Interactive computer-assisted analysis of chromosome 1 colocalization with nucleoli

0 1994 Wiley-Liss, Inc. Cytometry 16:313-323 (1994)

Interactive Computer-Assisted Analysis of Chromosome 1 Colocalization With Nucleoli

Isabelle Leger,' Martial Guillaud, Bruno Krief, and Gerard Brugal Equipe de Reconnaissance des Formes et Microscopie Quantitative, Universite Joseph Fourier, Grenoble, France

Received for publication June 7, 1993; accepted January 28, 1994

The applications of DNA cloning and fluorescent in situ hybridization (FISH) techniques have strengthened the hypothesis of an ordered chromatin structure in interphase nuclei, strongly suspected to vary with functional state. The nonrandom distribution of the centromeres and their dynamic rearrangement during the cell cycle have been well documented. A close proximity of specific centromeres to nucleoli has also been reported, but the functional meaning of this association is still unknown. In order to investigate whether the chromosome 1 centromere region to nucleolus association depends on the cell cycle and chromosome status, we combined FISH of probes specific for the lq12 region with Ki-67 nucleolar antigen fluorescent immunocytochemical (FICC) detection on the MCF-7 human breast cancer cell line and on the MRC-5 normal fibroblastic cell line. Both FISH and FICC signals were interactively localized in a one-step fluorescent microscopic observation and further analyzed

using the Highly Optimized Microscope Environment (HOME) graphics microscope workstation, which provided computerized interactive marking of lq12 to nucleolus associations (lql2-nu) at the individual nucleus and nucleolus levels. This study confirms that centromeric regions, other than those adjacent to the major ribosomal cistrons, contribute to the perinucleolar chromatin and demon- strate that, during the cell cycle, the heterochromatic band lq12 is dynamically rearranged with regard to both the nuclear volume and the nucleoli. A relationship between the association of the chromosome 1 pericentromeric region with nucleoli and the nucleolar transcriptional activity is also strongly suggested. 0 1994 Wiley-Liss, Inc.

Key terms: Combination of FISH and im- munocytochemistry, functional genome mapping, chromosome 1, nucleolus, interactive cytometry, HOME system

The recent applications of DNA cloning and fluorescent in situ hybridization (FISH) techniques have strengthened the hypothesis of an ordered chromatin structure, strongly suspected to vary with functional state, in the interphase nuclei (2,20-23,27,29,32). Us- ing specific antibodies or nucleic acid probes, several authors have demonstrated the nonrandom distribution and the cell cycle dependent rearrangement of centromeres in interphase nuclei (1,9,13,36,38). The proximity of specific chromosome centromeres and nucleoli has also been reported (3,14,16,23,24). The short distance between the secondary [nucleolus organizer regions (NORs)] and the primary (centromeres) constric- tions in acrocentric chromosomes is most likely responsible for the centromere to nucleolus proximity, which is the case for chromosomes 13,14,15,21, and 22 bearing the ribosomal DNA in humans. However, this

hypothesis does not apply when chromosomes such as 1 and 9 are involved (23,331 and when the number of chromosome to nucleolus associations is higher than the number of NOR-bearing chromosomes (25).

The aim of this work was thus to investigate whether the chromosome 1 centromeric region to nucleolus association depends on the cell cycle and chromosome status. For this purpose, FISH of probes specific for the heterochromatic band q12 of the human chromosome 1 was combined with fluorescent immunocytochemical

'Address reprint requests to Isabelle Leger, M.S., Equipe de Recon- naissance des Formes et Microscopie Quantitative, Universite Joseph Fourier, CERMO-Domaine Universitaire, 38041 Grenoble, France.

LEGEK ET AL. 314

(FICC) detection of the Ki-67 nucleolar antigen in order to analyze their colocalization (lql2-nu) during the cell cycle. The MCF-7 human breast cancer cells, known for chromosome 1 trisomy and aberrations (7,19), were compared to the MRC-5 normal fibroblastic cell line. Both FISH and FICC signals were interactively localized in a one-step microscopic observation and further analyzed using the Highly Optimized Mi- croscope Environment (HOME) graphics microscope workstation (4) whose application to fluorescent microscopy is reported here for the first time. Confocal laser scanning microscopy (CLSM) has also been used to verify the three-dimensional (3D) organization of this lql2-nu association.

MATERIALS AND METHODS Cell Culture

The MCF-7 human breast cancer cell line was main- tained in Dulbecco’s modified Eagle’s medium (DMEM) supplemented with 10% fetal calf serum (FCS), 4 mM L-glutamine, 0.6 mgil insulin, 0.15 pgiml amphotericin B (all from Boehringer Mannheim, Meylan, France), nonessential amino acids ( l%, viv), and 24 IUiml pen- icillin-60 pgiml streptomycin solution (both from Gibco BRL Life Technologies, Cergy Pontoise, France). Cells were grown as monolayers in plastic flasks (25 cm’) at 37°C in a humidified atmosphere containing 5% CO, and were passaged weekly using 0.25% trypsin-0.01% EDTA in Ca2+- and Mg2+-free phosphate buffered sa- line solution (PBS).

Combination of FICC and FISH FICC of Ki-67 antigen. The cells were rinsed for 2

min in PBS, fixed for 5 min in cold acetone at -20°C, and rinsed in PBS. The cells were stained with the Ki-67 monoclonal antibody (Immunotech, Marseille, France) using an indirect FICC technique before the FISH procedure. Briefly, slides were first incubated for 30 rnin at room temperature (RT) with the Ki-67 antibody diluted 1:50 in PBSiTween-20 (T20) (0.1%) and washed 3 times with PBSiT20 (0.1%). Slides were then incubated for 30 min a t RT with a F(ab’)2 fragment of goat anti-mouse IgG conjugated with Texas Red (TR) (Immunotech) diluted 1 3 0 in PBSiT2O (0.1%).

Postfixation. The slides were incubated at RT for 5 rnin in a postfixation buffer (50 mM MgC1, in PBS) and then for 10 rnin in 4% paraformaldehyde diluted in the postfixation buffer. The slides were rinsed in PBS for 5 min.

DNA probes. pUC 1.77 designates a clone of the plasmid vector pUC9 containing a 1.77 kb human Eco R1 fragment as an insert. The insert was isolated from the human satellite DNA fraction IIiIII and represents a tandemly organized repetitive sequence in the ql-2 heterochromatic region of chromosome 1 (6) (probe supplied by T. Cremer, Anthropogenetica Laboratory, Uni- versity of Heidelberg, Germany).

pl-79 designates a clone of the plasmid vector pSp65 containing a 0.9 kb human Bam H1 fragment as an

insert. The insert is specific for a 1p36.3 telomeric sequence embedded in a large Ah-rich chromosome region (5) (probe supplied by D.C. Ward, Department of Human Genetics, Yale University School of Medicine, New Haven, CT).

Probe labeling procedures. The plasmids were nick-translated with either biotin-1 1 dUTP (Enzo Di- agnostics, New York, NY) or digoxigenin-ll dUTP (Boehringer Mannheim) with a nick-translation system kit (Gibco BRL Life Technologies) according to the suppliers’ instructions.

FISH. In situ hybridization was carried out as described by Emmerich et al. (8). Probes were diluted in a hybridization medium containing 60% formamide in 2 x SSC (1 x SSC:sodium salt citrate:150 mM sodium chloride, 15 mM sodium citrate, pH 7.0). The final DNA concentration of probes was 1 kg/ml for pUC 1.77 and 2 pgiml for pl-79. An aliquot of 10 ~1 of the hybridization medium with probe(s) was added to the slides sealed with a coverslip; the nuclear target DNA and the DNA probe were simultaneously denatured at 72°C for 10 min on a heat plate. Hybridization was carried out overnight in a wet chamber a t 37°C. The slides were then washed 5 min in 2 x SSC a t RT, 3 x 5 min in 0.1 x SSC at 60”C, and 5 min in 2 x SSC at RT. The nonspecific immunologic sites were blocked for 10 min with 5% nonfat dry milk (NFDM) in 4 x SSC. The probe was detected either with avidin-fluorescein isothiocyanate (FITC) (Biosys Vector Laboratories, Compiegne, France) diluted 1:400 in 5% NFDM (30 rnin incubation) or with an anti-digoxigenin antibody (Fab fragment) linked to FITC (Boehringer Mannheim) diluted 1:200 in 5% NFDM (30 min incubation).

DNA Counterstaining The slides were washed twice in PBS and mounted

with a solution of 9:l glyceroliPBS containing 1,4-di- azabicyclo L2.2.21 octane (DABCO; Sigma Chimie, St. Quentin Fallavier, France) as antifading compounds and 250 ngiml 4’,6-diamidino-2-phenylindole (DAPI; Sigma Chimie).

Selection of Cells in Cell Cycle Phases Cells in the various cell cycle phases were visually

selected through a TR specific block filter (BP 546 * 12, FT 580, LP 590) (Carl Zeiss, Oberkochen, Germany) according to the characteristics Ki-67 patterns in MCF-7 described by Guillaud et al. (12) and Santiste- ban and Robert-Nicoud (30). Only “typical” nuclei were selected and assigned to the different cell cycle stages.

As nuclear size increase is a reflection of cell pro- gression from telophase to G, phase (lo), the measured areas of the interactively ordered nuclei were taken into account to verify the adequacy of the proposed chronological ordering. The presence of double FISH spots, reflecting the duplication of chomatids of chromosomes 1, could also provide a control to ascertain the proposed S and G, cells to actually be in the post-DNA replication state.

ANALYSIS O F DNA-NIJCLEOLIJS ASSOCIATION 315

Interactive Cytometry Fluorescent samples were observed through a x 100

iris oil immersion objective, NA = 1.30. The filter block used for simultaneous FISH (FITC) and FICC (TR) observation was BP 480-4951570-590, BP 605- 690, BP 520-5401615-645 (Omega Optical, Brattle- boro, VT).

Cytometry was performed on 50 cells from each cell cycle phase using the HOME (AxioHOME, Carl Zeiss) (4). The HOME station was composed of an epi-illumi- nation microscope using a HBO 100 W mercury lamp (Axiophot, Carl Zeiss), a DELL 486 computer con- trolled by a mouse, and a 1 in. high resolution computer display built in the microscope which superimposes a menu bar, drawing messages onto the optical image. Interactive morphometric measurements were performed on fluorescent samples: the nucleolar and nuclear outlines were drawn with the mouse. From these drawings, the basic morphometric parameters were automatically calculated by the HOME system. Among them, only the nuclear and the nucleolar areas were used in further analyses. In addition, a menu de- signed from macros and corresponding icons allowed the differential marking of cells in 4 cell cycle phases characterized by typical Ki-67 patterns (early GI, late G,, mid S, and G,) and the differential typing of the nucleoli according to their number of colocalized FISH spots (0, 1, 2, or 31, subsequently referred to as nu-0, nu-1, nu-2, and nu-3, respectively. Colocalized FISH spots are required to touch a nucleolus. Although the system made it possible to display histograms and scat- tergrams of the data, the files were transferred to Stat View I1 Microsoft (Abacus Concepts, Inc., Berkely, CA) for statistical analyses based on the PLSD Fisher’s test.

CLSM Analysis CLSM 10 (Carl Zeiss) was used to examine the colo-

calizatiun between FISH spots and nucleoli in 3D in 30 nuclei. A 0.5 pm Z scan step size was used for the optical sectioning. The X-Z section level was fixed on the FISH image through a FISH spot and was used for the optical sectioning of both FISH and FICC images. The two sets of sections were then superimposed and coded with false colors (green for FISH and red for FICC).

Mathematical Model Since the microscopic images are 2D representations

of 3D objects, we developed a mathematical model to compute the probability for an observed specific 2D configuration to be random. This random probability was then compared to the observed colocalization frequency to test objectively the hypothesis of a nonrandom vs. random association between FISH spots and nucleoli. The model is given in the Appendix.

Control Experiments Negative controls stained in the absence of either

probe or primary antibody were included in the proce-

dure. No nonspecific labeling was detected on these slides. Parallel double-staining with the anti-Ki-67 antibody and serum from a patient with systemic autoimmune disease containing specific nucleolar antibodies (17) was used in order to ascertain the nucleolar localization of the Ki-67 labeling. This serum was selected since it contains only nucleolar antibodies and no ad- ditional autoantibodies against a variety of nuclear and cytoplasmic antigens (17).

The MRC-5 normal human embryonic fibroblastic cell line was used as a reference diploid cell line. The telomeric pl-79 probe for chromosome 1 was used to control whether the colocalization between this probe and the nucleolus-subsequently referred to as tell-nu association-paralleled the lql2-nu association.

RESULTS Ki-67 Nucleolar Labeling

Double-staining with the anti-Ki-67 antibody and serum from a patient with systemic autoimmune disease containing nucleolar antibodies demonstrated that both nucleoli labelings were perfectly matched as seen in Figure 1A,B. The Ki-67 antigen labeling was selected for this study because it provided brighter and more contrasting signals and less background than the nucleolus specific serum, thus facilitating the localization of probes with regard to the nucleoli through the dual band pass filter (FITCITR).

FISH Signal Analysis The specificity and the efficiency of the FISH were

not reduced by the FICC procedure since the percentage of positive cells for FISH and the heterogeneity of the two cell lines with respect to the FISH spots number were comparable to those obtained on control slides processed for FISH only (data not shown).

MCF-7 cell line. FISH positivity was observed in 98% of cells when the pUC 1.77 probe was used. Most had three FISH spots in agreement with the previously demonstrated trisomy of the MCF-7 cell line for chromosome 1 (7,191. Only the nuclei with three FISH spots were further analyzed.

FISH positivity was observed in 95% of cells when the pl-79 probe was used. Most had two FISH spots in agreement with a previously demonstrated telomeric deletion of one chromosome 1 in the MCF-7 cell line (19). Only the nuclei with two FISH spots were further analyzed.

MRC-5 cell line. FISH with pUC 1.77 and pl-79 resulted in 70% and 60% positive cells, respectively. Most had two FISH spots in agreement with the dip- loidy of this normal reference cell line. Only nucLi with two spots for each probe were further analyzed.

Selection of MCF-7 Cells in Cell Cycle Phases The characteristic Ki-67 patterns made possible non-

ambiguous assignment of cells in 4 cell cycle phases as follows:

ANALYSIS OF DNA-NUCLEOLUS ASSOCIATION 317

Table 1 Distribution of the Cell Population (1,200 nuclei) With Regard to the Number of FISH Colocalized Spots With

Nucleoli Per Nucleus

No. FISH colocalized mots Der nucleus pUC 1.77 probe pl-79 probe

Cell line 0 1 2 3 0 1 2 3 MCF-7 0% 27% 46% 27% 80% 15% 5% MRC-5 51% 27% 22% 49% 40% 11%

1. The selected “early GI” nuclei were small, slightly elongated, and showed numerous small and irregularly shaped nucleoli and a constellation of scat- tered Ki-67 speckles in the nucleoplasm (Fig. 1C).

2. The selected “late GI” nuclei had less numerous small and regularly shaped nucleoli but no longer had speckled labeling (Fig. 1D).

3. The selected “mid S” nuclei were round and had large nucleoli with well-defined contours and increased size (Fig. 1E).

4. The selected “G2” nuclei were large and showed a small number of large and distorted nucleoli with blurred contours (Fig. 1F).

An artifactual morphological distortion of nucleoli was obtained in the MRC-5 normal human embryonic fibroblastic cells used as diploid reference. It has already been noted that this cell line is more fragile with regard to the FISH procedure than the MCF-7 (19). Consequently, the MRC-5 cells were not classified according to cell cycle phases.

Band lq12 and Telomeric Overall Association to Nucleoli

The distributions of the lql2-nu (pUC 1.77 probe) and tell-nu (pl-79 probe) associations among 1,200 cells of the MCF-7 and MRC-5 lines are reported in Table 1.

Using the pUC 1.77 probe, it was observed that one or three chromosome 1 centromeric regions of the MCF-7 cells were associated with the nucleoli at the same frequency (27%) but two FISH spots were colocalized in most of the cells (46%). Therefore, all the MCF-7 nuclei, while only half (49%) of the MRC-5 cell population, had at least one lql2-nu association. The

FIG. 1. A , B Double-staining of MCF-7 cells with the anti-Ki-67 antibody (A) and with serum from a patient with systemic autoimmune disease containing nucleolar antibodies (B). C-F Combination of FISH with the centromeric pUC 1.77 probe (green) and FICC of Ki-67 antigen (red) in MCF-7 nuclei. Typical selected nuclei in “early GI” (C), “late GI” (D), “mid S (E; the arrow indicates a double FISH spot), and “G,” (F). G Confocal optical sectioning of a FISH (pUC 1.77) and FICC (Ki-67 antigen) labeled MCF-7 nucleus. The two sets of sections are superimposed and coded with false colors (green for FISH and red for FICC). The arrow indicates the Z section orientation (top to bottom). H: The HOME computer display built in the microscope superimposes a menu bar, drawing, and messages.

question thus arises whether this discrepancy resulted from the number of chromosomes 1 differing in these cell lines.

Using the pl-79 probe, it was observed that 20% of MCF-7 cells had a t least one tell-nu association while the lql2-nu association was observed in 100% of these same cells. It should be noted that the MCF-7 cells had only two chromosome 1 telomeric regions (one on a chromosome 1 and the other translocated onto another chromosome) (19). Therefore, only one nonrearranged chromosome 1, with respect to the centromeric and the telomeric regions, was present per MCF-7 cell. Whether the low percentage of MCF-7 nuclei with at least one tell-nu association (20%) resulted from the lack of two undamaged chromosomes 1 or from intrin- sic different behavior of centromeric and telomeric regions cannot be inferred from this experiment. How- ever, the observation that the tell-nu association frequency in MCF-7 was about twice less than in MRC-5 cells (20% vs. 51%) can be related to the presence of one undamaged chromosome 1 in MCF-7 vs. two in MRC-5.

CLSM Study of the lql2-nu Association Because in our experiments, FISH and FICC combi-

nation did not tolerate paraformaldehyde fixation, which is recommended for preservation of structural integrity, the acetone fixation compatible with both FICC and FISH was reduced (5 min instead of 20 min) and no proteinase K (normally increasing the DNA target accessibility for FISH) was used. Although the CLSM capabilities were probably not fully exploited because of the fixation conditions, nuclear and nucleolar morphology and structure were well preserved and the optical sections in 30 nuclei confirmed that the lql2-nu associations observed in 2D were actual colocalizations, as is shown in Figure 1G.

Centromere to Nucleolus Association Throughout the MCF-7 Cell Cycle

The HOME microscope workstation made i t possible to analyze the nuclear and nucleolar morphometry as well as lql2-nu associations in 200 cells (50 cells per cycle phase), as is shown in Figure 1H. The overall distribution of FISH colocalized spots per nucleus in this 200 cell subset was identical to that reported above for 1,200 cells.

lql2-nu association at the nuclear level. The increase in nuclear area from early GI to G:, provided a posteriori verification of chronological ordering of cells (Fig. 2A).

As is shown in Figure 2, the mean number of nucleoli per nucleus significantly decreased throughout the cell cycle (P = 0.0001) (Fig. 2B), while the mean number of lql2-nu associations per nucleus significantly increased from early G, to mid S, where it was maximum, and then decreased from mid S to G2 (P = 0.01) (Fig. 2C). Statistical analysis showed that the number of lql2-nu associations per nucleus was related neither

318

5.

4.5 '

4.

3,5 '

3.

2,s .

\leiin huclear area

LEGER ET AL

A

early ( i I late <;I mid S (2 Cell cycle phases

Mean number of nucleoli per nucleus

ps 1

u11y (ii latc G I mid S c2 Cell cvcle phwes

Mean number of nucleoli per nucleus

5,s , D

7 '

1 2 3 Number of lql2-nu

associations per nucleus

FIG. 2. Chromosome 1 centromere to nucleolus association a t the nuclear level in MCF-7 cells. (The error bars represent i-2 S.D.) A: Evolution of the nuclear area throughout the cell cycle. B,C: Cell cycle dependent evolution of the mean number of nucleoli per nucleus

Mean number of lql2-nu per iiucleus C

1 1

early G I late Gl lllld s G.? h

Cell cycle phases

Mean fraction of the nuclear area occupied by nucleoli E

1 1 3

Number of lq12-nu associations per nucleus

(B) and of the mean number of lql2-nu associations per nucleus tC). D,E: The mean number of nucleoli per nucleus (D) and the fraction of the nuclear area (96) occupied by nucleoli (E) in nuclei with 1, 2, and 3 lql2-nu associations.

20.

18.

16.

14.

12.

10.

a. 63


Mean number of FISH colocalized spots per nucleolus A B

I

cnrly Cil latc (il mid S ct! Cell cycle phases

Mean nucleolar area

30 351

C

01 f 1

nu4 nu- 1 nu-2 nu-3 lypes of nucleoli

FIG 3. Chromosome 1 centromere to nucleolus association at the nucleolus level in MCF-7 cells. (The error bars represent + 2 S.D.) A,B: Cell cycle dependent evolution of the mean nucleolar area (A)

to the number of nucleoli per nucleus (P = 0.5) (Fig. 2D) nor to the mean fraction of the nuclear area (%) occupied by nucleoli (P = 0.4) (Fig. 2E).

lql2-nu association at the nucleolar level. As is shown in Figure 3, the mean nucleolar area increased throughout the cell cycle (P = 0.0001) (Fig. 3A). Since the number of nucleoli was decreasing meanwhile, as is shown in Figure 2B, this strongly suggests that inter- nucleolar fusions occurred as already reported (1 1,371. The mean number of lq12 FISH colocalized spots per nucleolus increased significantly (P = 0.0001) (Fig. 3B) during the cell cycle. Interestingly, the nucleoli with the highest number of lq12 FISH colocalized spots were the largest, whether the total cycling cell population is considered (Fig. 3C) or the cell subpopulations in each cell cycle phase (data not shown). Identical results were obtained on the MRC-5 cell population (data not shown).

cnrly GI lnte (;1 mid S (72 Cell cycle phases

and of the mean number of centromeric FISH colocalized spots per nucleolus (B). C: The mean area of the different types of nucleoli (with 0, 1, 2, and 3 centromeric FISH colocalized spots).

As is shown in Figure 4, the different types of nucleoli (nu-0, nu-1, nu-2, and nu-3) had a rather parallel increase in size from early G, to mid S, but the area of the nucleoli associated with a t least one lq12 increased more abruptly from mid S to G2 than the area of nucleoli without any lq12 colocalization (Fig. 4A).

Throughout the cell cycle, the number of the nu-0 and nu-1 nucleoli decreased while the number of nu-2 and nu-3 increased (Fig. 4B). From early G, to late GI, the nu-2 and nu-3 did not only result from the coalescing of two nu-1 or one nu-1 with a nu-2, respectively, since less nu-1 disappeared (seven) than is required to explain the corresponding increased number of nu-2 and nu-3 (nine nu-2 and one nu-1) by coalescence. Therefore, some nu-2, if not all, de novo appeared from one or two FISH spots colocalization with either a pre- existing nu-1 or nu-0, respectively. The increased number of nu-2 from late G, to S, while no nu-1 disap-

320 LEGER

140-

120-

100-

80.

60-

40-

ET AL.

A ~2 Cell qrlr pha\es earlyG1 IateCl mid S

Vurrihcr of nucleoli per 50 nurlei

B o 201zsb earlyG1 lateGI midS G2 Cell qrle p h a w

FIG 4. Cell cycle dependent evolution of the mean area (A) and the number (B) of nucleoli without (01, with one (01, with two (A), or with three ( 0) colocalized FISH spots in MCF-7 cells. This result represents the analysis of the nucleoli from 50 nuclei selected in each cell cycle phase.

peared, supports this observation. From mid S to G,, the coalescing of nu-1 or nu-1 and nu-2 to form the appearing nu-2 and nu-3 cannot account for the decreased number of nu-1, since many more nu-1 disappeared (30) than the number which would be sufficient to generate the observed number of appearing nu-2 and nu-3 ( 3 and 1, respectively). Therefore, about a quarter of nucleoli which were associated with one band lq12 at mid S just lost this association when moving to G,.

A similar evolution of the different types of nucleoli was observed when different sets of 30 nuclei randomly chosen among the 50 selected ones in each cell cycle phase were analyzed. Thus results are likely indepen- dent from sampling.

DISCUSSION The FICC-FISH combined procedure reported on

here offers several advantages. First, the Ki-67 antigen detection, which makes it possible to detect the nucleoli, also provides a suitable cell cycle marker (12,311. Indeed, after the required denaturation step for FISH, DNA is lost (28 ) and the counterstaining stoichiometry

is lost; it is no longer possible to use the DNA content as a cell cycle marker. Second, the technique is suitable for fluorescent microscopic imaging, either conventional or confocal. The HOME system (41, used for the first time in a fluorescent mode, provided computerized interactive marking and editing of associations between the band lq12 and the nucleoli at the individual nucleus and nucleolus levels. This system provides several advantages when compared to conventional image analysis; e.g., 1) accurate graphic tools are available to interactively flag and count relevant objects simply by clicking with a mouse-driven cursor, and 2) a variety of morphometric features are derived from drawing accurate t o 0.1 km (4). Marking and drawing are fast and thus suitable for fluorescent samples which fade. Dou- ble measurements of the same object are avoided since a marker persists on cells already analyzed. Finally, the system can be fitted with a dual band pass filter for simultaneous detection of several fluorescent signals, thus obviating any filter change and subsequent image shifting. The system was used in combination with a FISH-FICC dual-staining technique to investigate the localization of chromosome 1 specific sequences with respect to the nucleoli in a one-step fluorescent microscopic observation.

Several authors have already reported a specific close proximity between the chromosome 1 and nucleoli (23,33) in various cell lines. Stahl et al. (33) described a practically constant association for both ho- mologous segments while Manuelidis and Borden (23) reported two patterns of association in large neurons: either both homologs were closely apposed to a nucleolus or one homolog was on the nuclear membrane and the other abutted a nucleolus. On the contrary to chromosomal regions closely adjacent to NORs, the position of the chromosome 1 pericentromeric region would not be expected to reside near a nucleolus where the ribosome gene transcription, processing, and preribosomal particle formation take place. It has been hypothesized that, as the nucleoli form around the NOR chromatin, they grow in size and overlap any nearby chromosomal region (25). This could be most probable for malignant cells, such as MCF-7, since their nucleolar number and area are high when compared to nontransformed cells (15). However, this hypothesis is in clear contradiction with our results which show that the number of lql2-nu associations per nucleus depends neither on the number of nucleoli nor on the nuclear fractional area occupied by nucleoli. Two other alternative expla- nations were already proposed: either a minor ribosomal cistron site is inserted in this chromosomal region (33) or the chromosome 1 pericentromeric region acts as a general organizing center of the compartmenta- tion of chromatin in interphase nuclei (23).

The question also arises from our experiments whether the systematic cen-1 association observed in MCF-7 cells could be attributed to a dehydration artifact from alcohol fixatives such that the lq12 region could be stuck onto the nucleoli because of some overall


nuclear shrinkage. This artifact is unlikely to account for the observed high frequency of lql2-nu association since 1) the mathematical probability of a random colocalization to lop5) is 2 or 3 decades inferior to the observed frequency (10- 2, in 2D configurations; and 2) optical sectioning confirmed that the associations between FISH spots and nucleoli observed in 2D were actual colocalizations where the lq12 region was either in the perinucleolar region or more often inside the nucleolus. This result confirms that pericentromeric regions other than those adjacent to the major ribosomal cistrons could contribute to the perinucleolar chromatin. Moreover, some nu-2 de novo appeared in the population and some nu-1 were losing their specific association with the lq12 region during the cell cycle thus suggesting that the chromosome 1 pericentromeric region is dynamically rearranged with regard to both the nuclear volume and the nucleoli.

It can be emphasized from our results that in the MCF-7 cell line: 1) the number of lql2-nu associations per nucleus was related to the cell cycle phases and maximum in S phase where the high transcription activity, related to histone synthesis, requires concom- itantly increased rate of rDNA transcription for ribosomal biogenesis; and 2) nucleoli which were associated with the band lq12 enlarged much more from S to G, than those nucleoli which were not associated. From these observations and given that the nucleolar area is related to transcriptional activity (18), a relationship between the association of the lq12 region with nucleoli and the nucleolar transcriptional activity is therefore strongly suggested. This relationship could result from the integration of minor ribosomal cistrons sites, suspected t o exist near the centromere of chromosome 1 (26), in the nucleoli. Nevertheless, it cannot be disregarded that the lql2-nu association might fa- cilitate the gathering of the 5s ribosomal cistrons and the nucleoli given that these cistrons are located in the telomeric regions of the long arm of chromosome 1 in humans (34). The chromosome 1 pericentromeric region would thus act as an organizer center of the in- terphasic chromatin thus making it possible for the 5s rRNA-which is synthesized outside the nucleolus-to join the preribosome at an early stage of maturation (35 ) .

In any case, understanding the nucleolus-associated chromosome 1 heterochromatic region requires further investigations. The persistence or not of this association after actinomycin-D inhibitor treatment will confirm or invalidate the relationship between transcriptional activity and lql2-nu colocalization. The combination of FICC detection of the nucleoli with the FISH of 5s rRNA genes will make clear whether the 5s cistron actually sticks to the nucleolus. Moreover, the association between the lq12 region and the nucleoli should have to be investigated during the cycle of un- transformed cells in which the changes with regard to number, shape, and size are less complex than in the MCF-7 cells.

ACKNOWLEDGMENTS The authors thank Dr. Victoria von Hagen for her

critical revision of the manuscript and Dr. Michele Brugal for her valuable work in documentation re- search.

LITERATURE CITED 1.

2.

3.

4.

5 .

6.

7.

8.

9.

10.

11.

12.

13.

14.

15.

16

17

18

19

20

Bartholdi MF: Nuclear distribution of centromeres during the cell cycle of human diploid fibroblasts. J Cell Sci 99:255-263, 1991. Borden J, Manuelidis L: Movement of the X chromosome in epi- lepsy. Science 2421687-1691, 1988. Brenner S, Pepper D, Berns MW, Tan E , Brinkley B R Kineto- chore structure, duplication, and distribution in mammalian cells: Analysis by human autoantibodies from scleroderma pa- tients. J Cell Biol 91:95-102, 1981. Brugal G, Dye R, Krief B, Chassery J-M, Tanke H, Tucker JH: HOME: Highly Optimized Microscope Environment. Cytometry 13:109-116, 1992. Buroker N, Bestwick R, Haight G, Magenis RE, Litt M: A hyper- variable repeated sequence on human chromosome lp36. Hum Genet 77:175-181, 1987. Cooke HJ, Hindley J: Cloning of human satellite I11 DNA: Dif- ferent components are on different chromosomes. Nucleic Acids Res 10:3177-3197, 1979. Devilee P, Thierry R, Kievits T, Kolluri R, Hopman A, Willard H, Pearson P, Cornelisse C: Detection of chromosome aneuploidy in interphase nuclei from primary breast tumors using chromosome- specific repetitive DNA probes. Cancer Res 48:5825-5830, 1988. Emmerich P, Laos P, Jauch A, Hopman A, Wiegant J, Higgins MJ, White BN, van der Ploeg M, Cremer C, Cremer T: Double in situ hybridization in combination with digital image analysis. Exp Cell Res 181:126-140, 1989. Fergusson M, Ward DC: Cell cycle dependent chromosomal movement in pre-mitotic human T-lymphocyte nuclei. Chromosoma 101557-565, 1992. Giroud F: Cell nucleus pattern analysis: Geometric and densito- metric featuring, automatic cell phase identification. Bio Cell 44: 177-188, 1982. Gonzalez SP, Nardone RM: Cyclic nucleolar changes during the cell cycle. I. Variations in number, size, morphology and position. Exp Cell Res 50599-615, 1968. Guillaud P, du Manoir S, Seigneurin D: Quantification and topo- graphical description of Ki-67 antibody labelling during the cell cycle of normal fibroblastic (MRC-5) and mammary tumor iMCF-7) cell lines. Anal Cell Pathol 195-39, 1989. Haaf T, Schmid M: Centromeric association and non-random distribution of centromeres in human tumour cells. Hum Genet 81: 137-143, 1989. Haaf T, Steinlein C, Schmid M: Nucleolar transcriptional activity in mouse Sertoli cells is dependent on centromere arrangement. Exp Cell Res 191:157-160, 1990. Hadjiolov AA: Ribosome biogenesis in the life cycle of normal and cancer cells. In: The Nucleolus and Ribosome Biogenesis. Springer-Verlag Wien, Cell Biology Monographs, New York, 1986, pp 192-195. Hadlaczky G, Went M, Ringertz N R Direct evidence for the nonrandom localization of mammalian chromosomes in the interphase nucleus. Exp Cell Res 167:l-15, 1986. Hernandez-Verdun D, Robert-Nicoud M, Geraud G, Masson C: Behaviour of nucleolar proteins in nuclei lacking ribosomal genes. A study by confocal laser scanning microscopy. J Cell Sci 98:99-105, 1991. Kurata S, Koga K, Sakaguchi B: Nucleolar size in parallel with ribosomal RNA synthesis at diapause termination in the eggs of Boinbix mori. Chromosoma 68:313-317, 1978. Leger I, Thomas M, Ronot X , Brugal G: Detection of chromosome 1 aberrations by fluorescent in situ hybridization (FISH) in the human breast cancer cell line. Anal Cell Pathol5:299-309,1993. Lichter P, Cremer T, Borden J , Manuelidis L, Ward D: Delinea- tion of individual human chromosomes in metaphase and inter-

322 LEGER ET AL.

phase cells by in situ hybridization using recombinant DNA li- braries. Hum Genet 80224-234, 1988.

21. Manuelidis L: Different central nervous system cell types display distinct and non random arrangements of satellite DNA sequences. Proc Natl Acad Sci USA 81:3123-3127, 1984.

22. Manuelidis L: Individual interphase chromosome domains revealed by in situ hybridization. Hum Genet 71:288-293, 1985.

23. Manuelidis L, Borden J: Reproducible compartmentalization of individual chromosome domains in human CNS cells revealed by in situ hybridization and three-dimensional reconstruction. Chro- mosoma 96:397-410, 1988.

24. Moroi Y, Hartman AL, Nakade PK, Tan EM: Distribution of ki- netochore (centromere) antigen in mammalian cell nuclei. J Cell Biol 90:254-259, 1981.

25. Ochs RL, Press RI: Centromere autoantigens are associated with the nucleolus. Exp Cell Res 200:339-350, 1992.

26. Pardo D, Luciani J, Stahl A: Localisation par hybridation in situ des genes des ARN 28 S e t 18 S dans les chromosomes somatiques humains. Ann Genet 18:105-109, 1975.

27. Popp S, Scholl H, Jauch A, Stelzer E, Cremer C, Cremer T: Dis- tribution of chromosome 18 and X centric heterochromatin in the interphase nucleus of cultured human cells. Exp Cell Res 189:l- 12, 1990.

28. Raap A, Marijnen J, Vrolijk J, van der Ploeg M: Denaturation, renaturation and loss of DNA during in situ hybridization procedures. Cytometry 7235-242, 1986.

29. Rappold G, Cremer T, Hager H, Muller C, Yang T: Sex chromosome positions in human interphase nuclei as studied by in situ hybridization with chromosome specific DNA probes. Hum Genet 67:317-325, 1984.

30. Santisteban MS, Robert-Nicoud M: Chromatin texture and pro- liferation associated antigen patterns. A study by fluorescence digital imaging microscopy. Anal Cell Pathol 4:223, 1992.

31. Santisteban MS: Etude in situ de la relation structure-fonction de la chromatine par analyse d'images en fluorescence. Thesis, Uni- versite Joseph Fourier, Grenoble, France, 1993.

32. Shardin M, Cremer T, Hager H, Lang M: Specific staining of human chromosomes in Chinese hamster x man hybrid cell lines demonstrates interphase chromosome territories. Hum Genet 71: 281-287, 1985.

33. Stahl A. Hartung M, Vagner-Copodano A, Fouet C: Chromosomal constitution of nucleolus-associated chromatin in man. Hum Genet 35:27-34, 1976.

34. Steffensen D, Duffey P: Localization of 5s ribosomal genes on human chromosome 1. Nature 252:741-743, 1974.

35. Steitz JA, Berg C, Hendrick JP, La Branche-Chabot H, Metspalu A, Rinke J, Yario T: A 5 s rRNAIL5 complex is a precursor to ribosome assembly in mammalian cells. J Cell Biol 106:545-556, 1988.

36. Vourc'h C, Taruscio D, Boyle AL, Ward DC: Cell-cycle-dependent distribution of telomeres, centromeres and chromosome-specific subsatellite domains in the interphase nucleus of mouse lymphocytes. Exp Cell Res 205:142-151, 1993.

37. Wachtler F, Schwarzacher HG, Ellinger A: The influence of the cell culture on structure and number of nucleoli in cultured human lymphocytes. Cell Tissue Res 225155-163, 1982.

38. Weimer R, HaafT, Kruger J, Poot M, Schmid M: Characterization of centromere arrangements and test for random distribution in Go, G,, S, G, and early S' phase in human lymphocytes. Hum Genet 88573-682. 1982.

APPENDIX Model of the probabilities of colocalization of FISH

chromosome 1 centromeric signal and nucleoli, in interphase nuclei, is described.

Biological Problem In the MCF-7 cell line, all nuclei which have been

analyzed had at least one FISH spot colocalized with a

nucleolus. This result suggests that the association between the chromosome 1 centromere and the nucleoli was not random. However, i t must be borne in mind that the microscopic images are 2D representations of 3D objects (see Fig. A-1). In order to confirm the hypothesis of a nonrandom association between FISH spots and nucleoli, we have developed a mathematical model which computes the probability that an observed specific 2D configuration is random.

Probabilistic Model The model proposed here is based on three assump-

1. A nucleus can be considered the interior of a sphere.

2. A nucleolus can be considered the interior of a sphere.

3. The 2D image of a nucleus represents an orthog- onal projection of the nucleus onto a plane.

Let S = the surface of the nucleus; R = the radius of the nucleus; D = the diameter of the nucleus; V = the volume of the nucleus; C = the center of the nucleus.

Let si = the surface of the ith nucleolus; r, = the radius of the ith nucleolus; d, = the diameter of the ith nucleolus; vi = the volume of the ith nucleolus; c, = the center of the ith nucleolus.

tions:

Let dist, = the distance from ci to C. S, R, D, si, ri, d,, dist, are measured with the HOME

The volume of the nucleus, under assumption 1, is: system.

4 V = -mR3

3

The volume of the nucleolus, under assumption 2, is:

The volume of the truncated cylinder above the nucleolus is (see Fig. A-2):

vol' = d h , , .

h, is the opposed side of the right-angled triangle (C, c, g); R is the hypotenuse; dist is the adjacent side.

So we have h,:

h, = d R 2 - Dist2.

The volume Vol of the total cylinder is: Vol = 2 *

Let P = the probability of a spot to be colocalized vol'.

with the nucleolus. We have:

Vol p = - V

Generalization to Three FISH Spots and More Than One Nucleolus

In this study, the authors have observed three FISH

Let a specific 2D configuration [Config(A)] be char- spots per nucleus and up to six nucleoli per nucleus.

ANALYSIS OF DNA-NUC1,EOLUS ASSOCIATION

OBSERVATION

323

OBSLRVED IMAGE

HIS spot ROJeCh

Nucleolus

FIG. A-I. The observed image can be the projection of three different situations: the FISH spot is above the nucleolus (case 1); the FISH spot is inside the nucleolus (case 2 ) ; the FISH spot is below the nucleolus (case 3).

C

FIG. A-2. The volume of the cylinder can be approximated by two identical truncated cylinders, one above the plane of the projection (represented here) and another below the projection plane.

acterized by: R, S, D, V = the parameters of the nucleus under consideration; k = the number of the nucleolus in the nucleus and for each nucleolus; ri, si, di,

v,, dist, = the parameters of the ith nucleolus; N, (0, 1, 2,3) = the number of FISH colocalized spots in the ith nucleolus; x, = the number of FISH colocalized spots with the ith nucleolus.

For a nucleus with k nucleoli and three spots, the probability of obtaining the Config(x1, x2, . . . , xk) is:

i = k I1 pixi) . 1 = 1

The probability of observing a colocalized spot with a nucleolus in image A (Fig. A-2) is equal to the volume of the cylinder divided by the volume of a nucleus.

In our experiments, the MCF-7 nuclei after the acetone fixation were rather flat on the slide side and approximately spherical on the opposite side (see Fig. 1G in text). Therefore, a more specific model could be elaborated considering a half sphere rather than a com- plete one. In that specific case, the probability of observing a colocalized spot with a nucleolus in image A is equal to the volume of the truncated cylinder above the plane of the projection (which is equal to half of the total cylinder) divided by the half volume of the nucleus; this probability is thus equal to that obtained with the previous model.

Documents

Interactive computer-assisted analysis of chromosome 1 colocalization with nucleoli