At metaphase, the antigen seems to be distributed in a reticulate structure .... mean Ki-67 content in A&T cells is higher than that in G, cells. Ki-67-negative cells ...
Cytometry 12:455-463 (1991)
0 1991 Wiley-Liss, Inc.
Ki-67 Labeling in Postmitotic Cells Defines Different Ki-67 Pathways Within the 2c Compartment' Stanislas du Man~ir;,~ Philippe G ~ i l l a u dEmmanuel ,~ Camus, Daniel Seigneurin, and Gerard Brugal Equipe de Reconnnaissance des Formes et Microscopie Quantitative, TIM3, U.S.R. CNRS B 00690, CERMO Universite Joseph Fourier, 38041 Grenoble, France (S.d.M., E.C., G.B.); Equipe de Cytologie Quantitative, Laboratoire de Genetique, Histologie et Biologie de la reproduction, Faculte de medecine de Grenoble, 38700 la Tronche, France (P.G., D.S.) Received for publication October 5, 1990; accepted January 23, 1991
Simultaneous quantification of DNA and Ki-67 proliferation-associated antigen was performed using fluorescence image cytometry. In the MCF-7 cell line, the Ki-67 antigen content increases during the cell cycle, and its intranuclear distribution pattern varies. Quantitative evolution of Ki-67 content as a function of nuclear area makes it possible to define several pathways followed by cells going through the 2c compartment. 1) In some cells, the amount of Ki-67 antigen remains constant during GI (Ki-67 stable pathway), and a characteristic speckled pattern can be observed. 2) In the larger fraction of cells analyzed, there is a postmitotic decrease in the Ki-67 (Ki-67 decrease pathway) content. In this pathway, labeling is located in the nucleo-
G, phase was originally defined as a gap between mitosis and DNA synthesis (30). At present, some authors hypothesize that events taking place from mitosis to S phase (i.e., G, phase) in the cell cycle occur sequentially in a cause-effect relation (24,371. For others, there are no GI-specific events (9). In support of the first hypothesis, it has been observed that control of postembryonic cell proliferation occurs in the 2c compartment (before S phase) (30). In fact, extracellular factors determine whether a quiescent cell will begin to cycle (enter S phase) and also whether a cycling cell in the 2c compartment will continue to cycle or will revert to quiescence (30,311. The sequence of biochemical events that occurs when a cell moves out of Go has been extensively studied, but, for cycling cells, the sequence of events from mitosis to S phase is not well known. On the other hand, in support of the second hypothesis, G, phase is the most variable in duration of the cell cycle phases in somatic cells (31).Indeed, the GI phase is so
plasm in small nuclei, is located in nucleoli in intermediate-sized nuclei, and is absent from larger nuclei (Go). A progressive increase in Ki-67 content (Ki-67 increase pathway) was observed from intermediate-sized nuclei to S phase nuclei. From these results, we hypothesize that the Ki-67 stable pathway is the G, phase of newly formed cells going directly to S phase in local optimal conditions of growth and that Ki-67 decrease pathway and Ki-67 increase pathway correspond to cells whose progression to S phase is regulated by extracellular factors. Key terms: Cell cycle, cell proliferation, immunocytochemistry, image cytometry, G, phase
short in early embryos that it cannot be observed at all (31). What happens to slowly growing cells with a long G, duration? If i t can be shown that there are specific points in G, phase at which cells are arrested, and if 2c cells are arrested in a particular out-of-cycle state, i.e., Go, this would support the cause-effect hypothesis. If,
'Supported by Association pour la Recherche sur le Cancer, grant NO6106 and la Ligue Nationale contre le Cancer and le comite Iserois du GEFLUG. 'Supported by La Ligue DQpartementalecontre le Cancer de Haute Savoie. 'Supported by La Ligue Departementale contre le Cancer de I'lsere. 4Address reprint request to S. du Manoir, Equipe de Reconnaissance des Formes et Microscopie Quantitative, laboratoire TIM3, U.S.R. CNRS B 00690, CERMO, Universite Joseph Fourier, BP 53X, 38041 Grenoble, France.
456
DU MANOIR ET AL.
on the other hand, it can be shown that the cells can be arrested a t any time during G, phase, this would support the second hypothesis that there is no specific G, event, hence no cause-effect. Proliferation markers may provide some insight into these mechanisms, lending support to one or the other of these hypotheses. It is generally accepted that Ki-67 mouse monoclonal antibody is a proliferating marker that recognizes a nuclear antigen present in human proliferating cells and absent from Go cells (16,171. This monoclonal antibody has been used in both clinical and research laboratories to estimate the growth fraction of human tumors by immunochemistry (2,4,11,13-15,18,22). Ki-67 antigen content during the cell cycle has been studied by many authors (16,17,32). Gerdes et al. (17) showed that this antigen is present in all phases of the cell cycle, and also that stimulated lymphocytes passing from Go to G, lacked the Ki-67 antigen in G,T (transition from Go to G,) and G,A phases (G, with low cellular RNA content) but became Ki-67 positive in G,B phase (G, with high cellular RNA content) of the first cell cycle (16). Furthermore, Sasaki et al. (32) showed that Ki-67 content increases with cell cycle progression, reaching a maximum value in mitosis. When cells exit from mitosis, staining intensity seems to decrease, as has been established by microscopic examinations (5,34,36). The topographic distribution of Ki-67 also varies during the cell cycle. This antigen is localized principally in nucleoli in GI phase and is localized in the nucleoplasm later in the cell cycle (21). In mitosis, a peripheral labeling of all chromosomes is observed (17). At metaphase, the antigen seems to be distributed in a reticulate structure surrounding the chromosomes (36). At telophase, a topographic redistribution seems to occur (5,34,36). Verheijen et al. (35) showed using electron microscopy that Ki-67 antigen is localized predominantly in the nucleolar cortex and in the dense fibrillar components during interphase, but they did not succeed in identifying this antigen biochemically. The questions addressed here are 1) whether there are different Ki-67 patterns in G, phase and 2) whether they reflect cell commitment to continue to cycle. To attempt to answer these questions, we studied Ki-67 patterns in postmitotic cells in culture. The main problem is to order these Ki-67 patterns along G, phase chronologically: Image cytometry can be used to study simultaneously the topographic distribution and the quantitative evolution of Ki-67 labeling, DNA content, and morphological nuclear parameters. These morphological parameters permit the chronological ordering of Ki-67 patterns during G, phase.
MATERIALS AND METHODS Cell Culture MCF-7 cells are a cancerous cell line derived from a human mammary tumor (33). The cells were cultured at 2 x lo4 cells/ml on tissue culture chamber slides
(Lab-Tek 4 chambers) in 50% Dulbecco’s modified Eagle’s medium (DMEM) (Gibco) + 50% Ham’s F-12 medium (Gibco) supplemented with 10% heat-inactivated fetal calf serum (FCS) (Gibco), 2 mM L-glutamine (Flow Laboratories, U.K.), 100 IU1ml penicillin, and 100 pg/ml streptomycin. Cells were grown at 37°C in a humidified, 95% airl5% CO, atmosphere until 7 0 4 0 % confluence was reached. The culture medium was replaced by fresh culture medium every 2 days.
Cytochemical Protocols Ki-67 immunostaining. The cells were rinsed for 10 min in Hank’s balanced salt solution (Pasteur Diagnostics, France), fixed for 20 min in cold acetone at -2O”C, and rinsed in phosphate-buffered saline (PBS). The cells were stained with the Ki-67 monoclonal antibody (Immunotech, France; reference 256) using a n indirect immunof luorescence technique. Briefly, slides were first incubated for 30 min a t room temperature with the Ki-67 antibody diluted 1:20 with PBS, then washed twice with PBS, then incubated for 30 min a t room temperature with a n F(ab’), fragment of goat antimouse IgG conjugated with fluorescein isothiocyanate (FITC) (Immunotech, France; reference 219) diluted 1 : l O O with PBS. Negative controls stained in the absence of primary antibody showed no nonspecific staining. DNA staining. After Ki-67 antigen revelation, the slides were washed twice in PBS and stained in the dark at room temperature for 40 min with a 2 pM bisbenzimidazole (BBI, Hoescht 33342, Serva) solution. Slides were then washed in PBS and mounted with a n antifading medium composed of 90 ml glycerol + 10 ml PBS containing 10 mg 1,4-diazabicyclo[2,2,2]octane (Dabco, Sigma; reference D-2522).
Image Cytometry Image cytometry was performed using the Samba 2005 cell image analyzer system (Alcatel-TITN Co., Grenoble, France) coupled to a silicon intensified target (S.I.T.) camera (LH 4036 Lhesa electronique Co., Cergy-Pontoise, France) and a n epiillumination microscope (Zeiss Axioplan). Measurements were made a t x 800 through the following filter sets: bandpass filter BP 365112 nm, chromatic beam splitter FT 395 nm, longwave pass filter LP 397 nm (Zeiss No. 487901) for BBI images; and BP 485120 nm, FT 510 nm, BP 520560 nm (Zeiss No. 487917) for FITC images. Measurement procedures. Signal correction. The optoelectronic distortions were corrected (for both BBI and FITC images) by using a homogeneity reference (8).The correction method consists of dividing each acquired image, point by point (numeric images have 512 x 512 pixels), by the numeric image of a homogeneous reference. The digital image resulting from this division was multiplied by 255. The homogeneous reference used was a n agarose gel containing a constant amount of DNA per unit area. The gel was put on a microscopic slide and BBI
Ki-67 LABELING IN POSTMITOTIC CELLS
stained. This method has the advantage of taking into account the heterogeneity of the excitation intensity of the light and the heterogeneity of the optics and the camera response to each point of the target. To evaluate the accuracy of the measurements, the DNA content of 3-month-old mice liver cells was measured after BBI staining. The coefficient of variation of BBI integrated fluorescence (see parameters below) for 2c (diploid) liver cells was 4.5%. Image segmentation. Nuclear masks were defined by grey level thresholding on the BBI images. Measurements of DNA content (BBI integrated fluorescence) and labeled Ki-67 antigen content (FITC integrated fluorescence) were made through the BBI segmentation mask. To determine the threshold value between Ki-67 negative and positive cells, 50 cells defined as Ki-67 negative by the operator were analyzed. Discrimination between positive and negative cells was made by thresholding on the intranuclear standard deviation of FITC fluorescence (threshold = maximum value of intranuclear standard deviation of Ki67 integrated fluorescence-negative cells). This procedure has been demonstrated a s the most efficient to discriminate between weakly labeled and negative cells (26). Parameters. Six parameters were calculated for each stain per nucleus: 1)nuclear area (NA), which is to the nuclear area derived from the BBI images (1 pixel = 0.0416 Fm'); 2) integrated fluorescence (IF), which is a measure of the nuclear DNA content (BBI filter combination) or the amount of Ki-67 antigen (FITC filter combination); 3) mean fluorescence (MF), which is a measure of DNA or Ki-67 concentrations (MF = IFINA); and 4) intranuclear standard deviation (SD), 5) intranuclear skewness (SKE), and 6) intranuclear kurtosis (KUR), which are descriptors of the fluorescence intensity histogram for each nucleus image (6). Selection of cells in the different cell cycle phases. Cells in prophase and metaphase (P&M) and in anaphase and telophase (A&T) were visually selected. In prophase nuclei, zones of more intense BBI fluorescence corresponding to condensed DNA make it possible to distinguish prophase cells from G, cells. This was confirmed a posteriori by BBI parameters: mean of BBI-MF and mean of BBI-intranuclear standard deviation of P&M cell subpopulation were significantly higher than those of G, cell subpopulations. Observations of cells under contrast illumination make it possible to distinguish G, cells from anaphase-telophase cells. Postmitotic G, sister cells were selected as 2c DNA content cells and analyzed in pairs (27). Two nuclei were assigned to a pair if and only if 1)they were spatially close to one another; 2) they were equal in size (NA) within a tolerance of 15%;3) they had a n equal BBI integrated fluorescence within a tolerance of 15%; 4)they had a similar chromatin pattern a s evaluated by three independent observers; and 5) they were not surrounded by any candidate nuclei having similar fea-
457
FIG.1. Fluorescence micrographs of MCF-7 double-labeled nuclei BBI fluorescence (A,C) and Ki-67 labeling (FITC) (B,D).Several nuclear Ki-67 patterns can be observed: a in B: Ki-67 labeling is restricted to nucleoli in some nuclei; b in B: Ki-67 labeling is distributed throughout the entire nucleus in the largest nuclei; d in D: a peripheral Ki-67 labeling of chromosomes occurs in mitosis; a speckled labeling can be observed in some small nuclei (c in B). Note pairs of adjacent cells, with similar Ki-67 labeling patterns and nuclear sizes. They are obviously sister cells. A,B x 800. Bar = 5 pm.
tures. Cells in S and G, phases were identified on the basis of their DNA content (GI-S boundary = 2 . 2 5 ~ ; the boundary between the first and second part of S phase = 3c; S-G, boundary = 3.75~).The amount of Ki-67 (FITC) was plotted against DNA content (BBI) (see Fig. 3). A bivariate distribution of the nuclear area (mean value of BBI stained area of the two sister cells) vs. Ki-67 content (FITC integrated fluorescence) of postmitotic sister cells was plotted (see Fig. 4). Typical images of BBI and Ki-67-labeled nuclear were stored on hard disk and used to illustrate the relationship between nuclear area and the amount of Ki-67 (see Fig. 5).
RESULTS Variety of Ki-67 Patterns Figure 1shows MCF-7 nuclei simultaneously labeled with BBI (Fig. lA, C) and with the Ki-67 antibody (Fig.
DU MANOIR ET AL.
458
ioa
0
0
(A)
8a v) d m
60 ru
0 Ll
B
z
G2 P&M
40 20 .
,
2c
,
,
3c
.
.
.
.
4c
DNA content
0 iC
4C
BBI I.F. 141
0
FIG 3 . Bivariate distribution of DNA content (BBI integrated fluorescence) vs. Ki-67 content (FITC integrated fluorescence) of cell cycle phase subpopulations of MCF-7 cells. Subpopulations corresponding to cell cycle phase A&T (anaphase-telophase) GI, first part of S (Sa),second part of S (Sb),G, and P&M (prophase-metaphase)are represented by their confidence ellipses. These ellipses delineate the 95% confidence region of the mean of phase cell subpopulations. Ki-67 content increases from G, to G,, reaching a maximum in P&M. The mean Ki-67 content in A&T cells is higher than that in G, cells.
Ki-67-negative cells shows that they have a DNA content corresponding to 2c ? 0 . 2 5 ~(mean ? standard deviation) (Fig. 2B).
ru
0
2c
4c
BBI I.F. FIG 2. DNA histograms (BBI integrated fluorescence) of Ki-67positive (A) and -negative (B) MCF-7 cells. DNA content of Ki-67negative cells is equal to 2c i 0 . 2 5 ~ (mean -t standard deviation). The percentage of Ki-67 positive cells is 92.9%.
lB, D). Different patterns of Ki-67 labeling can be distinguished. The Ki-67 antigen is localized either exclusively in the nucleoli (a in Fig. 1B) or in both the nucleoli and the nucleoplasm (b in Fig. 1B). A strong perichromosomal staining is present during mitosis (d in Fig. lD), and a speckled labeling is observed in a few small nuclei (c in Fig. 1B). The different Ki-67 labeling intensities are clearly discernible.
Simultaneous Quantification of Ki-67 and DNA in Exponentially Growing MCF-7 Cells The percentage of Ki-67-positive cells in MCF-7 cells was 92.9%. DNA histograms of positive and negative cells are shown in Figure 2A, B. The DNA histogram of
Quantification of Ki-67 During MCF-7 Cell Cycle Phases To measure the Ki-67 antigen content in each cell cycle phase, six groups of MCF-7 cells, corresponding to prophase and metaphase (P&M), anaphase and telophase (A&T), G, (Ki-67-positive cells only), first part of S, second part of S, and G, phases were formed (see Materials and Methods). Figure 3 shows bivariate distribution DNA content vs. Ki-67 content of the cell subpopulations in each phase. Mean 5% confidence ellipses are shown for each subpopulation. From one phase to the subsequent phase in the cell cycle, respective means of Ki-67 content were statistically different (Student’s t test, P < 0.05). The mean of Ki-67 antigen content in the A&T subpopulation is reduced about twofold (2.5) compared with the mean in the P&M subpopulation. The Ki-67 content is approximately halved from the A&T subpopulation to the G, subpopulation. From G, phase to the end of S phase, the Ki-67 content increases slightly (value of first part S phase = 1.25 x G, value, value of second part S phase = 1.87 x G, value). Later, from S to G, phases, the Ki-67 content increases markedly (G, value = 3.3 x G, value). The Ki-67 content reaches a maximum a t the beginning of mitosis (P&M) (P&M value = 5 x G, value). The BBI IF values of cells in the P&M subpopulation are lower than those of G, cells [this underevaluation of DNA content is due to the extreme chromatin condensation
459
Ki-67 LABELING IN POSTMITOTIC CELLS
8 Paired cells Putative pathways
25
30
35
40
45
-< Anaphase Telophase
GI phase
Nuclear Area (pf)
50
&4 Beginning of S phase
FIG 4. Bivariate distribution of nuclear area (mean value of BBI stained area of the two sister cells) vs. Ki-67 content (FITC integrated fluorescence) of MCF-7 pairs of sister cells. Ki-67 values are not randomly distributed. Three types of quantitative evolution with nuclear enlargement are represented by the three lines.
can be identified by DNA measurements. Quantitative evolution of Ki-67 makes it possible to define the different Ki-67 pathways that the cells follow in the 2c compartment (from telophase to the Go state or to S Analysis of MCF-7 Cells by Pairs From Mitosis phase). to the Beginning of S Phase Just after telophase, a decrease in Ki-67 content occurs for the largest fraction of the MCF-7 population From telophase to the end of GI, nuclear size in(Ki-67 decrease pathway). Ki-67 labeling along this creases due to DNA decondensation. Thus, nuclear pathway is only nucleoplasmic in smaller nuclei, is nuarea is a valuable parameter that allows the cells to be ordered by age from anaphase to the end of G, phase. cleolar in intermediate-sized nuclei, and is absent in Figure 4 shows Ki-67 content as a function of the mean the largest nuclei (Go). From some intermediate-sized nuclear area of each pair of cells. Ki-67 values are not nuclei to those with nuclear area corresponding to the randomly distributed, and continuous Ki-67 content beginning of S phase, Ki-67 content begins to increase evolution a s a function of nuclear area can be seen. and the labeling is located in nucleoli (Ki-67 increase Three different types of Ki-67 content evolution with pathway). In about 10% of the cells, after telophase and during nuclear enlargement can be observed: constant, decreasing, and increasing. This content remains con- G, phase, the Ki-67 content remains constant (Ki-67 stant in the first type throughout GI phase (Fig. 4, stable pathway). Nucleoli structures appear in smaller upper curve, bold line); in the second type, the Ki-67 nuclei in the Ki-67 stable pathway and subsequently in content decreases in small nuclei (Fig. 4, lower left the Ki-67 decrease pathway. A characteristic finely dashed curve), in the third type, from intermediate nu- speckled labeling pattern is observed in these cells (see clear area value to nuclear area value corresponding to also Fig. 1B). Both daughter cells follow the same postthe beginning of S phase, the Ki-67 content increases mitotic pathway. (Fig. 4, right dashed curve). At the beginning of S phase, the amount of Ki-67 appears to be identical for DISCUSSION all the cells analyzed. We have presented the results of a simultaneous In Figure 5, the Ki-67-labeled nuclei images are located according to nuclear area and according to Ki-67 quantification of Ki-67 and DNA on MCF-7 cell lines. content. Cells in anaphase and telophase can be recog- We confirm that Ki-67-negative cells have a 2c DNA nized by their morphological characteristics. Cells with content as previously shown (17,21,28). Our results no Ki-67 labeling are Go cells, and cells in S phase confirm that Ki-67 content increases with cell progres-
during metaphase, which reduces the accessibility of the fluorochrome to DNA andlor increases the quenching, as described by Hamada and Fujita (23)].
460
DU MANOIK ET AL.
FIG 5. Bivariate illustration of nuclear area vs. Ki-67 content of Ki-67 patterns. Digital images of Ki-67- and BBI-labeled nuclei were stored and analyzed. Images of Ki-67-labeled nuclei were plotted according to nuclear area and Ki-67 content. Putative Ki-67 pathways followed by cells going through the 2c compartment are presented.
sion through the cell cycle as shown by us (21) and by other authors (28,321.
Ki-67 Patterns An increase in Ki-67 is associated with phase-characteristic topographic distributions of labeling within the nucleus. During S phase, labeling becomes nucleoplasmic. It is located throughout the nucleus in G,
phase. In prophase-metaphase, labeling intensity is maximum and located around chromosomes. A study of Ki-67 labeling in the 2c compartment is of considerable interest in that the sequence of events from mitosis to S phase (when it occurs) is poorly documented. This is unfortunate because the exit from the cell cycle occurs principally in the 2c compartment. In anaphase and telophase, nuclei are intensely and homogeneously labeled. Four other patterns resulting
Ki-67 LABELING IN POSTMITOTIC CELLS
from antigen localization can be distinguished in this compartment: 1) labeling is restricted to the nucleoplasm (in the smallest nuclei corresponding to the early postmitotic cells), 2) in some nuclei the Ki-67 labeling is nucleolar, 3) in others it is finely speckled, and 4) some nuclei are not labeled a t all (Go). From telophase to mid-G,, the chromosomes decondense and nuclear size increases. In telophase, nucleoli are organized from prenucleolar bodies. As a general rule, one can say that immediately after this reorganization, nucleoli are functional, i.e., they may produce preribosomal particles (20). Simultaneously with nucleogenesis, Ki-67 appears in the nucleolus, as has been shown here and elsewhere (21,35).
461
Some authors report that a rapid decrease in Ki-67 content can be observed in young postmitotic cells (5,34,36). The different Ki-67 patterns in the 2c compartment described above are consistent with Van Dierendonck et al.’s (34). Although he proposed a putative sequence of Ki-67 pattern from mitosis to early G, phase, no DNA and Ki-67 measurements were made. Therefore, differences between nuclei of similar area but different Ki-67 content could not be discerned. Thus he was unable to detect the different Ki-67 pathways.
Interpretations and Hypotheses of Ki-67 Pathways From the checkpoint for young postmitotic cells, cells Ki-67 Pathways can follow either the Ki-67 stable pathway or the Ki-67 Nuclear size is a reflection of cell progression from decrease pathway. The Ki-67 stable pathway links mitelophase to S phase (19). Thus the nuclei can be chro- tosis to S phase directly; therefore, i t is G, phase (these nologically ordered according to nuclear area (Figs. 4, cells have a 2c DNA content). 5). Figures 4 and 5 show that there are three different The largest fraction of postmitotic cells in these exKi-67 pathways through this compartment associated ponentially growing cultures follow the Ki-67 decrease with characteristic Ki-67 patterns. In the first, the Ki- pathway. Later, if cells progress toward S phase, the 67 stable pathway, nuclei have a particular speckled Ki-67 content increases (Ki-67 increase pathway). At pattern, and Ki-67 labeling intensity remains constant least two hypotheses can be advanced to explain these from telophase up to S phase. In the second, the Ki-67 observations. Either 1) the cells that follow the Ki-67 decrease pathway, labeling is successively nucleoplas- decrease pathway completely loose the Ki-67 antigen mic, nucleolar only, and then absent from nuclei. In the to enter Go phase (in that case the Ki-67 increase paththird, the Ki-67 increase pathway, some nuclei are of way may be followed by cells leaving the Go state, intermediate size and have nucleolar labeling, and which are stimulated to enter into the cycle) or 2) the their Ki-67 content rises to a level identical to that of cells that follow the Ki-67 decrease pathway in a first the largest nuclei of the Ki-67 stable pathway. Thus, G, subphase then recover this antigen in a second G, just after telophase, there is a checkpoint, after which subphase to proceed toward the S phase (in that case, cells must follow one of two possible cell pathways (ei- the intersection between the Ki-67 decrease and inther Ki-67 stable pathway or Ki-67 decrease pathway) crease pathways may be a checkpoint of the cycle to go through the 2c compartment. where the cells require some extracellular factors to proceed). Previous Studies on Ki-67 Labeling It appears that cycling cells remain constantly Ki-67 in 2c Compartment positive in this compartment. Several studies on estabVan Dierendonck et al. (34) have extensively studied lished exponentially growing cell cultures and of stimKi-67 patterns in MCF7-cells. Our previous work (21) ulated lymphocytes have shown that cycling cells are and Van Dierendonck e t al.’s study are in accordance continuously Ki-67 positive in the 2c compartment [for with pattern evolution in S, G,, and metaphase, de- example, 99.3% of MCF-7 cells are positive in some scribed above, in spite of the observation that some cultures (34, see also 16,2811. Furthermore, no decrease nuclei a t the beginning of S phase are Ki-67 negative. in the percentage of Ki-67-positive MCF-7 cells occurs Indeed, Van Dierendonck used simultaneous bromode- after 24 h incubation with tamoxifen (an antiestrogen oxyuridine (BrdU) and Ki-67 immunoenzymatic detec- compound that arrests cells in the 2c compartment) tion and noted that some BrdU-positive cells are Ki-67 (34). This suggests that there is a slow disappearance of negative (34). This interpretation is in contradiction to antigen in 2c-blocked cells (34). Campana e t al.’s (7) work on lymphocytes, in which Discrepancies Between Ki-67-PositiveFraction double BrdU and Ki-67 labeling was used (but with and Growth Fraction Determined by BrdU immunofluorescent detection). Campana concludes Continuous Labeling that all BrdU-positive cells are also Ki-67 positive. The difference between these results may be due to methBaisch and Gerdes (1)first noted unexpected differodological bias, since Van Dierendonck used double ences between the percentage of Ki-67-positive cells revelation with diaminobenzidine and aminoethylcar- (KF) and that obtained for the growth fraction (GF) as bazole, and low Ki-67 content may be masked by BrdU determined by continuous BrdU labeling in nutritionlabeling and hematoxylin counterstain, which he used. ally deprived cultures (7 days without medium reIn addition, fluorescent revelation is known to be more newal) in the U937 cell line. Van Dierendonck et al. (34) confirmed this result in MCF-7 cell lines for exposensitive.
DU MANOIR ET AL.
462 Ki-67 S t a b l e P a t h w a y G I phase of s h o r t d u r a t i o n
\ G I / S boundary
G r a n s i e n t quiescent compartment i n w h i c h c e l l s are s e n s i t i v e s t o growth tactors
FIG6. Working hypothesis. Relations between Ki-67 pathways and cell cycle events in the 2c Compartment.
nentially growing cells (KF = 99.3%, GF = 85%);for 5% serum MCF-7 cell culture (KF = 98%; GF = 62%), and for tamoxifen-treated MCF-7 cell cultures (KF = 84.3%; GF = 26%). These discrepancies are due mainly to cells being Ki-67 positive in the 2c compartment, at least in Baisch’s experiment and in the tamoxifen blockage experiment.
Discrepancies Between Progression Growth Factor Sensitivity and Growth Fraction Determined by BrdU Continuous Labeling The results reported above may be related to Ferrari et al.’s (12) report, in which unexpected differences between growth fraction and cell sensitivity to progression factors is described. He reported that, in WI-38 human fibroblast culture, 7 days after reaching confluence (GF determined by L3H1thymidine continuous labeling), cells remained sensitive to platelet-poor plasma (PPP), which contains a progression factor. These cells are not in a truly Go state, since they are PPP sensitive, but in a quiescent state and are sensitive to external growth factors. Twelve days after reaching confluence, these cells became PPP insensitive and are thus in a truly Go state. The discrepancies between growth fraction and Ki67-positive fraction on the one hand and growth fraction and progression factors cell sensitivity on the other suggest that low content Ki-67 cells in the 2c compartment are in a transient, quiescent state and can progress in the cell cycle in response to external growth factor. Our working hypothesis (Fig. 6) is that the Ki-67 stable pathway is a G, phase under optimum local conditions of growth. On the other hand, the Ki-67 decrease pathway can be subdivided into 1)mitosis exit (nucleoplasmic labeling); 2) transient quiescent compartment (nucleolar labeling, low intensity), in which cells would be sensitive to extracellular growth or differentiation factors [for example, estrogen for MCF-7 cell (2911; and 3) Go state (Ki-67-negative cells). Finally, we suggest that the Ki-67 increase pathway is
probably the growth factor-independent pre-DNA-synthetic part of G, phase (37). Alternatively, Hotlzer e t al. (25) have demonstrated in cell cultures of normal and cancerous cells that two different types of cell cycles exist: 1) ‘‘quanta1 cell cycle,” in which cells are sensitive to exogeneous differentiation factors, and 2) “proliferative cell cycle,” which only amplifies the numbers of cells. It is possible that the Ki-67 stable pathway we observed corresponds to a “proliferative cell cycle” and that the Ki-67 decrease pathway followed by the Ki-67 increase pathway corresponds to a “quanta1 cell cycle.”
PERSPECTIVES Further investigation using simultaneous Ki-67 detection and RNA in situ hybridization of genes with cell cycle-dependent expression is necessary to provide landmarks in the Ki-67 pathways described here. For example, c-myc, which is not expressed in Go cells and the ras or actine gene (3,101, which is expressed from late G,, may be used to localize Go-G, transition and late G,, respectively. Further studies combining timelapse videomicroscopy (37) to determine cell age from the last mitosis and further Ki-67 labeling analysis will allow us to determine whether cells from the Ki-67 stable pathway have a shorter 2c residence time (cell age) than cells from the Ki-67 decrease pathway. If our working hypothesis is correct, Ki-67-positive nuclei will be shown to be either cycling or in a transient quiescent state, and negative cells will be in Go. The results presented here permit cell progression in the 2c compartment to be visualized, and some of the checkpoints can be clearly identified by using morphological features. ACKNOWLEDGMENTS We are grateful t o Dr. Michele Brugal for documentation assistance, and to Jeff Cooke and Dr. Victoria von Hagen for help with the manuscript, and IMMUNOTECH for providing Ki-67 antibody.
LITERATURE CITED 1. Baisch H, Gerdes J: Simultaneous staining of exponentially grow-
2.
3. 4.
5.
6.
ing versus plateau phase cells with the proliferation-associated antibody Ki-67 and propidium iodine: Analysis by flow cytometry. Cell Tissue Kinet 20:387-391, 1987. Barnard NJ, Hall PA, Lemoine NR, Kadar N: Proliferative index in breast carcinoma determined in situ by Ki-67 immunostaining and its relationship to clinical and pathological variables. J Pathol 152:287-295, 1987. Baserga R: Measuring parameters of growth. In: Cell Growth and Division: A Practical Approach, Baserga R (ed). IRL Press, Oxford, 1989, pp 1-16. Boumeester VK, Blonk DI, Van der Kwast TH: Immunocytochemical staining of proliferating cells in fine needle aspiration smears of primary and metastatic breast tumours. Br J Cancer 57509-511, 1988. Braun N, Papadopoulos T, MQller-Hermelink HK: Cell cycle dependent distribution of the proliferation-associated Ki-67 antigen in human embryonic lung cells. Virchows Archiv B Cell Pathol 56:25-33, 1988. Brugal G: Image analysis of microscopic preparation. Methods Achiev Exp Pathol 1l:l-33, 1984.
Ki-67 LABELING IN POSTMITOTIC CELLS 7. Campana D, Coustan-Smith E, Janossy G: Double and triple staining methods for studying the proliferative activity of human B and T lymphoid cells. J Immunol Method 107:79-88, 1988. 8. Camus E, Santisteban Otegui MS, Monet JD, Brugal G Quantification de la fluorescence en cytologie par video-microfluorometrie. Innov Tech Biol Med 11:96-106, 1990. 9. Cooper S: The continuum model and c-myc synthesis during the division cycle. J Theor Biol 135:393-400, 1988. 10. Denhardt DT, Edwards DR, Parfett CL: Gene expression during the mammalian cell cycle. Biochim Biophys Acta 865:83-125, 1986. 11. DCle W, Dieckmann KP, Loy V: Immunohistological determination of proliferative activity in seminomas. J Clin Pathol41:304307, 1988. 12. Ferrari S, Calabretta B, Battini R, Stephen C, Cosenza SC, Owen TA, Soprano KJ, Baserga R: Expression of c-myc and induction of DNA synthesis by Platelet-poor plasma. Exp Cell Res 174:25-33, 1988. 13. Gerdes J: An immunohistological method for estimating cell growth fractions in rapid histopathological diagnosis during surgery. Int J Cancer 35:169-171, 1985. 14. Gerdes J, Dallenbach F, Lennert K, Lemke H, Stein G: Growth fraction in malignant non-Hodgkin’s lymphomas as determined ~n situ with the monoclonal antibody Ki-67. Hematol Oncol 2: 365-371, 1984. 15. Gerdes J, Lelle RJ, Pickartz H, Heidenreich W, Schwarting R, Kurtsiefer L, Stauch G, Stein H: Growth fractions in breast cancers determined in srtu with monoclonal antibody Ki-67. J Clin Pathol 39:977-980, 1986. 16. Gerdes J, Lemke H, Baisch H. Wacker HH, Schwab U, Stein H: Cell cycle analysis of a cell proliferation-associated human nuclear antigen defined by the monoclonal antibody Ki-67. J Immuno1 133:1710-1715, 1984. 17. Gerdes J , Schwab U, Lemke H. Stein H: Production of a mouse monoclonal antibody reactive with a human nuclear antigen associated with cell proliferation. Int J Cancer 31:13-20, 1983. 18. Giangaspero F, Doglioni C, Rivano MT, Pileri S, Gerdes J, Stein H: Growth fraction in human brain tumors defined by the monoclonal antibody Ki-67. Acta Neuropathol 74:179-182, 1987. 19. Giroud F: Cell nucleus pattern analysis: Geometric and densitometric featuring, automatic cell phase identification. Biol Cell 44:177-188, 1982. 20. Goessens G Nucleolar structure. Int Rev Cytol87:107-156, 1984. 21. Guillaud P, du Manoir S, Seigneurin D: Quantification and topographical description of Ki-67 antibody labeling during the cell cycle of normal fibroblastic (MRC-5) and mammary tumour cell lines (MCF-7).Anal Cell Pathol 125-39, 1989. 22. Hall PA, Richards MA, Gregory WM, d‘Ardenne AJ, Lister TA, Stansfeld AG: The prognostic value of Ki-67 immunostaining in non-Hodgkin’s lymphoma. J Pathol 154:223-235, 1988.
463
23. Hamada S, Fujita S: Problem of size dependence in fluorescence DNA cytometry. Cytometry 10:394-401, 1989. 24. Hartwell LH, Weinert TA: Checkpoints: Controls that ensure the order of cell cycle events. Science 246:629-634, 1989. 25. Holtzer H, Biehl J , Antin P, Tokunaka S, Sasse J , Pacifici M, Holtzer S: Quanta1 and proliferative cell cycles: How lineages generate cell diversity and maintain fidelity. In: Calobin Gene Expression and Hematopoietic Differentiation, Stamatoyannopoulos G, Nienhuis AW (eds). Alan R. Liss, Inc., New York, 1983, pp 213-227. 26. Humbert C, Giroud F, Brugal G: Detection of S cells and evaluation of DNA denaturation protocols by image cytometry of fluorescent BrdUrd labelling. Cytometry 11:481-489, 1990. 27. Killander D, Zetterberg A: Quantitative cytochemical studies on interphase growth. Exp Cell Res 38:272-284, 1965. 28. Landgerg G, Tan EM, Roos G: Flow cytometric multiparameter analysis of proliferative cell nuclear antigenicyclin and Ki-67 antigen: A new view of the cell cycle. Exp Cell Res 187:lll-118, 1990. 29. Leung BS, Potter AH. Mode of estrogen action o n cell proliferation in CAMA-1 Cells: 11. Sensitivity of G1 phase population. J Cell Biochem 34:213-225, 1987. 30. Pardee AB: G, events and regulation of cell proliferation. Science 246:603-608, 1989. 31. Prescott DM: Cell reproduction. Int Rev Cytol 100:93-128, 1987. 32. Sasaki K, Murakami T, Kawasaki M, Takahashi M: The cell cycle associated change of the Ki-67 reactive nuclear antigen expression. J Cell Physiol 133:579-584, 1987. 33. Soule HD, Vasquez J , Long A, Albert S, Brennan M: A human cell line from a pleural effusion derived from a breast carcinoma. J Natl Cancer Inst 51:1409-1416, 1973. 34. Van Dierendonck JH, Keyzer R, Cornelisse JC, Van de Velde CJ: Nuclear distribution of the Ki-67 antigen during the cell cycle: Comparison with growth fraction in human breast cancer cells. Cancer Res 49:2999-3006, 1989. 35. Verheijen R, Kuijpers HJH, Schlingemann RO, Boehmer AL, Van Driel R, Brakenhoff GJ, Ramaekers FCS: Ki-67 detects a nuclear matrix-associated proliferation-related antigen. I. Intracellular localization during interphase. J Cell Sci 92:123-130, 1989. 36. Verheijen R, Kuijpers HJ, Van Driel R, Beck JL, Van Dierendonck JH, Brakenhoff GJ, Ramaekers FC: Ki-67 detects a nuclear matrix-associated proliferation-related antigen. 11. Localization in mitotic cells and association with chromosomes. J Cell Sci 92: 531-540, 1989. 37. Zetterberg A, Larsson 0: Kinetic analysis of regulatory events in GI leading to proliferation or quiescence of Swiss 3T3 cells. Proc Natl Acad Sci USA 825365-5369. 1985.