Pancreatic trypsinogen and cathepsin B in human pancreatic carcinomas and associated metastatic lesions.

Expression of pancreatic trypsinogen and cathepsin B in 23 surgically resected pancreatic ductal adenocarcinomas was evaluated immunohistochemically, using a monoclonal antibody against human pancreatic trypsinogen and a polyclonal antibody against human cathepsin B. Fifteen of 20 invasive tubular adenocarcinomas (75%) expressed pancreatic trypsinogen in a coarse granular pattern located in the supranuclear cytoplasm of the carcinoma cells. In addition, metastatic lesions, including those in peripancreatic lymph nodes and neural plexuses, expressed pancreatic trypsinogen. In contrast, three intraductal (non-invasive) papillary adenocarcinomas did not express pancreatic trypsinogen. Cathepsin B expression was recognised in 14 of 20 invasive tubular adenocarcinomas (70%) in a fine granular pattern located diffusely in the cytoplasm of the carcinoma cells, while none of the three intraductal papillary adenocarcinomas had detectable cathepsin B. These findings suggest that pancreatic invasive ductal adenocarcinomas express pancreatic trypsinogen and cathepsin B immunoreactive peptides, raising the possibility that pancreatic trypsinogen and cathepsin B may act independently of each other in the process of carcinoma invasion and metastasis, like other different classes of proteases involved in the proteolytic modification of the matrix barrier.

Epidermal growth factor (EGF) is perhaps one of the best- characterised cytokines that has been studied in the context of cell growth.EGF mediates its response through interac- tion with specific cell-surface receptors which are expressed as both low-and high-affinity forms, the latter purportedly being formed from the non-covalent association of two or more receptor monomers; activation of the EGF receptor complex stimulates tyrosine kinase activity, resulting in signal transduction (Carpenter & Cohen, 1990).
Overexpression of the EGF receptor is considered by some to be the hallmark of human squamous cell carcinomas (Ozanne et al., 1986) and, in certain tumours, correlates with poor clinical prognosis (Gullick et al., 1991).In tumours of the head and neck, EGF receptor expression has been ex- amined predominantly using immunocytochemical techniques (Partridge et al., 1988; Sakai et al., 1989; Karsley et al., 1990;  Shirasuna et al., 1991) and by Southern and Northern blot hybridisation (Yamamoto et al., 1986; Eisbruch et al., 1987;  Ishitoya et al., 1989; Ebrahim El-Zayat et al., 1991; Saranath  et al., 1992).A consensus of this work indicates that EGF receptor overexpression is not always attributable to gene amplification and/or mRNA overexpression and may not necessarily correlate to poor clinical prognosis.The biological significance and the factors that control EGF receptor ex- pression, therefore, remain an enigma.
The discovery that many tumours which express EGF receptors also produce transforming growth factor a (TGF-a), a potent agonist of the EGF receptor (Massague, 1983), resulted in the attractive autocine hypothesis of malignant cell growth and transformation (Sporn & Todaro, 1980).Indeed, there is compelling evidence that TGF-a is involved in oncogenesis because the protein is not only secreted by a variety of experimental and naturally occurring tumours and cell lines (Derynck et al., 1987; Anzano et al., 1989; Imanishi  et al., 1989) but also induces epithelial hyperplasia in keratinocytes transfected with the TGF-a gene and in trans- genic mice overexpressing TGF-a in the basal epidermal layer (Finzi et al., 1988; Sandgren et al., 1990).Complete malignant transformation, however, requires a second defect in the autocrine loop (Cross & Dexter, 1991).The nature of such a second defect is unknown but it seems likely to involve the signal transduction pathway, possibly by way of overexpres- sion of the EGF receptor.It is not clear whether tumour cells that overproduce TGF-a concurrently express elevated numbers of EGF receptors.
There is now good evidence that the development and progression of epithelial malignancy is associated with the abrogation of normal growth control mechanisms.We have shown, for example, that carcinogen-treated (Game et al.,  1990) and Ha-ras-transfected (Game et al., 1992) keratino- cytes are independent of EGF-induced growth regulation.These findings would appear to be contradictory to a growthstimulatory role of TGF-a in tumour development and have been explained by others in terms of the down-regulation of EGF receptors by endogenous TGF-a (Ciardiello et al.,  1989).It is now essential not only to extend the observations made in different experimental systems to a situation closer to human malignancy but, also, to correlate the different parameters in the same tumour cell lines.
The purpose of the present study, therefore, was to investi- gate the expression of EGF cell-surface receptors, TGF-a autocrine production and the response to exogenous EGF in cell lines derived from untreated human oral squamous cell carcinomas.

Materials and methods
Tissues/cultured cells Untreated primary human oral squamous cell carcinomas (Table I), normal gingival mucosa from the third molar region (n = 10) and normal buccal mucosa (n = 4) were divided and either fixed in neutral formalin (prior to being routinely processed to paraffin wax), snap frozen in liquid nitrogen or used to establish cultured cells; cultures of nor- mal oral keratinocytes were derived from gingivectomy specimens (Prime et al., 1990; Parkinson & Yeudall, 1991).
Biopsy material used to establish cell lines was soaked briefly in absolute alcohol (2-3 s) and then washed (x 2) in Dulbecco's modified Eagle medium (DMEM) containing  Parkinson & Yeudall (1991).bT, tongue; BM, buccal mucosa; FOM, floor of mouth; AP, alveolar process.'Untreated primary oral carcinomas were classified clinically according to their site of origin (S), tumour size (T), lymph node (N) and metastatic (M) involvement and tumour pathology (P).Each parameter was 'weighted' numerically and the maximum index of severity was used to express the STNMP clinical stage.The 5-year survival rate is 51.5% for grade 1, 25.3% for grade II, 21.5% for grade III and 8.3% for grade IV (Henk, 1985).dApproximately 1 x 107 cells were transplanted subcutaneously into 4-to 6-week-old, male athymic (nu/nu; Balb/C) mice.Animals were killed following tumour formation or after 6 months.T, tumorigenic; NT, non-tumorigenic.Clones of the parental cell lines showed a similar pattern of tumorigenicity in athymic mice with the exception of H400, where a clonal cell line was tumorigenic.eNA, information not available.
All cultures were incubated in a humidified atmosphere of 5% carbon dioxide/95% air at 37°C and the medium was changed twice weekly.In later culture passages, cell lines were grown in the absence of 3T3 fibroblast support and in DMEM containing 10% (v/v) FCS and free of all antibiotics.
Immunocytochemical staining Details of the monoclonal antibodies reactive with epitopes on the protein portion of the extracellular domain of the EGF receptor are presented in Table II.Immunocyto- chemical staining was performed using a biotin-streptavidin immunoperoxidase technique (StrAviGen; Biogenex) on acetone-fixed spot preparations of cell lines (Deacon et al., 1991) or trypsin-treated formalin-fixed paraffin sections of the primary tumours or normal human oral mucosa.In brief, cell preparations and sections were incubated sequentially with anti-EGF receptor antibody and biotinylated anti- mouse immunoglobulin [BioGenex; 1:100 dilution in 1O mM phosphate-buffered saline (PBS) pH 7.6 containing 1% (v/v) normal human serum, 1 h at room temperature).Sections were washed in dilution buffer between each layer.Reaction products were developed by immersing slides in 3,3'-diaminobenzidine reagent (5 min) and subsequently enhanced by treatment with 0.5% copper sulphate (w/v in saline) for 5 min.Stained cell preparations and sections were lightly counterstained in Meyer's haematoxylin and mounted in Xam.
Negative controls included replacement of the primary layer with non-immune mouse immunoglobulin (1 or 3 fig ml-1), a monoclonal antibody of irrelevant specificity but similar IgG subclass to that of the test monoclonal antibody (MRC OX-6, IgGI, anti-rat I-A; MRC OX-40, IgG2b, anti-rat T cells; 1 or 3 fig ml-'; Serotec) and PBS.Tissues were also stained for keratin (clone LP34, Dako, 1:200, 1 h) in order to clearly define small areas of tumour epithelium and to act as a positive 'tissue' control.
Spot preparations of cell lines stained with EGFR-1 were evaluated by one individual (J.B.M.) without prior knowledge of the results from the radioligand binding studies.Cell counts were performed at a magnification of x 400 and the highest antibody dilution resulting in staining of > 50% and > 5% of cells from each cell line was deter- mined.A minimum of 200 cells were evaluated from each stained preparation and positive cells identified as those exhibiting a clear brown membrane staining pattern.Sections of the primary tumours were examined subjectively without prior knowledge of the origin of each cell line.

Binding and incorporation assays
Previous studies have described methods to quantify EGF receptor number and affinity (Game et al., 1990).
Cell proliferation in response to exogenous EGF was measured using tritiated thymidine incorporation assays (Game et al., 1990), and this was extended to include exogenous TGF-a (1 ng ml-' and 10 ng ml-').Cells were seeded at low density (5 x 104 cells per well) into 24-well 'EGFR-1 shows poor reactivity with formalin-fixed paraffin sections compared with frozen sections.bE30 detects EGF receptors in fresh-frozen and formalin-fixed paraffin-embedded tissue.cOptimum dilution determined by chequerboard titration.dO.1% (w/v in PBS) Difco 1:250 grade trypsin; 30 min at room temperature.eSome sections were only subjected to a 2-hr incubation.
culture plates, and normal keratinocytes were seeded into 60-mm cell culture dishes (1 x 105 cells per dish).After 48-72 h growth in complete medium, the serum concentra- tion was reduced to 1% (v/v) FCS and EGF (0.01-1.0 ng ml-') was added.After 24 h, the ligand and medium were replaced with 0.5 ml of DMEM plus 1% (v/v) FBS and 10 ,Ll of methyl-[3H]thymidine (2 Ci mmol-', 2.5 tiCi per well; Amersham, UK).The cells were incubated in a humidified atmosphere of 5% carbon dioxide/95% air at 37°C for 2 h, washed (x 3) in ice-cold PBS and fixed in 5% (v/v) trich- loracetic acid at 4°C for 10 min followed by three washes with distilled water.The cells were solubilised with 0.5 M sodium hydroxide (1 ml per well) at 37°C for 2 h, and this solution was added to 4 ml scintillant prior to determining the radioactivity in a beta scintillation counter (LKB Rack- beta).
Autocrine production of TGF-.aTGF-o was isolated from conditioned media using ion- exchange chromatography and then quantified by competi- tion binding assays (Donnelly et al., 1993).The degree of inhibition of 50-jLl aliquots was compared with the inhibition values obtained using a range of unlabelled growth factor concentrations (EGF 0.083-0.42nM).TGF-a was distin- guished from EGF in conditioned media by immunoprecipi- tation, Western blotting and chemiluminescence, as previously described (Donnelly et al., 1993).
Neutralisation of TGF-c Serum-free culture medium was collected from 107 H103 cells at 80% confluence over 48 h, filtered through a 0.45-tim filter (Millipore) and added to H103 cells in culture.Cell proliferation was measured as previously described (Game et al.,  1990).In certain experiments, a 10-fold molar excess of a neutralising antibody to TGF-a (Ab-3, Oncogene Science) was added to culture medium to block ligand activity.Similar techniques were used to block the activity of exogenously added TGF-o.

Statistics
Statistical analyses were performed using the Mann-Whitney test with P <0.05 being taken as statistically significant.

Immunocytochemistry
Antibody E30, directed towards the extracellular domain of the EGF receptor, gave intense membranous staining of epithelial cells within normal gingival and buccal mucosa which progressively diminished from the basal toward the most superficial layers (Figure la).When the normal oral mucosa was subjected to a reduced (2 h only) incubation with the primary antibody, the buccal epithelium stained more intensely than that of gingiva.Six of seven primary carcinomas (T45 tissue was not available) demonstrated a consistent and strong membrane- staining pattern for EGF receptors (Figure lb and c).Invading islands of malignant epithelium showed maximal reactivity at the periphery adjacent to the connective tissue and gradual loss of reactivity in areas of squamous differentiation.A similar differentiation-associated loss of staining was also demonstrable in uninvolved overlying mucosal epi- thelium present in five specimens.Subjectively, staining appeared to be of a lower intensity in normal compared with tumour epithelium.A single tumour, the source of the cell line H376, showed poor patchy staining for the EGF recep- tor with the majority of cells being negative or weakly positive (Figure Id).
Seven cell lines (T45 was not examined) demonstrated a membrane pattern of staining.There was a great variation in the proportion of positive cells and no cell line showed 100% cellular reactivity at any of the antibody dilutions tested.
EGF receptor number and affinity The number and affinity of EGF receptors in the normal and SCC-derived human oral keratinocytes is shown in Table III.One of eight malignant lines overexpressed EGF receptors (H413: 1,209,872), while the remainder expressed similar numbers of EGF receptors to normal keratinocytes (malignant cell lines, mean 456,189; normal, 534,938).The pattern of total EGF receptor expression predominantly reflected the number of low-affinity receptors, although the affinity of this receptor type was higher in the malignant cells (mean kD= 3.8 nM) than normal 5 (kD =16.4 nM).Four SCC lines (H103, H314, H400, H413) did not express high-affinity EGF receptors; the remaining SCC lines and normal keratinocytes expressed high-affinity receptors of similar affinity.
Response to exogenous EGF The effect of EGF on tritiated thymidine incorporation in the normal and SCC-derived human oral keratinocytes is shown in Figure 2. EGF stimulated thymidine incorporation in both the normal and malignant keratinocytes in a dose-dependent manner (data not shown).The majority of the malignant cell lines were more sensitive to EGF stimulation than normal cells, particularly using lower EGF concentrations (< 0. lIg ml-').
The response of the normal and malignant keratinocytes to EGF (1.0 ng ml-l) correlated significantly to the total number of EGF receptors: cells that expressed more EGF receptors demonstrated an increased response to exogenous EGF and vice versa (Figure 3; r = 0.77, P <0.03; H413 was excluded from the statistical analysis).The cellular response to exogenous EGF was not examined in the context of the different EGF receptor affinities because only four malignant cell lines expressed high-affinity receptors and the number of low-affinity receptors broadly corresponded to the total number of EGF receptors.
TGF-c production Both normal and malignant keratinocytes produced TGF-a and not EGF, as demonstrated by Western blot analysis to anti-TGF-a and anti-EGF antibodies (data not shown).In general, cell lines of SCC origin produced more TGF-a (mean: 40.1 pg 10-6 cells 48 h-') than normal controls (15.2 pg 10-6 cells 48 h-') (Figure 4).
In general, cell lines producing more TGF-o expressed fewer EGF receptors (and vice versa), but there was no statistical correlation between these parameters.There was no relationship between the autocrine production of TGF-x and the response of the cell lines to exogenous EGF.Further, TGF-a production was unrelated to both the clinical grade of the original tumour and the tumorigenicity of the cultured cells in athymic mice.

Neutralisation of TGF-c
The addition of exogenous EGF (1 ng ml-') and TGF-a (10 ng ml-') for 24 h stimulated H103 cells by some 10% and 20% respectively, and this effect was partially blocked by the addition of a neutralising antibody to TGF-a; the inclusion of this antibody in conditioned medium failed to decrease [H]thymidine incorporation below control levels (Figure Sa).When this experiment was repeated over a 48-h period to examine the effect of endogenous TGF-a on actively growing   Figure 3 Correlation between the response of the normal and carcinoma-derived human oral keratinocytes to EGF [1.0 ng per ml of culture medium containing 1% (v/v) FCS] and the total number of EGF cell-surface receptors.H413 was excluded from the statistical analysis.cells (Figure Sb), the proliferation of H103 cells progressively decreased in serum-free medium and remained approximately constant in conditioned medium.The addition of anti-TGF-o antibody to both the serum-free and conditioned medium experiments failed to inhibit cell proliferation.

Discussion
This study correlated the expression of EGF cell-surface receptors, the response to exogenous ligand and the autocrine production of TGF-x in normal and carcinoma-derived human oral keratinocytes.
The immunocytochemical pattern of EGF receptor expres- sion in the tissues from which the cell lines were derived in the present study was differentiation related in that there was marked staining in basal epithelial cells and diminished reac- tivity in areas of squamous differentiation; this confirms previous observations (Partridge et al., 1988; Sakai et al.,  1990; Kearsley et al., 1990; Shirasuna et al., 1991).The maximum dilution of antibody staining 50% and 5% of cultured cells in the present study broadly corresponded to the profiles of EGF receptor number as shown in the radio- Figure 4 Autocrine production of TGF-a by the normal and carcinoma-derived human oral keratinocytes.The results reflect the mean of samples assayed in triplicate.
ligand binding studies, findings that are consistent with previous observations (Henzen-Logmans et al., 1992).The data indicate no major inconsistencies between immunoreac- tive and ligand-binding receptor profiles.These findings sug- gest that endogenous TGF-o does not block EGF receptor expression.This proposal is supported by the fact that neither the EGFR-1 nor the E30 monoclonal antibodies com- pete with EGF for the ligand binding site and, also, by the observation in the present study that there was no relation- ship between endogenous TGF-o production and EGF recep- tor expression.Nevertheless, discrepancies between the ligand-binding and antibody dilution data (H103 and H400) were evident in the present study, and it may be that radioimmunoassays would clarify this situation.
A number of studies have shown that overexpression of EGF receptors is common in human squamous cell car- cinomas.In the present study, we examined three cell lines (LICR-LON-HN2, HN5 and -HN6) for which the EGF receptor profile had been reported (Cowley et al., 1986).Our data broadly corresponded to those of Cowley et al. (1986), thereby verifying the EGF receptor data in the present study.The EGF receptor numbers of the normal oral keratinocytes in this study were slightly higher than previously reported (5.5 x 105 vs 2.8 x 105 receptors per cell; Cowley et al., 1986), but we believe this to reflect inter-experimental variation and, possibly, site variation; the normal keratinocytes of the pre- sent study were gingival in origin, whereas those of Cowley et al. (1986) were derived from normal human adult skin.Pre- liminary data in the present study indicate that there may be a site variation concerning the expression of EGF receptor within human oral mucosa.Gingival epithelium, for example, stained less than buccal mucosal epithelium using the ED30 monoclonal antibody, a finding that warrants further investi- gation with ligand binding studies.
The results of the present study indicate that overexpres- sion of EGF receptors may not be an invariable characteristic of human oral squamous cell carcinomas; only one (H413) of eight oral carcinoma cell lines overexpressed EGF receptors.Even if the EGF receptor numbers of the normal keratinocytes in this study were at the level previously documented (2.8 x 105 receptors per cell; Cowley et al., 1986), then still only one (H413) of eight malignant cell lines showed marked elevation of EGF receptor number and four (H103, H157, H357, H400) cell lines demonstrated minimal increases in EGF receptor numbers.The mechanism of EGF receptor overexpression is unknown.Recent evidence sug- gests that human carcinoma cell lines fail to down-regulate EGF receptors (Reiss et al., 1991; Gilligan et al., 1992).
Recent studies indicate that serine phosphorylation within the C-terminal domain of the receptor may be one mechan- ism to regulate EGF receptor activity (Countaway et al., 1992), and Theroux et al. (1992) have shown that mutational removal of this negative regulatory site causes potentiation of Other studies have shown that structural or numerical alterations of chromosome 7 are associated with enhanced expression of EGF receptors (Korc et al., 1986;   Woloschak et al., 1986).
In the present study, exogenous EGF at concentrations of 0.01-1.0ng ml-stimulated the normal and malignant keratinocytes in a dose-dependent manner.The variable re- sponse of the different cell lines to exogenous EGF may I-_ _ __ reflect the heterogeneous nature of the cell populations as indicated by the immunocytochemical data.The present -' '-----F F -- study was carried out using early-passage cultures which, by necessity, had not been cloned.However, it is suggested that high-passage cultures and/or cloning is likely to select for A B C D E F cells expressing a specific EGF receptor phenotype and may Experimental protocol not be a true representation of carcinoma cells in vivo.
Previous studies have shown that, in cell lines showing b overexpression of EGF receptors, cell growth is stimulated by lower concentrations of EGF (1.0 ng ml-') but inhibited using higher EGF concentrations (>1O ng ml') (Cowley et  al., 1984; Kawamoto et al., 1984; Kamata et al., 1986;   Rabiasz et al., 1992).Kawamoto et al. (1984) explained this biphasic phenomenon by the presence of heterogeneous receptors; high-affinity EGF receptors were relevant to growth stimulation, while low-affinity receptors accounted for growth inhibition.In the present study, there was no rela- tionship between EGF receptor affinity and the uptake of tritiated thymidine.What was evident in the cell lines with normal EGF receptor expression, however, was that the total EGF receptor number correlated positively to the response to 1.0 ng ml-' exogenous EGF.H413, which overexpressed 0 h 24 h 48 h EGF receptors, did not exhibit an enhanced response to exogenous EGF.The results suggest that overexpression of EGF receptors may not confer an enhanced sensitivity to Tritiated thymidine incorporation in H103 cells cul- exogenous ligand.The constitutive activation of the receptor h in medium containing 1% (v/v) FCS (A,D) in the (Weiner et al., 1989), however, cannot be excluded as a r presence of exogenous EGF (I ng ml-'; B), TGF- mechanism imparting a selective growth advantage to these ; C) or TGF-a and a 10-fold molar excess of cells.intibody to TGF-ca (D).Cells were also cultured with The criteria to validate autocrine growth factor control *nditioned medium from H103 in the absence (E) or include not only the stimulation of growth by the addition of of the neutralising antibody.The results are ex- ligand and the presence of cell-surface receptors specific for percentage of the control in which the ligand and/or ercentage of the control in which thelithe ligand, but also the secretion of the growth factor ligand omitted.b, Tritiated thymidine incorporation in ' iltured for up to 48 h in 10% FCS (A), serum-free and the inhlbitlon of growth by antlbodies whIch specifically and C) and conditioned medium (D and E).Cells block the biological action of that ligand.In the present Itured in the presence of a I0-fold molar excess of study, the majority of oral carcinoma-derived cell lines pro- .ntibody in serum-free medium (C) and conditioned duced more TGF-o than normal keratinocytes (exception The results are expressed as thymidine incorporation H400), findings which support previous observations in a ta points are the mean of triplicate wells in three variety of tumour cell lines (Anzano et al., 1989; Imanishi et   riments.Bar= standard deviation, where not shown al., 1989).In contrast to previous work in gastrointestinal °.
tumours (Yonemura et al., 1992; Modjtahedi et al., 1992), there was no obvious relationship in the present study between the autocrine production of TGF-oa and either the clinical grade of the original tumour or the tumorigenicity of uction by the EGF receptor.It is currently not the cell lines following transplantation to athymic mice.her a similar mechanism is active in human While there was no statistical correlation between TGF-cx production and EGF receptor expression in the present ect to receptor synthesis, EGF receptor gene study, there was a distinct trend that cell lines that produced in head and neck cancer is highly variable more TGF-cz expressed fewer EGF receptors.Similar findings et al., 1986; Eisbruch et al., 1987; Ishitoya et al.,  have been reported previously (Partridge et al., 1989) and not always associated with enhanced gene ex- suggest a down-regulation of EGF receptors following the ,rahim El-Zayat et al., 1991).Furthermore, a autocrine production of TGF-cz.In the present study has shown that EGF receptor gene amplification exogenous TGF-cz stimulated cell growth of H103, and this f Indian origin with oropharyngeal cancer does effect was partly blocked by the addition of a neutralising with the clinicopathological parameters of the antibody.However, the presence of neutralising antibody in Saranath et al., 1992).In the present study, the conditioned culture medium failed to decrease cell prolifera-57, H314, H357, H400 and T45 showed marked tion to any measurable extent, suggesting that the effects of of the EGF receptor gene, while there was an endogenously secreted ligand were minimal.It may be that ene amplification in H103, H376 and H413 (V.endogenous TGF-a is stimulatory only in conditions of active in preparation).The data indicate that gene cell growth, and our data cannot exclude this possibility, does not reflect cell-surface receptor numbers particularly as cell proliferation of H103 remained approxi- e possibility that there may be a transcriptional mately constant during 48 h in conditioned medium.How- ial defect.We have shown recently that chromo-ever, the effect of biologically active, membrane-bound TGF- )oints are present in close proximity to the EGF a (Brachmann et al., 1989; Wong et al., 1989) may be a on chromosome 7 in H376 and H413 (Patel et significant autocrine growth regulator.The neutralising Lt whether these defects contribute to the expres- activity of the anti-ligand antibody used in this study may receptors in these cell lines is currently un- require the exposure of a particular epitope which is possibly masked in the membrane-bound form of TGF-a.A more efficient way to block growth stimulation by both membranebound and secreted ligand may be the use of an anti-EGF receptor antibody (Modjtahedi et al., 1993).
In conclusion, the results of this study show that over- expression of EGF receptors was not an invariable characteristic of human oral squamous carcinoma-derived cell lines.The overproduction of TGF-a occurred commonly in the carcinoma-derived cell lines, but its role as an autocrine regulator of cell growth in vitro remains questionable.

Figure 1
Figure1The histological appearance of normal and untreated primary squamous cell carcinoma stained with the E30 anti-EGF receptor monoclonal antibody.a, Normal gingival tissue.b and c, Carcinoma tissue from which, b, H103 and, c, H400 were derived showing strong membranous staining and loss of reactivity associated with keratinocyte differentiation.d, Carcinoma tissue from which H376 was derived showing strong reactivity of the overlying epithelium and weakly positive tumour cells in the underlying connective tissue.Bar = 15 tLm.

Figure 2
Figure 2 Tritiated thymidine incorporation in normal and carcinoma-derived human oral keratinocytes following incubation with EGF [0.01-1.0ng per ml of culture medium containing 1% (v/v) FCS].The results are expressed as a percentage of the control in which EGF was omitted.Data points are the mean of quadruplicate wells in two separate experiments.Standard devia- tions were less than 5%.
Figure 5 a, tured for 24 t absence (A) c ca (10 ngmlneutralising a serum-free co presence (F) pressed as a I antibody was H103 cells cu medium (A/B were also cul anti-TGF-a a medium (E).per well.Dal separate expe s.d. was < 5

Table I
Features of primary human oral carcinomas and cell linesa

Table II
Details of mouse monoclonal antibodies to EGF receptor Formalin-paraffinYesd 2 h at room Biogenex (Bio sections temperature' then Diagnostics) overnight at 4'C

Table III EGF
cell-surface receptor expression in normal and malignant human oral keratinocytes aData are the mean of three or more separate experiments; cells were assayed at passage 15. bData are the mean of two samples; cells were assayed at passage 2. CND, not done.Standard error of the kinetic parameters (not shown) was < 10%.