Nuclear chromatin texture and sensitivity to DNase I in human leukaemic CEM cells incubated with nanomolar okadaic acid

Abstract

It is now known that the analysis of chromatin texture can be used in oncology as a sensitive detection method, either to define diagnostic classifications or to locate a lesion along a defined trend curve. However, the functional significance of these variations in textural features remains sometimes unclear. Several drugs have been shown to be able to modulate chromatin structure. Among them, the phosphatase inhibitor okadaic acid at low concentration can increase accessibility to DNA in chromatin of carcinoma cells. This paper demonstrates that short exposures (0–3h) to a 10‐nM dose of okadaic acid induced an increased sensitivity to DNase I digestion in human CEM leukaemic cell nuclei and that this sensitization was associated to variations of nuclear texture characteristics, as evaluated by image cytometry. CEM cells treated with okadaic acid for 0–3h displayed changes in chromatin supraorganization with a more homogeneous and fine chromatin texture, as compared to control cells. This suggests that the appearance of an open configuration of chromatin structure as evaluated by biochemical methods corresponds to a more decondensed texture of nuclei measured by image cytometry. Longer exposures (6–24h) of CEM cells to 10nM okadaic acid lead to apoptosis. As reported previously for camptothecin‐treated HL60 cells, okadaic acid‐treated CEM cells display biphasic nuclear chromatin texture changes, i.e. a decondensation phase followed by the appearance of typical apoptotic cells with a smaller nuclear area and a highly condensed chromatin. Finally, using the multidrug‐resistant CEM‐VLB cell line, it was confirmed that these multidrug‐resistant cells also display cross‐resistance to okadaic acid, as this compound was unable to induce either increased DNase I sensitivity, apoptosis, or altered nuclear texture in this particular cell line.

Introduction

Various cytological processes such as cell proliferation, differentiation, transformation, apoptosis etc., are accompanied by chromatin changes, usually identified on the basis of the distribution of euchromatin and heterochromatin within the nucleus. Descriptions of chromatin patterns typically include such terms as ‘fine, coarse, clumped, pale …’ and the subjectivity of this process precludes reproducibility. One way to achieve a quantitative, non‐subjective evaluation of the chromatin pattern consists in the analysis of the chromatin texture by image cytometry (for review, see Doudkine et al. 1995 ). Texture features can be roughly classified into categories as descriptive statistics of chromatin distribution, discrete texture features, range extreme, Markovian run length, and fractal features ( Doudkine et al. 1995 ).

These features have proven to have value in the description of nuclei both in physiological and in pathological cell states, and in tumours, dramatic changes in the nuclear chromatin appearance are common and have been associated with the progression of the disease ( Deligdisch et al. 1993 , Doudkine et al. 1995 , Bartels et al. 1998 ). However, as it is still difficult to establish a close correlation between the variation of textural parameters and the eventual visible changes in chromatin, current texture parameters appear abstract, sometimes hamper understanding, or lead to cryptic results. Moreover, the functional significance of the variations in textural features is sometimes unclear. For example, it has been proposed that changes in chromatin condensation supraorganization could be related to proliferation state ( Santisteban & Brugal 1995), transcription activities in cell nuclei ( Vidal, Russo & Mello 1998), DNA synthesis rate ( Dörmer & Abmayr 1979), chromosomal rearrangements ( Liautaud‐Roger et al. 1992 ), or activation patterns of genes ( Mello et al. 1995 , Fischer et al. 1998 ).

A recent prior study showed that apoptosis induced by a topoisomerase I inhibitor was associated with an early textural ‘decondensation’ of nuclei and this was attributed to the appearance of a more open configuration of the chromatin structure ( Palissot et al. 1996b ). Several drugs are able to modulate chromatin supraorganization. Among them, the phosphatase inhibitor okadaic acid at low concentrations can increase accessibility to DNA in chromatin ( Rieber & Rieber 1992). This paper examines the consequences of this increased DNA accessibility on textural features as measured by image cytometry.

Material and methods

Chemicals

Okadaic acid (OA) was obtained from Sigma Chemical Co (St Louis, MO, USA) and was dissolved (25μg/ml) in dimethylsuphoxide (DMSO). Cell culture RPMI 1640 medium and fetal calf serum were purchased from Gibco/BRL (Grand Island, NY, USA). Stock preparations from all reagents were stored at –20^{^}C. All were diluted to appropriate final concentrations in medium.

Cell culture

The human lymphoblastic leukaemia CCRF‐CEM cell line and its multidrug‐resistant variant CEM/VLB100 were obtained from W.T. Beck, Memphis, TN, USA and Z. Steplewski, Philadelphia, PA, USA respectively. These were maintained in RPMI 1640 medium supplemented with 10% heat‐inactived fetal bovine serum, 20μM L‐glutamine, 100 units/ml penicillin G, and 100μg/ml streptomycin. All cultures were maintained under a fully humidified atmosphere of 95% air−5% CO₂ at 37^{^}C. CEM‐VLB were grown in the presence of 0.4μg/ml vinblastin. All cells were passaged twice weekly and all experiments with resistant cells were performed on cells cultured in drug‐free medium for more than 48h. Exponentially growing cells were used in all experiments. Cell viability was assessed by the ability to exclude trypan blue (0.5% w/v, Sigma).

Treatment with okadaic acid

Cells were seeded at 2×10;⁵ cells/ml in medium with okadaic acid at 10nM. Cells were then returned to a 37^{^}C, 5% CO₂ incubator until required for analysis. Okadaic acid stock solution was diluted to ensure a final concentration of less than 0.03% DMSO. Control cultures were treated with an equivalent volume of DMSO in RPMI. Cell growth was assessed by the MTS/PMS colorimetric assay (Promega, Madison, WI, USA). Apoptosis induction was monitored morphologically as described by Roger et al. (1996) after staining of the cells with a mixture of acridine orange (Sigma, 100μg/ml) and ethidium bromide (100μg/ml) in phosphate‐buffered saline (PBS), and immediate examination under a fluorescence microscope (Zeiss, Strasbourg, France) at a 490‐nm excitation wavelength.

Sensitivity to DNase I

Cells were washed twice in PBS and resuspended in PBS to a final concentration of 10⁷ cells/ml. One volume of cell suspension was mixed with one volume of ice cold lysis buffer (320mM sucrose, 5mM MgCl₂, 10mM Tris–HCl, 1% Triton X100, pH 7.5), 3 volumes of ice cold distilled water, and incubated for 10min on ice. After centrifugation (1300g, 15min, 4^{^}C), the pelleted nuclei were suspended at 10⁷/ml in 0.5ml lysis buffer.

Fifty microlitres of DNase I (Boehringer Mannheim, Germany) solution (100 units/ml) in incubation buffer (20mM Tris–HCl, pH7, 1M NaCl, 10mM MnCl₂) were added and incubated for 10min at 37^{^}C. After centrifugation, nuclei were lysed in 5ml lysis buffer (800mM GuHCl, 30mM EDTA, 30mM Tris–HCl, 5% Tween‐20, 0.5% Triton, pH8) containing 95μl protease solution (20mg/ml, Qiagen, Hilden, Germany) for 1h at 50^{^}C.

Genomic DNA was purified on anion‐exchange resin according to the manufacturer's protocol (Qiagen). DNA was then incubated during 2h at 37^{^}°C with 0.1μg/ml RNase (Dnase‐free) (Boehringer) prior to electrophoresis on 1% (w/v) agarose gel for 90min at 80V. Equal amounts of DNA (5μg) were loaded in each lane as determined by absorbance measurement at 260nm; 1μg lambda/HindIII molecular weight marker (23130‐−564bp, Gibco) was run in parallel. Electrophoregrams were scanned using an image analysis system (Vilbert Lourmat, Marne la Vallée, France) and analysed on a Macintosh 6100/60 using the public domain NIH Image program (developed at the US National Institutes of Health and available on the Internet at http://rsb.info.nih.gov/nih‐image/).

Image cytometry

Cells were smeared by cytocentrifugation at 100g for 5min, air‐dried and fixed in alcohol–formalin mixture (95% ethanol, 3% formaldehyde in saline; 3:1) for 10min. After acid hydrolysis with 5M HCl for 30 min, they were stained by the Feulgen method as previously described ( Palissot et al. 1996a ). Image cytometry was performed with an image analysis system (SAMBA 2005, Unilog, Grenoble, France) coupled to a colour three CCD camera (XC‐007P, Sony Corporation, Japan) and a microscope (Axioskop, Karl Zeiss, Oberkochen, Germany). Measurements were made with a plan‐achromat ×40 objective. After mitoses counting, at least 200 interphase cell nuclei were analysed on each microscopic slide. Eleven parameters were computed from each nuclear image: one geometric feature (nuclear area, NA), one densitometric parameter (integrated optical density, IOD), and nine textural parameters. The IOD is the summation of all pixels in the image of nucleus and quantitatively measures the total intensity of staining by the DNA‐binding dye. Since DNA staining with Feulgen reaction was stoichiometric, IOD was directly related to total nuclear DNA content. The distribution of nuclei according to IOD was plotted to provide cell cycle distribution. The margins defining G₀G₁ and G₂ peaks were defined as modal IOD±15% ( Dufer et al. 1995 ). At least 100 reference cells (mouse hepatocytes or human lymphocytes) were measured in the same conditions for the calibration of the normal diploid (2c) value.

Textural features estimated by image cytometry are closely related to the location of a given cell in the cell cycle. Comparisons of textures should therefore be performed on cells selected within a given cycle phase. For this purpose, G₀/G₁ cells were isolated on the basis of their DNA–IOD value, and analysed for chromatin texture. Although they could display slightly different nuclear structures, G₀ cells were not analysed separately, as they represent a minor component of such cell populations in growth phase (less than 1% as evaluated by Ki‐67 staining, i.e. below the resolution of the analysis system) ( Gorisse et al. 1990 ). Nine textural parameters were calculated on the nuclear image after reduction to 16 grey levels by linear rescaling ( Giroud 1987). Four features were calculated on the grey‐levels co‐occurrence matrix: Local mean (LM), energy (E), entropy (ENT), and inertia (I). Five parameters were calculated on the run‐length matrix: short run‐length emphasis (SRE), long run‐length emphasis (LRE), grey levels distribution (GLD), run‐length distribution (RLD), and run‐length percentage (RPC) ( Haralick, Shanmugan & Dinstein 1973, Galloway 1975, Giroud 1987). A brief description of these parameters appears in Table 1.

Table 1.

Computed textural parameters

Data analysis and statistics

The distribution, mean, and sd of the nuclear parameters were calculated for each cell population. Significance of the differences between parameters values was estimated by one‐way analysis of variance. Multivariate analysis was performed using the SAMBA software (modified BMD program — BMD07M, University of California, Los Angeles, CA, USA). Classification by discriminant analysis was based on use of the F‐test to select variables. The classification rates of the cells into predicted groups were estimated by splitting the data sets into two equal parts, defined as learning and test sets. In these estimations, discriminant functions were obtained on learning groups only and test groups were classified on these bases. Canonical analysis was used to display the differences between groups. Canonical variables were calculated on learning sets and cells from test groups projected as 95% confidence ellipses around the mean cell profiles onto the plane defined by the two first canonical variables.

Results

Cell growth changes induced by okadaic acid

Effects of 10 nM OA on cell growth and the cell cycle of CEM and CEM‐VLB cells were examined. The toxin was found to inhibit the cell growth of CEM cells markedly, whereas CEM‐VLB cells remain unaffected ( Fig. 1). Lower doses (2.5 and 5nM) were assayed in pilot experiments, but did not lead to significant growth or nuclear texture changes (data not shown). Cell cycle phases were extracted by isolation of G₀/G₁ and G₂ peaks on image cytometry data and counting mitoses on the same slides. Short exposures to OA (0–3h) did not affect cell cycle distributions either in CEM and CEM‐VLB cells. On the contrary, after 12 and 24h of treatment, CEM cells displayed a significant increase in mitotic cells and G₂ cells (67% of interphase cells after 24h) with a concomitant appearance of typical apoptotic cells as detected by the acridine orange–ethidium bromide method ( Table 2).

Figure 1(G).

rowth of CEM and CEM‐VLB cells cultured for 3 days in the presence or absence of 10 nM okadaic acid (OA). Solid circles: CEM cells without OA; solid squares: CEM cells with 10nM OA; open circles: CEM‐VLB without OA; open squares: CEM‐VLB with 10nM OA.

Table 2.

Effects of okadaic acid treatment (10nM) on cell cycle distribution of CEM cells.

DNase I digestion

Nuclei were isolated from CEM and CEM‐VLB cells either treated or not for 2h with 10nM OA. The nuclei were then digested by DNase I, and the digestion products were electrophoresed on agarose gels. The ethidium bromide stained gel ( Fig. 2) demonstrates that in CEM cells with 5 units DNase I a smear pattern of DNA fragmentation was produced. However, no clear‐cut internucleosomal DNA cleavage was produced. In OA‐treated CEM nuclei, the DNA appears more sensitive to digestion. When assessed by scanning of DNA smears, the increase in smear integrated density in OA‐treated cells vs. untreated cells is 40.5%. In contrast, treatment with OA did not increase DNase I sensitivity in CEM‐VLB cells.

Figure 2(D).

Nase I sensitivity of nuclei from CEM and CEM‐VLB cells treated with 5 units of DNase I with or without pretreatment with 10nM OA for 2h. Lane 1: molecular weights control; lane 2: CEM cells+OA; lane 3: CEM cells alone; lane 4: CEM cells+OA+DNase I; lane 5: CEM cells+DNase I; lane 6: CEM‐VLB cells+OA; lane 7: CEM‐VLB cells alone; lane 8: CEM‐VLB cells+OA+DNase I; lane 9: CEM‐VLB+DNase I. This figure shows a representative experiment from a series of three.

Analysis of nuclear texture

Values of the nuclear area and nine textural features were estimated in G₀/G₁ cells during a 24‐h incubation of CEM and CEM‐VLB cells with 10nM OA and compared with values observed in the respective control cell populations. Among these features, four (LM, E, SRE, and NA) were selected on the bases of their discriminatory power between the cell populations in a linear discriminant analysis, and of their mutual independence. The variations of these parameters are shown in Fig. 3. In CEM cells, short exposures to OA (0–3h) induced a progressive and significant decrease in SRE (which corresponds to the appearance of a more fine and homogeneous texture) and an increase in E (underlining a different distribution of chromatin), without changes in LM (which gives a global estimate of the condensation level), or NA. After 6–24h of treatment, CEM cells displayed an increase in chromatin condensation, as evidenced by the increase in LM and SRE, and the decrease in NA. On the contrary, OA treatment did not induce any change in textural parameters values measured on CEM‐VLB cells. These changes could be further visualized after a canonical analysis of the data observed on the nine textural features. The representation of the experimental groups onto the two‐dimensional plane resulting from this canonical analysis is shown on Fig. 4. The first canonical variable (CV1) explained 87.5% of the variance between the groups and mainly allowed separation of cell types. This representation shows that the nuclear texture of sensitive cells nuclear texture displays biphasic changes: decondensation and spatial reorganization of a more homogeneous chromatin (increase of LRE and E) from 0 to 3h, followed after 6–24h by condensation (increase in SRE, LM and RPC), whereas resistant cells nuclei do not display any significant alteration.

Figure 3 Variations of.

some representative cytometric parameters in CEM and CEM‐VLB cells treated with 10nM OA for 0–24h. Each point represents mean±sem of data from three independent experiments. Solid circles and solid line: CEM cells; open circles and dotted line: CEM‐VLB cells.

Figure 4(C).

anonical analysis of cytometric data obtained from textural features. Canonical plane was computed and 95% confidence ellipses around the mean cell profile of each population were projected onto this plane. Panel (a): CEM cells; panel (b): CEM‐VLB cells; panel (c): feature vectors. The length and the direction of a vector is directly related to the strength of the corresponding feature in the calculation of the canonical variable. Only the five most significant features are displayed.

Discussion

The present study has shown that short exposures to a nanomolar dose of OA induced an increased sensitivity to DNase I in human CEM leukaemic nuclei and that this sensitization was associated to variations of nuclear texture characteristics. OA‐treated CEM cells for 0–3h displayed a more homogeneous and fine chromatin texture than control cells. This suggests that the appearance of an open configuration of chromatin structure as evaluated by biochemical methods corresponds to a more decondensed texture of nuclei measured by image cytometry. It has been postulated that chromatin texture could be described in terms of condensation (compactness of chromatin segments), distribution (relative proportions of compactness levels), and organization (topographic arrangements of condensation degrees) ( Giroud 1987). It is interesting to note that the OA‐induced textural changes concerned mainly SRE and E, two features related to the spatial organization and distribution of chromatin, respectively, rather than to decondensation per se ( Giroud 1987). LM, a feature giving a global estimate of the whole chromatin condensation, appeared less modified, as did the nuclear area.

From a pathological point of view, it is now acknowledged that the analysis of chromatin texture can be used in oncology as a sensitive detection method, either to define diagnostic classifications or to locate a lesion along a defined trend curve ( Bartels et al. 1998 ). For example, such texture analyses have been applied to classifications of tumours from breast ( Dufer et al. 1993 , Weyn et al. 1998 ), bladder ( van Velthoven et al. 1995 ), ovary ( Deligdisch et al. 1993 ), prostate ( Jorgensen et al. 1996 ), and colorectal cancer ( Hamilton et al. 1995 ), as well as to the description of individual tumours by definition of textural signatures ( Bartels et al. 1998 ).

The biological significance of these changes in nuclear texture appears sometimes unclear and could be often associated with multifactorial causes. It has been shown that nuclear texture was associated with the rates of synthesis of DNA ( Dörmer & Abmayr 1979) and proteins ( Dörmer, Abmayr & Giaretti 1984). Textural changes have also been described during cell differentiation ( Giroud et al. 1988 , Dufer et al. 1989 ), or proliferation ( Giroud et al. 1988 , Santisteban & Brugal 1995). Moreover, texture alterations can be linked to chromosomal abnormalities ( Liautaud‐Roger et al. 1992 ) and oncogene activities ( Mello et al. 1995 ), and may be influenced by exogenous stimuli such as drug effects ( Colomb, Dussert & Martin 1991, Vidal et al. 1998 ).

In this work, it was showed that the induction of chromatin structure relaxation by a short OA treatment was actually associated with texture changes. These texture changes occurred between 1 and 3h OA incubation. These modifications could not be ascribed to different percentages of cells in G₁, S, or G₂ phases in the population analysed. Indeed, treatment with OA for such periods did not change the cell cycle phase distribution of nuclei ( Table 2). Moreover, texture analyses were performed on G₀/G₁ selected cells. Such an increase in DNase I sensitivity was previously reported on A431 carcinoma cells treated for 90min by 10nM OA ( Rieber & Rieber 1992). At this concentration, OA inhibits phosphatase 2 A, whereas it inhibits phosphatases I and 2B at higher levels ( Schonthal 1995). The mechanisms involved in these chromatin alterations remain controversial. These changes are reflected by a greater endonuclease susceptibility in the chromatin treated with OA correlated with a decrease in a 80‐kDa AT‐rich DNA binding protein and an apparent hyperphosphorylation of a 70‐kDa nuclear matrix protein ( Rieber & Rieber 1992). Histone phosphorylation (H1, H2A) and acetylation (H3) levels could modulate chromatin condensation ( Chadee et al. 1995 , Taylor et al. 1995 ). Pre‐treatment of cells with phosphatase inhibitors p‐(chloromercuri)‐benzenesulphonic acid (CMBS) or OA both lead to an enhanced DNase I sensitivity and to histone H2A phosphorylation ( Feng et al. 1991 , Rieber & Rieber 1992, Ajiro et al. 1996 ). However, other mechanisms, such as oncogene levels, could be implicated as well. Cells expressing various oncogenes displayed increased histone phosphorylation and a relaxed chromatin structure ( Chadee et al. 1995 ). Moreover, transfection of CHO cells with mutant p53, HPV 16‐E6, or cyclin G transgenes results in the disruption of higher order chromatin structure ( Smith et al. 1998 ).

The consequences of these changes in chromatin supraorganization might be of importance. Longer exposures of CEM cells to 10nM AO lead to mitotic arrest and apoptosis ( Table 1), two phenomena that have been shown to result from independent mechanisms ( Lerga et al. 1999 ). As reported previously for camptothecin‐treated HL60 cells ( Palissot et al. 1996b ), AO‐treated CEM cells display biphasic nuclear chromatin texture changes, i.e. a decondensation phase followed by the appearance of typical apoptotic cells with a smaller nuclear area and a highly condensed chromatin. Previous studies indicate that a decondensation of chromatin would be a prerequisite change which permits subsequent access to the chromatin by factors involved in DNA replication or transcription ( Chadee et al. 1995 ), or apoptosis ( Ferlini et al. 1996 , Palissot et al. 1996b ). Indeed, the apoptotic pathways induced by the two drugs could be different. For example, the caspase I selective protease inhibitor YVAD‐cmk inhibited camptothecin‐induced morphology, but not OA‐induced morphology ( Jensen et al. 1999 ). However, it must be stressed that incubation of cells for 24h with a low dose of OA leads to apoptosis with bcl2‐downregulation ( Riordan et al. 1998 ), a phenomenon also observed with camptothecin ( Palissot et al. 1998 ).

Finally, using a previously untested cell line, the present study confirmed that CEM‐VLB multidrug‐resistant cells also display cross‐resistance to OA. OA was unable to induce either increased DNase I sensitivity, apoptosis, or altered nuclear texture. These cells therefore could represent a convenient negative control for experiments involving OA treatments. The mechanisms responsible for this cross‐resistance remain unclear. Resistance to OA has been described either as P‐gp‐dependent ( Chambers, Raynor & Kuo 1993) or P‐gp‐independent ( Nishio et al. 1990 ), and OA induced a P‐gp overexpression together with a decrease in topoisomerase II expression ( Tohda et al. 1994 ). However, multidrug‐resistance is a complex phenomenon and other mechanisms could be implicated as well. For example, a role for bcl2 could not be ruled out as (i) OA induced apoptosis by bcl2‐downregulation ( Riordan et al. 1998 ), (ii) bcl2 gene transfer prevented apoptosis induced by OA ( Benito et al. 1997 , Lerga et al. 1999 ), and (iii) drug‐resistant cells could constitutively overexpress the bcl2 protein and mRNA ( Palissot et al. 1998 ).

Acknowledgements

The financial support of the ‘Comités Départementaux de la Haute‐Marne, de l’Aube et de l'Aisne de la Ligue contre le Cancer' is gratefully aknowledged. S. Yatouji is the recipient of a fellowship from the Région Champagne‐Ardenne.