A total of 55 surgical specimens of oligodendroglioma were retrieved from the files of the Histopathology Department of the Royal Hallamshire Hospital for the period 1985–1997. Ethical approval for the study was obtained from the local research ethics committee. Eight of the cases showed focal areas of astrocytic differentiation. In all, 47 of the cases were first diagnoses and eight were recurrent tumours. All of the cases were reviewed by a neuropathologist for confirmation of diagnosis and histological grade according to the World Health Organisation (WHO) criteria (Reifenberger et al, 2000a, 2000b). In total, 25 cases were graded as WHO grade II and 30 as anaplastic tumours, WHO grade III. The biopsies were derived from 22 female and 33 male patients. World Health Organisation grade II tumours were from patients with a mean age of 37.4 years (s.d. 15.6), while WHO grade III tumours tended to be from older patients, of mean age 44.6 years (s.d. 14.5).

MOLT-4 human leukaemic lymphocytes (ATCC CRL-1582; Rockville, MD, USA) were cultured in CO₂-independent media (Leibovitz (L-15)) supplemented with 50 U ml⁻¹ penicillin G, 50 μg ml⁻¹ streptomycin sulphate and 10% foetal bovine serum (all from Invitrogen, Paisley, UK) and maintained at 37°C. Using a novel method for cell synchronisation (membrane elution), minimally disturbed early G1-phase cells were collected and followed during synchronous growth as previously described (Thornton et al, 2002; Helmstetter et al, 2003). Samples from these synchronous batch cultures were removed at specific cell cycle time points for determination of cell concentrations, cell sizes, and analyses of DNA content and protein. Cell density and size were determined using a Coulter Z2 Particle Count and Size Analyzer (Beckman-Coulter, Hialeah, FL, USA). HL-60 cells were obtained from the European Collection of Cell Cultures (ECACC). Asynchronous HL-60 cells were maintained between 1 and 5 × 10⁵ cells ml⁻¹ at 37°C in a humidified 5% CO₂ atmosphere in RPMI 1640 medium supplemented with 2 mM glutamine and 10% FCS.

pET15b-human (hs)Geminin (a generous gift from Anindya Dutta; Wohlschlegel et al, 2000) was expressed in Escherichia coli strain BL21(De3) and purified by Ni-NTA metal affinity chromatography following the manufacturers’ instructions (Qiagen, Crawley, UK). Recombinant hsGeminin was further purified using an (HPLC) Hi-load Q sepharose 16/10 column in NaPi buffer and eluted with varying concentrations of 1 M NaCl. Four rabbits were injected with 125 μg of purified hsGeminin protein, and received three boost injections over a period of 80 days following a standard protocol (Eurogentech, Seraing, Belgium). The sera were collected and affinity-purified on a CNBr column against 10 mg of recombinant hsGeminin protein, eluted with 0.1 M glycine (pH 2.5) and dialysed into PBS, 1% BSA and 0.1% sodium azide. An equal volume of sterile glycerol was added and affinity-purified polyclonal antibodies G92 and G95 were stored at −20°C. Antibody purification was quality controlled by SDS–PAGE for purity and by ELISA. Specificity of the affinity-purified antibodies was demonstrated by immunoblotting of cell lysates and by quenching all immunohistochemical staining after incubating antibodies diluted at working concentrations with equal or less than molar amount of recombinant hsGeminin protein for 1 h prior to a standard immunohistochemical staining protocol.

For cell cycle analysis of DNA content, cell samples were fixed in 80% ethanol and stained with propidium iodide as previously described (Helmstetter et al, 2003) before flow cytometric analysis. For determination of Ki67 and Geminin content, cells were fixed as described above, and permeabilised with nonionic detergent using a method modified from Gong et al (1995). After centrifugation, cell pellets were resuspended in D-PBS, and either 10 μl of FITC-conjugated Ki67 monoclonal antibody (clone B56) or the matched nonspecific staining control (FITC-conjugated IgG; each from BD PharMingen™, Oxford, UK), or 0.1 μg of affinity-purified rabbit polyclonal anti-Geminin antibody G95, or G95 and recombinant hsGeminin protein (1 : 10 ratio). After 2 h incubation at 4°C, cells were washed with D-PBS containing 1% BSA, concentrated via centrifugation and stained with propidium iodide. Analyses of light-scatter properties and DNA/protein content were performed using a FACSCalibur flow cytometer (BD Biosciences, San Jose, CA, USA). Cell doublets were excluded by gating on a dot plot of the width vs the area of DNA fluorescence intensity (Erlanson and Landberg, 1998). In most samples, 10⁴ cells were examined and data were stored and analysed using CellQuest™ software (BD Biosciences), Cylchred (V.1.0.0.1) and WinMDI (V.2.7).

Whole-cell lysates (2 × 10⁶ cells) were prepared for immunoblotting from asynchronous or synchronous batch cultures using a method modified from Harlow and Lane (1999). Cell pellets were resuspended by vigorous vortexing in sample buffer containing 3% SDS, 100 mM DTT, 60 mM Tris (pH 6.8), 0.01% bromophenol blue and 10% glycerol. Equal loading of MOLT-4 (or HL-60) cell lysates was achieved by using the extract from equivalent cell numbers in each lane. Immunoblots were performed as previously described (Stoeber et al, 2001) using affinity-purified rabbit polyclonal anti-Geminin antibodies, and antibodies obtained from commercial sources against the following antigens: Mcm2 (clone 46) and PCNA (Clone 24; BD Transduction Laboratories™, Lexington, KY, USA), Cyclin A (C-19; Santa Cruz; CA, USA). Total cell extracts from equivalent cell numbers were separated on 10% SDS–PAGE gels (Invitrogen), before transfer to Hybond-P membranes (Amersham, Little Chalfont, UK), incubation with primary and secondary antibodies (HRP-conjugated anti-mouse or anti-rabbit antibodies (DAKO, Glostrup, Denmark)), and visualisation of immune complexes using enhanced chemiluminescence (Amersham).

All available slides from each case were reviewed and a block was selected containing a representative area of oligodendroglioma. Immunohistochemistry was performed using antibodies against Mcm2, Ki67, activated caspase 3 and p21, as shown in Table 1 Table 1

Conditions for immunohistochemistry

Statistical analyses were performed using the Statistical Package for the Social Science (SPSS, V.10.1). Labelling indices between grades were compared using the Mann–Whitney U-test and correlation determined using Pearson's coefficient of correlation.

We raised rabbit polyclonal antisera G92 and G95 against bacterially expressed full-length human Geminin. In immunoblots of total cell lysates from asynchronously proliferating human MOLT-4 leukaemic lymphocytes (Figure 1AFigure 1

(A) Immunoblots of total cell lysates prepared from asynchronously proliferating MOLT-4 and HL-60 cells and rec. human Geminin with anti-Geminin antibody G95. Rabbit polyclonal affinity-purified antibody G95 detected a single protein with a molecular mass of ∼33 kDa in total lysates from MOLT-4 (lane 1) and HL-60 (lane 2) cells, and recognised nanogram quantities of rec. Geminin (lane 3; ∼34 kDa). (B) Bivariate flow cytometry of asynchronous MOLT-4 cells with G95 alone, or G95 plus rec. Geminin, further supports the notion that the antibody is specific for human Geminin. Note that cells in G1 (2C; grey dots) are Geminin-negative, whereas cells in S–G2–M (black dots) are positive for Geminin. Pre-incubation of G95 with rec. Geminin results in loss of cell cycle periodicity and little Geminin expression is detected at any stage of the cell cycle. (C) Schematic of membrane elution. Asynchronously growing MOLT-4 cell cultures are immobilised on surfaces such that gravity coupled with cell division results in release of one daughter cell, while the other remains surface-bound. Newborn early G1-phase cells are continuously released in the effluent and grow synchronously without evidence of disturbance. (D) Immunoblots of origin-licensing factors and control proteins in synchronously proliferating MOLT-4 cells. Protein levels of Mcm2, Geminin, Cyclin A and PCNA were determined in lysates from equivalent numbers of cells isolated from synchronous batch cultures at the indicated times.

Determination of the molecular and temporal events comprising the process of mammalian DNA replication often requires the use of synchronous cell populations. We have examined expression of the standard proliferation marker Ki67, the DNA replication-licensing factor Mcm2 and the repressor of origin-licensing Geminin during the proliferative cell cycle, using a novel, nonchemical method for cell synchronisation (membrane elution; Thornton et al, 2002; Helmstetter et al, 2003; Figure 1C). We have employed membrane elution because use of chemical methods for cell synchronisation may account in part for reported discrepancies in temporal periodicities and subcellular localisation of replication proteins such as Orc1 (Pak et al, 1997; Kreitz et al, 2001; Okuno et al, 2001; Li and DePamphilis, 2002) and Cdc6 (Fujita et al, 1999; Jiang et al, 1999; Mendez and Stillman, 2000; Petersen et al, 2000).

Newborn (early G1-phase) MOLT-4 cells isolated by membrane elution progressed through a full cycle of synchronous growth, as demonstrated by FACS analysis of the DNA content (Figure 1D). In total cell extracts prepared from these synchronous batch cultures, protein levels of the DNA replication-licensing factor Mcm2 increase linearly in early G1 phase up to a maximum in S phase before decreasing during the latter part of the cell cycle (Figure 1D). Moreover, cell synchronisation by membrane elution has revealed a similar cell cycle-dependent periodicity in Mcm5 and Mcm7 expression (Eward et al, manuscript submitted), findings that may have been previously obscured by metabolic perturbation due to the use of chemical synchronisation agents. Previous work has suggested that Geminin is present during S, G2 and early mitosis of the proliferative cell cycle before ubiquitination by the anaphase-promoting complex (McGarry and Kirschner, 1998; Wohlschlegel et al, 2000; Nishitani et al, 2001; Tada et al, 2001). Using our newly generated affinity-purified polyclonal anti-Geminin antibody G95 (Figure 1A), we found little to no expression of this repressor of origin licensing during G1 phase, with detectable levels at the G1/S transition and increasing linearly during the remainder of the cycle before mitotic degradation (Figure 1D). Notably, the Geminin expression profile is similar to Cyclin A, another oscillating cell cycle regulator which, like Geminin, is degraded at the metaphase to anaphase transition via the ubiquitin–proteasome pathway (Figure 1D) (Yam et al, 2002). In contrast to the proteins described above, protein levels of proliferating cell nuclear antigen (PCNA) remained constant during synchronous progression through the cell cycle (Figure 1D). Moreover, the fraction of MOLT-4 cells expressing the standard proliferation marker Ki67 during synchronous growth (85–95%; determined by FACS analysis) remained constant and was similar to the proportion that was Ki67-positive in asynchronous culture (data not shown). Geminin therefore provides a precise estimate of the S-G2-M growth fraction in dynamic cell populations and the ratio of Geminin/Ki67 provides information about the relative length of the G1 phase of the proliferative cell cycle (Figure 1D).

Ki67 and Mcm2 both demonstrated an increased labelling index in WHO grade III compared to grade II tumours (P<0.01), reflecting greater proliferation with increasing anaplasia (Table 2 Table 2

Labelling indices – descriptive statistics

Both Geminin antibodies, G92 and G95, gave strong, predominantly nuclear staining (Figure 2Figure 2

Geminin immunohistochemistry. (A) Low-grade oligodendroglioma showing a single Geminin-positive nucleus (arrow) stained with G95. (B) G92 showing a similar nuclear pattern of staining. (C) Anaplastic oligodendroglioma showing multiple nuclei stained with G95. Diffuse cell staining is observed in mitotic figures (arrows). (D) Anaplastic oligodendroglioma stained with G92.

We sought to determine whether Geminin behaved similarly to the cell cycle inhibitor p21. Although the p21 labelling index showed a trend towards increased labelling with higher grade, this was not significant. Similarly, although p21 showed a correlation with the Mcm2 proliferation marker, this was very weak (r=0.3, P<0.05) (Figure 6Figure 6

Scatter plot showing a poor relationship between p21 and Mcm2 labelling indices.

A comparison of Geminin/Ki67 ratios between low- and high-grade tumours reveals significant differences. The ratio is significantly higher for the grade III tumours (mean 0.39, s.d. 0.2) than grade II tumours (mean 0.25, s.d. 0.17; P=0.005) (Figure 8Figure 8

Box plot showing Geminin to Ki67 ratio for grade II and grade III oligodendrogliomas.