Drosophila grim induces apoptosis in mammalian cells

Grim induces death in mouse 3T3 fibroblasts

To test its ability to induce apoptosis in mammalian cells, we transiently expressed Drosophila grim in mouse 3T3 fibroblasts. The grim‐coding sequence was inserted into a vector that provided doxycycline‐inducible synchronous expression of the sequences inserted. To characterize grim expression, we developed an antiserum against the whole recombinant protein purified from Escherichia coli. Western blots detected a band of the expected size, specific for extracts from transfected cells, which coincided with the size of the endogenous Drosophila protein and of the recombinant grim protein purified from E.coli (Figure 1I). An additional smaller band specific for transfected cells was also observed. A similar downshift was observed for both bands in extracts from cells transfected with a truncated grim mutant lacking amino acids 2–14 (see below) (Figure 1I). This confirms that the detected bands correspond to the transfected grim and shows that the smaller form results from partial cleavage of the protein at its C‐terminus. However, the smaller band was not observed when the concentration of the serine/cysteine protease general inhibitor phenylmethylsulfonyl fluoride (PMSF), included in the lysis buffer, was increased from the standard 1 mM to 2 mM (Figure 1I). We therefore conclude that grim proteolysis is taking place after cell lysis by a strong proteolytic activity present in the extracts. Immunofluorescence was used to identify grim‐expressing cells in transfected cultures. We found that cells with high grim levels show morphological traits characteristic of various phases of apoptosis including loss of substrate attachment, membrane blebbing and cytoplasmic loss (Figure 1A–H). Initially, when the predominant feature was membrane blebbing, nuclei were still negative for TUNEL staining (Figure 1B and F). Later, grim‐expressing cells adhered to neighbouring cells, showed nuclear DNA condensation and became positive for TUNEL staining, indicating DNA fragmentation (Figure 1C, G and J). Finally, cytoplasm and nuclei disaggregated completely into apoptotic bodies, which remained strongly positive both for grim and TUNEL (Figure 1D, H and K).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1.

Grim induces apoptosis in mouse 3T3 fibroblasts. The pictures show immunofluorescent detection of grim expression and DNA fragmentation by TUNEL staining. Grim expression is shown in red, TUNEL staining in green and Hoechst‐stained nuclei in blue. (A and E) Grim‐positive cells presenting apoptotic morphology (arrowheads) and a matching DIC + fluorescence‐composed image, respectively. (B–D) Cells representative of the successive steps of the apoptotic process. (F, G and H) DIC + fluorescence‐composed images matching respectively, (B), (C) and (D). (J and K) DIC + Hoechst‐composed images matching, respectively, (G) and (H). (I) Western blot detection of grim in whole protein extracts from Drosophila embryos (d), grim protein purified from E.coli extracts (p), and extracts from 3T3 fibroblasts transfected with either full‐length grim (g), grim^Δ (g^Δ) or the corresponding empty vectors (c) from cells lysed in either 1 or 2 mM PMSF as indicated.

To facilitate quantitation of the death induction effect, grim was co‐transfected with the same expression vector driving LacZ expression. Grim expression resulted in the elimination of two‐thirds of the transfected cells during the 24 h of culture after the addition of doxycycline (Figures 2A, and 3A and B). The reduction in blue cells specifically affected the high‐LacZ‐expressing cell class; >90% of cells in this class were killed in grim‐transfected cultures (Figures 2B, and 3A and B). In agreement with the observed cell death effect, grim‐transfected cultures showed a high percentage of blue cells with a characteristic apoptotic morphology (Figure 2C). All experiments were performed maintaining cells in 10% serum; lowering the serum concentration to 0.1% did not reduce the percentage of surviving cells (not shown). We conclude that grim kills 3T3 cells through apoptosis induction in a serum‐insensitive manner.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2.

Quantitation of cell death by LacZ co‐transfection and time‐lapse videomicroscopy. The frequency of total blue cells (A) or intense blue cells (B) was measured in co‐transfections of LacZ vector plus empty (control), grim or grim^Δ expression vectors. The moment of doxycycline addition (18 h after transfection) was taken as t₀ in the graphs. (C) The frequency of cells with apoptotic morphology within the blue cell class. Time‐lapse videorecording was used for the observation of the transfected cultures. Initiation of membrane blebbing invariably led to cell death; cell death events were therefore scored at the initiation of blebbing (D). Average total cell numbers during filming time were 245, 258 and 239 for control, grim and grim^Δ, respectively. (E) The average blebbing time for wild‐type and mutant grim forms.

• Download figure
• Open in new tab
• Download powerpoint

Figure 3.

Grim‐induced apoptosis requires caspase activity. The pictures show X‐Gal staining of 3T3 fibroblast cultures transfected with LacZ‐expressing vector and either empty or grim‐expressing vector as indicated. Vectors expressing p35 or OpIAP under the control of the elongation factor‐1 (EF‐1) promoter were co‐transfected, when indicated, in a 1:1 ratio to the grim‐expressing vector. After transfection, 50 μM zVAD.fmk, DEVD.cho or YVAD.cho was added to the culture medium as indicated. Cells were processed for LacZ staining 24 h after doxycycline addition.

The conserved N‐terminal region of grim is not necessary for apoptosis induction or execution

The only conserved sequence found in grim is a 14 amino acid N‐terminal motif, also present in hid and reaper (Chen et al., 1996). To test the relevance of this motif in our system, a truncated form of grim lacking amino acids 2–14 was expressed (grim^Δ, Figure 1I). The mutant form retained most of the grim killing ability, but cells appeared to tolerate slightly higher levels of the mutant protein without entering apoptosis (Figure 2A–C). We used time‐lapse videomicroscopy to characterize and compare the death‐inducing effects of wild‐type and mutant grim forms. Video analysis showed that between 4 and 10 h after doxycycline addition, the rate of cell death in grim‐transfected cultures was 14 times higher than in control‐transfected cultures, and demonstrated no differences between the wild‐type and mutant forms (Figure 2D). In addition, we determined the mean length of the blebbing period as a reference measure for the duration of the execution phase of apoptosis induced by both transfected forms. Again, no differences were observed between the mutant and wild‐type forms (Figure 2E). We conclude that the conserved N‐terminal domain is not required for either the initiation or the execution of apoptosis by grim. The same conclusion was reported when a similar grim truncation mutant was introduced in insect cells (Vucic et al., 1998).

Caspase activity is required to execute grim‐induced cell death

To test the involvement of caspases in grim‐induced death, we used specific inhibitors together with LacZ co‐transfection. Addition of the cell‐permeable broad specificity caspase inhibitor zVAD.fmk to the culture medium rescued all the LacZ‐expressing cells from grim‐induced death (Figures 3E and F, and 4A). At the same time and in agreement with previous evidence from insects (Chen et al., 1996), the wide‐range natural caspase inhibitor p35 also rescued cells from grim‐induced death (Figures 3G and 4A). In contrast, the group II‐caspase inhibitor DEVD.cho only showed a residual rescuing capacity (Figures 3D and 4A). In agreement with this result, the natural group II caspase inhibitor OpIAP also showed a very limited rescuing capacity (Figures 3C and 4A). Increasing the molar ratio of OpIAP‐expressing vector up to 3:1 versus the grim‐expressing vector did not improve the viability of grim‐transfected cells (data not shown). The group I caspase inhibitor YVAD.cho was unable to prevent grim‐induced death, either alone or in combination with DEVD.cho (Figures 3H and L, and 4A). Increasing YVAD.cho and DEVD.cho concentrations up to 250 μM did not elicit cell rescue by these inhibitors (data not shown). Western blot analysis showed that transfected grim was expressed at high levels in cultures that were either incubated or co‐transfected with the different inhibitors (Figure 4B), confirming that inhibition of caspase activity counteracts apoptosis execution by blocking a step downstream of grim action.

• Download figure
• Open in new tab
• Download powerpoint

Figure 4.

Quantification of cell rescue capacity of natural and synthetic caspase inhibitors. The bar graph (A) shows the survival ratio of cells transfected under the experimental conditions shown in Figure 3. The survival ratio was calculated as the ratio between the frequency of blue cells found in grim‐ and empty vector‐transfected cultures in each experimental condition. Western blot detection of p35‐ and OpIAP‐transfected proteins is shown above. (B) Western blot detection of grim protein in 3T3 fibroblasts transfected under control conditions and cultured in the presence of the different caspase inhibitors or co‐transfected with either p35 or OpIAP, as indicated.

The broad specificity caspase inhibitors zVAD.fmk and p35 prevented grim cleavage under conditions that allowed its proteolysis in cell extracts (Figure 4B). This result suggests that an endogenous caspase is responsible for grim cleavage. The similar shift observed in Western blot bands for both the cleaved and intact forms of the protein in the N‐terminal deletion mutant showed that the cleaved portion of the protein was the C‐terminus (Figure 1I). Three clustered aspartate residues near the grim C‐terminus (positions 126, 128 and 129) could then be used by a caspase as a target sequence. The downshift predicted by digestion at these sites is between 1 and 1.3 kDa; however, the observed size for the short form is ∼4 kDa less than that observed for the intact protein. Nevertheless, we have observed that C‐terminal deletions of the protein provoke downshifts greater than expected, which are in the range of the observed downshift for the proteolysed form of the protein (data not shown). These results show that besides activating caspases, grim itself can be processed as a consequence of their proteolytic activity. Even though OpIAP and DEVD.cho did not block grim‐induced apoptosis, they were both functionally active in repressing endogenous caspases in either treated or transfected cells, since they prevented grim cleavage (Figure 4B). Therefore, it is likely that caspases insensitive to OpIAP and/or DEVD.cho were sufficient to achieve grim‐induced apoptosis.

Interestingly, zVAD‐treated cells did initiate apoptosis but were blocked at the blebbing stage, did not complete the apoptotic programme during the observation time, remained intact in the culture dish, did not become TUNEL‐positive and did not suffer nuclear disaggregation (Figure 3I and J). The same cellular phenotype has been described previously for cells in which the apoptosis induced by c‐myc, E1A and Bak had been blocked by zVAD (McCarthy et al., 1997). In contrast, grim‐expressing p35‐rescued cells continued to divide at a similar rate to that of control p35‐transfected cells and did not show membrane blebbing (Figure 3K), suggesting that p35 may inhibit apoptotic processes independent of the activity of caspases sensitive to zVAD inhibition. These results demonstrate that grim induces apoptosis by activating endogenous mammalian caspases.

Grim does not induce unspecific toxicity in mammalian cells

Overexpressed exogenous molecules can generate unspecific toxicity leading to a stress response, including apoptosis induction. In order to determine whether the observed induction of apoptosis by grim was originated through an unspecific toxic effect, we analysed two basic metabolic parameters of grim‐transfected p35‐rescued cells. We measured DNA and protein synthesis by the incorporation of [³H]thymidine and [³⁵S]methionine. In spite of them expressing very high levels of grim (Figure 4B), we found that they synthesized DNA and protein at the same rates as mock‐transfected cells (Figure 5). Interestingly, simply p35‐transfected cells showed protein synthesis rates similar to control and p35+grim‐transfected cells, but showed very low levels of thymidine incorporation, suggesting that p35 inhibits DNA synthesis and that grim reverses this effect. Since caspase inhibition by p35 involves its cleavage, caspase activation by grim may lower the effective amount of p35, thereby allowing normal DNA synthesis rates. These results show that, as long as apoptosis is blocked by p35, even high levels of grim protein do not affect basic parameters of cell metabolism such as DNA or protein synthesis. We conclude that grim kills mouse fibroblasts by specifically inducing apoptosis pathways.

• Download figure
• Open in new tab
• Download powerpoint

Figure 5.

Grim‐expressing p35‐rescued cells do not show symptoms of cellular stress. Graphs (A) and (B) show, respectively, the [³H]thymidine and [³⁵S]methionine incorporation rates by cells transfected with the LacZ‐expressing vector and p35 and grim empty vectors (control), p35‐expressing vector and grim empty vector (p35), or p35‐ and grim‐expressing vectors (grim + p35). The horizontal axis shows the number of cells lipofected in each assayed culture well. Cells received 12 h pulses of either isotope starting 30 h after lipofection. Serum‐starved and cycloheximide‐treated cells were used as negative controls for the incorporation of [³H]thymidine and [³⁵S]methionine, respectively. Replica wells were stained for LacZ to determine the actual transfection level under each experimental condition. The percentage of transfected cells did not vary among the different cell concentrations and were as follows: control, 52 ± 4%; p35, 46 ± 6%; grim + p35, 49 ± 5%. CHX, cycloheximide.

Death induction by grim is inhibited by bcl‐2 and is independent of Fas signalling or p53 activity

Although the Fas pathway involves activation of an extracellular signalling mechanism, it has been shown to mediate the apoptosis‐inducing effect of c‐myc in fibroblasts by an autocrine mechanism (Hueber et al., 1997). To test the implication of the Fas–FasL pathway in grim death induction, we transfected it into primary mouse embryo fibroblasts (MEFs) derived from lpr (Fas‐deficient) mice. We found that the ability of grim to kill 3T3 fibroblasts was conserved in control wild‐type MEFs and that cell survival was not increased when grim was transfected into lpr/lpr primary fibroblasts (Figure 6A). Fas signalling is, therefore, not required for grim pro‐apoptotic activity, suggesting that grim activates a mitochondrial death pathway.

• Download figure
• Open in new tab
• Download powerpoint

Figure 6.

Death induction by grim is counteracted by bcl‐2 expression and does not require either Fas signalling or p53 activity. (A) The survival ratio of grim‐transfected p53‐ (p53 −/−) or Fas‐ (lpr/lpr) deficient MEFs and the respective control MEFs of the same genetic background. (B) The survival ratio of 3T3 fibroblasts co‐transfected with grim and LacZ expression vectors, and either empty or hbcl‐2‐pEF‐BOS expression vectors. The grim‐expressing vector was included at different molar ratios to the bcl‐2‐expressing vector as indicated. The survival ratio was calculated as described in the legend to Figure 4. Western blot detection of bcl‐2 and grim proteins in each experimental condition is shown above the corresponding bars. Both proteins were detected on the same filter by consecutive incubation with bcl‐2 and grim antibodies, respectively.

One of the endogenous pathways leading to mitochondrial‐mediated cell death is that triggered by p53 activation in response to DNA damage (Clarke et al., 1993; Lowe et al., 1993; Polyak et al., 1997). We transfected grim into p53 null mutant MEFs to determine whether p53 mediated grim‐induced apoptosis. Cell viability was not increased in p53‐deficient primary fibroblasts as compared with control wild‐type MEFs (Figure 6A), showing that grim does not require p53 activity to induce apoptosis.

Mitochondrial pathways leading to apoptosis induction are generally inhibited by the anti‐apoptotic protein bcl‐2. To test whether grim apoptosis induction was sensitive to bcl‐2 levels, we co‐transfected it with grim into 3T3 fibroblasts. Overexpression of bcl‐2 significantly reduced grim killing efficiency, increasing the relative viability of grim‐transfected cells, without affecting the efficiency of grim expression (Figure 6B). The bcl‐2 rescue effect was dose sensitive, and was suppressed by increasing the amount of grim‐expressing vector (Figure 6B). These results show that the balance between grim and bcl‐2 determines whether or not the cell enters into apoptosis, and suggest that grim activates death by interacting with components of the mitochondrial pathways.

Grim progressively localizes in mitochondria during apoptosis progression

The sensitivity of the grim‐induced death to bcl‐2 levels suggested that grim acts by activating a mitochondrial pathway. To determine the site of grim action, we explored its subcellular localization in transfected cells that do not yet show morphological symptoms of apoptosis. Pre‐apoptotic fibroblasts simultaneously show grim in two different cytoplasmic localizations: diffuse cytosolic and a punctate pattern. However, while cells simply grim‐transfected showed almost exclusively a cytosolic localization, cells treated with zVAD.fmk showed predominantly the punctate pattern (Figure 7). The same pattern is observed in grim‐expressing p35‐rescued cells (data not shown). Co‐localization of grim protein with a mitochondrial‐specific antibody showed that the punctate pattern corresponded to mitochondria (Figure 7). In contrast to the pre‐apoptotic cells, blebbing cells simply grim‐transfected, while still presenting the diffuse localization, also consistently showed the mitochondrial localization (Figure 7). Again, zVAD‐blocked blebbing cells showed almost exclusively a mitochondrial expression pattern (Figure 7). We conclude that grim localizes initially in the cytoplasm but accumulates progressively in the mitochondria, correlating with apoptosis progression. It is possible that grim translocation to mitochondria is the event that triggers the apoptotic pathway. In that case, the increased mitochondrial localization of grim in zVAD‐treated cells would be the result of the apoptosis blockade by zVAD at a point downstream of grim incorporation in the mitochondria.

• Download figure
• Open in new tab
• Download powerpoint

Figure 7.

Grim subcellular localization. Confocal optical sections detecting the subcellular localization of grim (red), mitochondria (green) and co‐localization of both signals (orange–yellow), as indicated. Stainings were performed on grim‐transfected cells in the absence or presence of zVAD.fmk, as indicated.

Expression vectors

Grim‐coding sequences were obtained by RT–PCR from Canton‐S 12–24 h Drosophila embryos using primers pairs G1–G3 for grim and G2–G3 for grim^Δ: G1, TTCCTTCCGCGGCCGCCATGGCCATCGCCTATTTCAT; G2, TTCCTTCCGCGGCCGCCATGGCCAGAAGCTATCAGCA; G3, CGGGATCCTTAGTTCTCCTTGGAGGTG. Primers introduced NcoI–BamHI sites that were used to clone the fragments in pET‐11d (Novagen, Abingdon, UK). Sequencing confirmed the published coded protein sequence (Chen et al., 1996). A novel expression vector was constructed from the retroviral vector pBPSTR1 (Paulus et al., 1996) by substitution of tTA for rtTA. LacZ‐, grim‐ and grim^Δ‐coding sequences were cloned into the responding cassette of the new vector and sequenced again before their use for lipofection. Human bcl‐2‐ and baculovirus OpIAP‐ and p35‐coding fragments were cloned in the pEF‐BOS expression vector (Mizushima and Nagata, 1990).

Cell culture and lipofection

NIH‐3T3 fibroblasts were cultured in Dulbecco‘s modified Eagle's medium (DMEM) supplemented with 10% newborn calf serum. Cells were lipofected as recommended by the manufacturer (Gibco, Gaithersburg, MD). For co‐transfections, the different expression or empty plasmids were added at equal molar ratios, except when indicated. At 18 h after lipofection, doxycycline was added to a final concentration of 2 μg/ml. At the indicated times, cells were fixed in 0.2% glutaraldehyde, washed in phosphate‐buffered saline (PBS) and X‐gal‐stained following standard protocols.

Measurement of DNA and protein synthesis

Cells were seeded in 96‐well culture plates and lipofected with the different expression vectors. For thymidine incorporation, 1 μCi of [methyl‐³H]thymidine (sp. act. 2.0 Ci/mmol, Amersham) was added per well 30 h after lipofection. Incorporation of radiolabelled thymidine was analysed 12 h later. Starved replicas were totally deprived of serum 12 h before labelling started. For methionine incorporation, 1 μCi of [³⁵S]methionine (sp. act. >1000 Ci/mmol, Amersham) was added per well 30 h after lipofection. A period of 12 h of incorporation was allowed before measurement of protein synthesis. Replicas were assayed in the presence of 10 μg/ml cycloheximide. The results from three independent experiments for each condition and cell concentration were averaged.

Antibodies

The rabbit anti‐grim antibody was obtained against the full‐length grim protein purified from E.coli extracts. The rabbit anti‐OpIAP antibody was obtained against a synthetic peptide corresponding to OpIAP amino acids 65–81 coupled to keyhole limpet hemocyanin (KLH). In both cases, rabbits were immunized according to standard protocols by multiple intradermic injections. The total IgG fraction from grim antiserum was purified on a protein G‐coupled affinity column by FPLC. Anti‐hbcl‐2 mouse monoclonal antibody was purchased from DAKO (Glostrup, Denmark). Chicken anti‐p35 antiserum was purchased from Promega (Madison, WI). Anti‐mitochondrial antibodies were obtained from the serum of a patient with primary biliary cirrhosis, shown to recognize the E2 polypeptide of the mammalian mitochondrial pyruvate dehydrogenase complex (G.Roy, personal communication).

Immunofluorescence and TUNEL assay

For immunofluorescence and TUNEL, cells were cultured in microscope slide chambers (Cultek). Cells were washed in PBS, fixed in 4% paraformaldehyde for 15 min at room temperature, pre‐incubated in 10% goat serum, 0.1% Tween‐20 in PBS for 1 h, incubated in 10 μg/ml anti‐grim IgG for 1 h, washed three times in PBS‐Tween and incubated for 1 h in Cy3‐conjugated anti‐rabbit secondary antibody (Jackson ImmunoResearch, West Grove, PA). TUNEL was performed using the MEBSTAIN Apoptosis kit as recommended by the manufacturer (Immunotech) and developed with streptavidin–fluorescein. For double mitochondrial and grim staining, anti‐mitochondrial serum was used as a primary antibody at a 1:1000 dilution and revealed with an anti‐human Cy2‐conjugated secondary antibody (Amersham) diluted 1:400, and anti‐grim IgGs were revealed with biotinylated anti‐rabbit secondary antibodies and streptavidin–Cy3. Optical sections where obtained using an Ar–Kr laser and a TCS‐NT Leica confocal imaging systems.

Western blot

Transfected cells were lysed 24 h after doxycycline addition by incubating for 30 min at 4°C in SDS sample buffer in the presence of either 1 or 2 mM PMSF. Proteins were separated by SDS–15% PAGE and transferred to membranes which were then probed with the following antibodies: anti‐grim, 3 μg/ml total IgG; anti‐hbcl‐2 monoclonal antibody diluted 1:1000; anti‐p35 whole serum diluted 1:1000; and anti‐OpIAP whole serum diluted 1:300. In all cases, sample loading was corrected according to the percentage of blue cells observed, such that the signal obtained is normalized for the effective number of transfected cells in each experimental condition.

Time‐lapse videomicroscopy

Cells were maintained in a 37°C chamber in HEPES‐buffered culture medium and recorded with a digital video camera controlled by a computerized support. Time‐lapse video was controlled by Scion‐image (NIH‐image for PC) software, set to record one image every 30 s for 6 h. Digital videos were mounted and viewed for a total duration of 1 min (360×).

Mouse strains

MEFs were prepared by standard protocols from 13.5 day post‐coitum from p53 −/− (Donehower et al., 1992) and lpr/lpr mouse embryos and from embryos of their respective similar genetic backgrounds. MEFs were lipofected two passages after the first plating.