Int. J. Cancer: 56,314-319 (1994) 0 1994 Wilcy-Liss, Inc.

Publication of the International Union Against Cancer Publication de I'Union Internationale Contre le Cancer

ANALYSIS OF HLA-DR EXPRESSION ON KERATINOCYTES IN CERVICAL NEOPLASIA Nicholas COLEMAN' and Margaret A. STANLEY Department of Pathology, University of Cambridge, Tennis Court Road, Cambridge CB2 lQP, UK.

We have investigated the expression in vivo and in vitro of HLA-DR in pre-invasive squamous-cell neoplasia of the uterine cervix. lmmunohistochemistry of cervical biopsies demon- strated HLA-DR expression by cervical keratinocytes in 50% of cases of high-grade squamous intra-epithelial neoplasia, al- though the molecule was rarely expressed in low-grade squa- mous intra-epithelial neoplasia and was absent from normal ectocervical cells. HLA-DR-positive high-grade lesions were associated with significantly greater numbers of T lymphocytes in the immediately sub-epithelial stroma than were the HLA-DR- negative high-grade lesions. In vitro HLA-DR expression was absent from normal ectocervical epithelium, and from the HPV type I6 containing cell lines W I2 (representing low-grade squamous intra-epithelial neoplasia) and CaSki and SiHa (each representing high-grade squamous intra-epithelial neoplasia), both in monolayer and in organotypic raft culture. HLA-DR expression was induced in all cell types following recombinant interferon gamma treatment. Our data suggest that the expres- sion of HLA-DR by keratinocytes in some high-grade cervical lesions in vivo may be due to local induction of the molecule by pro-inflammatory cytokines released by immunocompetent cells. The functional significance of HLA-DR positivity in enabling an effective host immune response to neoplastic cervical keratino- cytes remains unclear. o 1994 Wiley-Liss, Inc.

Evidence suggests that human papilloma viruses (HPVs) have an important aetiological role in the development of squamous-cell carcinoma (SCC) of the uterine cervix (Zur Hausen, 1991). However, additional factors are involved in determining the eventual clinical outcome following HPV infection in the genital tract. Host immune responses may be of critical importance in determining progression or regression of HPV-associated intra-epithelial neoplasms in the cervix: the risk of developing cervical neoplasia is increased by up to 14-fold after systemic immunosuppression (Porreco et al., 1975). Changes in local populations of immunocompetent cells have been documented in the cervix in HPV infection and in cervical intra-epithelial neoplasia (CIN). Several workers havc described depletion of Langerhans cells in CIN (Tay et al., 1987), and there have been reports of reduced numbers of intra-epithelial lymphocytes in low-grade CIN lesions, with local infiltration of lymphocytes in high-grade lesions (Viac et al., 1990). The significance of these morphological observa- tions remains unclear, however, as do the local molecular and cellular interactions which underlie them.

Epithelial cells may play an important role in modulating local immunological events (Nickoloff et al., 1990), and may undergo profound changes in their phenotype and functional activities after neoplastic transformation (McMichael and Bodmer, 1992). Immunohistological studies of invasive cervi- cal SCC have demonstrated aberrant expression by the neoplas- tic epithelial cells of major-histocompatibility-complex (MHC) molecules, cell-surface glycoproteins which mediate essential functional interactions between cells during an immune re- sponse. Total or allelic loss of class-I-MHC molecules was documented in 30% of 67 cases of cervical SCC (Connor and Stern, 1990), and induction of MHC-class-I1 antigens was seen in 83% of 53 cases (Glew et al., 1992), suggesting a potential role for cervical keratinocytes in invasive SCC as antigen- presenting cells (APCs). Less is known about the significance

of MHC molecules in pre-invasive cervical squamous neopla- sia, in which the expression and/or modulation of keratinocyte MHC molecules may contribute significantly to neoplastic progression or regression. We have investigated the in vivo expression of HLA-DR, the principal class-I1 MHC restriction determinant for antigen presentation (Ottenhoff et al., 1985), by keratinocytes in the normal cervix, and in low- and high- grade cervical squamous intra-epithelial neoplastic lesions. In addition, we have explored the mechanisms behind HLA-DR expression in cervical squamous neoplasia by examining the in vitro behaviour of normal cervical keratinocytes and of well- established cervical keratinocyte cell lines representing model systems for low- and high-grade squamous intra-epithelial lesions of the cervix.

MATERIAL AND METHODS Clinical specimens

Cervical punch biopsies were taken from patients with abnormal colposcopic appearances and/or recent evidence of abnormal cervical cytology. There was no clinical evidence of cervical inflammation, nor of infection with micro-organisms other than HPV. Patients undergoing laparoscopic steriliza- tion were used as controls: these women showed no evidence of on-going or previous cervical disease. All biopsies were immediately mounted in OCT compound (BDH, Poole, UK) and snap-frozen in liquid nitrogen, before being stored for up to 6 months in liquid nitrogen prior to use. Serial 6-pm frozen sections were cut, and routine haematoxylin and eosin (H&E) staining was performed on every eighth level. On the basis of the H&E appearance, the biopsies were classified as histologi- cally normal (n = 12), or as representing low-grade (n = 14) or high-grade (n = 16) squamous intra-epithelial lesions. The morphological changes in the low-grade lesions were equiva- lent to those of koilocytosis and/or CIN1, and in the high- grade lesions to CIN2 and CIN3. There was no difference in patient age, smoking history or contraceptive use between the three groups studied. After morphological evaluation the remainder of the tissue cut from each section underwent immunohistological examination.

Immunohistology HLA-DR expression was detected using 2 primary mono-

clonal antibodies (MAbs) against non-polymorphic determi- nants, L234 (mouse IgGz,, Becton-Dickinson, Mountain View, CA) and DK22 (mouse IgGza, Dako, High Wycombe, UK).

'To whom correspondence and reprint requests should be ad- dressed.

Abbreviations: HPV, human papillomavirus; CIN, cervical intra- epithelial neoplasia; MHC, major histocompatibility complex; SCC, squamous-cell carcinoma; APC, antigen-presenting cell; H&E, haema- toxylin and eosin; MAb, monoclonal antibody; GMEM, Glasgow modified Eagle's medium; SFM, serum-free medium; rIFNy, recombi- nant interferon gamma; rTNFa, recombinant tumor necrosis factor alpha; ICAM-I intercellular adhesion molecule 1.

Received: May 8,1993 and in revised form July 20,1993.

HLA-DR EXPRESSION IN CERVICAL NEOPLASIA 315

Each antibody was used at a dilution of 1:200, and each showed identical staining patterns. T lymphocytes were stained using the anti-CD3 MAb T3-4B5 (mouse IgG1, Dako) at a dilution of 1 : lOO. Immunoperoxidase staining was performed using stan- dard avidin-biotin protocols, and peroxidase activity was visual- ized using diaminobenzidine (DAB).

We devised an arbitrary grading scale in order to obtain a semi-quantitative representation of HLA-DR staining in the biopsies. A total score was awarded which represented the sum of the intensity and the extent of staining for each biopsy. The 2 parameters were scored as follows. Intensity: 0, no staining; 1, weak staining; 2, moderate staining; 3, intense staining. Distribution: 0, patchy basal positivity; 1, diffuse basal positiv- ity; 2, full-thickness positivity.

Lymphocytes were counted within the squamous epithelium of the biopsies and in the 200 pm of stroma immediately subjacent to the basement membrane. An eyepiece graticule was used in conjunction with a high-power ( ~ 4 0 ) objective lens, and a minimum of 10 fields was assessed per case. The mean number of intra-epithelial and stromal T lymphocytes was determined for each biopsy, and from these values the number of cells per pm2 of epithelium or stroma was calcu- lated. Differences between groups of biopsies were assessed statistically using the Wilcoxon rank-sum test. p values, where given, are 2-sided.

Cell culture Cultures of normal cervical epithelium were established

from hysterectomy specimens in which there was no cytological or histological evidence of cervical neoplasia, using protocols previously described (Stanley and Greenfield, 1991).

The cervical keratinocyte cell line W12 was derived directly from a cervical wart, and contains approximately 100 copies per cell of HPV16 DNA, which is present predominantly in the episomal form (Stanley et al., 1989). W12 cells are non- tumorigenic when transplanted into a nude mouse, and are able to re-form an epithelium which resembles low-grade cervical disease (CIN1). As such, W12 constitutes a unique in vitro reagent, representing an early stage in HPV16-related cervical disease.

SiHa (Fried1 et al., 1970) and CaSki (Patillo et al., 1977) are established tumorigenic cervical keratinocyte cell lines, de- rived from SCCs of the cervix. Each contains integrated HPV16 DNA, with 1 and 300 to 500 copies per cell respec- tively.

FACScan analysis of keratinocytes We used FACScan analysis to assess the expression of

HLA-DR on the surface of cervical keratinocytes, together with the effects of pre-treatment with varying doses (10 to 500 units/ml) of exogenous human recombinant interferon gamma (rIFN-y, Genzyme, Maidstone, UK). Prior to analysis all cells were weaned onto serum-free keratinocyte growth medium, containing 5 ngiml epidermal growth factor and 35 to 50 pgiml bovine pituitary extract (GIBCO Paisley, UK). The primary antibody used was the anti-HLA-DR MAb L234 (mouse IgG,), diluted 1:lOO in PBS containing 1.0% (wiv) BSA. Negative-control cell preparations were made using an isotype-matched irrelevant primary MAb (anti-neurofila- ments, mouseIgG,, SeraLab, Crawley Down, UK), and also by omitting the primary layer. Five thousand cells per sample were evaluated using a Becton-Dickinson FACScan cytom- eter.

Organotypic “raft culture The differential expression of HLA-DR by the various cell

types was further investigated using organotypic cell culture. The “raft” technique allows cells growing on a collagen “gel”

to differentiate at an air:liquid interface, enabling them to produce stratified epithelia closely resembling the natural lesions from which the cells originated, and for which they represent in vitro models (Rader et al., 1990). Gels were produced according to the manufacturer’s instructions (Colla- gen Corporation, Palo Alto, CA), and when the keratinocytes had reached confluence, the collagen was raised onto a metal grid to allow differentiation, according to protocols previously described (Stanley and Greenfield, 1991). rIFNy was added at varying doses (50 to 300 units/ml) to the medium for the final 24 to 72 hr of culture. After elevation for 10 days the collagen “raft” was removed from the grid, and representative strips were cut, mounted in OCT compound and snap-frozen in liquid nitrogen; 6-pm frozen sections were cut and immunohis- tochemical analysis of HLA-DR expression was performed as described for the tissue biopsies.

RESULTS Expression of HLA-DR in cervical biopsies

In 12 biopsies of normal ectocervix the keratinocytes showed no positivity for HLA-DR. Constitutive expression by endothe- lial cells in the sub-epithelial stroma was noted, however, as was surface positivity of Langerhans cells and occasional stromal and intra-epithelial lymphocytes (Fig. la) .

Two of fourteen low-grade cervical lesions showed some patchy staining of keratinocytes for HLA-DR. This was moder- ate in intensity and generally basal in distribution, although some superficial koilocytes also showed surface expression (Fig. 1b).

Eight of sixteen high-grade lesions showed positive expres- sion of HLA-DR, often with staining of the full thickness of the epithelium. Intensity was variable, but in some biopsies very strong staining was observed, generally with a cytoplasmic distribution (Fig. 1c). Using a semi-quantitative grading proto- col the expression of HLA-DR in the high-grade group was found to be significantly greater than that in the normal group, although there was no significant difference between the high- and low-grade groups or between the normal and low-grade groups (Table I).

The expression of HLA-DR in the high-grade group was further investigated by counting the number of T lymphocytes, as defined by surface expression of the CD3 molecule. The number of lymphocytes in the immediately sub-epithelial stroma of high-grade lesions was significantly greater in HLA-DR-positive than in HLA-DR-negative cases (Table 11). HLA-DR-positive high-grade lesions also contained more intra-epithelial lymphocytes than their HLA-DR-negative coun- terparts, although the difference was not significant (Table 11).

Expression of HLA-DR by cell lines Five separate experiments were performed using normal

ectocervical cells from 5 subjects, and the W12, CaSki and SiHa cell lines; representative results are presented in Figure 2. There was “background” binding of the isotype-matched control antibody to all keratinocytes (shaded area), and this could not be blocked despite pre-incubation with 20% normal rabbit serum. Fluorescence profiles of unstimulated cells overlapped those produced by the control antibody in all cases (illustrated for NCx and W12 in Fig. 2). This suggested that none of the cells expressed HLA-DR constitutively, although it could not be discounted from these experiments that low-level constitutive expression of HLA-DR by the unstimulated cells was masked by “background” fluorescence. All cells could be induced to express HLA-DR (producing fluorescence levels greater than that of controls) after pre-incubation with rIFNy (non-shaded area) for 48 hr, although there was no demon- strable induction after 24 hr. Each cell type demonstrated a

316 COLEMAN AND STANLEY

FIGURE 1 - Representative staining patterns of tissue biopsies stained for HLA-DR. (a) Normal ectocervix. Constitutive expression by Langerhans cells and stromal endothelial cells, with negative keratinocytes. (b) Low-grade lesion. Focal, moderately strong surface staining of basal epithelial cells and koilocytes. (c) High-grade lesion. Intense cytoplasmic staining of the neoplastic epithelium. Bar: upper panel, 80 pm; lower panel, 40 pm.

TABLE I - SEMI-QUANTITATIVE REPRESENTATION OF HLA-DR STAINING IN TISSUE BIOPSIES

Biopsy n n + Mean Range

Normal 12 0 0

High-grade 16 8 2.072 0-5 Low-grade 14 2 0.29’ 0-2

An arbitrary grading scale was employed (for details, see “Material and Methods”). n, number of cases; n+, number of positive cases; mean, mean score (derived from all cases assessed); range, range of scores. ‘Low-grade vs. normal, not significant.- 2High-grade vs. normal, p < 0.05; high-grade vs. low-grade, not significant (Wilcoxon rank-sum test;p values are 2-sided.)

similar dose-response relationship to rIFNy stimulation, with maximal induction of HLA-DR requiring approximately 300Uiml of rIFNy (Fig. 2).

HLA-DR expression in organotypic tissue culture The phenotype of the epithelia produced by NCx, W12 and

CaSki cells in “raft” culture resembled normal cervix (Fig. 3a), low-grade disease (Fig. 3c) and high-grade disease (Fig. 3e) respectively. In no case was HLA-DR expressed by the stratifying cells. Expression of HLA-DR could, however, be induced on all cells by rIFNy at 300U/ml, but to induce strong expression necessitated an exposure period of 48 to 60 hr, by which time the proliferative capacity of the keratinocytes had been inhibited to such an extent that the phenotype of the

TABLE II - CD3r I.YMPHOCYTE COUN’I‘S IN TIIF El’ITHELlUS1 A N D SUB-EPI‘lHtLI A1 STRO.MA OF HLA-DR-POSITIVE AND -NEGATIVE

I IlGH-GRADE L.ESIOSS

DR- (cells/mmz) (n = 8) ~ ~ ~ s ’ ; m = m ~ ] Significance

Eoithelium 327 415 N/S (153-419) (282-603)

Subeoithelial 1467 2211 B < 0.05 strbma (687-1 893) (1 140-3 192)

Results are expressed as numbers of cells per mm2 of sectional area. Mean values are given, with the range of scores in parenthe- sis. Results (DR+ vs. DR-) were compared using the Wilcoxon rank-sum test; p values are 2-sided. DR-, high-grade lesions negative for HLA-DR; DR+, high-grade lesions positive for HLA-DR; n, number of cases.

epithelium that they produced had changed dramatically. In most cases only a monolayer of HLA-DR-positive cells re- mained (Fig. 3b, d, f).

DISCUSSION

Class-I1 MHC molecules are polymorphic cell-surface glyco- proteins which serve as restriction elements for the recognition of antigenic peptides by CD4-positive T lymphocytes bearing specific receptor molecules (Scheherazade and Germain, 1992). Expression of class-I1 MHC molecules by neoplastic keratino- cytes was seen in 83% of cases of invasive SCC of the cervix

HLA-DR EXPRESSION IN CERVICAL NEOPLASIA 317

K3813\FLI\Rw C *I CaSki

FIGURE 2 - FACScan analysis of surface HLA-DR expression on the 4 cell types examined. Ordinate, number of cells; abscissa, fluorescence intensity (logarithmic scale). Control (shaded area), isotype-matched control antibody; no IFN, unstimulated keratino- cytes; 300 U/ml IFN, keratinocytes stimulated for 48 hr with rIFNy at 300 U/ml. The results are representative of 5 separate experiments. None of the cells shows constitutive expression of HLA-DR, and fluorescence profiles of unstimulated cells overlap those produced by the isotype-matched control antibody (illustrated for NCx and W12). However, all cells can be induced to express HLA-DR after pre-incubation for 48 hr with rIFNy. Each cell shows a similar response to increasing doses of rIFNy, with maximal levels of HLA-DR expression obtained by 300 U/ml.

(Glew et al., 1992), an observation which suggests a possible functional role for such molecules in immune responses to cervical carcinoma. Our results suggest that, whereas HPV infection without neoplasia (as represented by low-grade squamous intra-epithelial lesions) is usually not sufficient for induction of HLA-DR on cervical keratinocytes in vivo, HPV- associated neoplastic transformation, as in high-grade squa- mous intra-epithelial lesions, often is. It is interesting that MHC class-I1 expression in cervical SCC showed no correla- tion with the presence of HPV-type-16 DNA, as detected by Southern hybridization (Glew et al., 1992). The critical deter- minant of HLA-DR expression by cervical keratinocytes in vivo would appear to be neoplastic transformation rather than HPV infection alone. It is not possible, however, from such morphological studies to determine the mechanisms leading to HLA-DR induction, nor to comment on the functional signifi- cance of the expression of such a molecule.

Surface expression of HLA-DR has been documented in a number of types of non-lymphoid neoplastic cell (Zaloudik et at., 1988; Zuk and Walker, 1987), and a variety of mechanisms may lead to the induction of the molecule. Pro-inflammatory

cytokines such as IFNy and TNFa are able to induce HLA-DR expression on a variety of human cell types (Nickoloff et al., 1990), and the release of such immunomodulatory cytokines by host immune cells in the immediate vicinity of a neoplasm has been suggested to contribute to HLA-DR expression in carcinoma of the breast (Zuk and Walker, 1987). There is also evidence of constitutive expression of HLA-DR by neoplasti- cally transformed cells, such as melanoma cells (Real et al., 1985). This has been postulated to be related to direct induction by the neoplastic process (Albino et al., 1986), or to neoplastic “fixation” of a population of cells which are nor- mally MHC-class-11-positive (Houghton et al., 1982).

In order to address the mechanisms involved in HLA-DR expression on cervical keratinocytes, we investigated in vitro models of cervical intra-epithelial neoplasia. No constitutive HLA-DR expression could be demonstrated on any of the cell lines investigated, either in monolayer or in organotypic tissue culture. In both systems, however, HLA-DR could be induced by the pro-inflammatory cytokine IFNy, which is produced in vivo by activated T lymphocytes. These data suggest that the presence of HLA-DR on cervical keratinocytes in high-grade

318 COLEMAN AND STANLEY

FIGURE 3 - Expression of HLA-DR by cervical keratinocytes in organotypic tissue culture, with or without the addition of rIFNy. (u, b) NCx; (c, d) W12; (e, f ) CaSki; (u, c, e) no rIFNy; (b, d, f ) 300 Uiml rIFNy, 60 hr. Bar, 40 ym.

squamous intra-epithelial lesions is more likely to be a result of induced expression rather than a constitutive effect of neoplas- tic transformationperse. Such a hypothesis is supported by our in vivo evidence of a significantly greater number of T lymphocytes in the HLA-DR-positive cases of high-grade intra-epithelial neoplasia than in the HLA-DR-negative high- grade cases.

Our findings arc at slight variance with those of Glew et al. (1992) in invasive SCC of the cervix. Using an immunohisto- chemical approach, these authors were unable to demonstrate an association between MHC-class-I1 positivity of neoplastic cells and the numbers of tumour-infiltrating lymphocytes. They postulated that class-I1 positivity in cervical SCC may be due at least partly to transformation of the normally MHC-class- 11-positive metaplastic cells at the transformation zone of the cervix (Glew et uZ., 1992), although they performed no in vitro experiments to test their hypothesis. While our findings of lack of HLA-DR expression by the 2 transformed cell lines CaSki and SiHa d o not negate the possibility of “constitutive” expression in some cases of cervical neoplasia, they do suggest that induction of MHC-class-I1 molecules such as HLA-DR may also be of importance, a t least in squamous intra-

epithelial lesions of the cervix. Induction of HLA-DR may still occur in invasive SCC, despite the lack of correlation with numbers of tumour-infiltrating lymphocytes. One potential mechanism may involve direct production of cytokines, such as TNFa, by a population of the malignant cells, with consequent autocrine or paracrine induction of MHC-class-I1 molecules on secreting or neighbouring neoplastic cells. Such a possibility is supported by evidence of TNFa production by malignant epithelial cells in vivo (Naylor et uf., 1990).

It was consistently observed that induction of HLA-DR on cervical keratinocytes by rIFNy in vitro is achieved with difficulty, requiring at least 48 hr of cytokine treatment in organotypic tissue culture, by which time a profound reduction in keratinocyte proliferative activity has occurred. The situa- tion is different from that seen with ICAM-1, an adhesion molecule critical in lymphocyte interactions with target cells (Springcr, 1990). ICAM-1 is constitutively expressed on SiHa and CaSki cells, and can be up-regulated within 24 hr by rIFNy on all types of cervical keratinocyte studied, without any impairment of viability or proliferative capacity in vitro. We have observed ICAM-1 expression on HLA-DR-negative cervi- cal keratinocytes in 5/ 16 cases of high-grade cervical intra-

HLA-DR EXPRESSION IN CERVICAL NEOPLASIA 319

epithelial lesions, and in 7/14 cases of low-grade cervical intra-epithelial lesions (Coleman et al., 1993). Whether this differential expression and in vitro inducibility of immunocom- petent molecules on cervical keratinocytes exert any significant effects in the local cervical micro-environment is unclear, as is the functional consequence of HLA-DR expression by the keratinocytes. Whereas some workers have shown that HLA- DR-positive keratinocytes can present antigen to lymphocytes and induce specific responsiveness (Cunningham and Noble, 1989), others have demonstrated that HLA-DR-positive kera- tinocytes can induce antigen-specific unresponsiveness in T lymphocytes (Gaspari a n d Katz, 1991). This apparent di- chotomy is not easily resolved, and the association of keratino- cyte HLA-DR positivity with lymphocytic infiltration in the

REFER

ALDINO, A.P., HOUGHTON, A.N., EISINGER, M., LEE, J.S., KANroR, R.R., OLIFF, A.I. and OLD, L.J., Class-11-histocompatibility-antigen expression in human melanocytes transformed by Harvey murine sarcoma virus (Ha-MSV) and Kirsten MSV retroviruses. J. exp. Med.,

COLEMAN, N., GREENFIELD, I.M., HARE, J., KRUGER-GRAY, H., CHAIN, B.M. and STANLEY, M.A., Characterization and functional analysis of the expression of intercellular adhesion molecule-1 in human papillomavirus-related disease of cervical keratinocytes. Am J. Pathol., 143,355-367 (1993). CONNOR, M.E. and STERN, P.L., Loss of MHC-class-I expression in cervical carcinomas. Int. J. Cancer, 46,1029-1034 (1990). CUNNINGHAM, A.L. and NOBLE, J.R., Role of keratinocytes in human recurrent herpetic lesions. Ability to present herpes simplex virus antigen and act as targets for T-lymphocyte cytotoxicity in vitro. J. Clin. Invest., 83,490-496 (1989). FRIEDL, F., KIMURA, I.T.O. and ITO, Y., Studies on a new cell line (SiHa) derived from carcinoma of uterus. I. Its establishment and morphology. Proc. Soc. exp. Biol. Med., 135,543-545 (1970). GASPARI, A.A. and KATZ, S.I., Induction of in vzvo hyporesponsiveness to contact allergens by hapten-modified Ia+ keratinocytes. J. Immu- nol., 147,4155-4161 (1991). CLEW, S.S., DUGGAN-KEEN, M., CABRERA, T. and STERN, P.L., HLA-class-11-antigen expression in human papillomavirus-associated cervical cancer. Cancer Rex, 52,4009-4016 (1992). HOUGHTON, A.N., EISINGER, M., ALBINO, A.P., CAIRNCROSS, J.G. and OLD, L.J., Surface antigens of melanocytes and melanomas. Markers of melanocyte differentiation and melanoma subsets. J. exp. Med., 156, 1755-1766 (1982). MCMICHA~L, A.J. and BODMER, W.F., A new look at tumour immunol- ogy, CSHL Press, London (1992). NAYLOR, M.S., MALIK, S.T.A., JOBLING, T., STAMP, G. and BALKWILL, F.R., Demonstration of tumor necrosis factor in human ovarian cancer by in situ hybridisation. Europ. J. Cancer, 26,1027-1030 (1990). NICKOLOFF, B.J., GRIFFITHS, C.E. and BARKER, J.N., The role of adhesion molecules, chemotactic factors, and cytokines in inflamma- tory and neoplastic skin disease-1990 update. J. invest. Dermatol., 94,

OTTENHOFF, T.H., ELFERINK, D.G., HERMANS, J. and DE VRIES, R.R., HLA-class-I1 restriction repertoire of antigen-specific T cells. I. The main restriction determinants for antigen presentation are associated with HLA-D/DR and not with DP and DQ. Hum. Zmmunol., 13, 105-116 (1985).

164,1710-1722 (1986).

151s-158s (1990).

ectocervix does not necessarily represent evidence of an effective host immune response in high-grade cervical intra- epithelial neoplasia.

ACKNOWLEDGEMENTS

We gratefully acknowledge financial support from the Impe- rial Cancer Research Fund, and expert technical assistance from Mrs. B. Wilson. Clinical specimens were kindly provided by Mr. M.J. Hare, Hinchingbrooke Hospital, Huntingdon, Cambs. UK. We thank Dr. G. Higgins for critically reading the manuscript.

ENCES

PATILLO, R.A., HUSSA, R.O., STORY, M.T., RUCKERT, A.C.F., SHALABY, M.R. and MATTINGLY, R.F., Tumor antigen and human chorionic gonadotrophin in CaSki cells; a new epidermoid-cancer cell line. Science, 196,1456-1458 (1977). PORRECO, R., PENN, I., DROEGGEMULLER, W., GREER, B. and MA- KOWSKI, M., Gynaecologic malignancies in immunosuppressed organ homograft recipients. Obstet. Gynecol., 45,359-364 (1975). RADER, J.S., GOLUB, T.R., HUDSON, J.B., PATEL, D., BEDELL, M.A. and LAIMINS, L.A., In vitro differentiation of epithelial cells from cervical neoplasias resembles in wivo lesions. Oncogene, 5, 571-576 (1990). REAL, F.X., HOUGHTON, A.N., ALBINO, A.P., CORDON, C.C., MELA- MED, M.R., OETTGEN, H.F. and OLD, L.J., Surface antigens of melanomas and melanocytes defined by mouse monoclonal antibodies: specificity analysis and comparison of antigen expression in cultured cells and tissues. CancerRes., 45,44014411 (1985). SCHEHERAZADE, S.-N. and GERMAIN, R.N., How MHC-class-I1 mol- ecules work: peptide-dependent completion of protein folding. Immu- nol. Today, 13,4346 (1992). SPRINGER, T.A., Adhesion receptors of the immune system. Nature (Lond.), 346,425-434 (1990). STANLEY, M.A., BROWNE, H.M., APPLEBY, M. and MINSON, A.C., Properties of a non-tumorigenic human cervical keratinocyte cell line. Int. J. Cancer, 43,672-676 (1989). STANLEY, M.A. and GREENFIELD, I.M., Culture of human cervical epithelial cells. In: R.1. Freshney (ed.), Culture of epithelial cells, pp. 135-158, Wiley-Liss, New York (1991). TAY, S.K., JENKINS, D., MADDOX, P., CAMPION, M. and SINGER, A., Subpopulations of Langerhans’ cells in cervical neoplasia. Brit. 1. Obstet. Gynaecol., 94,lO-15 (1987). VIAC, J., GUERIN-REVERCHON, I., CHARDONNET, Y. and BREMOND, A., Langerhans cells and epithelial cell modifications in cervical intra- epithelial neoplasia: correlation with human papillomavirus infection. Zmmunobiology, 180,328-338 (1990). ZALOUDIK, J., MOORE, M., GHOSH, A.K., MECHL, Z. and REJTHAR, A., DNA content and MHC-class-I1 antigen expression in malignant melanoma: clinical course. J. clin. Pathol., 41,1078-1084 (1988). ZUK, J.A. and WALKER, R.A., Immunohistochemical analysis of HLA antigens and mononuclear infiltrates of benign and malignant breast. J. Pathol., 152,275-285 (1987). ZUR HAUSEN, H., Human papillomaviruses in the pathogenesis of anogenital cancer. Virology, 184,9-13 (1991).