Developmental Cell Death in Dictyostelium Does Not Require Paracaspase*

Apoptotic cell death often requires caspases. Caspases are part of a family of related molecules including also paracaspases and metacaspases. Are molecules of this family generally involved in cell death? More specifi-cally, do non-apoptotic caspase-independent types of cell death require paracaspases or metacaspases? Dictyostelium discoideum lends itself well to answering these questions because 1) vacuolar it These were tested systems, developmental the S. cerevisiae meta- caspase (GenBank TM accession number NP_014840), caspase of Danio rerio (GenBank TM accession number AB047003), amino acids 151–402 of mouse caspase 1 (GenBank TM accession number P29452), and amino acids 1–350 of a Cyanobacteria “pseudocaspase” (NP_487716). These probes were used for blastp search on the NCBI site (www.ncbi.nlm. nih.gov/BLAST/, limited to Dictyostelium proteins, default parameters, expect maximum 10), psi-blast on the NCBI site (default parameters, until convergence), wu-blast on the Sanger site (www.sanger.ac.uk/ Projects/D_discoideum/blast_server.shtml, on all of the contigs produced by the international consortium, January 2003 update, default parameters), and tblastn on the Jena site (genome.imb-jena.de/Dictyo-stelium/, on all available sequences; shotgun Baylor (cid:4) GCSJ (cid:4) Sanger, ESTs, GenBank TM , mtDNA, rDNA, expect 10).

Apoptotic cell death often requires caspases. Caspases are part of a family of related molecules including also paracaspases and metacaspases. Are molecules of this family generally involved in cell death? More specifically, do non-apoptotic caspase-independent types of cell death require paracaspases or metacaspases? Dictyostelium discoideum lends itself well to answering these questions because 1) it undergoes non-apoptotic developmental cell death of a vacuolar autophagic type and 2) it bears neither caspase nor metacaspase genes and apparently only one paracaspase gene. This only paracaspase gene can be inactivated by homologous recombination. Paracaspase-null clones were thus obtained in each of four distinct Dictyostelium strains. These clones were tested in two systems, developmental stalk cell death in vivo and vacuolar autophagic cell death in a monolayer system mimicking developmental cell death. Compared with parent cells, all of the paracaspase-null cells showed unaltered cell death in both test systems. In addition, paracaspase inactivation led to no alteration in development or interaction with a range of bacteria. Thus, in Dictyostelium, vacuolar programmed cell death in development and in a monolayer model in vitro would seem not to require paracaspase. To our knowledge, this is the first instance of developmental programmed cell death shown to be independent of any caspase, paracaspase or metacaspase. These results have implications as to the relationship in evolution between cell death and the caspase family.
Caspases, a subset of cysteine proteases, play an essential role in apoptotic death in animals (1,2). The caspase family includes not only caspases but also other members called paracaspases and metacaspases. Paracaspases were identified through their remote homology with caspases, and metacaspases were identified in turn through their remote homology with paracaspases (3). Within eukaryotes, paracaspases are found in animals and Dictyostelium discoideum and meta-caspases are found in plants, fungi, and some protists (3,4). The role of these caspase-related proteases is largely unknown. In particular, it is not known whether they are involved in cell death. A thorough investigation into the role of the human paracaspase, MALT1, showed that MALT1 as a complex with Bcl10 or as a fusion protein was able to activate NF-B (3, 5), but no direct involvement of MALT1 in cell death effector mechanisms could be demonstrated (3). From a more general point of view, investigating the role of caspase family members in cell death may be difficult in animals where they are numerous and may have redundant roles.
Model organisms that harbor only one member of the caspase family, thus excluding any possibility of functional redundancy within this family, would seem more appropriate for functional investigation. Thus, the yeast Saccharomyces cerevisiae has only one metacaspase gene, Yor197w, and no caspases or paracaspases (3). Yeast cell death with some traits of apoptosis occurs in certain cdc48 mutants (6) or upon treatment with low doses of H 2 O 2 (7). Overexpression of Yor197w increases this cell death, and disruption of the Yor197w gene prevents it (8), strongly suggesting that a metacaspase may be required for apoptotic-like cell death in this organism.
The protist Dictyostelium bears only one paracaspase (pcp) 1 and neither metacaspase nor caspase genes (for review see Refs. 3 and 4) (see "Results"). Paracaspases share with caspases the conserved Cys-His catalytic diad, but the catalytic Cys resides within a context that is different from the QACXG prototypic caspase sequence, probably leading to a different specificity. The Dictyostelium paracaspase does not bear the death domain and immunoglobulin domains found in metazoan paracaspases (3,9). Cell death can be observed in Dictyostelium in vivo upon starvation, which triggers Dictyostelium cells to aggregate, differentiate, and develop into 1-2-mm high multicellular fruiting bodies. Each of these is made of a mass of spores supported by a stalk where cells are massively vacuolated and dead (10). This developmental cell death can be mimicked and studied in detail in vitro using Dictyostelium HMX44A cells, which upon starvation and addition of the morphogen differentiation-inducing factor (DIF) differentiate as a monolayer from vegetative to "stalk" cells (11), undergoing caspase-independent (12) autophagic vacuolar (13,14) cell death.
We obtained paracaspase-null (pcpϪ) clones by homologous recombination in this haploid organism in four distinct Dictyo-stelium strains. We tested their cell death behavior both in vivo in stalks for three strains and in vitro in the monolayer system for all four strains. Compared with parent cells, all of the pcpϪ cells showed undiminished cell death in vivo and in vitro. In addition, paracaspase inactivation led to no alteration of development or difference as to interaction with a range of bacteria. Thus, in Dictyostelium, programmed cell death does not require paracaspase.

EXPERIMENTAL PROCEDURES
Cells and Cell Culture-D. discoideum axenic strain HMX44A was cloned from HMX44 cells adapted to axenic growth (a kind gift from J. G. Williams). Upon starvation HM44 and HMX44A cells did not morphogenize and produced only little of the differentiation factor DIF-1. Upon the addition of DIF-1, these starved cells differentiated to stalk cells, i.e. dead cells, as a monolayer (11,13). HMX44A, AX2, and K-AX3 cells were grown in HL5-modified medium (13). The thymidine auxotroph JH10 cells (15) were grown in HL5 supplemented with 100 g/ml thymidine (Sigma). HMX44A, AX2, and KAX3 pcpϪ cells were selected and grown in HL5 supplemented with 10 g/ml blasticidine (Invitrogen), and JH10 pcpϪ cells were selected and grown in HL5. All of the cultures were at 23°C.
Preparation of pcpϪ Cells-The paracaspase genomic sequence (Gen-Bank TM accession number AF316600) (3) was amplified by PCR from AX2 genomic DNA with forward primer (5Ј-CCGCGGATAAAAATAG-AAAGATAAAAA-3Ј) and reverse primer (5Ј-TTGTTTTTTCTTTCAGT-3Ј). The PCR product was cloned into pGEMT-easy vector (Promega). The bsr cassette of vector pBsR503 (16) flanked by EcoRV restriction sites was inserted into the HpaI site of the paracaspase sequence cloned into pGEMT-easy. The DNA was digested with AccI (leading to an extra 400-bp fragment because of an AccI site in the pGEM polylinker) and SacII (a restriction site present in the forward primer) (Fig. 1). The 2280-bp insert was purified from gel fractionation and electrotransfected into HMX44A, AX2, or K-AX3 cells, which were subsequently selected with blasticidine (Invitrogen).
The thymidine auxotroph JH10 cells (15) were transfected with a construct containing the 3.2-kb Thy1 (17) cassette from the pJH60 vector (a gift from J. A. Hadwiger). The Thy1 cassette was inserted in the paracaspase/bsr vector above after BamHI removal of the bsr cassette. The resulting 4080-bp insert was purified and electrotransfected into JH10 cells. Cells were selected for their ability to grow without thymidine.
Resistant cells were cloned by limiting dilution in microplates. Putative pcpϪ cell clones were first analyzed by PCR using the primers described above or the forward primer (5Ј-GATTCCACTGGTGCCATA-ATCAAATAC-3Ј) and reverse primer (5Ј-CTCCCCATCTCTACAACAA-TCTACG-3Ј) and confirmed by reverse transcription-PCR (data not shown) and Southern blots.
Southerns Blots-10 g of genomic DNA was digested by HindIII-AccI or SpeI. Southern blots were done using Biodyne B transfer membranes (Pall Gelman). Hybridization was performed at 60°C in Ex-pressHyb hybridization solution (Clontech) according to the manufacturer's instructions. Southern blots were probed with a 445-bp HindIII-SspI paracaspase fragment (from a HindIII site 314 bp 5Ј of HpaI to a SspI site 131 bp 3Ј of HpaI) and with a blasticidine probe (a 450-bp XhoI-BglII fragment of pUCBsRBamHI) or with a Thy1 probe (a 1.5-kb ClaI fragment of pJH60).
Development Assays-For development on filters, cells were washed twice with SB (Soerensen buffer 50 times: 100 mM Na 2 HPO 4 , 735 mM KH 2 PO 4 , pH 6.0) and resuspended in SB. Cells were starved on filter pads (18) (with the exception that cells were starved in SB), the pads were soaked with SB, and 50 l of cells at 5 ϫ 10 6 cells were spotted on the filters. The same conditions were used also for development on 1% agarose dishes (19). On both filters and agarose, upon incubation at 23°C fruiting bodies were obtained in 1-2 days.
For development on bacteria, bacteria previously grown on SM/5 plates (18) were harvested in SB and spread on SM/5 plates. Cells were plated (50 l of cells at 5 ϫ 10 6 cells) on the bacteria lawn and incubated for 3-5 days at 23°C. Plates were examined for plaques formed by Dictyostelium cells and for development with a binocular photomicroscope (Zeiss). Monolayer assays, cell fixation and staining, microscopy, and image processing were done as described previously (14).
Dictyostelium Genome Search-To search the Dictyostelium genome for molecules of the caspase family, we used as probes the caspase domain (amino acids 301-600) of the human paracaspase (GenBank TM accession number AAG38589), the already known Dictyostelium para-caspase (GenBank TM accession number AAG38592), the caspase domain (amino acids 251-500) of the Caenorhabditis elegans paracaspase (GenBank TM accession number AAG38591), the S. cerevisiae metacaspase (GenBank TM accession number NP_014840), caspase 3 of Danio rerio (GenBank TM accession number AB047003), amino acids 151-402 of mouse caspase 1 (GenBank TM accession number P29452), and amino acids 1-350 of a Cyanobacteria "pseudocaspase" (NP_487716). These probes were used for blastp search on the NCBI site (www.ncbi.nlm. nih.gov/BLAST/, limited to Dictyostelium proteins, default parameters, expect maximum 10), psi-blast on the NCBI site (default parameters, until convergence), wu-blast on the Sanger site (www.sanger.ac.uk/ Projects/D_discoideum/blast_server.shtml, on all of the contigs produced by the international consortium, January 2003 update, default parameters), and tblastn on the Jena site (genome.imb-jena.de/Dictyostelium/, on all available sequences; shotgun BaylorϩGCSJϩSanger, ESTs, GenBank TM , mtDNA, rDNA, expect 10). (November 2003), the whole chromosome shotgun approach to Dictyostelium genome sequencing has reached an ϳ8-fold coverage, leading to the probability of finding a given gene estimated to be higher than 0.98 (20) 2 The various Dictyostelium genome search approaches listed under "Experimental Procedures" most often yielded the already known Dictyostelium paracaspase with significant "e" values and sometimes yielded other Dictyostelium sequences, however, with very borderline expect values. Only one of the latter sequence, AAO51557, showed a caspase-like domain but considerably truncated and with very low homology. Altogether, the only unambiguous member of the caspase family found in the thoroughly sequenced Dictyostelium genome remains the paracaspase gene AF316600/ AAG38592 identified previously (3).

Paracaspase Gene Inactivation Does Not Grossly Alter Dictyostelium Developmental Cell Death-Currently
We disrupted this paracaspase gene by homologous recombination in the K-AX3 strain (Fig. 1) and checked the impact thereof on development under three experimental conditions: on filters in starvation buffer, on agar in starvation buffer, and on a lawn of Klebsiella aerogenes. In each of these three conditions, development appeared similar in parent K-AX3 and in two independently obtained pcpϪ clones (Fig. 2). Testing on a lawn of bacteria checks not only development but also vegetative growth and ability to ingest and process bacteria and to resist to them, all of them clearly not dependent on paracaspase. In particular, stalk development appeared unaltered, suggesting that stalk cell death is unaffected by paracaspase gene inactivation.
Because of mostly unpublished reports that Dictyostelium mutant phenotypes can be strain-dependent, we disrupted the paracaspase gene also in strain AX2 cells (Fig. 1). In this case as well, development as tested on filters and on agar and in particular stalk cell development appeared unaltered by the disruption (data not shown). In both K-AX3 and AX2 cells, the paracaspase gene had been disrupted using a blasticidine resistance cassette. To eliminate possible bias resulting from expression of this cassette, we disrupted the paracaspase gene in another strain, JH10, using the Thy1 cassette and selecting for thymidine prototrophy (Fig. 1) (15). The resulting JH10 pcpϪ cells tested on filters and on agar also showed unaltered stalk development (data not shown). Thus, in three different Dictyostelium strains, stalk development, thus presumably stalk cell death, did not require an intact paracaspase gene.
To more directly check stalk cell death, K-AX3 stalks were examined after triple staining with fluorescein diacetate (FDA), propidium iodide (PI), and calcofluor. Both parent and pcpϪ stalks showed the usual pattern of dying vacuolated cells stained with FDA and later with PI and synthesizing cellulose as shown through calcofluor staining (Fig. 3). Thus, by these criteria, stalk cell death in vivo does not require an intact paracaspase gene.
Paracaspase Gene Inactivation Does Not Alter DIF-induced Monolayer Cell Death-Stalk cell death in vivo can be mimicked in vitro in a monolayer system where cell death is induced by starvation and DIF (11). Induction of vacuolated dead cells in the monolayer system was not impaired in both independently derived pcpϪ K-AX3 clones compared with parent cells (Fig. 4). Very similar results were obtained with parent and pcpϪ AX2 and JH10 cells (data not shown). In these strains as well, cell death in the monolayer system was not impaired in pcpϪ cells.
The stages of Dictyostelium cell death have been described in detail (13,14) using cells of another strain, HMX44A, as a monolayer. Compared with the in vivo situation or with K-AX3 cell monolayers, this system could permit the detection of more subtle alterations of the cell death process. Therefore, we inactivated by homologous recombination the paracaspase gene in Dictyostelium HMX44A cells and the resulting pcpϪ clone (Fig.  1) was functionally investigated in more detail. Upon induction by starvation and addition of DIF, the morphological stages on the cell death pathway (emergence and demise of paddle cells, cellulose synthesis, paddle-to-round cell transition, actin depolymerization, vacuolization) seen in parent HMX44A cells (14) were also observed in pcpϪ cells (Fig. 5). To control for any clonal effect, the homologous recombination process in HMX44A cells was repeated in an additional, separate round of transfection and yielded another pcpϪ clone, which could also be induced to die (data not shown).
Whereas the HMX44A strain was derived from the initial V12M2 Dictyostelium isolate, the AX2, K-AX3, and JH10 strains were derived from the NC4 isolate (see dictybase.org/ strain_history.htm). Thus, in pcpϪ cell clones belonging to four distinct Dictyostelium strains of two distinct derivations, the paracaspase gene was not required for DIF-induced vacuolar cell death in the monolayer system.
Paracaspase Is Not Required for Staurosporine-induced Cell Alterations or for Growth and Development on Some Bacteria- Most if not all animal cells would die when subjected to the protein kinase inhibitor staurosporine, and this death is often caspase-mediated (21,22). If staurosporine-induced cell death requires caspases in animal cells, staurosporine-induced cell death or other effects might require the caspase-related paracaspase in Dictyostelium. We subjected vegetative K-AX3 cells to 1 M staurosporine in HL5 medium for 15 h. Unexpectedly, the cells acquired a flat morphology with marked extension of filopodia often appearing as networks (data not shown). Similar although less marked effects could be seen at 0.1 M staurosporine. These cells were apparently not dead, because upon replacement of staurosporine-containing HL5 with fresh HL5, they reverted to vegetative cell morphology (data not shown). Thus, in Dictyostelium cells and under conditions similar to the ones that induce death in animal cells, staurosporine did not induce cell destruction but induced reversible alterations in cell morphology. We then subjected K-AX3 pcpϪ cells and also parent and pcpϪ HMX44A cells to staurosporine and observed the same alterations (data not shown). Thus, in Dictyostelium cells, under these conditions staurosporine induces reversible alterations and these do not require paracaspase.
Because of the relationship between cell death and some defense reactions, we wondered whether paracaspase could be involved in interactions between Dictyostelium and bacteria. Therefore, we tested a range of bacteria (Escherichia coli, Aeromonas hydrophila, Serratia marcescens, Salmonella typhimurium, Pseudomonas aeruginosa, Agrobacterium tumefaciens, Xanthomonas campestris oryzae, Bacillus megaterium, Shewanella putrefaciens, Erwinia carotovora carotovora, and Erwinia chrisanthemi) for their ability to support vegetative growth and subsequent plaque development of parental or pcpϪ K-AX3 cells. We found no detectable difference linked to the absence of the paracaspase gene (data not shown).

DISCUSSION
The purpose of this work was to check whether the paracaspase molecule was necessary for Dictyostelium cell death. The experimental answer seems to be that it is not. To our knowledge Dictyostelium is the only developmentally competent eukaryote so far, in which through a combination of natural absence of caspases and metacaspases and experimental inactivation of paracaspase, it has been possible to show the persistence of cell death in the genetic absence of all known caspases and paracaspases and metacaspases. Strengthening this conclusion, these results were obtained with four Dictyostelium strains derived from two distinct initial Dictyostelium isolates using two selection agents and two cell death procedures. However, there are some limitations to these conclusions. There is still a very remote possibility that a gene encoding a member of the caspase family lies in the few unexplored segments of the Dictyostelium genome. Also, we cannot exclude a role in Dictyostelium cell death of more remote members of the caspase-hemoglobinase fold protease family (4) nor of some other proteases that might have similar enzymatic activity. Clearly, other cysteine proteases could be involved in this cell death. Changes in protease activity have previously been found associated with Dictyostelium cell differentiation including that of stalk cells (23). However, we could not inhibit Dictyostelium cell death using even high concentrations of several calpain inhibitors (data not shown). From another point of view, we cannot exclude the possibility that, in particular through the previously indicated domain differences, animal paracaspases might play a role in some instances of cell death.
Dictyostelium cells inactivated for the paracaspase gene showed unchanged cell death but were also not grossly affected as to vegetative growth, co-existence with some bacteria, staurosporine-induced alterations, and overall development. Although our purpose was to explore the relationship between paracaspase and cell death, leading to the conclusion that paracaspase was not required for Dictyostelium cell death, paracaspase clearly did not seem to be required for a number of other key functions in Dictyostelium. We did not investigate further the function of this paracaspase.
These findings have at least two implications. First, in the face of complexity of apoptosis (24) and of multiplicity of nonapoptotic types of cell death, moreover often intertwined within the same dying cell (25), there is a need for "cleaner" cell death models exhibiting only one type of cell death. These models would often belong to non-animal kingdoms and would ideally lend themselves to a trans-kingdom approach of cell death mechanisms (26). Within such cell death models, D. discoideum can be used to explore the mechanism of vacuolar autophagic cell death. The present results contribute to fully validate it as a model for this caspase-independent cell death.
A second implication bears on the role of the caspase family in cell death mechanisms. To answer this question, two organisms, yeast and Dictyostelium, have been explored that show the favorable experimental situation of expressing only one member of this family. However, yeast cell death was apparently dependent on this caspase-extended family member (8) but Dictyostelium cell death was not. A number of factors may explain these contrasting results. From a molecular point of view, molecular proximity to caspases of the non-caspases should matter. Paracaspases are closer to caspases than metacaspases (3,4). However, in apparent paradox, the Dictyostelium paracaspase is not involved in cell death (this report), whereas the yeast metacaspase is involved (8). Also, peculiarities of given metacaspases and paracaspases could be important such as the absence of some domains indicated above. From a phylogenetic point of view, the earlier emergence from the main eukaryotic lineage of Dictyostelium compared with yeast (27) and the possibly distinct mode and earlier time of entry into eukaryotes of metacaspases compared with paracaspases (28) may be significant. Also, the circumstances of cell death were quite different, namely development in Dictyostelium and not so in yeast. It could well be, however, that the most important factor is the type of caspase-independent cell death, apoptotic-like in yeast and not so in Dictyostelium. Although clearly more information in more organisms is needed, this report shows that, at least in Dictyostelium, developmental cell death can occur in the absence of any member of the caspase family, making a constitutive link throughout evolution between this caspase family and programmed cell death unlikely. Rather, from the caspase family, only or mostly bona fide caspases were recruited to cell death. This caspase recruitment seems to have occurred when animals emerged in evolution, perhaps mostly as a tool to contribute together with cell death phagocytosis to the disappearance of dead cells, which is remarkably more thorough and rapid in apoptosis in the animal kingdom than in most other types of cell death.