Minichromosome Maintenance Proteins as Biological Markers of Dysplasia and Malignancy¹

Dysplasia is conventionally defined in morphological terms by a loss of normal orientation of epithelial cells, accompanied by alterations in cellular and nuclear size, shape, and staining characteristics (1). Dysplasia is graded according to the degree of these cellular abnormalities and is widely accepted to be an intermediate stage in the progression from normal tissue to neoplasia, as evidenced by the identification of premalignant dysplastic conditions such as CIN.³

One feature that is common to most neoplasms is their rapid proliferation rate compared to normal tissues. The presence of mitoses in increased numbers and in an abnormal distribution are criteria used in identifying dysplastic epithelia, suggesting that there is also an increase in the proliferative rate of these preinvasive lesions. However, attempts to test this hypothesis by defining the cell cycle kinetics in dysplastic human tissues have been limited by the biological markers available.

In this study, we have investigated the proliferative status of a range of human tissues, using antibodies against nuclear proteins involved in regulating DNA replication as novel markers of cellular proliferation. Using monoclonal and polyclonal antibodies against MCM and Cdc 6 proteins, we have investigated the expression of these molecules in a range of normal tissues, hyperplasias, dysplasias, and neoplasms using frozen and formalin-fixed, paraffin-embedded tissues.

The MCM proteins, first discovered in yeast, are a family of closely related proteins with striking sequence homology. Similar classes have been found in Xenopus, murine, and human cells, with significant conservation of gene sequences (2, 3, 4). The MCM proteins form a prereplicative complex by binding to DNA sites at which the origin recognition complex and Cdc 6 proteins have already sequentially bound. This complex acts as a “license” permitting DNA replication and then dissociates irreversibly limiting replication to once per cell cycle (5). The proteins forming the prereplicative complex are considered to be essential for DNA replication in all types of eukaryotic cells examined (6, 7).

In studies using cell lines, our group and others have shown that the level of origin recognition complex proteins remains stable throughout the cell cycle and even in quiescence. However, MCM and Cdc 6 proteins are present only during the cell cycle and are lost from the cell during quiescence and differentiation (8, 9, 10). The expressions of MCM and Cdc 6 proteins can, therefore, be used as specific markers of proliferating cells.

Previous markers used to detect proliferation include PCNA and Ki67, but these have recognized limitations. PCNA is a M_r 36,000 nuclear protein that acts as an auxiliary factor for DNA polymerase δ and is, thus, involved in DNA repair mechanisms as well as replication (11). Staining for PCNA can be affected by a variety of factors, including fixation time (12). Numerous studies have demonstrated a correlation between expression of Ki67 and the growth fraction of proliferating cells (13). However, the function of the Ki67 antigen remains largely unknown, and there is evidence that the molecule is not essential for cell proliferation (14). Moreover, levels of expression of the Ki67 antigen can be altered by external factors, such as nutrient deprivation (15). Ki67 has been shown in a large number of studies to have some use as an indicator of cell proliferation in clinical specimens, although there have been inconsistencies between studies concerning the value of Ki67 as an indicator of prognosis in neoplasia (13). Despite the fact that such discrepancies may at least partly be due to methodological differences, the clinical usage of Ki67 has remained relatively limited.

Here, we report the use of monoclonal and affinity-purified polyclonal antibodies against MCM and Cdc 6 proteins as powerful proliferation markers to characterize the proliferative compartment of normal tissues and provide evidence that the majority of dysplastic and neoplastic cells are in cell cycle. We have extended our group’s previously published findings on the cervix (8) to a range of other tissues in which dysplasia and malignancy commonly occur.

The method for producing polyclonal antibodies against MCM and Cdc 6 proteins was as described previously (7). The antibodies used in this study were raised in rabbits using a standardized immunization protocol and included polyclonal antibodies against Mcm 3, Mcm 4, Mcm 5, Mcm 6, Mcm 7, and Cdc 6. Mouse monoclonal antibody against Mcm 2 (BM28) was purchased from Transduction Laboratories (Lexington, KY). The tissues stained with these antibodies are detailed in Table 1.

Frozen and formalin-fixed, paraffin-embedded human tissues (210 cases total; Table 2) were obtained from diagnostic biopsy or resection specimens from patients at Addenbrooke’s Hospital NHS Trust (Cambridge, United Kingdom). The tissue was used in accordance with ethical guidelines approved by the Local Research Ethics Committee. Samples from fresh resection specimens were snap-frozen in liquid nitrogen and stored at −70°C in cryovials.

Five-μm paraffin-embedded sections were cut onto aminopropyltriethoxysilane-coated slides, dewaxed in xylene, and taken through a series of ethanols to water. The tissues were pressure-cooked in 0.08 m citrate buffer for 2 or 10 min to facilitate antigen retrieval. Following washes in TBS, endogenous peroxidase activity was quenched by incubation in 0.6% hydrogen peroxide in TBS for 30 min. Sections were then washed in TBS and blocked with 10% goat serum (DAKO, Ely, United Kingdom) in TBS for 2 h.

Polyclonal antibodies against Mcm 5 and Cdc 6 were preabsorbed overnight at 4°C with 10% BSA in TBS. The antibodies to the other MCM proteins did not require preabsorption. Final dilutions of primary antibody were made with TBS containing 1% BSA and 0.1% Triton X-100. One hundred μl of antibody were added to each section, and the slides were incubated at 4°C overnight in a humidified chamber. The following concentrations were found to produce optimal staining on the tissues studied: Mcm 2 (BM28), 1:100; Mcm 3, 1:200; Mcm 4, 1:200; Mcm 5, 1:400; Mcm 6, 1:200; Mcm 7, 1:200; Cdc 6, 1:200; PCNA, 1:750; and Ki67, 1:100.

The slides were then washed in TBS containing 0.025% Triton X-100 and incubated for 1 h with biotinylated goat antirabbit secondary antibody (DAKO). A strepavidin-horseradish peroxidase system (DAKO) with the substrate diaminobenzidine was used to develop the stain. The slides were then lightly counterstained with Harris’ hematoxylin, dehydrated in ethanol, and cleared in xylene. Coverslips were applied with DEPEX mounting medium (Gurr, BDH, Poole, Dorset, United Kingdom).

For all tissues studied, appropriate negative controls were performed by omitting the primary antibody and/or substituting with rabbit preimmune serum. Serial sections of each specimen were stained with MCM proteins, PCNA, and Ki67 to provide comparative data.

Five-μm frozen sections were cut onto aminopropyltriethoxysilane-coated slides, fixed in acetone for 10 min, and then soaked in a methanol-0.6% hydrogen peroxide solution for 30 min to quench endogenous peroxidase activity. Sections were washed in TBS and blocked with 10% goat serum in TBS for 2 h. Primary antibodies were applied at the same dilutions as above. The rest of the protocol was identical to that for paraffin-embedded tissues.

A semiquantitative indication of the extent of staining was obtained by calculating a labeling index for each protein stained. At least 200 nuclei were assessed per case. Results were expressed as a percentage of positively stained nuclei out of the total number of nuclei counted in representative microscopic fields.

The median and range of the labeling indices were calculated, and data were compared using the Wilcoxon rank sum test. The P s determined were two-sided, and P < 0.05 was considered to be significant.

Total cell extracts from frozen specimens of normal and tumor tissues were prepared by boiling in lysis buffer [2% SDS, 2% 2-mercaptoethanol, 5 mm EDTA, 20% glycerol, and 100 mm Tris-HCl (pH 6.8)]. Following protein quantitation assay (Bio-Rad protein quantitation kit; Hemel Hempstead, Herbs, United Kingdom) on the lysates, they were separated by SDS-PAGE on a 10% gel, loading 100 μg of protein per well. The proteins were transferred to a nitrocellulose membrane, which was then blocked overnight with blocking buffer (TBS, 10% milk protein, and 1% Tween 20) at 4°C on an orbital shaker.

The blot was incubated with primary anti-Mcm 5 polyclonal antibody at 1:500 for 90 min and washed in blocking buffer. A secondary goat antirabbit horseradish peroxidase-labeled antibody (DAKO) at dilution of 1:10,000 was added for 60 min, and the reaction was developed using ECL, a chemiluminescent substrate (Amersham Life Science, Little Chalfont, Bucks, United Kingdom).

Five-μm frozen sections were fixed for 5 min in 4% paraformaldehyde (BDH) 0.1% glutaraldehyde (25% stock; Sigma, Poole, Dorset, United Kingdom) in PBS and then quenched with washes of 1 mg/ml sodium borohydride in PBS. The sections were permeabilized for 10 min in 0.2% Triton X-100-0.04% SDS in PBS and blocked for 30 min using 5% BSA (fraction V; Sigma) in the Triton X-100-SDS-PBS mix. All antibodies were diluted in this blocking solution.

The sections were incubated with polyclonal anti-Mcm 5 antibody for 2 h at 37°C. After washes in PBS, secondary antibody (1:100 fluorescein antirabbit antibody; Amersham), with propidium iodide as a DNA counterstain, was added to the sections for 90 min at 37°C. After thorough washing in PBS, the sections were mounted in 1 mg/ml phenylenediamine (Sigma) in 90% glycerol and PBS and viewed on a Bio-Rad 1024 confocal microscope.

Normal and tumor tissue pairs were stained and analyzed under identical conditions. For each section, the nuclear MCM fluorescence signal was quantitated as an indicator of MCM protein levels per cell. These levels were controlled for nuclear DNA content, as assessed by intensity of propidium iodide staining. This image analysis was performed using the NIH Image computer program.

In all normal stratified squamous epithelial tissues examined (ectocervix, skin, larynx, and esophagus), a similar pattern of immunostaining for the MCM proteins was observed (Table 3). Staining was restricted to the nuclei of basal epithelial cells (>50% basal layer staining), together with some cells in the immediate suprabasal layers. The more superficial differentiating cells of each epithelium were negative in all cases (Figs. 1,A, 2,A, and 3 D). Overall, Mcm 5 protein expression was fairly constant, with ∼15% of cells showing nuclear staining in all the stratified squamous epithelia examined.

In the cervix, metaplastic squamous cells only showed staining of the basal layer, with very occasional staining of inflammatory cells in the cervical stroma. In normal skin, epithelial cells in sebaceous glands and the roots of hair bulbs showed nuclear staining, but there was no staining of connective tissue elements or inflammatory cells. In the epidermal hyperplastic condition psoriasis, Mcm 5 protein was expressed in the lower three layers of epidermis, although there was no staining in more superficial layers (Fig. 1 B).

There was nuclear staining of 70% of cells in the lower third of colonic crypts for Mcm 2 and Mcm 5 but <5% expression in the upper third, resulting in ∼50% positive staining of crypt cells overall (Table 4). No staining was seen in the connective tissue or mononuclear cell components of the normal mucosa (Fig. 4 A). In the stomach, nuclei expressing Mcm 2 and Mcm 5 were also seen predominantly in the lower third of gastric glands, with staining of <5% of nuclei in the more superficial portions. Approximately 25% of gastric glandular cells stained overall.

Normal endometrium showed variable immunostaining with Mcm 5, depending on the stage of the menstrual cycle. Twenty-four % of glandular epithelial cells expressed Mcm 5 in secretory-phase endometrium, and 65% expressed Mcm 5 in the proliferative phase (Fig. 5 C). Around 5% of stromal cells in the proliferative sample were positive for Mcm 5.

In normal lung parenchyma, ∼15% of alveolar pneumocytes were positive for Mcm 5, and 10% were positive for Mcm 2. Inflammatory cells and septal blood vessels were negative (Fig. 3 A), although bronchial respiratory epithelium showed parabasal staining for Mcm 5, with ∼25% of nuclei staining overall.

Interestingly, in small epithelial inclusions noted in four cases of normal ovary, >80% of cells were positive for Mcm 5. The surrounding ovarian stroma was negative (Fig. 5 G).

Normal kidney showed ∼2% positive staining in renal tubules but no expression in inflammatory or stromal cells (Fig. 5,B). In prostate, only ∼10–15% of normal glandular epithelial cells, including some myoepithelial cells, were positive for Mcm 2 and Mcm 5 (Fig. 5 A).

Sections of normal transitional epithelium showed staining predominantly of basal cells, ∼20% of which showed nuclear expression of Mcm 2 and Mcm 5 (Table 4). Up to 5% of cells in the superficial two layers, including occasional umbrella cells, were also positive with these markers (Fig. 4 D). In five cases of acute cystitis, a similar pattern of staining was seen, with no increase in the frequency of stained cells.

Reactive lymph nodes expressed Mcm 2 and Mcm 5 proteins in 30–40% of follicular cells and 5% of cells in the paracortex.

With Mcm 2, Mcm 5, and Cdc 6, positively stained nuclei were present above the basal layer in all CIN lesions (Table 3). The percentage of positive nuclei in the superficial five layers of the lesion correlated well with the degree of dysplasia. Around 50% of cells in the most superficial layers of CIN I lesions (Fig. 2,D) were positive for Mcm 5 and Cdc 6, with >90% nuclear staining in superficial cells in CIN III lesions (Fig. 2 B). CIN III lesions showed >90% overall immunostaining for Mcm 2, Mcm 5, Mcm 7, and Cdc 6.

The cutaneous squamous dysplastic lesions examined were solar keratosis and Bowen’s disease (Table 3). In solar keratosis, overall nuclear staining for Mcm 5 was 68%, with positive staining of 40% of cells in the most superficial five layers (Fig. 1,C). In Bowen’s disease, there was 90% nuclear staining for Mcm 5 throughout the epithelium (Fig. 1 D).

Nuclear staining for Mcm 5 was seen in >85% of cells for all grades of colonic dysplasia (Fig. 4,B and Table 4). The median frequencies of positive staining were 88% for mild dysplasias, 94% for moderate dysplasias, and 88% for severe dysplasias.

A case of well differentiated (grade 1) noninvasive TCC showed expression of Mcm 5 in 40% of cells, with staining confined to the lower half of the epithelium (Fig. 4,E and Table 4). In a poorly differentiated (grade 3) noninvasive TCC, 85% of cells were positive with antibodies against Mcm 5, with positive cells seen at all levels of the epithelium.

We stained parallel tissue sections of malignant neoplasms for prereplicative complex proteins, PCNA and Ki67. In almost all cases, the labeling indices for Mcm 2 and Mcm 5 were greater than the labeling indices calculated for PCNA and Ki67 (Figs. 1, G–I and Table 5). Moreover, there was much less variability in the labeling indices for the MCM proteins than for PCNA and Ki67, which showed considerable variation between cases for all neoplasms examined (Table 5).

In all SCCs examined (skin, cervix, lung, and esophagus), there was nuclear staining with anti-Mcm 2 and anti-Mcm 5 antibodies in >70 and 85%, respectively, of malignant cells (skin, Fig. 1, E and F; cervix, Fig. 2, C, E, and F; lung, Fig. 3,B; esophagus, Fig. 3, E and F; and Table 3). In general, moderately and poorly differentiated SCC contained a similar frequency of cells expressing Mcm 2 and Mcm 5, with well-differentiated SCC displaying a lower percentage of positive cells (Table 5). In well-differentiated SCC of the skin, there were unstained foci of differentiating cells adjacent to keratin pearls.

The percentage of positive nuclear staining with Mcm 2, Mcm 5, and Cdc 6 correlated with the degree of differentiation of the tumor, ranging from ∼40% of total cells in grade 1 lesions to 95% of total cells in grade 3 lesions (Table 4). This was also reflected in the staining of the superficial five layers of the tumors, in which there was staining of ∼40% of cells in grade 1 tumors and staining of >80% of cells in grade 2–3 tumors (Fig. 4 F). Further studies using a biochemical method of analysis for detection of MCM proteins in urine have been performed by our group and are detailed elsewhere.⁴

The majority of neoplastic cells were positive for Mcm 5, with a frequency of expression ranging from 70 to 95% in total (Table 4). There was a general increase in the frequency of nuclear staining with less differentiated tumours, with 70% of cells showing expression of Mcm 5 in well-differentiated adenocarcinomas, 85% showing expression in moderately differentiated adenocarcinomas (Fig. 4 C), and ∼90% showing expression in poorly differentiated adenocarcinomas.

Two moderately differentiated endometrial adenocarcinomas showed 59 and 79% positive staining with Mcm 5 (Fig. 5 F). A poorly differentiated adenosquamous carcinoma expressed Mcm 2 and Mcm 5 in >90% of nuclei.

The percentage of positive staining for Mcm 5 in borderline mucinous tumors was ∼70% (Fig. 5,H), and in mucinous cystadenocarcinomas (Fig. 5 I), the percentage was ∼90%. In contrast, benign mucinous cystadenomas showed expression of Mcm 2 in only 17–26% of nuclei and expression of Mcm 5 in only 20–40%. Serous cystadenocarcinomas showed >80% positive staining with Mcm 5.

All six renal cell carcinomas examined were moderately differentiated (grade II). All showed immunostaining of ∼70% of nuclei for Mcm 2 and Mcm 5, with ranges of 60–86% and 60–84%, respectively (Table 4). These cases included clear cell and granular cell variants (Fig. 5 E).

In three cases of prostatic adenocarcinoma, ∼40% of nuclei showed positive staining with Mcm 5 (Fig. 5,D and Table 4). There was a correlation between increased frequency of nuclear staining and tumor grade. Cases with combined Gleason grades of 6, 7, and 8 showed stainings of 31, 38, and 53% of nuclei, respectively.

In six cases of lymphoblastic lymphoma, 70–96% of neoplastic cells were found to express Mcm 2 and Mcm 5 proteins.

Upon immunoblotting, we observed a quantitative difference in the total Mcm 5 content of neoplastic tissues compared to their normal counterparts (Fig. 6). A strong signal was seen for samples of moderately differentiated colonic adenocarcinoma and grade 2 invasive bladder TCC, with no detectable signal for the equivalent normal tissues.

The percentage of nuclei in sections of cervix and colon staining positively by immunofluorescence was similar to that seen by immunoperoxidase (Fig. 7). Using direct measurements of fluorescence intensity as an indicator of Mcm 5 protein levels in individual cells, we observed overlapping ranges for squamous epithelial cells of normal cervix and CIN III (Fig. 8,A). When we controlled these values for nuclear DNA content, as indicated by propidium iodide intensity, we again observed no difference in the amount of Mcm 5 between normal and dysplastic/neoplastic cells from cervix, breast, and colon (Fig. 8 B). This suggests that there is no difference in the levels of bound nuclear Mcm 5 between normal and dysplastic/neoplastic cells expressing the protein.

The MCM and Cdc 6 proteins are important regulators in the process of eukaryotic DNA replication. In vitro studies have shown that these proteins are detectable throughout the proliferative phases of the cell cycle but are lost in differentiation and quiescence (8, 9), making them specific indicators of cell proliferation. As such, antibodies against these proteins are valuable reagents for identification of proliferating cells, and we have already described their value in the detection of proliferating cells in cervical smears (8).

One marked advantage of MCM and Cdc 6 proteins over other proliferation markers is the fact that their presence throughout the cell cycle has been well characterized. The prereplicative complex is essential for the initiation of DNA replication and represents the point of convergence of numerous signaling pathways involved in cell growth. The MCM and Cdc 6 proteins are, therefore, better markers of cells in cycle than other candidates, such as growth factor receptors or signal transduction molecules, the functions of which are inherently redundant. Moreover, antibodies against MCM and Cdc 6 proteins do not detect cells undergoing DNA repair, which is a feature of the existing “proliferation” marker PCNA. The proteins of the prereplicative complex are readily accessible in frozen tissues and cytological preparations (8), such that antigen retrieval methods are not necessary for their detection. The antibodies we have used have the additional benefit that they are able to localize MCM and Cdc 6 proteins in formalin-fixed, paraffin-embedded tissues.

Here, we have extended our analysis of MCM protein expression from the cervix (8) to other tissue systems, including skin, lung, colon, and bladder. We consistently observed that antibodies against MCM proteins identify a greater number of cells in cycle than do antibodies against PCNA and Ki67. Moreover, antibodies against MCM proteins are able to detect proliferating cells in paraffin and frozen tissues, unlike antibodies against PCNA, which only reliably label cells in paraffin tissues and require antigen retrieval for optimal staining.

We were only able to examine a relatively small range of tissues for the expression of Cdc 6 and Mcm 7 proteins, as only limited amounts of rabbit polyclonal antibodies against these molecules were available. Our group is currently preparing monoclonal antibodies against these proteins and other members of the prereplication complex. Nevertheless, the evidence that we have obtained suggests that Cdc 6 and Mcm 7 proteins are as reliable in indicating cell proliferation as Mcm 5 and Mcm 2. Indeed, we have not observed significant differences in the expression patterns of any of the prereplication proteins examined in this study, with all antibodies used producing essentially similar results.

Immunohistochemical detection of MCM protein expression in normal tissues shows that these molecules are restricted to the basal and parabasal compartments of stratified epithelia. Loss of MCM proteins and Cdc 6 occurs in differentiating cells that have lost the ability to proliferate. In all of the stratified squamous epithelia examined, there was ∼15% overall nuclear staining with Mcm 2, Mcm 5, and PCNA, suggesting that there is close regulation of the proliferative components of these normal epithelia. Similar findings regarding normal proliferative compartments have also been made by others in studies of the expression of PCNA (16), Ki67 (17), Mcm 2 (18), and Mcm 7 (19) in a range of tissues.

In our study, normal transitional epithelium of the bladder differed in that occasional surface cells showing nuclear staining for Mcm 2 and Mcm 5 were seen, although these accounted for ≤5% of the total surface urothelial cells. This observation may be related to the fact that transitional epithelium contains relatively few layers of cells. As a result of this, cells may transit through the epithelium in a relatively short period of time and reach the epithelial surface before the loss of MCM proteins associated with cell differentiation can occur.

Our findings in colon and endometrium confirm those made by others with markers such as PCNA (11, 20). In normal colon, expression of Mcm 2 and Mcm 5 occurs in the lower half of colonic crypts, with loss of these proteins from differentiated cells in the upper portion of the crypts. A cyclical change occurs in the endometrium, with the percentage of stained cells varying with the menstrual cycle, resulting in a lower frequency of Mcm 2 and Mcm 5 expression in the secretory phase.

Our observation that epithelial inclusion cysts of the ovary show a high frequency of expression of MCM proteins is interesting. These inclusions are thought to represent invaginations of the ovarian surface epithelium and are usually lined by a single layer of epithelial cells. It has been suggested that the majority of epithelial neoplasms of the ovary are likely to originate from these inclusions, rather than from the ovarian surface epithelium (21). Indeed, ovarian carcinoma antigens are expressed far more frequently in inclusion cyst epithelium than in surface epithelium (22). Our finding of a high frequency of expression of MCM proteins in these inclusions is intriguing and raises the possibility that a high level of proliferative activity may contribute to the development of neoplasia in such cells.

We have observed altered localization of MCM proteins in dysplastic epithelia. All grades of dysplasia show staining for Mcm 2 and Mcm 5 in >90% of basal cells. In addition, there is a strong correlation between the number of nuclei positive for Mcm 2 and Mcm 5 at the surface of dysplastic epithelia and the severity of the dysplasia. Only 40% of surface epithelial cells express MCM proteins in mild dysplasia, compared to >80% in severe dysplasia. This observation applies to most stratified epithelia studied, including those of the skin, bladder, and cervix.

Studies of histological sections inevitably provide only “static” representations of dynamic cellular processes. Nevertheless, as prereplication complex proteins have been shown to be present throughout the cell cycle and lost in quiescence and differentiation, our findings in 47 cases of epithelial dysplasia suggest that >80% of cells in high-grade dysplasias and >90% of basal cells in all grades of dysplasia are in cell cycle at any given time point. These observations suggest that dysplastic cells may be characterized in functional terms as remaining in cell cycle, due to deregulation of normal controls over cell proliferation. Interestingly, it has recently been shown that Mcm 7 interacts with the E6 protein of human papillomavirus type 18 and is also a substrate of the E6-AP/E3 ubiquitin ligase (23). The functional significance of such interactions in dysplastic epithelia remains uncertain. However, it may be speculated that bypass of normal pathways of ubiquitination within a cell by viral or cellular oncogenes may contribute to stabilization of proteins such as the MCMs within dysplastic and neoplastic cells.

It can be hypothesized that detection of prereplication proteins might enable the identification of dysplastic cells in clinical settings. The proteins may aid, for example, in the distinction of dysplastic cells from those showing reactive changes, which may be a very difficult exercise using morphological criteria. Our data in bladder, skin, cervix, and lymph node suggest that prereplication proteins show more restricted expression patterns in reactive/hyperplastic conditions than in dysplasia. Indeed, our group has shown elsewhere that urine may be effectively screened for TCC by biochemical immunoassay for the presence of MCM proteins. Nevertheless, further work is required to rigorously test the potential clinical utility of the prereplication proteins, by examining their expression in a larger range of reactive conditions.

We have also observed a difference in the staining patterns of well-differentiated, moderately differentiated, and poorly differentiated carcinomas, with a reduction in the number of positively staining nuclei in more differentiated tumours (Table 5). This was seen most clearly in well-differentiated SCC of the skin, where foci of unstained cells lacking expression of Mcm 2 and Mcm 5 proteins were identified in differentiated areas adjacent to keratin pearls. This observation is consistent with our findings in normal tissues and cell lines (8, 9) that there is gradual loss of MCM proteins in cells undergoing differentiation. Our results also agree with data from other groups showing an increase in the frequency of PCNA staining in less well-differentiated TCC of the bladder (24). The role of members of the prereplicative complex in the process of differentiation in epithelial cells has not been addressed experimentally, and the functional significance of our observations in carcinomas and normal epithelia currently remains unclear.

Our immunoblotting data show that there is a quantitative difference in the level of Mcm 5 protein between neoplastic and normal tissues. High levels of Mcm 5 were observed in carcinomas of bladder and colon, although there was no detectable protein in the equivalent normal tissues (Fig. 6). Our immunofluorescence data further suggest that the overall increase in the level of MCM proteins in neoplastic tissues is not due to overexpression of protein in individual cells. Instead, the increase can be attributed to a greater number of cells showing normal levels of expression. Frozen sections stained by immunofluorescence showed patterns of staining similar to those observed using immunoperoxidase (Figs. 7 and 8). Demonstration of antibody staining by fluorescence may enable automated detection of MCM and Cdc 6 proteins for future clinical applications.

In conclusion, we have utilized antibodies against proteins involved in regulation of DNA replication to define the proliferative compartment of normal tissues and to show that the majority of dysplastic and neoplastic cells are in cell cycle. Our data argue that dysplastic cells can be defined by functional as well as morphological criteria. The presence of proteins of the prereplicative complex in the nuclei of dysplastic cells is consistent with the notion that these cells remain in cycle, as a result of deregulation of normal controls over cell proliferation. Further work is required to assess the potential use of antibodies against members of the prereplicative complex in clinical practice. Potential applications include the identification and/or diagnosis of dysplastic and neoplastic conditions, assessment of prognosis in tissue sections, and cytological or biochemical analysis of cells shed into body fluids.

Many thanks to Kate Bird and Peter Laskey for technical assistance in this project, Jackie Marr for assistance with antibody production, and Mark Madine and Piotr Romanowski for contributions at the early stages of the project. Thanks also to Sarah Trewick for assistance with the immunoblotting experiment.