Clustering of apoptotic cells via bystander killing

Clustering of apoptotic cells via bystander killing by peroxides KYRILL REZNIKOV,* LARISSA KOLESNIKOVA,*,1 ALADDIN PRAMANIK,† KOICHI TAN-NO,* IRINA GILEVA,*,1 TATJANA YAKOVLEVA,* RUDOLF RIGLER,† LARS TERENIUS,* AND GEORGY BAKALKIN*,2 *Experimental Alcohol and Drug Addiction Research Section, Department of Clinical Neuroscience and †Department of Medical Biochemistry and Biophysics, Karolinska Institute S-171 76, Stockholm, Sweden ABSTRACT Clustering of apoptotic cells is a characteristic of many developing or renewing systems, suggesting that apoptotic cells kill bystanders. Bystander killing can be triggered experimentally by inducing apoptosis in single cells and may be based on the exchange of as yet unidentified chemical cell death signals between nearby cells without the need for cell-to-cell communication via gap junctions. Here we demonstrate that apoptotic cell clusters occurred spontaneously, after serum deprivation or p53 transfection in cell monolayers in vitro. Clustering was apparently induced through bystander killing by primary apoptotic cells. Catalase, a peroxide scavenger, suppressed bystander killing, suggesting that hydrogen peroxide generated by apoptotic cells is the death signal. Although p53 expression increased the number of apoptoses, clustering was found to be similar around apoptotic cells whether or not p53 was expressed, indicating that there is no specific p53 contribution to bystander killing. Bystander killing through peroxides emitted by apoptotic cells may propagate tissue injury in different pathological situations and be relevant in chemo-, ␥-ray, and gene therapy of cancer.—Reznikov, K., Kolesnikova, L., Pramanik, A., Tan-No, K., Gileva, I., Yakovleva, T., Rigler, R., Terenius, L., Bakalkin, G. Clustering of apoptotic cells via bystander killing by peroxides. FASEB J. 14, 1754 –1764 (2000)

Key Words: intercellular communications 䡠 cell death 䡠 reactive oxygen species

Clustering of apoptotic cells has been described in different tissues such as Drosophila wing imaginal discs (1), the capillary network in remodeling rat ocular tissue (2), stem cell zones in small intestinal crypts (3), and proliferating glial cell populations in the hippocampus (4). The occurrence of apoptotic clusters may reflect either simultaneous passage of a group of clonally related cells through differentiation with synchronous entry into apoptosis (5, 6) or killing of viable neighbor cells by primary apoptotic cells, a process named bystander killing (2–5, 7, 8). 1754

Bystander killing could be triggered experimentally in Drosophila imaginal discs after induction of apoptosis in single cells (1). Cells individually killed in vitro by exposure to ␣-particles or X-rays increased the incidence of apoptosis (9 –11) and accumulation of the p53 tumor suppressor protein (12), a major transcription regulator of apoptosis (reviewed in ref 13), in non-hit bystander cells (9 –12). Bystander killing effects may be critical for enzyme prodrug (7, 8) and p53 substitution (14 –16) cancer gene therapy since current techniques for introduction of transgenes into tumors are not capable of transducing all neoplastic cells (14). Bystander killing may also contribute to the propagation of tissue injury in chronic diseases and after low levels of ionizing irradiation (9, 12), and appears to be a basis for the propagation of brain injury in cerebral ischemia (17). Bystander killing may be based on exchange of chemical cell death signals between nearby cells. These signals may reach their targets by diffusion through extracellular spaces or via gap junctions. There is some evidence for the release of toxic signals from cells individually exposed to radiation, followed by the induction of apoptosis of unirradiated cells. Cell death signals spread through the culture medium and specialized cell– cell communication via, for example, gap junctions was not required for bystander killing (10, 11). The chemical nature of these signals remains unknown. In the present study the phenomenology and mechanisms of the clustering of apoptotic cells were examined in vitro in cell culture monolayers. Experiments were designed with the emphasis on the role of p53, which mediates effects of different signals on 1 Permanent address: State Research Center of Virology and Biotechnology ‘Vector’, Institute of Molecular Biology, 633159 Koltsovo, Novosibirsk region, Russia. 2 Correspondence: Section of Alcohol and Drug Addiction Research, Department of Clinical Neuroscience, CMM L8 – 01, Karolinska Institute, S-171 76, Stockholm, Sweden. E-mail: [email protected]

0892-6638/00/0014-1754/$02.25 © FASEB

apoptosis and was suggested to induce bystander killing (14 –16). Clustering of apoptotic cells was registered in p53-negative Saos-2 cell monolayers. After transfection with p53, no difference in apoptotic clustering was found around p53-transfected and nontransfected apoptotic cells, indicating no p53 specific contribution in bystander killing. However, apoptotic cell clustering was inhibited by catalase, a peroxide scavenger, and intracellular concentrations of hydrogen peroxide in apoptotic cells estimated by using fluorescence correlation spectroscopy (FCS) in living cell cultures appeared to be very high, in the micromolar range. These observations suggested that hydrogen peroxide that readily penetrates cell membranes (18, 19) can function as an extracellular mediator of cell death.

37°C), fixed with 3.7% formalin, counterstained for 30 min with 0.4 ␮g/ml Hoechst 33258 (Hoechst, Frankfurt, Germany), and mounted on the slides with Moviol 4 – 88 (Hoechst). The DNA-intercalating dye propidium iodide (fluoresces red) enters only dead or dying cells with damaged membrane integrity; chromatin condensation in small, round-shaped nuclei and their fragmentation revealed by Hoechst staining (fluoresces blue) provide the evidence of their apoptotic nature (20). The TUNEL technique was used for in situ nick end-labeling detection of DNA fragmentation; an in situ cell death detection kit (Boehringer Mannheim, Germany) with fluorescein (fluoresces green) was used according to the manufacturer’s instructions. For immunofluorescence detection of p53-transfected Saos-2 cells, a monolayer culture seeded on coverslips was fixed in 3.7% formalin 48 h after transfection, probed with a monoclonal anti-p53antibody (clone DO-7; DAKO A/S, Glostrup, Denmark), exposed to secondary rabbit anti-mouse IgG fluoresceinconjugated antibodies (DAKO) in the presence of 0.4 ␮g/ml of Hoechst 33258 dye, and mounted on slides with Moviol.

MATERIALS AND METHODS

Determination of intracellular hydrogen peroxide

Cell culture and experimental treatments

Intracellular production of H2O2 was detected with the nonfluorescent 5-(and-6)-carboxy-2⬘,7⬘-dichlorodihydro-fluorescein diacetate (C-H2DCFDA; Molecular Probes), which easily passes through membranes during cell loading. The acetate moieties are cleaved by intracellular esterases, yielding the nonfluorescent molecule 5-(and-6)-carboxy-2⬘,7⬘-dichlorodihydro-fluorescein (C-H2DCF), which is trapped within the intracellular granules and cytoplasm due to its polarity. Hydrogen peroxide, together with intracellular peroxidases, is able to oxidize the trapped C-H2DCF to 5-(and6)-carboxy-2⬘,7⬘-dichloro-fluorescein (C-DCF), which fluoresces in green with a max at 525 nm (21–23). Saos-2 cells on coverslips at 100% confluence were loaded with 10 ␮M of CM-H2DCFDA and propidium iodide (20 ␮g/ml) in growth medium at 37°C for 1 h and the last 10 min, respectively. Cultures were fixed with 3.7% formalin for 30 s and mounted on the slides with Moviol. The peroxide concentrations were estimated in Saos-2 cells cultivated in 8-well chambers (Nalge, Nunc, Naperville, Ill.), 8 ⫻ 104 cells per well, in phenol red-free Iscove’s medium (Gibco) in the presence or the absence of 10% FBS for 3 days with daily changing of the medium. The medium collected on the third day was used for analysis of extracellular hydrogen peroxide. The cells were washed three times with phenol red-free Iscove’s medium, incubated for 2– 4 h at 37°C with 20 ␮M of C-H2DCFDA in fresh medium, and the intensity of the resultant C-DCF fluorescence in the cells and medium was measured by FCS. The average fluorescence intensity was calibrated with exogenous C-DCF (Molecular Probes) added to cell culture medium. The relationship of fluorescent intensity was linear in the range from 1 ␮M to 9 ␮M C-DCF in the medium.

The human osteosarcoma Saos-2 and cervical carcinoma HeLa cells were grown in Iscove’s medium (GibcoBRL, Gaithersburg, Md.) supplemented with 10% fetal bovine serum (FBS; Gibco) at 37°C in 5% CO2 humidified atmosphere. 2– 4 ⫻ 106 Saos-2 cells in 15 ml medium were seeded onto 19 mm coverslips in 10 cm plates (8 –10 slips/plate); after 24 h, slips were transferred to 12-well plates with 1 ml medium/well. Cells on slips in 12-well plates at 50 – 60% confluence were transfected with 0.4 ␮g of CMV wild-type p53 expression plasmid or CMV vector plasmid and 2 ␮l of lipofectamine (Gibco) in 0.2 ml of OPTI-MEM medium (Gibco) per well. A 0.2 ml aliquot of Iscove’s medium supplemented with 20% FBS per well was added after 5 h of incubation and the transfection medium was replaced with 0.4 ml of Iscove’s medium containing 10% FBS after 21 h additional incubation. Coverslips were processed for analysis of apoptotic and p53-positive cells 24 h later. HeLa cells at 70% confluence were labeled with 25 mM carboxyfluorescein diacetate, succinimidyl ester (CFDA se) using VybrantTM CFDA se cell tracer kit (Molecular Probes, Leiden, The Netherlands) according to the manufacturer’s recommendations. Labeled HeLa cells were loaded over unlabeled Saos-2 cells growing on 19 mm coverslips at 80% of confluence in 10 cm plates (8 slips in one plate) at the ratio of one part of HeLa cells to 40 parts of Saos-2 cells. After 24 h slips were transferred to 12-well plates with 1 ml fresh medium/well and incubated for the next 24 h. Saos-2 cells at 80% confluence were treated with catalase (Sigma Chemical Co., St. Louis, Mo.; 400, 4000, 10000, or 20000 U/ml) or NG-monomethyl-L-arginine (NMMA; Sigma) for 48 h with change of medium after 24 h. Catalase (10000 U/ml) was added to cells growing on coverslips 24 h after transfection with CMV-p53 expression or CMV vector plasmid; cells were incubated with the enzyme for the next 48 h with change of medium after 24 h. Saos-2 cells seeded on coverslips at 80% confluence were treated with 2, 5, 10, or 50 ␮M hydrogen peroxide for 6, 12, and 24 h and the percentage of apoptotic cells was counted. Staining of apoptotic and p53-positive cells Saos-2 cell monolayer cultures seeded on coverslips were incubated in propidium iodide (Sigma) (20 ␮g/ml, 10 min at APOPTOTIC CELLS KILL VIABLE NEIGHBORS

Detection of hydrogen peroxide in the medium C-H2DCFDA does not undergo oxidation; thus, the acetate moieties were first cleaved by alkali treatment to yield nonfluorescent C-H2DCF, which is susceptible to H2O2-mediated oxidation catalyzed by peroxidases (21–23). The peroxidecontaining samples and the H2O2 standard solutions were incubated with 80 ␮M of C-H2DCF and 10 ␮g/ml of horseradish peroxidase (Sigma) for 30 min at room temperature; the intensity of the resultant C-DCF fluorescence was determined by FCS. The relationship of fluorescence intensity was linear to H2O2 concentrations from 0.25 ␮M to 2 ␮M. 1755

Breakdown of exogenous hydrogen peroxide was assessed when solutions of 100 ␮M H2O2 in the phenol red-free Iscove’s medium were loaded into Nunc chambers with Saos-2 cells at 80% confluence or without cells. Chambers were incubated at 37°C for 10 or 60 min in 5% CO2 humidified atmosphere and peroxide concentrations in the medium were measured. FCS FCS was performed with confocal illumination of a laser volume element of 0.2 fl in a ConfoCor instrument (Carl Zeiss, Oberkochen, Germany) built according to the principles described by Rigler et al. (24). As focusing optics a Zeiss Neofluar 40 ⫻ NA 1.2 objective for water immersion was used in an epi-illumination setup. For separating exciting from emitted radiation a dichroic filter (Omega 540 DRL PO2) and a bandpass filter (Omega 565 DR 50) were used. The C-DCF was excited with the 514.5 nm line of an Argon laser. The average fluorescence intensity was detected by an avalanche photo diode (EG & SPCM 200). The FCS experimental setup used in this paper has been described earlier (25). Quantitation of the percentage of apoptotic cells The percentage of apoptotic cells was assessed by fluorescent microscopy after vital staining with propidium iodide and postfixation counterstaining with Hoechst 33258 dye. For each experimental observation, the number of apoptotic cells per 900 cells was scored on three coverslips, 300 cells on each slip, besides the experiment with hydrogen peroxide treatment where the number of apoptotic cells per at least 500 cells was scored on 10 microscopic fields with 50 – 80 cells on each field. Data were used for calculation of apoptotic index (the number of apoptoses as a percentage of the total number of counted cells) and reported as means ⫾ se. Cell growth/cytotoxicity assays Saos-2 cells (105 cells/35 mm plate) were treated with 0.5 or 1 mM hydrogen peroxide for 24 h; adherent and nonadherent cells were pooled and analyzed for viability using a trypan blue exclusion assay. Morphological assessment of the bystander effect The number of apoptoses within a 110 ␮m radius from centers of 300 viable and apoptotic cells was counted on three coverslips for each experimental point under analysis. In cell cultures immunostained with p53 antibody, the number of apoptoses within 110 ␮m radius was counted separately for viable p53-negative and p53-positive cells, as well as apoptotic p53-negative and p53-positive cells. The Student’s t test was used to evaluate differences between means. Apoptotic cell distribution and cluster analysis The distribution of apoptoses was mapped in rectangular areas (2330 ␮m ⫻ 1000 ␮m) projected from five Saos-2 subconfluent cultures using camera lucida. In all maps, the randomness in spatial distribution of apoptoses was tested using the nearest neighbor distances and K-means clustering statistics as described for cell clustering in tissue sections (4). The nearest neighbor distances between apoptotic cells were analyzed using Clark-Evans statistics (26). This method is based on a comparison of the means of the nearest neighbor distances between the set of observed points and a completely 1756

Vol. 14

September 2000

random point pattern of the same density per unit of the area. The K-means clustering algorithm operates with a given number of clusters by moving the objects (points) in and out of clusters to give the minimal variability within clusters and the maximum variability between clusters. The optimal partition for different sizes of clusters was tested using the locations of points (centers of apoptotic nuclei) on maps with rectangular (X and Y) coordinates. The K-means cluster analysis was performed using the Statistica StatSoft program (Scandinavia AB, Uppsala, Sweden).

RESULTS Clustering of apoptotic cells in vitro Apoptotic human osteosarcoma Saos-2 cells were detected with propidium iodide (red fluorescence), by visualizing nuclear shrinkage/fragmentation with Hoechst 33258 staining (blue fluorescence), and by TUNEL staining (green fluorescence) (Fig. 1). Cells labeled with propidium iodide composed 92 ⫾ 2.4% of Hoechst-stained cells with nuclear condensation. Propidium iodide, which is a DNA-intercalating dye, enters cells with loss of membrane integrity and stains both necrotic and apoptotic cells. A high incidence of apoptotic cell labeling with propidium iodide indicated that apoptosis was the predominant form of cell death in Saos-2 cell cultures grown in control medium deprived of serum or p53 transfected. The TUNEL stain labeled 71 ⫾ 3.2% of cells with apoptotic morphology, demonstrating that DNA fragmentation had occurred in these cells. However, 29% of the morphologically apoptotic cells were not stained by TUNEL, probably because DNA fragmentation is a much later event in apoptosis than nuclear condensation (27). The extent of apoptosis assessed on coverslips by morphological criteria was 0.2–1%, 1– 4%, and 10 –20% in normal cell cultures, cells transfected with CMV wild-type p53 expression plasmid, and after 3 days serum deprivation, respectively. The flow cytometry analysis of cells with the subG1 DNA content provided essentially proportional but higher numbers since floating apoptotic cells were included in analyzed samples. Thus, ⬃4% of all cells were apoptotic in cell cultures grown under normal conditions or mock transfected with CMV vector plasmid, whereas ⬃16% apoptotic cells were registered in cultures transfected with CMV wild-type p53 expression plasmid (unpublished results). There was a tendency for clustering of apoptotic cells in subconfluent Saos-2 cultures grown in control medium (Fig. 2, left). The visual impression of apoptotic clustering in control Saos-2 cultures was confirmed with statistical analyses. Mapping the location of apoptotic cells, followed by nearest neighbor analysis, demonstrated a nonrandom distribution since the means of nearest neighbor distances

The FASEB Journal

REZNIKOV ET AL.

uted apoptotic cells; Table 1) from the centers of viable and apoptotic cells was also different (Fig. 3). The number of apoptoses around apoptotic cells was two to three times greater than around viable cells. In cultures transferred to serum-free medium, a massive entry into apoptosis was accompanied by prominent cell clustering (Fig. 2, right). Analysis of cultures deprived of serum for 24, 48, and 72 h showed that clusters consisted of cells at different stages of apoptosis at all these end points. Clusters composed of apoptotic cells with chromatin condensation in small round-shaped nuclei characteristics of early stages of apoptosis were observed (Fig. 4B). Many clusters contained apoptotic cells with fragmented nuclei and apoptotic bodies typical for the late stages of apoptosis (Fig. 4C). There were also clusters with a mixture of cells at different stages of apoptosis (Fig. 4D). Thus, apoptotic clusters were apparently formed due to synchronous and asynchronous entry of neighboring cells into apoptosis. To test the possibility that apoptotic clusters were formed by entry of sister cells into apoptosis (e.g., via aberrant mitosis), HeLa cells covalently labeled with CFDA se were loaded over confluent unlabeled Saos-2 cells; 24 h later, the occurrence of unlabeled apoptotic Saos-2 cells around labeled HeLa cells was registered. The number of apoptotic Saos-2 cells within the 110 mm radius from a center of an apoptotic HeLa cell (0.49⫾0.06) was significantly (P⬍0.001) higher than that around a viable HeLa cell (0.1⫾0.25). Thus, a majority of apoptotic cells in clusters did not derive from the same cell precursors and cell clustering was not a cell type-specific phenomenon. Catalase suppresses apoptosis clustering Figure 1. Fluorescence staining of two adjacent Saos-2 apoptotic cells with propidium iodide (upper panel), TUNEL technique (middle panel), and Hoechst 33258 dye (bottom panel). Scale bar represents 15 ␮m.

between apoptotic cells were less than expected by chance (Table 1). Consequently, the ratios of observed to expected mean distances (R) were less than 1 for all populations observed, indicating a departure from random distribution in the direction of aggregation (clustering) (26). For all populations observed, the deviations were statistically significant. Analysis of the location of apoptotic cells within rectangular coordinates of the apoptosis plotting maps using the K-means clustering algorithm revealed optimal partition of three different sizes of clusters with an average size of 7–9, 3, and 2 cells per cluster, respectively (Table 2). The occurrence of apoptoses within a 110 mm radius (equal to the mean expected distance between randomly distribAPOPTOTIC CELLS KILL VIABLE NEIGHBORS

Clustering of apoptotic cells may occur due to bystander killing by signal substances released from dying cells (9 –11). These substances may reach their targets by diffusion through extracellular spaces (10) or via gap junctions. Given that cell– cell communication via gap junctions is weak in a Saos-2 monolayer (28, 29), reactive oxygen species (ROS), or nitric oxide may be released, penetrate cellular membranes (18, 19) and induce apoptosis (30 –33). If hydrogen peroxide is released from the cells, catalase, which cannot diffuse into cells (34), would primarily metabolize hydrogen peroxide in the medium (32) and block apoptotic cell clustering. Indeed, catalase in the dose range of 4000 to 20,000 U/ml abolished clustering of apoptotic cells (Fig. 3); the number of apoptoses was decreased much more strongly around apoptotic cells than around viable cells. The nitric oxide synthase inhibitor NMMA (0.1 and 1 mM) did not affect the number of apoptoses. 1757

Figure 2. Clustering of apoptotic cells in subconfluent Saos-2 cultures grown in control medium (left) and 2 days after serum deprivation (right). Apoptotic cells were doublestained with propidium iodide (upper panels) and Hoechst 33258 dye (lower panels). Scale bar represents 150 ␮m.

Apoptotic cells produce high concentrations of hydrogen peroxide Generation of hydrogen peroxide was confirmed using the nonfluorescent C-H2DCFDA, which turns into the fluorescent dye C-DCF on cleavage of the acetate by intracellular esterases and oxidation by intracellular peroxides (21–23). While only few viable cells (0.9⫾0.2%) were stained by DCF, more than 40% of apoptotic cells labeled with propidium iodide (40.8⫾4.9%) were also labeled with DCF (Fig. 5). The reason that DCF fluorescence was not registered in all apoptotic cells could be technical, such as DCF leaking out the cell during the fixation procedure. More likely, however, the ROS are only involved in the activation phase of programmed cell death and are not required for the execution of the death program (35). Hydrogen peroxide concentrations were estimated in living cells by using FCS. FCS is a powerful

biophysical tool for examining molecular events with very high specificity (24). In this technique, the fluorescence of single dye molecules excited by a sharply focused laser beam is registered. The extremely tiny 0.2 fl volume element of the laser beam allows the measurement of molecular properties at specific coordinates within and around the cell (24, 25, 36, 37). Laser focus was placed on the upper surface of the cell, attached to the floor of a Nunc chamber, 2 ␮m below this level inside the cell, or 10, 20, and 40 ␮m above this level in the medium (Fig. 6). Similar average fluorescence intensity was registered in the control medium and medium from cells precultivated under normal conditions, produced through the spontaneous hydrolysis of C-H2DCFDA and oxidation of C-H2DCF to the fluorescent C-DCF. The fluorescence intensity was ⬃fourfold higher in the medium from cells deprived of serum for 3 days. When focus was placed intracellularly 2 mm below the upper cell surface, two types of responses were

TABLE 1. The nearest neighbor distances between apoptotic cells in five maps of a Saos-2 monolayera

Map

Apoptotic cells, N

Density (␳) N/␮m2 ⫻ 10⫺5

rE ⫾ ␴rE ␮m

robs ⫾ ␴robs ␮m

R

C

P

45 46 57 62 59

1.93 1.97 2.45 2.66 2.53

114 ⫾ 8.9 112 ⫾ 8.7 101 ⫾ 7.0 97 ⫾ 6.4 99 ⫾ 6.8

88 ⫾ 6.5 86 ⫾ 7.4 76 ⫾ 6.7 75 ⫾ 5.6 63 ⫾ 5.8

0.77 0.77 0.75 0.78 0.63

2.93 3.01 3.56 2.58 5.38

⬍0.01 ⬍0.01 ⬍0.01 ⬍0.01 ⬍0.001

1 2 3 4 5

a Each map was plotted with the location of apoptoses in a rectangular area (2330 ␮m ⫻ 1000 ␮m) of a Saos-2 monolayer. Density (␳) is the number of apoptotic cells per unit area; rE ⫾ ␴rE is the mean expected nearest neighbor distance ⫾ se for a random distribution (rE ⫽ 1v␳; ␴rE ⫽ 2 0.26136/vN 䡠 ␳); robs ⫾ ␴robs is the mean nearest neighbor distance observed ⫾ se; R (R ⫽ robs/rE) is a measure of the degree to which the observed distribution departs from random, R ⫽ 1. The significance of the departures from the random distribution was tested using the formula C ⫽ robs ⫺ rE/␴rE, where C is the standard variate of the normal curve (26). P is the probability that the observed distances between apoptotic cells are distributed randomly.

1758

Vol. 14

September 2000

The FASEB Journal

REZNIKOV ET AL.

TABLE 2. Optimal partition of apoptotic cell distribution in Saos2 monolayers using an iterative procedure with the K means clustering algorithma Map

Cells per cluster (mean)

1

7.5

2

9.2

3

9.5

4

8.9

5

6.6

FX 251 FY 13.3 FX 178 FY 15.1 FX 257 FY 19.6 FX 226 FY 22.7 FX 261 FY 14.4

3.0 2.9 3.2 3.0 3.1


2.3 2.2 2.1 2.0 2.0


a For each map (described in Table 1) apoptotic cells were identified within rectangular (X and Y) coordinates. Possible clustering was tested from 2 (minimum) to 31 apoptotic cells per cluster. FX and FY represent the ratio (F) of the between-clusters variance over the within-cluster variance for cells within X and Y coordinates, respectively. In all cases significance in the F test was P ⬍ 0.001.

Figure 4. Morphology of Saos-2 cells in apoptotic cell clusters during serum withdrawal. Saos-2 cells were cultured with (A) or without serum for 24 (B), 48 (C), and 72 h (D). Cells were stained with Hoechst 33258 dye. Nuclei of control cells (A). Apoptotic cells with condensed chromatin (B; arrowheads, early stages of apoptosis) or fragmented nuclei (C; arrows, late stages of apoptosis). Clusters containing apoptotic cells with condensed chromatin (arrowheads) or fragmented nuclei (arrows) (D).

registered. Viable cells characterized by slightly elongated polygonal morphology demonstrated low levels of fluorescence. Dying, presumably apoptotic cells characterized by a rounded shape, blebbings of plasma membrane and located 5–10 ␮m above the viable cell monolayer (27, 38, 39) generated 40- to 100-fold higher fluorescence. Three days of serum

Figure 3. Apoptoses within 110 ␮m radius from the centers of p53-negative or p53-positive viable (white bars) and apoptotic (black bars) cells in subconfluent Saos-2 culture transfected with CMV-p53 expression plasmid untreated or treated with catalase (CAT; 10 000 U/ml). Data are means ⫾ se. The number of apoptoses around p53-negative and p53-positive apoptotic cells significantly surpassed that around p53-negative and p53-positive viable cells, respectively, in untreated cultures whereas catalase treatment diminished the number of apoptoses around apoptotic cells. At 400 U/ml, catalase did not cause any significant effect on the number of apoptoses. At doses 4000 U/ml, 10000 U/ml, and 20000 U/ml catalase similarly inhibited clustering of apoptotic cells. No differences in the number of apoptoses around viable and apoptotic cells, respectively, were found between untransfected cultures (not shown), mock-transfected cultures (not shown), and the p53-negative cells in cultures transfected with CMV-p53 expression plasmid. There was no difference in any culture in the number of apoptoses around p53⫺ vs. p53⫹ viable cells or around p53⫺ vs. p53⫹ apoptotic cells. *P⬍0.00001. APOPTOTIC CELLS KILL VIABLE NEIGHBORS

Figure 5. Apoptotic cells generate hydrogen peroxide. Apoptotic Saos-2 cells stained with propidium iodide (red fluorescence) and loaded with the nonfluorescent dye C-H2DCFDA, which is converted by intracellular peroxides to the green fluorescent dye C-DCF. Cells in which both dyes are detectable appear yellow in superimposed images. Scale bar represents 35 ␮m. 1759

Concentrations of hydrogen peroxide were extrapolated from the calibration curves generated with the fluorescent C-DCF as a reference added into the medium. The estimated peroxide concentrations in the medium, 2 ␮m under or 10 ␮m above the surface of 3 day serum-deprived apoptotic cells after 2 h incubation with C-H2DCFDA, were ⬃5, 12, and 7 ␮M, respectively. These concentrations are in the same range as those released by activated macrophages (40, 41) or neutrophils (42, 43); here an apoptotic cell produced ⬃0.7 fmol of H2O2 equivalents/min into medium whereas macrophages and neutrophils released 0.05–2.6 fmol H2O2/min and cell (40 – 43). Hydrogen peroxide secretion and breakdown

Figure 6. The average intensity of intra- and extracellular C-DCF fluorescence in Saos-2 cell cultures. The fluorescence intensity was determined by fluorescence correlation spectroscopy (FCS) in cultures cultivated in the presence (black bars) or the absence (empty bars) of fetal bovine serum, as well as in medium only (the striped bar). The C-DCF fluorescence in the control medium was generated by the spontaneous deacetylation of C-H2DCFDA and subsequent oxidation of C-H2DCF. The laser focus was placed on the upper surface of a cell (CS) attached to the Nunc chamber, 10 and 20 ␮m above this level in the medium, or 2 ␮m below this level inside the cell. In panel A, fluorescence was registered inside a cell or in the medium. Viable cells were identified as cells with slightly elongated polygonal morphology whereas rounded cells with blebbing of the plasma membrane located ⬃10 ␮m higher than viable cells in the monolayer, were taken as apoptotic. Data (RU, relative units) are means ⫾ se. Student’s t test was used for statistical evaluation of the results; *P⬍0.05, **P⬍0.01, ***P⬍0.0001.

deprivation substantially increase the hydrogen peroxide production 3.5-fold by viable cells whereas presumably apoptotic cells from control and serumdeprived cultures did not differ in fluorescence intensity. The average fluorescence intensity was similar inside the cell 2 ␮m below the cell surface and at the cell surface level, but was 8 –10 times lower in the medium (Fig. 6B). However, the intensity 10 ␮m above the cell surface was only 1.8-fold lower than that inside the cell and 5-fold higher than 20 ␮m above the cell surface or in the medium. This gradient demonstrates that the fluorescent dye was constantly generated in the intracellular oxidative reactions and that there was a strong outflow of this dye from the apoptotic cells into the medium. 1760

Vol. 14

September 2000

The peroxide levels were similar in the control medium incubated without cells and the medium from cells grown under normal conditions (0.75 and 0.80 ␮M, respectively), but were lower in the medium from cells serum-deprived for 3 days (0.33 ␮M). To test whether peroxide is stabile in the medium, 100 ␮M solution of H2O2 in serum- and phenol red-free medium was loaded into a Nunc chamber with Saos-2 cells at 80% confluence or without any cells. Figure 7 demonstrates that peroxide concentrations rapidly declined and were practically undetectable after 60 min incubation in both cell-free medium and medium with cells. The short half-life in the medium did not allow determination of actual concentrations of peroxide in the medium secreted by apoptotic and viable cells. Hydrogen peroxide in micromolar concentrations induces apoptosis Despite a short half-life, hydrogen peroxide induced apoptoses registered 12 h after loading into the medium (Fig. 8). The number of apoptoses was increased 2.4- and 2.6-fold after addition of 10 and 50 ␮M of peroxide, respectively. No substantial differences between untreated and peroxide-treated cultures were revealed after 6 and 24 h. Thus, Figure 7. Degradation of 100 ␮M of exogenous hydrogen peroxide in the control phenol red-free Iscove’s medium without cells (black bars) and the medium with Saos-2 cells (empty bars). The medium was incubated for 10 or 60 min at 37°C. The ordinate shows the resulted concentrations of hydrogen peroxide.

The FASEB Journal

REZNIKOV ET AL.

peroxide triggered changes in cells that developed into apoptosis during the next 12 h. Apoptotic bodies formed had apparently disappeared from the coverslips by 24 h. At 0.5 and 1 mM concentrations, peroxide induced death of 50 –100% Saos-2 cells after 24 h incubation, as demonstrated with a trypan blue assay; only a few cells remained attached to the dish, and of those, few had apoptotic morphology. Apparently, the bulk of apoptotic cells had floated away into the culture medium. No difference in bystander killing around p53positive and p53-negative apoptoses The p53 tumor suppressor protein has been implicated in bystander cell death induced by adenovirusmediated p53 gene therapy (14 –16) as well as in bystander growth arrest (44). To assess whether p53 contributes to bystander killing, p53-deficient Saos-2 cells were transiently transfected with a CMV wildtype p53 expression plasmid (Fig. 9). The proportion of p53 immunopositive (p53⫹) cells to total cell number ranged from 1 to 4% 48 h after transfection in 10 experiments. The expression of p53 resulted in a 1.5- to 3-fold increase in the total proportion of apoptotic cells compared to that in mock transfection experiments. About one-fifth (20⫾5%) of the apoptotic cells in p53-transfected cultures were p53⫹, indicating activation of the p53-dependent pathway of cell death. The number of apoptoses around p53⫹ and p53⫺ apoptotic cells (within the 110 ␮m radius from cell centers) did not differ significantly and was much higher than that around p53⫹ and p53⫺ viable cells (Fig. 3). The data demonstrate that cells that commit apoptosis via p53dependent or p53-independent pathways induce bystander killing with similar efficiency; 27–29% of apoptoses located within 110 ␮m radius from apoptotic cells were p53 immunopositive independent of whether a centered cell was p53⫹ or p53⫺ (Fig. 9), demonstrating that neighboring apoptotic cells were not sister cells resulting from aberrant mitosis. Catalase inhibited clustering of apoptotic cells around both p53⫹ and p53⫺ apoptotic cells with the same Figure 8. Hydrogen peroxide induces apoptosis in Saos-2 cells. Hydrogen peroxide was added to the culture medium and the number of apoptoses after 6, 12, and 24 h of incubation was counted as the percentage of the total number of cells. The number of apoptoses after 12 h is shown as means ⫾ se. No effects were registered after 6 and 24 h of incubation at the indicated peroxide concentrations. Statistical significance was estimated by the chi-square test. *P⬍0.05, **P⬍0.02. APOPTOTIC CELLS KILL VIABLE NEIGHBORS

Figure 9. p53-immunopositive and p53-immunonegative cells in clusters of apoptotic cells. Neighboring apoptotic cells in a p53 transiently transfected Saos-2 culture stained with propidium iodide (upper row), with the p53 antibody (middle row), and with Hoechst 33258 dye (bottom row). Three apoptotic cells are p53-positive in the left middle panel, whereas only one apoptotic cell is p53-positive in the right middle panel. Scale bar represents 75 ␮m.

efficiency (Fig. 3). Apparently, p53 expression did not generate additional extracellular death signals. DISCUSSION Apoptotic cell clusters characteristically occur in many developing and renewing tissues (1– 6). In the monolayer Saos-2 cell culture, small foci of apoptotic clusters were spread throughout the surface of cultures grown on coverslips. These foci contained fewer than 10 cells per cluster. The optimal partitions identified three different sizes of clusters with 7–9, 3, or 2 cells per cluster. The distribution of primary apoptotic cells dying spontaneously may depend on the heterogeneity of the glass surface. However, the glass surface did not influence cell clustering because treatment with catalase, which does not affect the glass surface properties, abolished clustering. In addition, generation of primary apoptotic cells due to p53 expression, a process that does not depend on glass surfaces, also resulted in the formation of apoptotic cell clusters. Mosaics of proliferating, differentiating, or dying cells in tissues and cell cultures suggest the existence of discrete units with coordinated and cooperated 1761

decisions taken by a cell community about the fate of individual cells that form the cluster (4, 45–50). These decisions may result in proliferation, differentiation, or cell death. The borders between clusters may be relative and flexible, thus defining a cluster as a functional rather than a structural unit. This mechanism does not exclude the influence of external factors. Death of one or a few cells induced by external factors may generalize to a majority of cells in a cluster by intercellular communication. In any population cells receive information from every neighboring cell and patterns will develop, with peaks and valleys in concentration/activity of signal molecules or other information-bearing agents (51–53). The most plausible reason for apoptotic clustering in culture might be the variability in spatial distribution of cell death signals that cells receive from their neighbors. When the total pool of these molecules shared by adjacent cells surpasses a certain threshold level, it would initiate a synchronized death of cells in this group. Death signals released from cells can reach their targets through gap junction channels or extracellular spaces. Intercellular communication by means of gap junctions is weak in the Saos-2 monolayer (28, 29), which leaves the other alternative that the extracellular spread of soluble substances triggers apoptosis in neighboring cells. The synchronized changes in a limited group of adjacent cells suggest that these substances have a half-life comparable with their time of diffusion across a group of cells. Readily diffusible molecules with a half-life on the order of seconds and cell killing properties include nitric oxide, which has already been implicated in local cell– cell interactions (49, 50, 54), and hydrogen peroxide (18, 19). Among biologically relevant ROS, hydrogen peroxide is a likely candidate for a cell death signal messenger, since it is comparatively stabile with a half-life on the order of seconds (rather than fractions of the second for the hydroxyl radical); it can freely permeate cell membranes (18, 19) and cause cell death via apoptosis in a variety of cell lines including Saos-2 cells (present study) at low micromolar concentrations (30, 55– 60). Peroxides are downstream mediators of p53-dependent apoptosis, and peroxide scavengers can inhibit apoptosis induced by different stimuli (55, 56, 61 and references therein). Signaling pathways leading to hydrogen peroxide production remain unclear (55, 56, 61). How apoptosis is induced by ROS is also not yet known, although activation of the sphingomyelin/ ceramide apoptotic signaling pathway has been suggested (57). Concentrations of hydrogen peroxide produced by apoptotic cells were estimated to be very high, in the micromolar range, and similar to those released by activated macrophages and neutrophils (40 – 43). A gradient of the fluorescent dye C-DCF generated 1762

Vol. 14

September 2000

in the oxidative reaction with intracellular peroxide was registered around apoptotic cells. The concentration of this dye at a 10 ␮m distance from the cell was 5-fold higher than in the medium and only 1.8-fold lower than inside the cell. To maintain over time such a gradient in a volume that exceeds that of the cell ⬃10 –20 times, the rate of production of hydrogen peroxide by apoptotic cells likely surpasses the rate of C-DCF diffusion from the cell surface into the medium. Hydrogen peroxide at low micromolar concentrations killed Saos-2 (present study) and many other cells (55– 60) via apoptosis. Since hydrogen peroxide easily degraded in the medium (present study; 33), the effective concentrations around cells were probably much lower and the exposure time much shorter. Cumulative concentrations of hydrogen peroxide released by cells during entering and progression of apoptosis were probably sufficient to kill viable bystanders. This notion is supported by the observation that catalase, a peroxide scavenger, strongly inhibited bystander killing. Catalase is a 60 kDa protein and cannot simply diffuse into cells (34); it should therefore primarily metabolize hydrogen peroxide released from the cells (32). Macrophages and neutrophils produce large amounts of ROS (40 – 43), which lead directly to the destruction of the target cells in the course of an antibody-mediated cytotoxic response. By analogy, hydrogen peroxide emitted by apoptotic cells may function as an extracellular mediator of cell death that propagates tissue injury in different pathological situations. Various anticancer agents or ␥-ray irradiation kill tumor cells via genotoxic stress, which includes intracellular production of hydrogen peroxide (30, 35, 55, 56, 61, 62). The oxidant emitted from dying cells may promote elimination of viable neighbor cells by bystander killing. Treatments designed to promote bystander killing may be developed to increase efficacy of chemo-, ␥-ray, and gene therapy of cancer. This work was supported by grants from the Swedish Cancer Society (3935) and the Swedish Medical Research Council (12190) to G.B. and by a fellowship from the Karolinska Institute to K.R. and from the Royal Swedish Academy of Sciences to I.G. We thank Dr. B. Vogelstein for pC53-SN3 plasmid.

REFERENCES 1.

Milań, M., Campuzano, S., and Garcıá-Bellido, A. (1997) Developmental parameters of cell death in the wing disc of Drosophila. Proc. Natl. Acad. Sci. USA 94, 5691–5696 2. Lang, R., Lustig, M., Francois, F., Sellinger, M., and Plesken, Y. (1994) Apoptosis during macrophage-dependent ocular tissue remodeling. Development 120, 3395–3403

The FASEB Journal

REZNIKOV ET AL.

3. 4. 5.

6. 7. 8. 9.

10.

11.

12. 13. 14.

15.

16.

17.

18. 19. 20.

21.

22.

23. 24. 25.

Ijiri, K., and Potten, C. S. (1987) Cell death in cell hierarchies in adult mammalian tissues. In Perspectives on Mammalian Cell Death (Potten, C. S., ed) pp. 326 –356, Oxford University Press, Oxford Reznikov, K. Y. (1991) Cell proliferation and cytogenesis in the mouse hippocampus. Adv. Anat. Embryol. Cell Biol. 122, 1– 83 Hendry, J. N., and Potten, C. S. (1998) Letter to the Editor. Comments on the paper—Cell survival in irradiated microcolonies: how influential are neighbours? Int. J. Radiat. Biol. 73, 575–576 Wilson, J. W., Pritchard, D. M., Hickman, J. A., and Potten, C. S. (1998) Radiation-induced p53 and p21WAF-1/CIP1 expression in the murine intestinal epithelium. Am. J. Pathol. 153, 899 –909 Dilber, M. S., and Smith, C. E. (1997) Suicide genes and bystander killing: local and distant effects. Gene Therapy 4, 273–274 Paillard, F. (1997) Bystander effects in enzyme/prodrug gene therapy. Gene Therapy 4, 273–274 Narayanan, P. K., Goodwin, E. H., and Lehnert, B. E. (1997) Alpha particles initiate biological production of superoxide anions and hydrogen peroxide in human cells. Cancer Res. 57, 3963–3971 Mothersill, C., and Seymour, C. B. (1998) Cell– cell contact during gamma irradiation is not required to induce a bystander effect in normal human keratinocytes: evidence for release during irradiation of a signal controlling survival into the medium. Radiat. Res. 149, 256 –262 Prise, K. M., Belyakov, O. V., Folkard, M., and Michael, B. D. (1998) Studies of bystander effects in human fibroblasts using a charged particle microbeam. Int. J. Radiat. Biol. 74, 793–798 Hickman, A. W., Jaramillo, R. J., Lechner, J. F., and Johnson, N. F. (1994) Alpha-particle-induced p53 protein expression in a rat lung epithelial cell strain. Cancer Res. 54, 5797–5800 Ko, L. J., and Prives, C. (1996) p53: puzzle and paradigm. Genes Dev. 10, 1054 –1072 Bouvet, M., Ellis, L. M., Nishizaki, M., Fujiwara, T., Liu, W., Bucana, C. D., Fang, B., Lee, J. J., and Roth, J. A. (1998) Adenovirus-mediated wild-type p53 gene transfer down-regulates vascular endothelial growth factor expression and inhibits angiogenesis in human colon cancer. Cancer Res. 58, 2288 –2292 Favrot, M., Coll, J.-L., Louis, N., and Negoescu, A. (1998) Cell death and cancer: replacement of apoptotic genes and inactivation of death suppressor genes in therapy. Gene Therapy 5, 728 –739 Frank, D. K., Frederick, M. J., Liu, T.-J., and Clayman, G. L. (1998) Bystander effect in the adenovirus-mediated wild-type p53 gene therapy model of human squamous cell carcinoma of the head and neck. Clin. Cancer Res. 4, 2521–2527 Lin, J. H.-C., Weigel, H., Cotrina, M. L., Liu, S., Hansen, A. J., Hansen, T. W., Goldman, S., and Nedergaard, M. (1998) Gap-junction-mediated propagation and amplification of cell injury. Nat. Neurosci. 1, 494 –500 Chance, B., Sies, H., and Boveris, A. (1979) Hydroperoxide metabolism in mammalian organs. Physiol. Rev. 59, 527– 605 Halliwell, B., and Gutteridge, J. M. C. (1989) Free Radicals in Biology and Medicine, 2nd Ed, Clarendon Press, Oxford McGahon, A. J. Martin, S. J., Bissonnette, R. P., Mahboudi, A., Shi, Y., Mogli, R. J., Nishioka, W. K., and Green, D. R. (1995) The end of the (cell) line: methods for the study of apoptosis in vitro. Methods Cell Biol. 46, 153–185 Royall, J. A., and Ischiropoulos, H. (1993) Evaluation of 2⬘,7⬘dichlorofluorescein, and dihydrorhodamine 123 as fluorescent probes for intracellular H2O2 in cultured endothelial cells. Arch. Biochem. Biophys. 302, 348 –355 Morita, I., Smith, W. L., DeWitt, D. L., and Schindler, M. (1995) Expression-activity profiles of cells transfected with prostaglandin endoperoxide H synthase measured by quantitative fluorescence microscopy. Biochemistry 34, 7194 –7199 Haugland, R. P. (1996) Handbook of Fluorescent Probes and Research Chemicals (Spence, M. T. Z., ed) p. 491. Molecular Probes Europe BV, Leiden, The Netherlands Rigler, R. (1995) Fluorescence correlations, single molecule detection and large number screening. Applications in biotechnology. J. Biotech. 41, 177–186 Rigler, R., Pramanik, A., Jonasson, P., Kratz, G., Jansson, O. T., Nygren, P.-Å., Ståhl, S., Ekberg, K., Johansson, B.-L., Uhleń, S.,

APOPTOTIC CELLS KILL VIABLE NEIGHBORS

26. 27. 28.

29.

30. 31. 32.

33. 34. 35. 36.

37.

38.

39. 40.

41.

42.

43.

44.

45.

Uhleń, M., Jo¨rnvall, H., and Wahren, J. (1999) Specific binding of proinsulin C-peptide to human cell membranes. Proc. Natl. Acad. Sci. USA 96, 13318 –13323 Clark, P. J., and Evans, P. C. (1954) Distance to the nearest neighbor as a measure of spatial relationship in populations. Ecology 35, 445– 453 Collins, J. A., Schandl, C. A., Young, K. K., Vesely, J., and Willingham, M. C. (1997) Major DNA fragmentation is a late event in apoptosis. J. Histochem. Cytochem. 45, 923–934 Cavitelli, R., Beyer, E. C., Warlow, P. A., Robertson, A. J., Geist, S. T., and Steinberg, T. H. (1993) Connexin43 mediates direct intercellular communication in human osteoblastic cell networks. J. Clin. Invest. 91, 1888 –1896 Donahue, H. J. McLeod, K. J., Rubin, C. T., Anderson, J., Grine, E. A., Hertzberg, E. L., and Brink, P. R. (1995) Cell-to-cell communication in osteoblastic networks: cell line dependent hormonal regulation of gap junction function. J. Bone Mineral Res. 10, 881– 889 Lennon, S. V., Martin, S. J., and Cotter, T. G. (1991) Dosedependent induction of apoptosis in human cell lines by widely diverging stimuli. Cell Prolif. 24, 203–214 Nikotera, P., Bonfoco, E., and Brune, B. (1997) Mechanisms of nitric oxide-induced cell death: involvement in apoptosis. Adv. Neuroimmunol. 5, 411– 420 Rollet-Labelle, E., Grange, M.-J., Elbim, C., Marquetty, C., Gougerot-Pocidalo, M.-A., and Pasquier, C. (1998) Hydroxyl radical as a potential intracellular mediator of polymorphonuclear neutrophil apoptosis. Free Rad. Biol. 24, 563–572 Davies, K. J. A. (1999) The broad spectrum of responses to oxidants in proliferating cells: a new paradigm for oxidative stress. Iubmb Life 48, 41– 47 Sundaresan, M., Yu, Z-X., Ferrans, V. J., Irani, K., and Finkel, T. (1995) Requirement for generation of H2O2 for platelet-derived growth factor signal transduction. Science 270, 296 –299 Jacobson, M. D. (1996) Reactive oxygen species and programmed cell death. Trends Biochem. Sci. 21, 83– 86 ¨ ., Bartfai, T., and Rigler, R. Pramanik, A., Jureús, A., Langel, U (1999) Galanin receptor binding in the membranes of cultured cells measured by Fluorescence Correlation Spectroscopy. Biomed. Chromatogr. 13, 119 –120 Boonen, G., Pramanik, A., Rigler, R., and Ha¨berlein, H. (2000) Evidence for specific interactions between a kavain derivative and human cortical neurons measured by fluorescence correlation spectroscopy. Planta Med. 66, 7–10 Kravtsov, V. D. Daniel, T. O., and Koury, M. J. (1999) Comparative analysis of different methodological approaches to the in vitro study of drug-induced apoptosis. Am. J. Pathol. 155, 1327– 1339 Bedner, E., Li, X., Gorczyca, W., Melamed, M. R., and Darzynkiewicz, Z. (1999) Analysis of apoptosis by laser scanning cytometry. Cytometry 35, 181–195 van Dissel, J. T., Stikkelbroeck, J. J., Michel, B. C., Leijh, P. C., and van Furth R. (1987) Salmonella-typhimurium-specific difference in rate of intracellular killing by resident peritoneal macrophages from salmonella-resistant CBA and salmonellasusceptible C57BL/10 mice. J. Immunol. 138, 4428 – 4434 Maxeiner, H., Husemann, J., Thomas, C. A., Loike, J. D., El Khoury, J., and Silverstein, S. C. (1998) Complementary roles for scavenger receptor A and CD36 of human monocyte-derived macrophages in adhesion to surfaces coated with oxidized low-density lipoproteins and in secretion of H2O2. J. Exp. Med. 188, 2257–2265 Nathan, C. F. (1987) Neutrophil activation on biological surfaces. Massive secretion of hydrogen peroxide in response to products of macrophages and lymphocytes. J. Clin. Invest. 80, 1550 –1560 Ogino, T., Packer, L., and Maguire, J. J. (1997) Neutrophil antioxidant capacity during the respiratory burst: loss of glutathione induced by chloramines. Free Rad. Biol. Med. 23, 445– 452 Komarova, E. A., Diatchenko, L., Rokhlin, O. W., Hill, J. E., Wang, Z. J., Krivokrysenko, V.I,. Feinstein, E., and Gudkov, A. V. (1998) Stress-induced secretion of growth inhibitors: a novel tumor suppressor function of p53. Oncogene 17, 1089 –1096 Potten, C. S. (1981) Cell replacement in epidermis (keratopoiesis) via discrete units of proliferation. Intern. Rev. Cytol. 69, 271–318

1763

46.

47.

48. 49. 50. 51. 52. 53. 54.

55.

1764

Reznikov, K., Fu¨lo¨p, Z., Basco´, E., and Hajo´s, F. (1984) Mosaicism in the ventricular layer as the developmental basis of neocortical columnar organization: a 3H-thymidine study in newborn mice. Anat. Embryol. 170, 99 –105 Reznikov, K., and van der Kooy, D. (1995) Variability and partial synchrony of the cell cycle in the germinal zone of the early embryonic cerebral cortex. J. Comp. Neurol. 360, 536 – 554 Booth, C., and Potten, C. S. (1995) Cell proliferation effect of IL-11 on the growth of intestinal epithelial cells in vitro. Cell Prolif. 28, 581–594 Peunova, N., Kuzin, B., Roberts, I., O’Kane, C., and Enikolopov, G. (1996) Nitric oxide, cell multiplication, and cell survival. Cold Spring Harbor Symp. Quant. Biol. 61, 417– 426 Kuzin, B., Roberts, I., Peunova, N., and Enikolopov, G. (1996) Nitric oxide regulates cell proliferation during Drosophila development. Cell 87, 639 – 649 Turing, A. M. (1953) The chemical basis of morphogenesis. Phil. Trans. R. Soc. B237, 37–72 Edelstein-Keshet, L., and Erementrout, G. B. (1990) Contact response of cells can mediate morphogenetic pattern formation. Differentiation 45, 147–159 Kaneko, K., and Yomo, T. (1997) Isologous diversification: a theory of cell differentiation. Bull. Math. Biol. 59, 139 –196 Matsumoto, H. Hayashi, S., Hatashita, M., Ohnishi, K., Ohtsubo, T., Kitai, R., Shiuora, H., Ohnishi, T., and Kano, E. (1999) Nitric oxide is an initiator of intercellular signal transduction for stress response after hyperthermia in mutant p53 cells of human glioblastoma. Cancer Res. 59, 3239 – 3244 Simizu, S., Takada, M., Umezawa, K., and Imoto, M. (1998) Requirement of caspase-3(-like) protease-mediated hydrogen

Vol. 14

September 2000

56.

57.

58.

59.

60. 61. 62.

peroxide production for apoptosis induced by various anticancer drugs. J. Biol. Chem. 273, 26900 –26907 Johnson, T. M., Yu, Z.-X., Ferrans, V. J., Lowenstein, R. A., and Finkel, T. (1996) Reactive oxygen species are downstream mediators of p53-dependent apoptosis. Proc. Natl. Acad. Sci. USA 93, 11848 –11852 Goldkorn, T., Balaban, N., Shannon, M., Chea, V., Matsukuma, K., Gilchrist, D., Wang, H., and Chan, C. (1998) H2O2 acts on cellular membranes to generate ceramide signaling and initiate apoptosis in tracheobronchial epithelial cells. J. Cell Sci. 111, 3209 –3220 Hiraoka, W., Vazquez, N., Nieves-Neira, W., Chanock, S. J., and Pommier, Y. (1998) Role of oxygen radicals generated by NADPH oxidase in apoptosis induced in human leukemia cells. J. Clin. Invest. 102, 1961–1968 Dumont, A., Hehner, S. P., Hofmann, T. G., Ueffing, M., Droge, W., and Schmitz, M. L. (1999)Hydrogen peroxide-induced apoptosis is CD95-independent, requires the release of mitochondria-derived reactive oxygen species and the activation of NF-kappaB. Oncogene 18, 747–757 Olejnicka, B. T., Dalen, H., and Brunk, U. T. (1999) Minute oxidative stress is sufficient to induce apoptotic death of NIT-1 insulinoma cells. APMIS 107, 747–761 Polyak, K., Xia, Y., Zweier, J. L., Kinzer, K. W., and Vogelstein, B. (1997) A model for p53-induced apoptosis. Nature (London) 389, 300 –305 Qui, X.-B., Scho¨nthal, A. H., and Cadenas, E. (1997) Anticancer quinones induce pRb preventable G2/M cell cycle arrest and apoptosis. Free Rad. Biol. Med. 24, 848 – 854

The FASEB Journal

Received for publication October 8, 1999. Revised for publication March 6, 2000

REZNIKOV ET AL.