Mitotic Chromosomal Bcl-2: II. Localization to Interphase Nuclei

Abstract

We have previously shown, by immunofluorescence of fixed cells, that bcl-2 is found only in mitotic chromosomes in KB cultured human tumor cells expressing low levels of this oncoprotein. However, other studies showed that bcl-2 did not change its levels during the cell cycle when analyzed using Western blots. In this study we analyzed the distribution of bcl-2 during interphase, the point at which it is undetectable by immunofluorescence, using biochemical extraction, immunoprecipitation, and cell fractionation with Western blots. Interestingly, when carefully examined by immunofluorescence in fixed cells, the earliest point in the cell cycle showing bcl-2 localization was early G₂, in which bcl-2 could be found within the intact nucleus. In spite of showing no immunofluorescence reaction in fixed interphase cells, immunoprecipitation of gentle detergent extracts showed that bcl-2 from interphase cells reacted readily with the antibody used (#124) after extraction. However, immunoprecipitation using anti-bcl-2 followed by Western blots using anti-Bax showed that, unlike overexpressing cells, this bcl-2 was not complexed with Bax. Classical cell fractionation methods were used to separate nuclei from cytosol and cell membranes. Surprisingly, these experiments clearly showed that essentially all of the bcl-2 in interphase KB cells was present in the nucleus. Therefore, the lack of reaction in fixed cells with anti-bcl-2 antibody reflects either a masking or a conformational change of the reactive epitope in bcl-2 present within the nucleus. By correlation, this change may be related to the phosphorylation of bcl-2 that occurs just before mitosis. The nature of this novel yet highly conserved nuclear form of bcl-2 and the understanding of its function will require further study.

THE ACCOMPANYING ARTICLE (Schandl et al. 1999) describes the immunocytochemical detection of bcl-2 attached to mitotic chromosomes (Lu et al. 1994; Willing-ham and Bhalla 1994). This article also shows that KB cells contain a low level of bcl-2 that is detectable in fixed cells only in G₂/M, yet this protein is present at constant levels during the entire cell cycle when examined by Western blots. Furthermore, the bcl-2 seen in intact fixed cells could also be found when mitotic chromosomes were isolated from KB cells, and Western blots of these chromosomes showed the same hyperphosphorylated bands seen for bcl-2 in whole cell extracts of cells arrested in mitosis. A major question raised in this companion study was the location of bcl-2 in such cells in other phases of the cell cycle, when it is undetectable in fixed cells using immunocytochemical techniques. This article addresses this question through the use of classical cell fractionation methods. These experiments reveal the unexpected and unprecedented result that bcl-2 in such interphase cells is actually a nuclear protein.

Materials and Methods

Immunofluorescence, Immunoblotting, Antibodies, and Drug Treatments

Immunofluorescence detection of bcl-2 was performed as previously described (Schandl et al. 1999) using acetone fixation and MAb #124. Immunoblotting using chemiluminescence detection was performed as previously described (Schandl et al. 1996,1999). For primary antibodies from rat or rabbit, goat anti-rat or goat anti-rabbit IgG, respectively, conjugated to horseradish peroxidase (Jackson ImmunoResearch; West Grove, PA) was utilized for the second step. Anti-Bax antibody was #sc-493 (a rabbit polyclonal) from Santa Cruz Biotechnology (Santa Cruz, CA). Anti-tubulin was a rat monoclonal antibody (MAb) (#YL1/2) from Accurate Chemical (Westbury, NY). Cells were treated where indicated (e.g., Figure 2) with nocodazole (Sigma; St Louis, MO) at 0.3 μg/ml overnight.

Cell Extracts

Cell extracts were prepared as previously described (Schandl et al. 1996,1999). Insoluble fractions were pelleted and the protein was extracted from them as described below. In the soluble fractions, protein was first precipitated with trichloroacetic acid (TCA) and then resuspended in an equal volume of 3 M Tris base and protein extract buffer in a variation of a published protocol (Wang et al. 1996). Final amounts of protein extract were adjusted such that they represented an equal number of cells per volume of buffer as determined by hemocytometric count.

Immunoprecipitation

KB cells were scraped from the surface of flasks and then washed with PBS. Cells were then resuspended in cytoplasmlike buffer (CLB: 142.5 mM KCl, 5 mM MgCl₂, 10 mM HEPES, pH 7.2, 1 mM EGTA, 1 μg/ml leupeptin, 0.02% aprotinin, and 20 μg/ml PMSF) containing either 0.2% Triton X-100 or 0.2% Nonidet P40 as described in the text. CLB-cell suspensions were incubated in ice on a rocker for 30 min, vortexed briefly, and centrifuged to remove insoluble material (4C for 10 min at 14,000 × g). Whole cell extracts were then precleared by incubation with pansorbin (Calbiochem; LaJolla, CA) alone for 30 min on ice. For each sample, 100 μl of precleared protein extract was incubated on ice for 60 min with 20 μl of mouse anti-bcl-2 MAb #124 (initial concentration 270 μg/ml). Simultaneously, 50 μl pansorbin was incubated on ice with 50 μl (1.8 mg/ml) of affinity-purified rabbit anti-mouse IgG (Jackson ImmunoResearch). The contents of both incubations were then combined and incubated on ice for an additional 60 min. The pansorbin-antibody-protein complexes were then washed three times with CLB containing the appropriate nonionic detergent and stored in CLB at −20C. Bound proteins and antibodies were eluted from the pansorbin by boiling for 10-15 min in SDS sample buffer containing 5% β-mercaptoethanol. Supernatant was retained after centrifugation (3000 × g for 5 min) and mixed 1:1 with protein sample buffer before analysis by SDS-PAGE followed by immunoblotting as described.

Cell Fractionation

Ten flasks of KB cells were trypsinized and the cells were washed twice with Dulbecco's PBS (with Ca⁺⁺ and Mg⁺⁺). The cells were collectively resuspended in 2 ml of a hypotonic solution and incubated for 150-180 min on ice [40 mM KCl, 0.2 mM spermine, and 0.5 mM spermidine containing protease inhibitor cocktail at the recommended concentration (Boehringer Mannheim; Indianapolis, IN)]. The cells were then lysed with a tight-fitting Dounce homogenizer by 10-25 gentle strokes. Fractionation was effected by differential centrifugation similar to the methods of Evans (1978) optimized for the fractionation of KB cells. A fraction consisting mostly of nuclei, but also containing some whole cells and aggregates, was removed from the lysate by centrifugation (50 × g for 3 min) and is referred to as nuclear pellet #1. From the clarified lysate, nuclei were isolated with a second centrifugation at 500 × g for 5 min (nuclear pellet #2). Nuclei were then resuspended in a nuclear stabilization buffer [0.25-0.32 M sucrose with 1-2 mM CaCl₂ or MgCl₂ and Complete Protease Inhibitor Cocktail (Boehringer Mannheim)] and recentrifuged at 500 × g for 5 min. Purified nuclei were either removed for protein extraction or resuspended in nuclear stabilization buffer for further analysis. Cytoplasmic fractions obtained from the supernatant (termed “cytoplasm”) of the first centrifugation at 500 × g were clarified by a 600 × g centrifugation (5 min). Cytosolic proteins (supernantant was termed “cytosol”) were separated from membranous structures (pellet was termed “membrane”) with a final centrifugation of the cytoplasmic fraction at 14,000 × g (20 min). Protein extract was obtained from each fraction as described above. Slides were generated from each fraction using the Cytospin centrifugation apparatus (Shandon; Pittsburgh, PA). The absence of nuclei in all except nuclear fractions was verified. Cytospin preparations were fixed in a 3.7% formaldehyde solution and stained with hematoxylin for morphological analysis.

Results

The Earliest Appearance of bcl-2 in KB Cells During the Cell Cycle

We have previously shown that in KB cells there is essentially no detectable reaction of MAb #124 in fixed cells in interphase (Willingham and Bhalla 1994). Because one can distinguish the various stages of mitosis on morphological grounds, as well as the early nuclear changes seen before prophase in G₂ cells before the loss of the nuclear envelope, we carefully examined a large number of randomly growing KB cells for evidence of G₂ nuclei using phase-contrast microscopy. Condensation of chromatin in enlarged nuclei is a hallmark feature of this stage, and >100 examples of this morphology were found. The later stage of extensive chromatin condensation in G₂ has been demonstrated previously, in which the bcl-2 pattern is relatively diffuse in the entire G₂ nucleus (Willingham and Bhalla 1994). However, a few examples were found that appeared to be early steps in this process, one of which is shown in Figure 1. The bcl-2 pattern shown is not diffuse within the nucleus but is concentrated around nucleoli. Whether this is the product of entry of bcl-2 into the nucleus or the unmasking of existing nuclear bcl-2 cannot be determined from these images, but it does suggest the possibility that bcl-2 is located in the nucleus before this point, a possibility that is without precedent in the literature.

Reactivity of bcl-2 with MAb #124 Without Denaturation

One assumption about the ability of MAb #124 to detect bcl-2 on Western blots of KB cell extracts from interphase cells was that the bcl-2 was denatured by exposure to SDS and might not react with this antibody in its native state. To address this question, we prepared cell extracts from KB cells using gentle detergent extraction in the absence of SDS, and immunoprecipitated the bcl-2, detecting its presence later using SDS denaturation and Western blot analysis. Randomly growing cells mainly represent cells in G₁- and S-phase, as determined from the analysis of flow cytometric data using propidium iodide labeling of DNA content. Cells treated with nocodazole can be >80% arrested in G₂/M. Figure 2 clearly shows that MAb #124 specifically reacts with bcl-2 from these cells in both its native and denatured conformations, and without the requirement for the hyperphosphorylation that occurs at mitosis. This raises the possibility that bcl-2 is masked in fixed cells by interaction with some other component that is presumably removed by detergent extraction, yet it may also be unmasked even in fixed cells if it becomes hyperphosphorylated. This leads to the speculation that these two events might be linked in intact cells.

Figure 1

Detection of the earliest appearance of bcl-2 in G₂ nuclei by immunofluorescence. The morphology of G₂ nuclei can be detected by phase-contrast microscopy, characterized by chromatin clumping and condensation in enlarged nuclei (B). Cells identified in this way were also examined for bcl-2 using immunofluorescence with MAb #124 (A). Unlike the more diffuse pattern of bcl-2 seen in nuclei later in G₂, these early G₂ nuclei show clustered bcl-2 localization that appears to be located around nucleoli (arrow). Other cells in interphase also show a very weak speckled pattern with this antibody, but the level of reaction is too low to be certain that it represents a pattern specific for bcl-2, even though the controls in which MAb #124 is deleted do not show it. Bar = 30 μm.

Association of bcl-2 with Bax

Bax is an apoptosis-controlling cytoplasmic protein of #21 kD that forms heterodimeric complexes with bcl-2 (Oltvai et al. 1993). Most studies of Bax-bcl-2 complexes have been performed in cells that overexpress large amounts of bcl-2 in the cytoplasm. Because we speculated that bcl-2 might be complexed with another protein, Bax was a likely candidate. We performed immunoprecipitations using MAb #124 with extracts of KB cells, as well as with GM697-bcl-2 cells, a cell type known to contain high levels of cytoplasmic bcl-2. We then analyzed these immunoprecipitates for the presence of Bax using a commercially available anti-Bax antibody on Western blots; the results are shown in Figure 3. As expected, the precipitated bcl-2 from GM697-bcl-2 cells contained large amounts of Bax, comparable to the total Bax seen when total cell extracts were analyzed by direct Western blots. KB cells showed a similar band of Bax in direct Western blots of whole cell extracts, yet the anti-bcl-2 immunoprecipitate showed no evidence of Bax (asterisk in Figure 3). This suggested that bcl-2 in KB cells was not complexed with Bax. One explanation for this result is that the two proteins are contained within different compartments in the cell.

Cell Fractionation of KB Cells

Because immunocytochemical methods were apparently unable to detect the bcl-2 present in fixed inter-phase KB cells, perhaps due to inaccessibility of the antigen, and because we knew that we could detect bcl-2 present in cell extract fractions using Western blots, we resorted to a classical cell fractionation scheme with cell extracts to try to determine roughly where this interphase bcl-2 might be located. The fractionation scheme we devised using differential centrifugation is shown in Figure 4. We monitored the efficiency of the homogenization using phase-contrast microscopy and the purity of the resultant fractions morphologically. These procedures produced relatively pure nuclear fractions in the first two spins with very few intact cells, and the cytoplasmic fraction was completely free of nuclei (results not shown).

Figure 2

Immunoprecipitation/Western blot detection of bcl-2 from KB cells extracted in Triton X-100 detergent. Randomly growing or mitotically arrested (nocodazole-treated) KB cells were extracted using Triton X-100 as described in Materials and Methods. These cell extracts were then subjected to immunoprecipitation using MAb #124, together with rabbit anti-mouse IgG attached to pansorbin. The immunoprecipitate pellets were then washed free of nonreactive cell components and subjected to SDS extraction and gel electrophoresis, followed by Western blotting using MAb #124 and an anti-mouse chemiluminescence detection system (Immp/Western lanes). Other samples of whole cell extracts in SDS were subjected to gel electrophoresis and immunoblotting, typical of a standard Western blot protocol (Western). Note that in the “blank” control lanes on the right side of this figure, no bands are evident on the whole cell extract Western in either random cell extracts (R) or in nocodazole-treated mitotically arrested cell extracts (N). In the adjacent Immp/Western lanes, bands corresponding to the heavy and light chains of MAb #124 present in the solid phase pellet are detected because of the anti-mouse IgG chemiluminescence detection system. Note that the light chain of MAb #124 (#30 kD) runs at a position higher than the bcl-2 bands we wished to detect. At left, the direct Western blots of whole cell extracts show the single band at 26 kD of bcl-2 (long arrow) present in randomly growing cells, and the adjacent lane shows the hyperphosphorylated bands of bcl-2 derived from mitotically arrested cells. In the immunoprecipitate analyzed by Western blot (Immp/Western), the bcl-2 band at 26 kD is clearly evident in the random cell extract, as well as the normal mobility and hyperphosphorylated bands of bcl-2 present in the Triton X-100 extract of mitotic cells (short arrows). These images clearly show that the bcl-2 extracted by Triton X-100 can react specifically in solution with MAb #124 before SDS treatment. Similar results (not shown) were obtained using NP-40 or Nonidet P-40 in place of Triton X-100.

Figure 3

Detection of Bax complexed with immunoprecipitated bcl-2. KB or 697-bcl-2 cells were extracted with detergent and subjected to immunoprecipitation with MAb #124 followed by Western blotting as shown in Figure 2, except that the Western blot utilized a rabbit polyclonal antibody to Bax as the first step, followed by an anti-rabbit chemiluminescence detection system. In addition, direct Western blots of SDS cell extracts were prepared (Cell Extract). The bcl-2 immunoprecipitates from 697-bcl-2 cells showed a specific Bax band at #21 kD, similar to that seen in the whole cell extract (arrow). The whole cell extract of KB cells also showed a Bax band. However, the bcl-2 immunoprecipitate from KB cells showed no specific Bax band (asterisk), suggesting that bcl-2 and Bax were not complexed in KB cells.

Figure 4

Differential centrifugation cell fractionation scheme. The details of this method are described in Materials and Methods.

As a marker for the cytoplasmic fraction, we chose tubulin, against which a highly specific commercial MAb is available (Wehland and Willingham 1983). As a control for the ability of this method to identify cytoplasmic bcl-2 (which is usually considered to be bound to cytoplasmic membranes such as mitochondria and ER), we used the MF2-bcl-2 cell line, a hamster cell that had been transfected with a human bcl-2 cDNA construct (shown by immunofluorescence in Figure 1B' of the accompanying article; Schandl et al. 1999). Figure 5A shows both the total proteins in each of these fractions derived from equal numbers of cells (amido black staining), and the Western blots using either anti-tubulin or anti-bcl-2. As can be seen in this figure, tubulin is present, as expected, mostly in the soluble cytoplasmic fraction (Figure 5A, Lane F). Bcl-2, on the other hand, is also cytoplasmic (Figure 5A, Lane D), but mainly associated with cytoplasmic membranes (Figure 5A, Lane E, arrow), as expected. Figure 5B demonstrates the results from a similar experiment using randomly growing KB cells. The total protein stains and the distribution of tubulin are similar to those shown in MF2-bcl-2 cells (Figure 5A). However, the bcl-2 distribution is dramatically different. Most of the bcl-2 in these randomly growing cells appears in the nuclear fractions (arrow in Figure 5B).

Because randomly growing cells are a mixture of interphase and mitotic cells (mostly interphase), we conducted a further experiment using KB cells that had been arrested at the G₁/S interface to remove the contribution of mitotic cells (Figure 6). This experiment showed that tubulin fractionates as a cytoplasmic (mostly cytosolic) protein, as expected, but that bcl-2 is exclusively nuclear. This is in stark contrast to the results shown in Figure 5A for the cytoplasmically distributed bcl-2 in MF2-bcl-2 cells. Therefore, bcl-2 in interphase KB cells is a nuclear protein, in agreement with the earliest distribution seen by immunofluorescence shown in Figure 1.

Figure 5

Analysis of cell fractions using Western blotting from randomly growing cells. Fractions generated using the scheme shown in Figure 4 from randomly growing MF2-Bcl2 (A) or KB (B) cells were examined using immunoblotting for tubulin (anti-tubulin) or for bcl-2 (anti-bcl-2). The total proteins in these fractions were also revealed using amido black staining. Whole cell extract (Lane A), nuclear fractions 1 and 2 (Lanes B and C), cytoplasm (Lane D), cytoplasmic membranes (Lane E) and cytosol (Lane F) were examined. Anti-tubulin demonstrated in both cell types that it was located in the cytosol. Anti-bcl-2 in MF2-bcl-2 cells showed that bcl-2 was mainly present in cytoplasmic membranes (arrow in A). Anti-bcl-2 in KB cells showed that bcl-2 was present mainly in the nuclear fractions from KB cells (arrow in B), and very little could be found in the cytoplasm.

Figure 6

Analysis of cell fractions using Western blotting from KB cells synchronized at the G₁/S interface. KB cells were synchronized using a double thymidine block as described in Materials and Methods and then fractionated according to the scheme shown in Figure 4. The cell fractions were then immunoblotted using anti-tubulin or anti-bcl-2. This figure shows that essentially all of the bcl-2 was found in the nuclear fractions (arrow).

Discussion

Bcl-2 was initially demonstrated in cell types that express high levels of the protein (Tsujimoto and Croce 1986). In these cells, bcl-2 showed the ability to inhibit apoptotic cell death, probably at the surface of mitochondria (Reed 1997). Most subsequent studies of this protein have utilized the same cell types that express high levels of bcl-2. Many cell types were believed to contain no bcl-2 when examined for bcl-2 mRNA using Northern blot analysis. However, interest increased in some of these cell types that express low levels of bcl-2 when it was realized that their bcl-2 underwent specific hyperphosphorylation in response to mitotic arrest agents, such as taxol (Haldar et al. 1994; Schandl et al. 1996). Because this was associated with subsequent apoptotic cell death, it was proposed that this hyperphosphorylation inactivated the apoptosis-inhibiting function of bcl-2 (Haldar et al. 1995). However, we have shown in an earlier report that this hyperphosphorylation is actually a cell cycle-regulated event that occurs just before normal mitosis and has no effect on subsequent apoptotic death in synchronized cells (Schandl et al. 1996). In addition, the levels of bcl-2 in such cells are well below those shown to affect apoptosis when expressed diffusely in the cytoplasm. This therefore suggests that the significance of various levels of bcl-2 may need to be reevaluated.

Mitotic Chromosomal bcl-2

Earlier studies from our laboratory and others showed that bcl-2 was detectable in association with mitotic chromosomes by immunocytochemistry (Lu et al. 1994; Willingham and Bhalla 1994). Our accompanying article (Schandl et al. 1999) shows that this may be a ubiquitous phenomenon seen in essentially all growing cells. The levels of bcl-2 in KB cells are #1/1000 of the levels seen in leukemic cells or in cells in which bcl-2 has been shown to be operative in inhibiting apoptosis. The location of bcl-2 in cells in which it affects apoptosis is mainly cytoplasmic, and this is in keeping with the location of the apoptotic machinery that includes proteases and mitochondria. At present, the actual molecular mechanism by which bcl-2 inhibits apoptosis in such cells is not clear, although the surface of the mitochondrion is a likely site for this action (Reed 1997; Yang et al. 1997). Nuclear functions and location for bcl-2 are unprecedented, except for the prior demonstration of bcl-2 in G2 nuclei just before mitosis (Lu et al. 1994; Willingham and Bhalla 1994). The clarification of this localization in cells expressing low levels has been hampered by the lack of available antibodies that react specifically with bcl-2 in fixed cells. MAbs #124 and #100 are unique in that they react with an accessible epitope in bcl-2 when it is diffusely distributed in the cytoplasm, as well as when it is hyperphosphorylated during mitosis. When extracted by detergent or denatured by SDS, these antibodies react readily with bcl-2. The reactive epitope for these antibodies resides in the amino acid sequence 41-54 near the N-terminus of bcl-2α and −β (Pezzella et al. 1990). These two known isoforms differ by a short substitution of sequence at their C-termini (Tsujimoto and Croce 1986). By analogy with a similar apoptosis-inhibiting member of the bcl-2 family, bcl-XL, for which the crystal structure has been derived (Much-more et al. 1996), this reactive epitope lies in a large unstructured loop of the protein, far removed from the BH (bcl-2 homology) domains that interact with other members of the bcl-2 family (such as Bax) to form heterodimers (Yin et al. 1994). Interestingly, this loop also contains several serine residues that are possible phosphorylation sites, the bulk of the phosphorylation of bcl-2 being known to occur on serine residues (Alnemri et al. 1992; Haldar et al. 1994) in this loop region (Chang et al. 1997).

Is This Mitotic Chromosomal Antigen Really bcl-2?

A major concern in our studies was the actual identity of the antigen reactive with MAb #124. Careful comparison of molecular mobility with authentic bcl-2 introduced through transfection of cloned cDNA shows that the reactive protein is essentially identical in size (Schandl et al. 1999). Western blots do not reveal other proteins that are reactive with this antibody, and the epitope is highly species-specific. Analysis of the proteins present in isolated chromosomes shows that this antigen is quantitatively recovered together with chromosomes and has the same patterns of hyperphosphorylation seen in intact cells (Schandl et al. 1999). Studies are under way to examine whether cell types can be found that would allow transfected bcl-2 to attach to chromosomes in this same fashion. Some preliminary results suggest that there are some circumstances where this can happen, but the control of this process is complex. Other studies are also under way examining post-translational modifications of bcl-2 in these cells, such as phosphorylation. Another possibility is that the bcl-2 present in these cells is actually an alternate isoform, perhaps resulting from a novel splice variant of bcl-2 mRNA. Preliminary results have not ruled out this possibility.

A Hypothetical Model for bcl-2 Accessibility to Antibodies

Although the reactivities of bcl-2 in different cellular compartments may appear confusing, the specificity and sensitivity of immunocytochemistry may reveal molecular details of bcl-2 interactions not previously identified. A model consistent with these observations is presented in Figure 7. In this hypothesis, the cytoplasmic compartment contains components, such as Bax, that can heterodimerize with bcl-2 at the BH domains, but these complexes leave the unstructured loop region containing the a.a. 41-54 epitope free to react with MAb #124. Therefore, bcl-2 in the cytoplasm is always detectable using antibodies to the loop region. In the nucleus, however, there may be other nuclear proteins (X) that interact selectively with the loop region of bcl-2. When complexed with (X), the loop region is inaccessible to antibodies such as #124. During G₂, bcl-2 in the nucleus (and perhaps also in the cytoplasm) undergoes hyperphosphorylation at sites in the loop region, leading to the dissociation of (X) and making the loop epitopes accessible to antibody. This association may also be released by treatments with detergents, such as Triton X-100, resulting in the reactivity of nuclear bcl-2 with antibodies in immunoprecipitation experiments using detergent extracts (such as shown in Figures 2 and 3). At the end of mitosis, when bcl-2 is dephosphorylated, nuclear bcl-2 can then reassociate with (X), leading to inaccessibility of the loop epitopes in fixed cells.

The ability of MAbs to detect interactions between specific domains of proteins using immunocytochemistry has been demonstrated previously (e.g., Ashizawa et al. 1991). At least one candidate protein for interaction with bcl-2 and nuclear components has been isolated (Naumovski and Cleary 1996). The identification of the specific protein(s) responsible for the inaccessibility of the loop region of nuclear bcl-2 will clearly require further study.

Figure 7

Hypothetical model for differential accessibility of bcl-2 loop epitopes in the nucleus and the cytoplasm.

Does bcl-2 Have a Second Function in the Nucleus?

Bcl-2 is a member of a highly conserved family of genes that vary only slightly in structure between nematodes and humans (Reed 1997). Therefore, bcl-2 is likely to have a central function in animal cell physiology. It is also possible, and is not without precedent, that such a highly conserved gene may have more than one function. Overexpression of bcl-2 is believed to be important in development and in differentiated cell function, probably through its effects on apoptotic thresholds. However, the ubiquitous distribution of chromosomal bcl-2 in growing cell types and its unique intracellular distribution suggest that bcl-2 may have another role in cell physiology. Some studies have suggested that bcl-2 may have a second function in the regulation of cell cycle progression, and that the domains of the bcl-2 molecule required for that function are different from those that inhibit apoptosis (Huang et al. 1997).

Protection from apoptosis at high levels of bcl-2 expression may involve mechanisms at the surface of mitochondria. During the cell cycle, however, the cell undergoes a dramatic change in organization, in that the nuclear contents are exposed to the cytoplasm in mitosis. Perhaps, this mixing of compartments could trigger the sensitive apoptotic machinery, were it not for a mechanism that locally protects the components of these otherwise foreign environments. It is well known that agents that prolong mitosis, usually through interference with microtubule function, can eventually lead to apoptotic cell death directly from mitosis (Collins et al. 1997). Perhaps the cell contains a trigger for apoptosis that is blocked by a protective mechanism for brief periods of mitosis but is released to induce cell death if the cell detects problems with the completion of mitosis. Perhaps, also, chromosomally attached bcl-2 could play such a protective role, being present in high concentrations at the interactive surface between chromosomes and cytoplasmic components. Recently, Fang et al. (1998) have shown that a bcl-2 mutant in which the phosphorylated “loop” region has been deleted protects cells from apoptosis caused by DNA-damaging agents but not by agents, such as taxol, that cause apoptosis after mitotic arrest. This might suggest that the bcl-2 on mitotic chromosomes may be localized at that site due to phosphorylation, and that this localization is required for protection from apoptosis during mitotic arrest. The resolution of these questions clearly will require further study.

Conclusions

This article and the accompanying article (Schandl et al. 1999) provide evidence that KB and other cell types contain a stable but low level of a form of bcl-2 that attaches to chromosomes during mitosis and is located in the cell nucleus during interphase. As previously shown, the phosphorylation of this form of bcl-2 and its reactivity with MAb #124 are correlated with the G₂- and M-phases of the cell cycle. Whether this form of bcl-2 plays any role in the control of apoptosis or the survival of cells during the cell cycle is unclear, but its presence in all growing human cells examined thus far suggests that its role is important.