Photorhabdus bacteria produce a number of toxins to kill their insect hosts. The expression of one of these, Makes caterpillars floppy (Mcf), is sufficient to allow Escherichia coli to persist within and kill caterpillars. Mcf causes shedding of the insect midgut epithelium and destructive blebbing of haemocytes suggesting it may trigger apoptosis. To investigate this hypothesis, here we examine the effects of E. coli -expressed Mcf on the mammalian cell lines COS-7, NIH 3T3 and HeLa cells. Cells treated with Mcf show apoptotic nuclear morphology, active caspase-3, DNA laddering after 6 h, and the presence of cleaved PARP after 16 h. These effects are prevented by the apoptosis inhibitor zVAD-fmk. Transfection of cells with constructs expressing only the NH 2 -terminal 1280 amino acids of Mcf, as a fusion with Myc, also triggered cell destruction. The expressed fusion protein was concentrated into the Golgi apparatus before cell death. These results confirm that the novel insecticidal toxin Mcf induces apoptosis but the precise intracellular pathway remains obscure.
Photorhabdus is an insect pathogenic bacterium that lives in the gut of nematodes that invade insects, termed entomopathogenic nematodes ( Forst et al., 1997 ). Once released into the insect blood system (haemocoel) the bacteria kill their new host using a variety of toxins ( ffrench-Constant et al., 2003 ). Both the bacteria and the nematodes then reproduce within the insect cadaver until new infective juvenile nematodes repackage the bacteria, leave the corpse and set off to find new hosts ( Forst and Clarke, 2001 ). As insect death is critical to the successful replication of Photorhabdus , the bacterium appears to employ numerous, potentially functionally redundant, insecticidal toxins ( ffrench-Constant et al., 2003 ). These include the high molecular weight Toxin complexes, which have oral activity to caterpillars ( Bowen et al., 1998 ) and the Makes caterpillars floppy (Mcf) toxin which has injectable activity ( Daborn et al., 2002 ).
The gene encoding Mcf was discovered by injection of Escherichia coli carrying individual cosmid clones into the larvae of Manduca sexta (Daborn et al., 2002). Normally injected E. coli are cleared from the insect by the insect immune system. However, E. coli carrying the mcf gene are not only capable of persisting in the insect but also kill the caterpillar via a rapid loss of body turgor, the ‘floppy’ phenotype (Daborn et al., 2002). The predicted amino acid sequence of Mcf shows limited similarity to known proteins in three domains (Daborn et al., 2002) (Fig. 1A). First, the NH2-terminus carries a Bcl2-homology 3-like (BH3-like) domain. Proteins carrying only a BH3 domain are pro-apoptotic (Budd, 2001). Second, the central domain is hydrophobic and has similarity to the translocation domain of Clostridium difficile toxin B (Hofmann et al., 1997). Finally, the C-terminus shows similarity to the C-terminus of an RTX-like toxin from Actinobacillus pleuropneumoniae (Schaller et al., 1999). Insect haemocytes treated with Mcf in the cell culture media die via membrane blebbing, and cells of the midgut epithelia of injected insects also shows rapid blebbing and TUNEL positive nuclei (Daborn et al., 2002). These morphological observations, together with the presence of the BH3-like domain, led us to hypothesize that Mcf is a protein that promotes apoptosis.
Apoptosis is mediated by a wide range of pathogen virulence factors and can result in the elimination of immune cells, evasion of host attempts to limit infection or even the proliferation of intracellular pathogens. Bacterial virulence factors inducing apoptosis include pore-forming toxins, toxins expressing enzymatic activity within the host cytosol and superantigens that target immune cells (Weinrauch and Zychlinsky, 1999). These virulence factors either promote or inhibit apoptosis in host cells via numerous specific mechanisms (Weinrauch and Zychlinsky, 1999). Morphologically, apoptosis is apparent as a series of rapid events that result in the dismantling of cellular components. This is evidenced by membrane blebbing, nuclear condensation and DNA fragmentation, with apoptotic debris being cleared by macrophages (Kaufmann and Hengartner, 2001).
Here we investigate the hypothesis that Mcf promotes apoptosis by examining the morphological changes occurring in mammalian cells after application of the toxin to the tissue culture media. Mammalian cells were chosen over insect cells because of the wider range of diagnostic reagents for apoptosis available. We investigate the activation of caspase-3 and cleaved PARP in Mcf treated cells, and examine which domain of the protein promotes apoptosis by expression of the N-terminus within transfected cells.
Morphological changes in Mcf treated cells
To determine if Mcf induces apoptosis, we initially examined its morphological effects on NIH 3T3 and COS-7 cells. Untreated NIH 3T3 cells formed a confluent monolayer (Fig. 2A). Following the application of Mcf to the tissue culture media at time 0 h, gaps appear in the confluent monolayer after 16 h, suggesting cytotoxicity (Fig. 2B). This effect can be reduced by pretreatment with the caspase inhibitor zVAD-fmk, suggesting that cell loss is caused by apoptosis (Fig. 2C). To examine the morphological changes seen in cells treated with Mcf we examined COS-7 cells transfected with GFP-actin, which is incorporated into growing actin filaments thus allowing visualization of actin-containing structures in living and fixed cells (Fig. 3). Within two hours of treatment, cells showed signs of shrinkage (Fig. 3B and F). Six hours after Mcf treatment, some cells show zeiosis (membrane blebbing), a typical apoptotic characteristic, indicating that cells are dying while still attached to the coverslip, and subsequently detach into the culture medium (Fig. 3G, arrow). Cells were destroyed at a sufficient rate that after 6–7 h few cells remained visible on treated coverslips (Fig. 3D and H) whereas in contrast GFP-actin expressing cells treated with empty vector alone displayed the same morphology as at time 0 (data not shown). These data provide morphological evidence that Mcf induces apoptosis.
In addition to morphological changes occurring within whole cells, we also examined nuclear apoptotic events after Mcf treatment. The nuclei of treated cells showed the typical apoptotic morphology (Fig. 4A). The percentage of cells undergoing apoptosis after Mcf treatment was compared to that induced by a known apoptosis-inducer, staurosporine, and shown to be similar (Fig. 4B). Again, pretreatment with zVAD-fmk reduced Mcf mediated apoptosis (Fig. 4B). In addition, nucleosomal DNA ladders were detectable in NIH 3T3 cells treated with Mcf (Fig. 4C). These were visible within 6 h of treatment and maximal at 24 h.
Together, these results indicate that Mcf rapidly induces the classical symptoms of apoptosis, including zeiosis, nuclear fragmentation and DNA laddering.
Biochemical markers of apoptosis
To further confirm that Mcf treatment triggers apoptosis we assayed for the cleavage of poly (ADP-ribose) polymerase (PARP) and activation of caspase-3. Poly (ADP-ribose) polymerase is a chromatin-associated enzyme with a role in DNA repair, whose degradation is an indicator of apoptosis (Oliver et al., 1998). Both NIH 3T3 and HeLa cells treated with Mcf showed the production of cleaved PARP (Fig. 5A). Caspases are present as inactive proenzymes whose activation is important in apoptosis (Thornberry and Lazebnik, 1998). Caspase-3 activation can be detected in a colorimetric assay, and was strongly induced within NIH 3T3 cells by Mcf (Fig. 5B). Six hours of treatment with Mcf leads to strong activation of caspase-3, comparable to that observed with the positive control, staurosporine. Activation further increased at 9 h, but showed a decline at 16 h, most likely the result of massive apoptosis and cell loss. Again, caspase-3 activation over 16 h could be inhibited by pretreatment with zVAD-fmk. These results show that two important biochemical markers of apoptosis are produced by Mcf treatment and together with the morphological observations confirm that apoptosis is induced.
Expression of the NH2-terminus of Mcf in transfected cells
To test if the NH2-terminus (N-terminus) of Mcf, which contains the BH3-like domain (Fig. 6A), is sufficient to cause apoptosis we expressed the first 1280 amino acids of the Mcf N-terminus as a Myc tagged fusion protein in transfected cells (pMcfN1280). To examine the morphology of cells expressing the Mcf N-terminus we co-transfected cells with GFP-actin. Cells expressing the empty expression vector alone showed a normal morphology and faint diffuse cytoplasmic staining indicating the presence of the Myc epitope tag (Fig. 6A and B). In cells expressing the N-terminus of Mcf, cell bodies shrank around the nucleus (Fig. 6C–F). Interestingly, the nuclei themselves however, appear unaffected (Fig. 7A). Finally, to determine where the transfected N-terminus was expressed we used an anti-Gpp130 antibody to highlight the Golgi apparatus (Fig. 7B). The Myc-tagged N-terminus of Mcf co-localizes with Gpp130 (Fig. 7C), showing that the transfected protein accumulates in the Golgi apparatus. These data suggest that the transfected BH3 domain containing N-terminus accumulates in the Golgi apparatus and is sufficient to cause a cytotoxic effect without inducing any clear apoptotic phenotypes such as nuclear disintegration.
Expression of the Mcf toxin by recombinant E. coli is sufficient to allow them both to persist within and to kill injected caterpillars (Daborn et al., 2002). Histopathological investigation of these infected caterpillars reveals that both the haemocytes and midgut epithelium are destroyed via rapid membrane blebbing (Daborn et al., 2002). As nuclei within the disintegrating midgut epithelium are TUNEL positive (Daborn et al., 2002), here we have investigated the hypothesis that Mcf triggers apoptosis directly. In order to take advantage of the wide range of biochemical markers available in mammalian cells we examined the effects of topically applied Mcf on several well studied cell lines, COS-7, NIH 3T3 and HeLa cells. Treated cells show rapid shrinkage of the cell body, characteristic membrane blebbing and a reduction in cell number. These changes are concomitant with the appearance of DNA ladders, fragmenting nuclei and an increase in apoptotic index. Extracts from treated cells also show the appearance of cleaved PARP and activated caspase-3. Importantly all of these effects are inhibitable by the caspase inhibitor zVAD-fmk. Finally, to distinguish between Mcf mediated apoptosis in attached cells versus apoptosis triggered by Mcf mediated cell detachment (anoikis), we note that the assays for DNA laddering were performed on attached cells only. The apoptosis inhibitor zVAD-fmk also prevents the detachment of Mcf treated cells. These data suggest that apoptosis is triggered before cell detachment and is not a consequence of the loss of cell-substrate contact following toxin treatment. In conclusion, all of these observations support the hypothesis that Mcf triggers apoptosis in attached mammalian cells.
The observation that Mcf added to tissue culture medium promotes rapid apoptosis in cells raises the important question of how it does so. Previous analysis of the unusual domain structure of Mcf suggested two potential protein functions (Daborn et al., 2002). The first N-terminal domain contains a BH3-like consensus sequence and suggests that Mcf may promote apoptosis. The second central domain is similar to a region involved in translocation of C. difficile toxin B, which may infer that the toxin can be translocated within cells. One possibility for Mcf delivery is therefore that somehow the protein is able to translocate across the plasma membrane, enter the cytosol and hereby interact with the intrinsic apoptosis machinery (Hengartner, 2000). Bcl-2 proteins control apoptotic cell fate at the mitochondrion, and BH3 only proteins are proposed to act as apoptosis sensors, relaying an apoptotic signal to Bcl-2 and its pro-apoptotic homologues, Bax and Bak (Scorrano and Korsmeyer, 2003). The presence of a BH3 domain within Mcf may usurp the normal cellular machinery and trigger untimely apoptosis by interfering with the homeostasis between Bcl-2/–X and Bax/Bak. It will therefore be important for us to determine if the interactions between these proteins are altered after Mcf-treatment, and whether the apoptosis effector, cytochrome c, is released from mitochondria.
We have attempted to determine whether the BH3-like domain of Mcf induces apoptosis by directly transfecting its NH2-terminal region into NIH 3T3 cells. The N-terminal construct containing the BH3-like domain was expressed and most of the protein localized to a compartment identified as the Golgi body by labelling with anti-Gpp130 antibody. Several bacterial toxins, including cholera and Shiga toxins, are routed from the cell surface through the Golgi apparatus and to the endoplasmic reticulum before translocation to the cytosol (Sandvig and van Deurs, 2002). However, at this stage, the significance of the localization of the transfected N-terminus to the Golgi body is not clear. Following expression of the N-terminus the cell body of transfected cells shrunk around the nucleus but the nuclei themselves did not show apparent fragmentation. Our current data have not so far demonstrated apoptosis as a result of transfection with pMcfN1280. However, the observed retention of this fragment within Golgi bodies could indicate that it may not be delivered to the cytosol and it may not be in the correct cellular compartment to interact with mitochondrial proteins.
These observations raise several questions about the potential translocation of Mcf within the cell. First, if internalization of Mcf is necessary for toxin action, is transport through the Golgi a necessary part of toxin activation? Second, are there other parts of the Mcf protein that are responsible for the nuclear fragmentation observed when the whole toxin is applied to the outside of cells. We note that several protein toxins are translocated from the cell surface through the Golgi apparatus and to the endoplasmic reticulum before translocation to the cytosol (Sandvig and van Deurs, 2002), we will therefore begin to examine the fate of the Mcf toxin after interaction with the cell surface.
In conclusion, here we have documented that the novel insecticidal protein Mcf triggers apoptosis in mammalian tissue culture cells, although the precise point at which Mcf induces apoptosis remains unclear. Bacteria use a number of different mechanisms to promote apoptosis including superantigens, activation of second-messenger pathways, alteration of host membrane permeability, inhibition of protein synthesis and direct interactions with host cell death machinery (Weinrauch and Zychlinsky, 1999). However certain pro- and antiapoptotic host cell mechanisms, such as the activation of the Bcl-2 family of proteins, are currently not recognized as being targeted by bacterial toxins (Weinrauch and Zychlinsky, 1999). It will therefore be important to investigate if the BH3-like domain of Mcf is functional and if it represents a unique and novel mechanism of triggering apoptosis by a bacterial toxin.
Preparation of Mcf toxin
For assays of toxin activity cytosolic fractions were prepared from E. coli strain BL21 transformed with either pUC18-mcf (Mcf) or pUC-18 (control). Escherichia coli were grown for 24 h and the bacterial cells pelleted from 500 ml of culture by centrifugation at 10 000 r.p.m. for 30 min. After disposing of the supernatant the pellet was snap cooled at − 80°C for 10 min to weaken the bacterial cell walls and then resuspended in 25 ml of phosphate-buffered saline (PBS). The mixture was then sonicated on ice (amplitude 90% on 0.3 s off 1.0 s) for 10 s three times with a two min rest between. The sonicated product was then centrifuged at 10 000 r.p.m. for 15 min, the supernatant decanted and ultracentrifuged at 50 000 r.p.m. for 2 h. The supernatant was taken as the cytosolic fraction and total protein content was quantified using the Bradford Assay (Sigma). The integrity of the expressed Mcf in the cytosolic fraction was verified by Western analysis with an anti-Mcf monoclonal antibody (data not shown).
Mammalian cell culture, transfection and immunofluorescense
The mammalian cell lines COS-7 (African green monkey kidney cells), NIH 3T3 (Swiss mouse fibroblasts) and HeLa cells (human negroid cervix epitheloid carcinoma) were from the European Collection of Animal Cell Cultures (ECACC, Porton Down, Salisbury, UK). All cells were cultured in Dulbeccos Modified Eagle's Media (DMEM) supplemented with 10% fetal calf serum, 2% (10 ml) 1× penicillin and streptomycin and 1% non-essential amino acids (Sigma) and grown at 37°C, 95% air/5% carbon dioxide (v/v). Cell lines were grown as monolayers and subcultured at 80–100% confluence. For toxin assays, monolayers were washed with sterile PBS and then detached from the substratum using 1× trypsin-EDTA. Detached cells were washed with complete culture medium and seeded into plates at the required densities before incubation with toxin preparations or transfections with DNA constructs. Cells were treated with dilutions of 1: 400 Mcf cytosolic fraction in complete tissue culture media (3.5 µg ml−1 total protein) and an equivalent quantity of cytosolic fraction from the pUC-18 transformed control.
For cell transfection, the mammalian expression vector pRK5myc (a kind gift from Karen Knox, MRC) was used to express the Myc epitope tag. This vector was also modified to produce an N-terminal fragment of Mcf as a Myc-fusion protein. The N-terminus (McfN1280) was cloned by PCR using rTth DNA polymerase (Applied Biosystems) using DNA isolated from a cosmid containing full length Mcf. pMcfN1280 was generated by PCR using the sense primer AGGAGGATCCATGGCTTC TATATCCAAAG and CTTCATCTAGAGTGCCGACATTTTGTCAT CCA as the antisense primer. PCR was conducted with 60°C annealing temperature and a three min extension time.
For transfection, NIH 3T3 cells were seeded at a concentration of 1 × 105 cells ml−1 into six well plates containing ethanol-sterilised borosilicate glass coverslips (BDH) and vectors co-transfected with an GFP-actin construct using GeneJuice Transfection Reagent (Novagen) according to the manufacturer's instructions. The GFP-actin construct chosen was pEGFP-actin, a mammalian expression vector (Clontech) expressing the EGFP-actin fusion protein of the human codon optimised variant of green fluorescent protein (EGFP) and the gene encoding human cytoplasmic β-actin.
For immunofluorescense, samples were fixed with 4% paraformaldehyde (w/v) in PBS, permeabilised with 0.2% Triton X-100 and then blocked with 10% normal donkey serum. Cells were stained with an anti-myc primary antibody (Invitrogen) and Gpp130 – a transmembrane protein localized to the cis-Golgi (Covance Research Products). These were followed with a Cy3- conjugated donkey anti-mouse secondary to detect the Myc protein, and Cy5- conjugated donkey anti-rabbit secondary antibody, to detect the Gpp130 (Jackson Laboratories). Fluorescence images of the samples were obtained using a Zeiss LSM-510 confocal laser-scanning microscope (Zeiss LSM-510 system with inverted Axiovert 100 M microscope).
COS-7, HeLa or NIH 3T3 cells were seeded at 1 × 105 cells ml−1 into six well plates containing ethanol-sterilised coverslips and cells were grown to 90–95% confluence before treatment with the toxin preparations. The Mcf toxin preparation was then incubated with the cells for 1, 3, 6, 9, 16 or 24 h. For caspase inhibitor assays, cells were preincubated for 30 min with 100 µM zVAD-fmk (Calbiochem Cat no. 627610) before addition of Mcf preparation. As positive controls for apoptosis assays, cells were treated with 10 µM staurosporine and incubated for 3 h.
To derive apoptotic indices, cells were stained with 0.12 µg µl−1 Hoechst 33258 (Sigma) and visualized on a fluorescence microscope. Sample sizes of ∼1000 cells were counted and the proportion of cells demonstrating apoptotic nuclei in the total population expressed as a percentage. To detect nucleosomal DNA laddering, confluent T75 flasks of NIH 3T3 cells were incubated for 1, 3, 6 and 24 h with 5 µl ml−1 Mcf cytosolic fraction; pUC-18 cytosolic fraction was incubated with cells for 24 h as the control. Cells were detached (Trypsin-EDTA) and pelleted (1500 r.p.m. for 5 min). DNA purification and electrophoresis were preformed as described elsewhere (Metcalfe et al., 1999).
For the detection of caspase-3 activation, NIH 3T3 cells were incubated with Mcf toxin and controls as described above. Cells were harvested, then lysed in 100 µl NET buffer pH 7.6 (150 mM NaCl, 1 mM EDTA, 50 mM Tris, 0.1% NP40). Samples were kept on ice before centrifugation at 14 000 r.p.m. for 10 min. Protein concentration was measured using the Bradford Assay (Sigma) and samples diluted to equivalent protein concentrations. Caspase-3 substrate Ac-DEVD-pNA (Calbiochem) was added to a final concentration of 200 µM. Each sample was transferred to a 96-well plate and the absorbance increase at 405 nm, as a result of the cleavage of pNA, was measured over 60 min using a Thermomax kinetic microtitre plate reader (Molecular Devices) at 37°C.
For the detection of PARP cleavage, cell samples were treated as described previously, harvested and lysed in a NET buffer and the protein concentration of samples determined using the Bradford Assay (Sigma). Samples were diluted as required in 2× SDS-PAGE sample buffer and heated at 100°C for 10 min. Equivalent amounts of protein (15 µg per sample) were loaded onto a vertical 8% SDS-polyacrylamide gel with a 5% stacking gel layer and electrophoresed at 120 V for 90 min. Gels were electroblotted onto PVDF membrane (Bio-rad) at 15 V for 20 min. Membranes were soaked briefly in 0.1% Ponceau S Red stain (Sigma) and 5% acetic acid to check for protein transfer and the stain then removed by rinsing with water. The membranes were then blocked overnight in 5% milk protein solution at 4°C and then immunoblotted with 0.6 µg ml−1 monoclonal anti-PARP (R and D Systems) for 90 min. Detection was carried out using an alkaline-phosphatase conjugated anti-mouse IgG (Sigma), 1:10 000 dilution and developed using the 5-bromo-4-chloro-3-indolyl phosphate/nitro blue tetrazolium (BCIP/NBT) liquid substrate system (Sigma).
This work is supported by a grant from the BBSRC Exploiting Genomics Initiative to R.ff.-C. A. D. is supported by a BBSRC CASE studentship with Syngenta, Jealotts Hill. We thank members of the D. Clarke and S. Reynolds laboratories at Bath for suggestions and discussion and B. Reaves for the provision of antibodies and practical advice.