Abstract
Bcl-2 is present in a cytoplasmic distribution in cells that express high levels of this oncoprotein. In contrast, using immunocytochemistry in cells expressing low levels of bcl-2, such as KB human carcinoma cells, we and others have shown that bcl-2 is present on the surface of mitotic chromosomes. However, monoclonal antibodies reactive with an epitope representing amino acids 41-54 of the bcl-2 sequence did not detect bcl-2 in other phases of the cell cycle. This study extended those earlier findings to determine if bcl-2 was expressed as a cyclin or if this pattern was an artifact of immunocytochemistry. Immunofluorescence studies in several other human cell lines showed the same mitotic distribution of bcl-2. Other studies using flow cytometry also showed selective mitotic phase detection of bcl-2. A comparison of available commercial antibodies showed that, in spite of reactivity with denatured bcl-2 on Western blots, clear reactivity with bcl-2 in fixed cells was found only with those reactive with the (a.a. 41-54) epitope. With RNase protection and Western blot analyses, cells synchronized at various stages of the cell cycle showed constant levels of bcl-2 mRNA and protein. Analysis of bcl-2 using Western blots showed a band with the same apparent molecular weight as that seen in comparison with authentic bcl-2 overexpressed in the cytoplasm. The retention of bcl-2 on chromosomes in unfixed, permeabilized preparations was influenced by protease treatment, phosphate, and pH. Studies using isolated chromosomes showed that much of the bcl-2 in these cells was attached to chromosomes in mitosis, had the expected molecular weight, and was phosphorylated in the same manner as that seen in whole-cell extracts. These results show that bcl-2 is not a cyclin and that the bcl-2 localized on chromosomes is the same molecule seen by immunoblotting. These results suggest that the reactive (a.a. 41-54) epitope present in bcl-2 is somehow modified or masked in interphase.
Bcl-2 is a small intracellular nonglycosylated protein that was originally identified as the product of a gene located near the 14:18 chromosomal translocation breakpoint found in human follicular lymphoma (Tsujimoto and Croce 1986). Subsequently, bcl-2 was shown to be capable of inhibiting the apoptotic pathway when overexpressed in cells, a death pathway that was blocked at a central point common to many different inducers (Nunez et al. 1990; Sentman et al. 1991; Miyashita and Reed 1993). Processing of the primary transcript of the bcl-2 gene, normally located on human chromosome 18, leads to the generation of two alternate transcripts that code for two isoforms of this protein, a and β. α is the commonly detected isoform, with a molecular weight of #26 kD, whereas β-protein is rarely detected in intact cells, with a molecular weight of #21 kD (Tsujimoto and Croce 1986). Several other genes have since been identified that have similarities to bcl-2, most of which have some role in controlling apoptotic thresholds in cells. Among the many members of this bcl-2 family of genes are Bax (Oltvai et al. 1993), Bcl-X (Boise et al. 1993), and Bad (Yang et al. 1995), which variously heterodimerize with other family members dependent on specific domains within the protein that share homologies with bcl-2 (BH domains). The striking ability of bcl-2 to block apoptosis in most cells has led to intense interest in its mechanism of action (Yang and Korsmeyer 1996). In spite of this interest, however, the exact molecular mechanism involved is still not clear. It is also not clear whether bcl-2 has any other functions in cells beyond its inhibition of apoptosis (Reed 1997).
Most of the studies of bcl-2 have been performed in cells that express high levels of the protein (e.g., Alnemri et al. 1992). In an attempt to understand its function through detecting its location, several studies have demonstrated that bcl-2 is a cytoplasmic protein in these overexpressing cells, and many have suggested an association with the cytoplasmic face of intracellular membranes, presumably mediated by the hydro-phobic domain present at the C-terminus of the α-isoform (Monaghan et al. 1992). Studies in cells that do not overexpress bcl-2 have been more limited. Our laboratory and others have previously shown that the distribution of bcl-2 in such low-expressing cell types is not simply diffuse in the cytoplasm. These studies showed that bcl-2 specifically associates with mitotic chromosomes and that the distribution of bcl-2 in interphase cells was unclear because it was undetectable by immuncytochemical methods in fixed cells (Lu et al. 1994; Willingham and Bhalla 1994). The immunocytochemical studies previously reported mainly utilized monoclonal antibodies (MAbs) that recognize a unique epitope in human bcl-2 near the N-terminus, generated through the immunization of a mouse with a synthetic peptide encompassing amino acids 41-54 in the human bcl-2 sequence (Pezzella et al. 1990). Two such MAbs (#124 and #100) are commercially available. By analogy with the deduced crystal structure of a related protein, BC1-XL, this epitope lies within a large unstructured loop with an unknown function (Much-more et al. 1996). This loop domain has also been implicated as a site for phosphorylation within bcl-2 (Chang et al. 1997).
This article and the accompanying article (Schandl et al. 1999) report further studies into the distribution of bcl-2 in low-expressing cells, especially the question of the location of bcl-2 in these cells during interphase. The results in this and previous articles (Schandl et al. 1996) demonstrate that bcl-2 protein and mRNA levels do not vary with the cell cycle and that bcl-2 remains associated with isolated mitotic chromosomes.
Materials and Methods
Cell Culture
Cells were obtained from American Type Culture Collection (ATCC; Rockville, MD) unless otherwise specified and were cultured continuously in RPMI-1640 (OVCAR-3, SK-N-SH, MDA-MB-231, MCF7, GM697/neo, GM697/bcl-2, PC-3, DU145, LNCaP, WI-38, Detroit 532, Swiss 3T3, NIH3T3, MF2/BCL2) or Dulbecco's modified Eagle's (KB, HeLa) medium supplemented with 10% fetal bovine serum (FBS) and containing 50 U/ml penicillin G and 50 μg/ml streptomycin (P-S). Media and supplementation were obtained from Irvine Scientific (Santa Ana, CA), Gibco/BRL Laboratories (Grand Island, NY), or Sigma (St Louis, MO) as available. Cell cultures were incubated at 37C with an atmosphere containing 5% CO2/95% air. MF2/BCL2 cells were generated by lipofection of DDT1-MF2 hamster leiomyosarcoma cells (Fan et al. 1993) with human bcl-2-α cDNA under the control of a CMV promoter (to be presented elsewhere) and were grown continuously in RPMI medium supplemented with FBS and P-S as above and containing F-12 nutritional supplement. GM697/neo and GM697/bcl-2 human leukemia cell lines were a gift from Dr. John Reed (Miyashita and Reed 1993; Willingham and Bhalla 1994), and phytohemagglutinin (PHA)-stimulated peripheral blood monocyte metaphase chromosome spreads and primary human amniotic cells were generously provided by Dr. Eduardo Cantu (Medical University of South Carolina). All cells were grown in T-75 cell culture flasks (Corning Glass Works; Corning, NY) unless otherwise specified. All cells formed monolayers except the human leukemia cell lines (GM697/neo and GM697/bcl-2), which were grown in suspension. To subculture, adherent cells were treated with 2 mg/ml trypsin containing 0.1% ethylenediaminetetraacetic acid (EDTA; Sigma).
Antibodies
Primary antibodies used for immunofluorescence and/or Western blot analysis were obtained from commercial sources. Mouse MAbs to bcl-2 were MAb #124 (Dako; Glostrup, Denmark), MAb #100 (Zymed; So. San Francisco, CA), #4D7 (Pharmingen; San Diego, CA), #6C8 (Pharmingen), and #sc-492 (Santa Cruz Biotechnology; Santa Cruz, CA). Rabbit polyclonal anti-bcl-2 was #PC68 (Calbiochem; La Jolla, CA).
Immunofluorescence
Immunofluorescence for bcl-2 was performed at 23C as previously described (Willingham and Bhalla 1994). Briefly, cells cultured in 35-mm dishes were washed and fixed with 80% acetone/20% dH2O at 23C. After fixation for 10 min, nonspecific antibody binding sites were blocked with 1% BSA-PBS before cell incubation in primary antibody. The primary antibodies were prepared at a concentration of 10 μg/ml in a 1% BSA-PBS solution. Antibody binding was detected with a secondary antibody, affinity-purified goat anti-mouse IgG conjugated to rhodamine (Jackson ImmunoResearch; West Grove, PA) at a concentration of 25 μg/ml. The signal was further amplified with a 30-min incubation in a tertiary antibody, affinity-purified rabbit anti-goat IgG conjugated to rhodamine (Jackson ImmunoResearch), which was prepared at 25 μg/ml in 1% BSA-PBS. Each antibody incubation transition was preceded by three washes in PBS and a brief incubation in 1% BSA-PBS. Cells were postfixed with 3.7% formaldehyde prepared in PBS for 15 min and were coverslipped under 90% glycerol/10% PBS. The fluorescence signal was detected using a Zeiss Axioplan fluorescence microscope with a 546-nm excitation filter and LP 590-nm barrier under a X40 or X100 (N. A. 1.3). PlanNeofluar objective or a X63 (N. A. 1.4) Planapochromat objective.
Flow Cytometric Detection of DNA Content
Two × 106 washed cells were fixed in 1% formaldehyde for 15 min on ice. Then they were washed with PBS, centrifuged, and resuspended in 1 ml PBS. To dehydrate, 9 ml ice-cold 80% ethanol was added to the cell-PBS suspension and cells were then stored at −20C. To determine the DNA content of the cells, they were washed twice with PBS and once with 1% BSA-PBS, then resuspended in 1 ml of propidium iodide stain solution [100 μg propidium iodide, 100 Kunitz units RNase (Sigma) and 10 ml PBS]. They were incubated in this solution for 30 min at room temperature (RT) in the dark. Cells were analyzed using a Coulter Epics XL-MCL flow cytometer and the data were further analyzed using Multicycle software.
Flow Cytometric Detection of bcl-2
Cells arrested at the G1/S boundary of the cell cycle as described below, were released and collected at G1/S plus 0, 2, 4, 6, 8, 10, 12, or 24 hr. They were trypsinized for removal from flasks and washed with PBS. The cell pellet was fixed (2-3 × 106 cells) in 1% formaldehyde for 15 min on ice, then washed with PBS and dehydrated in ice-cold 80% ethanol/20% dH2O. To rehydrate, cells were washed twice in PBS, then once in 1% BSA-PBS. To label for bcl-2, 100 μl of antibody #124 (Dako) at 1:100 in 1% BSA-PBS was added and cells were incubated overnight at 4C. Negative control samples were incubated in 1% BSA-PBS in the absence of any primary antibody. Cells were then washed by adding 5 ml of a 1% BSA-PBS solution. The pellet was resuspended in 100 μl of affinity-purified goat anti-mouse IgG conjugated to fluorescein isothiocyanate (FITC; Jackson ImmunoResearch) at 1:30 (#50 μg/ml) in 1% BSA-PBS and incubated for 30 min in the dark at RT. Cells were then washed by adding 5 ml 1% BSA-PBS and the pellet was resuspended and incubated for 30 min at RT in the dark in 100 μl of tertiary antibody, affinity-purified rabbit anti-goat IgG conjugated to FITC (Jackson ImmunoResearch) at 1:30 in 1% BSA-PBS, for further amplification of the signal. Subsequently, the cells were washed with 5 ml 1% BSA-PBS, the supernatant removed, and the pellet resuspended in 1 ml of propodium iodide solution to stain the DNA, and incubated for 30 min in the dark at RT. Bcl-2- and DNA-associated fluorescence was then examined using a Coulter Epics V flow cytometer (Coulter Corporation; Hialeah, FL) with an argon laser set to excite at 488 nm and equipped with Blue 4.0 and MultiCycle software for data analysis (Phoenix Flow System; San Diego, CA).
Protein Extraction
Cells grown in monolayer were trypsinized, washed twice in PBS, and resuspended in a protein extraction buffer at 4C [1.0% Triton X-100, PBS (without Ca++, Mg++) and protease inhibitor cocktail at the concentration suggested by the manufacturer (Boehringer Mannheim; Indianapolis, IN)]. Suspension cultures were recovered by centrifugation, washed twice in PBS, and also resuspended in protein extraction buffer. After resuspension, extracts were vortexed vigorously and then centrifuged at 14,000 × g (4C) for 20 min. Insoluble material (pellet) was discarded and the extract was stored at −20C. Protein concentrations were determined with a microprotein determination kit (Sigma).
Immunoblotting
Protein extract was combined with sample buffer with β-mercaptoethanol in a 1:1 ratio. The protein was loaded at equal protein concentration as determined by a microdetermination kit (Sigma) onto 10, 12, or 15% SDS-polyacrylamide gels and electrophoresis was performed at 150-200 V. All steps were performed at RT. Samples were transferred overnight in the presence of an ice cooling chamber at 30 V to PVDF membranes according to the BioRad (Richmond, CA) protocol, using a transfer buffer containing 14.4 g/liter glycine, 3.03 g/liter Tris base, and 20% methanol, but no SDS. Membranes were blocked for 3-6 hr in Blotto (5% nonfat dry milk in PBST). The protein of interest was localized by an overnight incubation in primary antibody at a concentration of 0.3 μg/ml in 3% BSA-PBST (3% BSA in PBS containing 0.05% Tween-20), followed by three additional blocking incubations in Blotto (15 min each). Then the membranes were incubated in a secondary antibody for 2 hr: affinity-purified goat anti-mouse conjugated to horseradish peroxidase (Jackson ImmunoResearch) at a concentration of 0.1 μg/ml in 3% BSA-PBST. After two additional blocking steps in Blotto (15 min each) and two washes in PBST (10 min each), the reactive protein band(s) was identified using a chemiluminescent substrate for horseradish peroxidase (Amersham; Arlington Heights, IL). Light was detected on 20,000 ASA black-and-white, high-contrast Polaroid film. Exposure time was dependent on the individual experiment but ranged from 5 sec to 5 min.
Cell Cycle Synchronization
Flasks of KB cells were synchronized at the G1/S boundary by two methods. In the first procedure, cells were preblocked with 2 mM thymidine for 16 hr. Then the thymidine was removed and cells were washed three times with RPMI-1640 medium, Next, the cells were allowed to grow for 4-6 hr in the absence of drugs before treatment with 2 mM hydroxyurea. After 1-16 hr, >90% of the cells were at the G1/S boundary as determined by flow cytometry. Protein samples collected at this time (G1/S + 0 hr) were considered to be at G1/S arrest (synchrony was verified by flow cytometry as described). In the G1/S release experiments, the remainder of the cells were washed with RPMI-1640 medium several times to completely remove the G1/S synchronizing drug. Protein samples were collected at G1/S + 3, 6, 7, 8, 9, 10, 11, 12, and 24 hr for analysis by Western blot. At each time point, cells were also collected and fixed for analysis by flow cytometry. In a second method, KB cells were synchronized at G1/S with a double thymidine block. Cells were incubated in 2 mM thymidine for 24 hr, released from synchrony for 6 hr, and then incubated a second time in 2 mM thymidine for 16 hr. Protein and flow cytometry samples were obtained at G1/S arrest and G1/S release at various times. Mitotic arrest was performed by overnight incubations in one of the following drug conditions: 0.85 μg/ml paclitaxel (taxol), 0.3 μg/ml nocodazole, 0.1 μg/ml colchicine, or 1 μg/ml vinblastine. All synchrony reagents were obtained from Sigma.
Ribonuclease Protection Assay
Preparation of DNA Template. The template for the preparation of a bcl-2 riboprobe was constructed from the recombinant transfer plasmid pVL1393-bcl-2, which contained the entire coding region of bcl-2-α cDNA (a gift from Dr. Carlo Croce) (Tsujimoto and Croce 1986; Alnemri et al. 1992). From this plasmid, a 320-BP EcoR1/Sac II DNA fragment from −57 to approximately +263 relative to the ATG start site of the bcl-2 gene was separated and subcloned into EcoR1 and Sac II sites of the pBluescript II SK vector (Stratagene; La Jolla, CA). The recombinant vector was linearized using EcoRI, leaving the T3 promoter on the 3′ end of the bcl-2 cDNA fragment, and was purified by electrophoresis followed by electroelution. The linearized vector was used as the template for the synthesis of an anti-sense 320-base RNA probe.
Preparation of 32P-labeled RNA Probe. Radiolabeled cRNA probe was prepared by incorporation of α-[32P]-UTP (800 Ci/mmole, 10 mCi/ml) using the MAXiscript In Vitro Transcription Kit following the supplied protocol (Ambion; Austin, TX). The 32P-labeled cRNA probe was size-fractionated on a 5% polyacrylamide denaturing gel containing 8 M urea. The band containing the RNA probe was cut out from the gel and eluted overnight at 37C using 350 μl of elution buffer (0.5 M NH4OAc, 1 mM EDTA, 0.1% SDS) provided by the RPA II Ribonuclease Protection Assay Kit (Ambion). In addition, a 32P-labeled 125-BP β-actin RNA probe was prepared as a control by the same method using a cDNA template provided by Ambion.
Ribonuclease Protection Assay. The ribonuclease protection assay (RPA) was performed by using the RPA II Ribonuclease Protection Assay Kit (Ambion). In brief, total RNA was extracted by using a guanidium isothiocyanate method (Chomczynski and Sacchi 1987) and quantitated by measuring absorbance at 260 nm. RNA samples extracted from synchronized KB cells, including RNA obtained from the G1-, S-, G2-, and M-phases of the cell cycle, as well as from unsynchronized GM 697 lymphoma cells as a positive control, were hybridized with 1.6 × 105 cpm of radiolableled probe under conditions previously published by others (Melton et al. 1984). The integrity of the RNA was confirmed by analytical gel electrophoresis. The RNA samples were first co-precipitated with α-[32P]-RNA probes. A total of 20 μl hybridization buffer was added to each co-precipitated sample and incubated at 42C for 18-20 hr. The remaining single-stranded RNA was destroyed by the addition of an RNase A and T1 mixture to a final concentration of 2.5-3.0 U/ml and incubated at 37C for 30 min. β-Actin mRNA was chosen as the internal standard. Two × 104 cpm of α-33P-labeled β-actin RNA probe was used for each total RNA sample. The control reactions including 20 μg yeast RNA hybridized with both bcl-2 and β-actin RNA probes and degraded by RNase, 20 μg yeast RNA hybridized with bcl-2 RNA probe without degradation, and 20 μg yeast RNA hybridized with β-actin RNA probe without degradation. RNA fragments protected from RNAse were analyzed on a 5% denaturing polyacrylamide gel, followed by autoradiography. The intensity of the radiolabeled fragments was quantitated by densitometry of the autoradiographs.
Cytoskeletal Extraction
Cells were grown on 35-mm dishes that were treated with 1 mg/ml poly-L-lysine (Sigma) for 30 min at 37C and then arrested in mitosis with vinblastine as described. Then they were washed five times with PBS and incubated on a rocker in cytoskeletal extraction buffer (CEB: 0.1 M 2-[N-morpholino] ethanesulfonic acid, 1 mM EGTA, pH 8.0, 1 mM MgCl2, 4% PEG 8000) for 30 min at RT. Next, each dish was washed three times with PBS and incubated on a rocker for an additional 30 min, with cytoskeletal extraction buffer with 0.2% Triton X-100 added. As described in the text, the CEB pH was varied to determine the optimal pH at which bcl-2 associated with chromosomes by the addition of NaOH or HCl (further experiments were performed at pH 8.0). In addition, the following agents were added to the CEG incubations as delineated in Results: 20 mM Na3PO4 (Sigma); 0.5 μg/ml proteinase K (Sigma), 10 μg/ml DNase I (Sigma), and 2.5 μg/ml RNase A (Sigma). After cytoskeletal extraction under each of these conditions, dishes were washed five times with PBS and then fixed with 80% acetone/20% dH2O or 3.7% formaldehyde for 10 min as indicated in the text. Cells were again washed with PBS and then immunostained for bcl-2.
KB Chromosome Isolation on 35-mm Dishes
KB cells were arrested in mitosis with an overnight incubation in 0.1 μg/ml colchicine. Mitotic cells were shaken from flasks and poured into 50-ml centrifuge tubes. These cells were then collected by centrifugation at 250 × g for 5 min. The supernatant was removed and 5 ml dH2O was added to the unrinsed cells, which were then incubated in a 37C water bath for 45 min. Such hypotonically swollen cells were then spun onto poly-L-lysine-coated dishes at 500 × g for 15 min. Dishes were then washed three times with PBS (without Ca++, Mg++) containing 0.1% Triton X-100 before fixation with 80% acetone/20% dH2O. Chromosome preparations were then assayed by rhodamine immunofluorescence for the presence of bcl-2. Presence of chromosomal material was verified by DAPI staining as follows. Dishes were washed once with DAPI-methanol (4′6-diamidine-2′phenylindole dihydrochloride) at a concentration of 1 μg/ml. Next, they were covered with DAPI-methanol at 0.1 μg/ml and incubated for 15 min at 37C. Finally, the stain solution was decanted and the dishes were washed once with methanol and cover-slipped under 90% glycerol/10% PBS and viewed using a 340/380-nm excitation filter and an LP 430-nm barrier filter on a Zeiss Axioplan fluorescence microscope and a X40 (N. A. 1.3) Neofluar oil immersion objective.
Cell Fractionation Chromosome Isolation for Western Blot Analysis
Ten flasks of KB cells were synchronized overnight in mitosis using 1 μg/ml vinblastine. Media from mitotically arrested cells were then collected and any cells present were spun into a pellet. Cells remaining attached to flasks were trypsinized or scraped for removal from the flask substrate and spun into a pellet. All cells were then combined and washed twice with PBS. After the second centrifugation wash, cells were resuspended in 10 times the pellet volume of a hypotonic solution [40 mM KCl, 0.2 mM spermine, 0.5 mM spermidine, and protease inhibitor cocktail (Boehringer Mannheim)]. Cells were then incubated on ice for 120-180 min. After swelling, cells were pelleted at 100 × g for 10 min, the supernatant was removed, and 5 times the pellet volume of chromosome isolation buffer [CIB: 20 mM NaCl, 80 mM KCl, 15 mM Tris-HCl, pH 7.2, 0.5 mM EGTA, 2 mM EDTA, 0.15% β-mercaptoethanol, 0.2 mM spermine, 0.5 mM spermidine, and protease inhibitor cocktail (Boehringer Mannheim)] was added. Cell membranes were lysed by drawing the swollen cells through a 22-gauge needle and chromosome monodisperson was assessed using 50 μg/ml propidium iodide (Sigma). Contaminating nuclei were sedimented with a 250 × g spin at 4C for 1 min. Supernatant was then layered on and centrifuged through a 10-60% sucrose/CIB step gradient. The chromosome fraction was identified by staining an aliquot of each gradient interface with propidium iodide at a concentration of 50 μg/ml. Once identified, the chromosomal fraction was collected and protein was precipitated with TCA and then resuspended as described elsewhere. Equal concentrations of protein were loaded onto SDS-polyacrylmide gels before electrophoresis and were further analyzed by Western blot.
Results
Immunofluorescence Detection of Localization of bcl-2 at Various Levels of Expression
Initially, we wished to determine the generality of the chromosomal distribution seen in KB and OVCAR-3 cells (Willingham and Bhalla 1994). In addition, because KB cells express relatively low levels of bcl-2 protein, and because lymphoid and other cells can express much higher levels, we wished to examine the differences in immunofluorescence patterns at these various levels.
A variety of human cultured cells were examined by immunofluorescence using anti-bcl-2 (#124) as summarized in Table 1, including cell lines derived from carcinomas (epidermoid, ovarian, prostate, breast), neuroblastoma, leukemia, embryonic fibroblasts, and primary amniotic cells. In all cases, when cells were found to be in mitosis, their chromosomes showed intense bcl-2 localization. In addition, PHA-activated peripheral blood mononuclear cells were also examined. These cells showed a more variable reaction in their chromosomes, some being very bright and others showing no reaction. Cells of nonprimate origin (Swiss 3T3 mouse fibroblasts, NIH3T3 mouse fibroblasts, and DDT1-MF2 hamster osteosarcoma cells) showed no reactivity with MAb #124, in keeping with its known human-specific epitope reactivity.
To clarify the relationship between cytoplasmic bcl-2 and chromosomal bcl-2, we examined cells that were known to express variable levels of bcl-2. KB cells express very low levels and show essentially no detectable immunofluorescence labeling in interphase cells, yet they showed easily detectable chromosomal labeling in mitosis (Figure 1A'). When nonhuman cells were transfected with human bcl-2 cDNA using a mammalian expression vector construct, such as with DDT1-MF2 cells transfected with human bcl-2 cDNA (MF2-Bcl-2 cells), only the transfected human bcl-2 can be detected using MAb #124, and high-level expression of this bcl-2 led to a diffuse bright cytoplasmic distribution (Figure 1B'). This cytoplasmic pattern is similar to that seen in human cells that overexpress endogenous bcl-2 (Monaghan et al. 1992). Some cells express an intermediate level of bcl-2, such as some subclones of MCF-7 human breast carcinoma cells, and these show a diffuse cytoplasmic distribution of bcl-2 in addition to the strong chromosomal signal seen in G2 and mitotic cells (Figure 1C'). Therefore, the cytoplasmic and chromosomal patterns can exist at the same time in the same cell.
Immunofluorescence analysis of chromosomal localization of bcl-2 using MAb #124
Detection of bcl-2 Expression by Flow Cytometry
Because one might argue that our failure to detect bcl-2 in KB cells in interphase was due to a failure of interpretation of a weak diffuse cytoplasmic signal, we examined the detection of bcl-2 using a method that did not depend on morphological interpretation. Using flow cytometry of fixed and permeabilized KB cells with MAb #124, we examined cells in various phases of the cell cycle after release of their arrest at the G1/S interface by washing away a synchronizing drug (hydroxyurea). Cells in early S-phase or 8 hr after release showed no specific reactivity with MAb #124 (Figure 2). We determined the time of mitosis in this synchronized population using propidium iodide flow cytometric analysis. The majority of these cells entered G2/M approximately 10-12 hr after G1/S release. When these cells were examined for bcl-2 detection using MAb #124 at 10 hr after release, a sharp bcl-2-specific peak of increased fluorescence could be seen (Figure 2, arrow), a peak that was absent at 0 or 8 hr after release. Therefore, flow cytometry detected the same sudden appearance of specific reactivity seen using this antibody by immunocytochemistry in mitotic cells.
Immunofluorescence localization of bcl-2 in cells expressing various levels of bcl-2. KB cells (A,A') express low levels of bcl-2 and immunofluorescence in such cells shows specific labeling only in mitotic chromosomes, such as the cell shown in prophase (arrow, A'). Note the lack of labeling in adjacent cells in other phases of the cell cycle. MF2-bcl-2 cells (B,B') are from a hamster cell line transfected with an expression vector construct coding for the human bcl-2 α protein. Such cells show a diffuse cytoplasmic distribution of bcl-2 in interphase (arrow, B'). Note the “nuclear sparing” pattern of this diffusely distributed cytoplasmic protein. MCF-7 cells (C,C') are from a human breast cancer cell line that shows variable expression of bcl-2 in a cytoplasmic distribution, but also shows that bcl-2 in such cells can also be found concentrated on chromosomes shown here in a nucleus in G2 (arrowhead) or a mitotic cell in metaphase (arrow). A-C are phase-contrast images of the same fields shown in immunofluorescence images in A-C Bar = 35 μm.
Comparison of MAb #124 with Other Commercial Antibodies to bcl-2
To determine if the reactivity seen in mitotic cells with MAb #124 was an artifact of crossreaction with some other component or due to some nonspecific property of the antibody preparation, we tested various other commercially available antibodies to human bcl-2 using both immunofluorescence and Western blot analyses. The results are summarized in Table 2. MAb #100 is another clone reactive with the same epitope recognized by MAb #124 (amino acids 41-54 of the human bcl-2 sequence), and it showed the same pattern of reactivity seen with MAb #124 (Pezzella et al. 1990). Some antibodies (6C8 and PC68) to other synthetic peptide or recombinant bcl-2 epitopes showed reactivity to bcl-2 in Western blots but did not give strong reactions using fixed cells, either by immunofluorescence or by flow cytometry, or by using cells that expressed low or high levels of bcl-2. Other antibodies (4D7 and SC-492) showed high levels of background that did not correlate with the known levels of bcl-2 present in the cell types tested as detected by Western blots with MAb #124. Therefore, we were not able to utilize other commercially available antibodies to clarify the chromosomal distribution of bcl-2 detected by MAbs #100 and #124, because they were not effective in recognizing the conformation of bcl-2 present in fixed cells.
Detection of bcl-2 using MAb #124 and flow cytometry in synchronized KB cells. KB cells were synchronized at the G1/S boundary using thymidine and hydroxyurea as described in Materials and Methods, then released by washing away the synchronizing drug. At 10-12 hr after this release, propidium iodide staining of DNA content demonstrated that these cells entered G2/M. At 0, 8, and 10 h after release, cells were also collected for detection of bcl-2 in cell suspensions using formaldehyde and ethanol fixation and then incubation with MAb #124 and fluorescein double indirect labeling. Negative control samples consisted of the deletion of the MAB #124 step. As shown in these histograms, the cells displayed background fluorescence at 0 and 8 hr after release, similar in intensity to the negative control. A distinct unique peak appeared at 10 hr (arrow), indicating the sudden appearance of bcl-2 reactivity in cells at G2/M, in agreement with the immunofluorescence results. Therefore, the lack of detection of bcl-2 in interphase shown in Figure 1A' is not due to an interpretative artifact related to a weak, diffuse distribution, because flow cytometry measures the labeling present in each whole cell.
Variations in bcl-2 Protein and mRNA During the Cell Cycle
One interpretation of the mitotic phase pattern seen in KB cells was that the levels of bcl-2 protein might vary dramatically with the cell cycle, i.e., bcl-2 might be a cyclin. We examined the levels of bcl-2 protein using Western blot analysis in G1/S arrested and mitotically arrested KB and HeLa cells (a closely related cell line) and found no significant variation in total amounts of bcl-2 as a function of the cell cycle (Fig 3; and Schandl et al. 1996). However, a more slowly migrating form of bcl-2 was seen during mitosis, suggesting that a variation in phosphorylation occurred. This finding has been reported and discussed previously (Haldar et al. 1994; Schandl et al. 1996) and is believed to represent a normal phosphorylation event involving bcl-2 that is probably unrelated to subsequent apoptosis or an anti-apoptotic function of bcl-2 (Haldar et al. 1995; Schandl et al. 1996). Therefore, the lack of detection of bcl-2 in fixed interphase KB cells was not due to its absence from the cell.
In addition to protein levels, we also examined the levels of bcl-2 mRNA to determine whether mRNA analysis would reveal variations that could not be detected in a longer-lived protein population. Using RNase protection assays, we examined the levels of bcl-2 mRNA in synchronized KB cells released from G1/S arrest. No change in bcl-2-specific mRNA levels was detected during the cell cycle in KB cells (Figure 4). Collectively, these results indicate that the cell cycle-dependent changes we have detected in bcl-2 by immunocytochemistry are not regulated at the level of transcription nor at the level of bcl-2 protein amounts, but are more likely to be controlled by cell cycle-dependent structural differences that influence the accessibility of the epitope.
Utility of commercially available antibodies to bcl-2 a
+/-, weak or negative; +, detectable; + + +, strong reaction.
Molecular Characteristics of bcl-2 in KB Cells
Because the detection of bcl-2 in chromosomes was highly dependent on the specificity of the epitope reactive with MAb #124, we wished to examine whether the form of bcl-2 reactive in KB cells might have slightly different molecular properties in comparison with bcl-2 in overexpressing cells. We evaluated the relative molecular weight of bcl-2 from these different cell types, as well as the relative levels of bcl-2 in extracts from equal numbers of cells using Western blot analysis. Figure 5 shows that the transfected leukemic cell line GM697-Bcl2, a cell showing a diffuse cytoplasmic immunofluorescence pattern in interphase (Willingham and Bhalla 1994), contains between 400 and 1000 times more bcl-2 than that found in KB cells. Further, a careful side-by-side comparison of the relative mobilities of the bcl-2 from these two cell types showed no obvious difference in molecular size. As shown in Figure 3 and in previous studies (Schandl et al. 1996), however, there is a major difference in bcl-2 properties that correlates closely with the reactivity of MAb #124 with bcl-2 in fixed KB cells, i.e., the presence of a hyperphosphorylated form of bcl-2 at G2/M. However, this hyperphosphorylation is not necessary for reaction of MAb #124 with bcl-2 after its extraction and denaturation on Western blots (Figure 3).
Another approach to the analysis of the molecular nature of chromosomal bcl-2 in cells is to examine partially permeabilized cell preparations to see if chemical environmental changes could alter the attachment of bcl-2 to chromosomes. We performed these experiments using a modified cytoskeletal extraction technique which leaves detergent-stable anchored protein antigens in place. Using this approach with immunofluorescence detection, we found that bcl-2 attached to chromosomes was somewhat stable to gentle detergent treatment, and we further examined its stability to pH, protease, and phosphate concentration. In data not shown, the bcl-2 signal intensity in these preparations was optimal at pH 8.0, was noticeably weaker at pH 7.0 or pH 9.3, and was undetectable below pH 6.0. Proteinase K at 0.5 μg/ml destroyed the bcl-2 signal in 30 sec. RNase A had no effect, whereas DNase I slowly decreased the signal, presumably related to the loss of its attached chromatin. We also examined free phosphate on the premise that it might inhibit phosphatases that would interfere with the hyperphosphorylated state of mitotic bcl-2. There was a slight but statistically significant (p < 0.05) increase in bcl-2 signal intensity in the presence of 20 mM phosphate as measured by microphotometry (data not shown).
Western blot detection of bcl-2 in KB and HeLa cells synchronized at G1/S or mitosis. Extracts of equal numbers of KB and HeLa cells synchronized either at G1/S (A) or at mitosis (B) were prepared, and subjected to Western blot analysis using MAb #124. The total levels of bcl-2 protein at these two different stages of the cell cycle are approximately the same in both cell types. Note, however, that part of the bcl-2 in mitotic cells shows a slower mobility (arrowhead) compared to the major 26-kD lower band (line).
Quantitation of bcl-2 mRNA levels in synchronized KB cells using a sensitive RNase protection assay. An RNase protection assay was developed for the detection of bcl-2-specific mRNA as described in Materials and Methods. KB cells were synchronized at the G1/S interface, then released for various times, with G2/M corresponding to 10-12 hr after release. (B) Images of an autoradiograph from random (R) or synchronized cells released for 0, 3, 7, 8, 11, 12, or 24 hr. An internal control for loading and specificity was a probe specific for β-actin shown in the lower group. (A) Quantitation of the bands from this autoradiograph on an expanded scale to show the remarkable stability of these mRNA levels throughout the cell cycle. These results represent the combination of data from four separate experiments.
Identification of bcl-2 in Isolated Chromosomes
To study the physical association of the bcl-2 antigen with chromosomes, we isolated metaphase chromosomes from KB cells as described in Materials and Methods and analyzed then using immunofluorescence with MAb #124 or with Western blot analysis. Figure 6 shows that bcl-2 remains attached to chromosomes during the isolation procedure and remains reactive with this antibody. When these isolated chromosomes were examined using Western blot analysis, bcl-2 could be detected in this fraction, and a portion of it demonstrated hyperphosphorylation, as shown by its retarded mobility in SDS gels (Figure 7). This experiment shows that bcl-2 is strongly attached to mitotic chromosomes, remains associated with them during chromosome isolation, and maintains the same properties present in the intact cell. These observations suggest that the detection of this antigen on chromosomes in fixed cells is not due to a crossreaction with some other unrelated protein.
Discussion
Bcl-2 was initially demonstrated in the cytoplasm of cells expressing high levels (e.g., Monaghan et al. 1992). In contrast, immunocytochemical analysis of bcl-2 in cells expressing low levels showed the detection of bcl-2 only during G2 and mitosis, especially on the surface of mitotic chromosomes (Lu et al. 1994; Willingham and Bhalla 1994). In the present study, we examined (a) whether bcl-2 localization on chromosomes reflects a morphological artifact, (b) whether bcl-2 is expressed as a cyclin, and (c) whether bcl-2 remains associated with chromosomes after isolation.
Western blot comparison of bcl-2 levels and mobility in overexpressing cells and KB cells. Extracts from equal numbers of GM697-bcl2 cells and KB cells were compared through serial dilutions by Western blot analysis using MAb #124. Bcl-2 bands for GM697 cells diluted by a factor of 1:200, 1:400, and 1:1000 are compared with constant amounts of KB cell extract. It can be seen that GM697-bcl2 cells contain between 400 and 1000 times more bcl-2 compared to KB cells. The relative mobility of the bands between these different cell types is essentially the same at the resolution level of this 15% acrylamide SDS gel. The positions of molecular weight standards are shown at left.
Detection of bcl-2 on isolated chromosomes from mitotically arrested KB cells. KB cell synchronized using vinblastine were extracted and a chromosomal fraction was isolated as described in Materials and Methods. Some of these isolated chromosomes were attached to slides and labeled using MAb #124 and rhodaminelabeled antibodies, as well as with DAPI to visualize DNA content. Phase-contrast (A), DAPI fluorescence (B), and bcl-2 (rhodamine) fluorescence (C) are shown. Note that bcl-2 is still present on these isolated chromosomes. Controls deleting the MAb #124 step showed no fluorescence on the rhodamine channel (not shown). Bar = 12 μm.
Appearance of bcl-2 in Chromosomes Is Not a Morphological Artifact
The bright decoration of chromosomes by bcl-2 detected by immunofluorescence in fixed cells might reflect a concentration of protein that gives a false impression of exclusivity to this chromosomal location, i.e., the signal in the cytoplasm might be significant but diffuse and difficult to see. Flow cytometry results shown in this article essentially rule out that interpretive possibility. The immunoreactivity of bcl-2 protein clearly varies with the cell cycle. Furthermore, the chromosomal pattern seen with KB and OVCAR-3 cells is present in all of the growing cultured human cell types examined and is not the result of an artifact unique to one or two cell types.
Western blot analysis of isolated chromosomes from KB cells. Chromosomes isolated as shown in Figure 6 were extracted and analyzed using Western blots with MAb #124. A Western blot using this same antibody and a whole KB cell extract from randomly growing cells was run in adjacent lanes. Note the presence of the bcl-2-specific band in the isolated chromosomes and the presence of the more slowly migrating hyperphosphorylated band of bcl-2 (arrow) similar to that typically seen in mitotic cell extracts.
Crossreactive Antigens: Alternative Antibodies
One approach to rule out the possibility that the chromosomal pattern seen represents a crossreactive antigen is through the use of an antibody that recognizes parts of the bcl-2 molecule other than the (a.a. 41-54) epitope. Unfortunately, as shown in this article, the commonly available commercial antibodies to this protein are not useful for fixed cells, with the notable exception of those that react with this unique epitope (a.a. 41-54). As noted below, this epitope lies in an unusually accessible part of a rather small protein molecule, and it is this variation in accessibility that may account for the variation in the detectability during the cell cycle.
Bcl-2 Variations in the Cell Cycle: Bcl-2 Is Not a Cyclin
The variation in antibody reactivity for bcl-2 in fixed cells during the cell cycle suggested that bcl-2 might vary in its levels, i.e., it might be a cyclin. Furthermore, it was possible that some of the variation was due to the synthesis of new bcl-2 with properties different from that already synthesized. Experiments in this study employing Western blots and mRNA analysis clearly show, however, that neither of these possibilities is likely.
MAb #124 Reactivity in Fixed Cells Correlates with bcl-2 Hyperphosphorylation
Two of the properties of chromosomal bcl-2 change at G2/M. One is that the protein becomes hyperphosphorylated and slows its mobility on SDS gels. This is a normal event that happens during the cell cycle at G2 (Schandl et al. 1996). Second, the protein becomes accessible to MAb #124, initially appearing in a more diffuse pattern of G2 nuclei (Schandl et al. 1999). That these two properties change at the same point in the cell cycle is not likely to be a coincidence. The unstructured loop containing the MAb #124-reactive epitope has been suggested as the major domain for phosphorylation in bcl-2 (Chang et al. 1997; Fang et al. 1998). Because many of the steps of progression in the cell cycle are known to be mediated by phosphorylation events, usually through cyclin-dependent protein kinases, it is tempting to speculate that chromosomal bcl-2 may be a substrate for such a kinase.
Chromosomal and Cytoplasmic bcl-2: Two Distribution Patterns for the Same Gene Product?
The levels of expression of bcl-2 vary widely among cell types. In cells expressing low levels, the predominant pattern of the protein detectable in fixed cells appears to be chromosomally related. As the levels of expression increase this pattern is retained, but an additional diffuse cytoplasmic pattern appears. In cells overexpressing bcl-2, which contain as much as 1000 times more bcl-2 protein, the cytoplasmic distribution is the dominant detectable form, although these cells also contain the chromosomal pattern in mitosis. For example, in results not shown here, one can separate the cytoplasm of cells such as GM697-Bcl2 cells from mitotic chromosomes using mechanical shear, and the chromosomes remaining can be seen to be decorated with bcl-2 using immunofluorescence.
The assumption that these two patterns represent the same protein has to be interpreted with caution. Although it is unlikely that the chromosomal localization seen is a trivial crossreactive artifact, it may represent a different form of the bcl-2 protein, even though it is coded for by the same gene. The antibody-reactive epitope is located at the N-terminal portion of the bcl-2 molecule and, by analogy with the known crystallographic structure of the closely related protein, Bcl-XL, this region is part of a large freely accessible loop (Muchmore et al. 1996). The BH domains responsible for the interaction with other bcl-2 family members (such as Bax) lie in other parts of the molecule. Even the hydrophobic anchor domain at the C-terminus of bcl-2 α is far-removed from this loop region. Therefore, it is possible that the antigen present on chromosomes is an isoform of bcl-2 that retains the N-terminal epitope region but differs in other domains that determine its cellular location. However, the molecular weight of the resulting protein is almost identical to that of bcl-2α, at least by its relative mobility on SDS gels. The only other known splice variant of bcl-2, bcl-2β, has a lower molecular weight (Tsujimoto and Croce 1986) and none has been detected in any of our Western blots. The splice site shared by the mRNA for both a and β lies near the 3′ end of the mRNA (C-terminal region). Therefore, any other alternative splice variants utilizing this splice site would still react with MAb #124, because they would retain the same N-terminal domain.
Bcl-2 in Interphase in KB Cells: Where Is It?
We have been able to detect a constant level of bcl-2 in KB cells throughout the cell cycle using Western blots, and yet immunofluorescence in fixed cells only detects the protein during G2/M. One might assume that bcl-2 in G1- and S-phase cells would be located in the cytoplasm, because transfection of cells with bcl-2 cDNA leads to a cytoplasmic expression pattern. However, there is no direct evidence of this location in interphase in cells expressing low levels. Because the antibody used fails to react with this antigen in this part of the cell cycle in fixed cells, immunocytochemistry alone cannot resolve this question. The accompanying article (Schandl et al. 1999) describes experiments that address this question using classical cell fractionation techniques.
Footnotes
Acknowledgements
Supported in part by grants from the American Cancer Society (CB-144 to MCW) and the National Cancer Institute (CA71851 to WF).
We wish to thank Dr John Reed and Dr Eduardo Cantu for specific cells, Dr Carlo Croce for bcl-2 cDNA, Josef Vesely for expert technical assistance, and Kristy K. Young for helpful suggestions and comments.