-40.

7. Eichner R, Kahn M. Differential extraction of keratin subunits and filaments from normal human epidermis. J Cell Biol 1990; 110: 1149-68.

8. Whitbread L, Powell B. Expression of the intermediate filament keratin gene, K15, in the basal cell layers of epithelia and the hair follicle. Exp Cell Res 1998; 244: 448-59.

9. Schirren CG, Rütten A, Kaudewitz P, Diaz C, McClain S, Burgdorf W. Trichoblastoma and basal cell carcinoma are neoplasms with follicular differentiation sharing the same profile of cytokeratin intermediate filaments. Am J Dermatopathol 1997; 19: 341-50.

10. Yoshikawa K, Katagata Y, Kondo S. Biochemical and immunohistochemical analyses of keratin expression in basal cell carcinoma. J Dermatol Sci 1998; 17: 15-23.

11. McKee P. Pathology of the Skin. 2nd edition. London: Mosby-Wolfe, 1996.

12. Kore-Eda S, Horiguchi Y, Ueda M, Toda K, Imamura S. Basal cell carcinoma cells resemble follicular matrix cells rather than follicular bulge cells: immunohistochemical and ultrasructural comparative studies. Am J Dermatopathol 1998; 20: 362-9.

13. Kirchman T, Prieto V, Smoller B. CD34 staining pattern distinguishes basal cell epithelioma from trichoepithelioma. Arch Dermatol 1994; 130: 589-92.

14. Smoller B, van de Rijn M, Lebrun D, Warnke R. bcl-2 expression reliably distinguishes trichoepitheliomas from basal cell carcinomas. Br J Dermatol 1994; 131: 28-31.

15. Vigneswaran N, Haneke E, Peters K. Peanut agglutinin immunohistochemistry of basal cell carcinoma. J Cutan Pathol 1987; 14: 147-53.

16. Basarab T, Orchard G, Russell-Jones R. The use of immunostaining for bcl-2 and CD34 and the lectin Peanut Agglutinin in differentiating between basal cell carcinomas and trichoepitheliomas. Am J Dermatopathol 1998; 20: 448-52.

17. Swanson P, Fitzpatrick M, Ritter J, Glusac E, Wick M. Immunohistologic differential diagnosis of basal cell carcinoma, squamous cell carcinoma, and trichoepithelioma in small cutaneous biopsy specimens. J Cutan Pathol 1998; 25: 153-9.

18. Wick M, Swanson P. Cutaneous Adnexal Tumors. A Guide to Pathologic Diagnosis. Chicago: ASCP Press, 1991.

19. Kanitakis J. Solid cutaneous tumors. In: Diagnostic Immunohistochemistry of the Skin. Kanitakis J, Vassileva S, Woodley D, eds. London: Chapman & Hall Med., 1998: 279-99.