Nuclear localisation is crucial for the proapoptotic activity of the HtrA-like serine protease Nma111p

Higher organisms make use of programmed cell death (apoptosis) during development, morphogenesis and homeostasis to sculpt and maintain distinct tissues and organs. Cell-death programs also exist in lower eukaryotes, and the yeast Saccharomyces cerevisiae, for example, undergoes cell death upon viral infection as well as during chronological and replicative ageing (Herker et al., 2004; Laun et al., 2001; Reiter et al., 2005). Moreover, apoptosis in yeast can be stimulated by oxidative and osmotic stress, acetic acid, nitrogen oxide, the yeast mating type factor α or decreased actin dynamics (Almeida et al., 2007; Gourlay and Ayscough, 2006; Ludovico et al., 2001; Madeo et al., 1999; Severin and Hyman, 2002; Silva et al., 2005). In addition, the basic molecular machinery executing cell death is evolutionarily conserved and orthologues of caspases (Yca1p), the serine protease Omi (also known as HtrA2) (Nma111p; also known as Ynm3p), apoptosis-inducing factor (Aif1p), endonuclease G (Nuc1p) and the inhibitor of apoptosis protein (IAP) survivin (Bir1p) have been identified (Buttner et al., 2007; Madeo et al., 2002; Walter et al., 2006; Wissing et al., 2004). Yeast apoptotic death comes along with the typical diagnostic features of apoptosis, such as phosphatidylserine externalisation, DNA condensation and fragmentation, production of reactive oxygen species (ROS), cytoskeletal perturbations, histone H2B phosphorylation, cytochrome-c release from mitochondria, and dissipation of the mitochondrial transmembrane potential (Ahn et al., 2005; Fannjiang et al., 2004; Gourlay and Ayscough, 2005; Ludovico et al., 2002; Ludovico et al., 2001; Madeo et al., 1997).

The serine protease Nma111p belongs to the family of HtrA (high temperature requirement A) proteins, which are defined by a characteristic combination of a catalytic serine-protease domain with at least one PDZ (postsynaptic density 95/disc large/zona occludens) domain (Clausen et al., 2002; Pallen and Wren, 1997; Ponting, 1997; Vande Walle et al., 2008). Nma111p harbours an internal duplication of the HtrA-like sequence, with the N-terminal repeat retaining the catalytic triad residues of HtrA-like serine proteases (Clausen et al., 2002; Pallen and Wren, 1997; Ponting, 1997). The C-terminal repeat, by contrast, contains an incomplete serine-protease site and is supposed to be non-functional. Bacterial HtrA-family members have been implicated in stress tolerance and pathogenicity, whereas human and Drosophila Omi are mitochondrial proteins that contribute to apoptosis through caspase-dependent and -independent processes (Challa et al., 2007; Hegde et al., 2002; Igaki et al., 2007; Suzuki et al., 2001; Suzuki et al., 2004; Vande Walle et al., 2008; Verhagen et al., 2002). Nma111p is able to promote apoptotic cell death by, at least in part, degradation of Bir1p, the only identified IAP in S. cerevisiae (Fahrenkrog et al., 2004; Walter et al., 2006). In addition, this protein is implicated in lipid homeostasis and exhibits chaperone activity (Padmanabhan et al., 2009; Tong et al., 2006). In this context, both the apoptotic and the chaperone activity of Nma111p depend on its serine-protease activity, although the actual catalytic serine has remained controversial (Fahrenkrog et al., 2004; Padmanabhan et al., 2009).

Nma111p is a nuclear protein that interacts with nuclear-pore complexes (NPCs) (Fahrenkrog et al., 2004). Both Yca1p, the yeast caspase, and Bir1p, the only identified substrate for Nma111p, are also nuclear proteins (Walter et al., 2006). Therefore, some of the processes that occur in the cytoplasm during mammalian apoptosis seem to occur in the nucleus during yeast cell death. The importance of the nucleus for the yeast apoptotic programme is further supported by the notion that Aif1p and Nuc1p translocate from mitochondria to the nucleus upon induction of apoptosis (Buttner et al., 2007; Wissing et al., 2004).

Here, we identify and characterise the nuclear-localisation signal (NLS) of Nma111p and show that nuclear localisation is a prerequisite for the apoptotic activity of this protein. Nma111p exhibits a bipartite NLS and Nma111p nuclear import is mediated by the nuclear-import receptor Kap95p. In heterokaryon assays, we found that Nma111p is not shuttling between the nucleus and the cytoplasm. Mutations in the NLS of Nma111p lead to reduced sensitivity of the mutant cells to hydrogen peroxide (H₂O₂) treatment as well as to a prolonged life span during chronological ageing. Moreover, the Nma111p N-terminal HtrA repeat, which localises to the nucleus, is sufficient to promote apoptosis, whereas the C-terminal HtrA repeat, which lacks an NLS and an active catalytic site, fails to trigger cell death. Together, our data show that nuclear localisation, the N-terminal HtrA repeat and serine-protease activity are required for the proapoptotic activity of Nma111p.

Nma111p contains an HtrA-like serine-protease domain and is required for efficient apoptosis in yeast in response to several stresses, including elevated temperature and oxidative stress (Fahrenkrog et al., 2004). Interestingly, Nma111p is the only known HtrA-like protein that is localised primarily to the cell nucleus. In this study, we seek to determine whether Nma111p is an exclusively nuclear protein, how Nma111p nuclear localisation is mediated, and whether changes in nuclear localisation are important for Nma111p function in response to oxidative stress and/or during chronological ageing.

To determine the subcellular localisation of Nma111p under normal and apoptotic conditions, we expressed GFP fusion proteins in cells grown to log phase and observed fluorescence after exposure to the oxidising agent H₂O₂. A positive control for nuclear localisation, the SV40 `classical NLS' fused to GFP (cNLS-GFP), is primarily nuclear in the absence of H₂O₂ (Fig. 1A). However, the cNLS-GFP becomes more cytosolic after exposure to the oxidising agent, possibly as a result of diffusion out of the nucleus due to the increased nuclear-pore size in apoptotic cells (Mason et al., 2005). A different localisation pattern was observed for Nma111p. In the absence of H₂O₂, Nma111p-GFP fluorescence was nuclear (Fig. 1A) (Fahrenkrog et al., 2004). After exposure to 3 mM H₂O₂, Nma111p-GFP localisation remained unchanged, exhibiting intense fluorescence within the nucleus and lacking any detectable cytosolic staining. Thus, Nma111p is predominantly present within the nucleus under steady-state conditions in both the absence and presence of a concentration of H₂O₂ that induces apoptosis in yeast.

Although the steady-state localisation of Nma111p-GFP fluorescence is within the nucleus in both the presence and absence of H₂O₂, it is possible that Nma111p undergoes nucleocytoplasmic shuttling under either or both conditions. The appearance of nuclear localisation under steady-state conditions might simply be due to a higher rate of Nma111p protein import compared with export, resulting in a much greater protein concentration within the nucleus than the cytoplasm (DeLotto et al., 2007; Feng and Hopper, 2002; Selitrennik et al., 2006). To determine whether Nma111p undergoes nucleocytoplasmic shuttling, we performed a heterokaryon shuttling assay (Feng and Hopper, 2002) on cells expressing an Nma111p-GFP fusion protein. Briefly, we expressed Nma111p under control of the GAL1 promoter in a wild-type haploid yeast strain until a detectable amount of Nma111p-GFP was visible in the nucleus. We then repressed further Nma111p-GFP transcription by adding glucose and allowed the existing Nma111p-GFP to equilibrate within the cells. Next, we introduced haploid cells of the opposite mating type that contained a kar1-1 allele. Cells with this kar1 mutation are able to undergo cytoplasmic fusion with cells of the opposite mating type as an early stage of diploid-zygote formation, but are unable to complete karyogamy with a partner cell (Conde and Fink, 1976), thus generating a heterokaryon with two distinct nuclei derived from two distinct populations of haploid cells. Because one nucleus is from a cell expressing Nma111p-GFP and no new Nma111p-GFP is being synthesised, the only way the second nucleus can become fluorescent is by importing Nma111p-GFP that was exported or diffused out of the first. Thus, shuttling is detected through the observation of fluorescence in both nuclei of the zygote. Indeed, examination of heterozygotes in which one donor nucleus harbours Cca1-GFP, which shuttles to assist in tRNA export (Feng and Hopper, 2002), reveals the appearance of fluorescence in both nuclei, indicating that Cca1-GFP has been exported from one nucleus and some Cca1-GFP protein has been imported into the second (Fig. 1B). Conversely, the histone protein H2B does not shuttle (Mosammaparast et al., 2001), so we observed fluorescence in only one nucleus of the heterokaryon. When we observed the pattern of fluorescence generated by Nma111p-GFP in heterokaryons, we only saw Nma111p in a single nucleus (Fig. 1B, top row). Thus Nma111p-GFP does not shuttle between the nucleus and cytoplasm under steady-state conditions.

Nma111p is essential for efficient apoptosis in the presence of 3 mM H₂O₂ (Fahrenkrog et al., 2004). Although we observed that Nma111p does not shuttle under steady-state conditions, the possibility remained that the protein is selectively exported under conditions that induce apoptosis. To examine whether Nma111p shuttling is induced under apoptotic conditions, we repeated our shuttling assay using Nma111p-GFP, this time exposing zygotes to 3 mM H₂O₂ for 2 hours (data not shown) and 4 hours (Fig. 1C) after initiation of mating. In both the presence and absence of H₂O₂, heterokaryons containing Nma111p-GFP retain fluorescence in only a single nucleus. These data indicate that Nma111p does not leave the nucleus, even under conditions that induce apoptosis.

In order to investigate the targeting of Nma111p to the nucleus, we examined the amino acid sequence of the protein for domains that might contain potential NLSs. In silico examination of the entire Nma111p sequence using the PSORT (Nakai and Horton, 1999) or PredictNLS (Cokol et al., 2000) algorithms failed to identify any potential cNLS sequences (data not shown). However, careful manual analysis of the N-terminal region of Nma111p revealed two short basic clusters of three residues each at amino acids 9-11 and 28-30 (Fig. 2A). The close juxtaposition of these basic residues is similar to the organisation of a prototypical cNLS (Lange et al., 2007). To determine whether these basic clusters are sufficient for mediating nuclear import of a reporter protein, we generated two chimeric polypeptides that respectively included the first 35 (Nma111^1-35-GFP) and first 83 (Nma111^1-83-GFP) amino acids of Nma111p fused to GFP. Expression of these chimeras in wild-type yeast resulted in predominantly nuclear fluorescence, with a low level of cytosolic staining that was excluded from the vacuole (Fig. 2B,C). Thus, the first 35 amino acids of Nma111p, containing the two short basic clusters, are sufficient to function as an NLS.

In order to determine whether either or both of the clusters of basic residues in the N-terminus of Nma111p are necessary for the NLS activity of this region, we performed site-directed mutagenesis on Nma111p^1-83-GFP to generate chimeric proteins that contained the first 83 amino acids of Nma111p with either: the lysine (K) and arginine (R) residues at amino acids 9-11 (mutated NLS1) altered to alanines to make Nma111p^1-83NLS1Δ–GFP; the K and R residues at amino acids 28-30 (mutated NLS2) replaced with alanines (Nma111p^1-83NLS2Δ–GFP); or both basic clusters altered (Nma111p^1-83-NLS1ΔNLS2Δ–GFP). Each Nma111p mutant was then expressed in wild-type yeast and observed for localisation of fluorescence (Fig. 2B). Expression of Nma111p^1-83NLS1Δ–GFP resulted in substantially more cytosolic fluorescence than Nma111p^1-83-GFP, but some nuclear accumulation of the fusion protein was retained. Nma111p^1-83NLS2Δ–GFP was similarly found in the cytoplasm and nucleus, with a subtly greater accumulation within the nucleus. However, alteration of both basic clusters from this short region at the N-terminus of Nma111p (Nma111p^1-83-NLS1ΔNLS2Δ–GFP) resulted in almost no nuclear accumulation above that found within the cytosol. Thus, whereas the alteration of either basic cluster affects the efficiency of nuclear import of this region of Nma111p, altering the entire bipartite NLS is necessary to severely reduce the accumulation of the GFP fusion within the nucleus.

The two basic amino acid clusters in the N-terminus of Nma111p, which are necessary for efficient nuclear import, have the characteristics of being a cNLS imported by the heterodimeric Kap60p-Kap95p importin complex. In order to determine whether Kap95p is necessary for Nma111p import, we expressed Nma111p^1-83-GFP in yeast cells expressing a temperature-sensitive kap95-3 allele and assayed for nuclear fluorescence at the permissive and restrictive temperatures. At the permissive temperature, Nma111p^1-83-GFP expressed in a kap95-3 mutant was present both in the cytoplasm and the nucleus, with some accumulation in the nucleus (Fig. 2C). However, after a 2-hour shift to 37°C, Nma111p^1-83-GFP in kap95-3 cells was redistributed exclusively to the cytosol, with little detectable nuclear fluorescence. We did not observe this redistribution of Nma111p^1-83-GFP in cells expressing mutant alleles of the karyopherins msn5 (Fig. 2C) or crm1 (data not shown). These observations indicate that Kap95p is the primary karyopherin for importing the N-terminal `cNLS' of Nma111p.

Although these experiments suggest that the clusters of basic residues found in the N-terminal 30 amino acids of Nma111p function as a bipartite NLS, we sought to determine whether these sequences were necessary for import of the full-length Nma111p protein. To this end, we constructed a chimeric GFP fusion containing the entire 997 amino acids of the Nma111p protein. Expression of this protein in wild-type cells resulted in entirely nuclear fluorescence, with essentially no detectable Nma111p-GFP visible in the cytoplasm (Fig. 3A, top row). We then generated mutants with the three K and R residues in either NLS1 or NLS2 replaced with three alanines. Replacement of either basic cluster with three alanine residues resulted in a redistribution of the Nma111p-GFP protein so that strong cytosolic fluorescence was observed with no detectable GFP accumulation within the nucleus (Fig. 3A, second and third row). Removal of both basic clusters also resulted in exclusively cytosolic fluorescence (Fig. 3A, bottom row). Thus, in the context of the full-length Nma111p protein, the loss of either basic cluster from the bipartite NLS results in a loss of nuclear Nma111p accumulation, suggesting that both NLS1 and NLS2 are necessary for Nma111p nuclear import.

We also examined the localisation of full-length Nma111p in yeast lacking functional Kap95p and Msn5p. Wild-type, msn5Δ and kap95-3 cells were transformed with a plasmid expressing Nma111p-GFP from its endogenous promoter. Cells expressing Nma111p-GFP were grown at 24°C and shifted to 37°C for up to 5 hours, then observed for intracellular fluorescence (Fig. 3B). As observed for the Nma111p^1-83 fragment, full-length Nma111p-GFP was predominantly nuclear in wild-type cells and in cells lacking Msn5p at both 24°C and 37°C. A similar nuclear localisation was observed at both the permissive and restrictive temperatures in cells containing a kap95-3 allele, indicating that nuclear import and accumulation of full-length Nma111p is not solely dependent on Kap95p.

Previously, we have shown that Nma111p is able to promote apoptosis and that this proapoptotic activity depends on its serine-protease activity (Fahrenkrog et al., 2004). To determine whether nuclear localisation of Nma111p is crucial to trigger cell death, we next mutated the NLS1, NLS2 or both NLSs in the plasmid pNOPPATA1L-Nma111 (Fahrenkrog et al., 2004), as described for the GFP fusion proteins, and transformed the resulting plasmids into Δnma111 cells. First, we determined the subcellular localisation of the wild-type and the resulting mutant ProtA-Nma111p fusion proteins by indirect immunofluorescence microscopy. ProtA-Nma111p was found to be a nuclear protein in wild-type and kap95-3 cells, whereas the replacement of either or both NLSs with three alanine residues resulted in a redistribution of the ProtA-Nma111p protein with strong cytosolic fluorescence and no detectable accumulation of the fusion proteins within the nucleus (supplementary material Fig. S1).

Next, ProtA-Nma111p, ProtA–Nma111p-NLS1Δ, ProtA–Nma111p-NLS2Δ and ProtA–Nma111p-NLS1ΔNLS2Δ cells were incubated with 0.4 mM H₂O₂ for 4 hours and analysed for apoptotic hallmarks. Apoptotic features include chromatin condensation and fragmentation, single-strand DNA breaks and accumulation of ROS. As shown in Fig. 4A, ProtA-Nma111p cells showed accumulation of ROS as indicated by dihydroxyethidium (DHE) staining after treatment with H₂O₂. DHE reacts with ROS and forms red fluorescent ethidium (Sharikabad et al., 2001). ProtA–Nma111p-NLS1Δ, ProtA–Nma111p-NLS2Δ and ProtA–Nma111p-NLS1ΔNLS2Δ cells, by contrast, showed less DHE staining as compared with ProtA-Nma111p cells. Quantification of DHE staining revealed that about 11% of ProtA-Nma111p cells were ROS positive, as compared with only 3-4% of the NLS-mutant cells (Fig. 4B).

Single-strand DNA breaks can be detected by the TUNEL assay (Gavrieli et al., 1992; Gorczyca et al., 1993). The TUNEL test detects free 3′ ends, which are generated by chromosome fragmentation, by attaching labelled nucleotides with terminal deoxynucleotidyl transferase. Consistent with the ROS staining, ProtA-Nma111p cells were TUNEL positive and ProtA–Nma111p-NLS2Δ cells were partially positive, whereas ProtA–Nma111p-NLS1Δ and ProtA–Nma111p-NLS1ΔNLS2Δ cells were all TUNEL negative (supplementary material Fig. S2A, top two rows). Furthermore, annexin-V/propidium iodide (PI) costaining was used to discriminate between early apoptotic (annexin-V positive, PI negative), late apoptotic (annexin-V positive, PI positive) and necrotic (annexin-V negative, PI positive) cell death. This annexin-V/PI costaining revealed that ProtA-Nma111p cells mainly undergo apoptosis, whereas ProtA–Nma111p-NLS1Δ and ProtA–Nma111p-NLS1ΔNLS2Δ cells were neither apoptotic nor necrotic (Fig. S2A, bottom two rows).

Cell survival of ProtA-Nma111p cells was further tested in a clonogenicity assay (Buttner et al., 2007). Treatment with 0.4 mM H₂O₂ for 4 hours resulted in the death of yeast cells expressing ProtA-Nma111p (survival rate of less than 20%), whereas ProtA–Nma111p-NLS1Δ and ProtA–Nma111p-NLS1ΔNLS2Δ cells were largely unaffected (survival rates ∼60%; Fig. 4C). Interestingly, ProtA–Nma111p-NLS2Δ cells were more sensitive to H₂O₂ as compared with the other NLS-mutant cells, with about 30% of cells surviving (Fig. 4C), consistent with the stronger nuclear accumulation of the mutant proteins (Fig. 2B). In the absence of H₂O₂, all ProtA-Nma111p constructs showed similar survival rates of between 80-90% (Fig. 4C). Together, these data indicate that the nuclear localisation of Nma111p is required for its function in response to oxidative stress.

Chronologically aged yeast cells show features of apoptotic death and are considered to have undergone physiologically induced apoptosis (Fabrizio and Longo, 2008). We therefore investigated whether the nuclear localisation of Nma111p is relevant for cell death during chronological ageing. We observed that disruption of either NLS1 or NLS2 of Nma111p does not significantly delay the onset of cell death in chronologically aged cells after 14 days in culture, whereas, by contrast, combined disruption of the two NLSs significantly increased survival of the cells (Fig. 5A and supplementary material Fig. S3B). The survival rates were reproduced in 9-12 independent experiments. When these yeast cells were tested for apoptotic markers after 7 days in culture, Nma111p and Nma111p-NLS2Δ cells showed typical hallmarks of apoptosis, such as the production of ROS (as detected by DHE staining) or DNA condensation (as detected by DAPI staining), whereas Nma111p-NLS1Δ and Nma111p-NLS1ΔNLS2Δ cells lacked apoptotic markers (Fig. 5B,C; supplementary material Fig. S3A). Therefore, nuclear localisation of Nma111p is also fundamental for cell death during chronological ageing.

Nma111p belongs to the HtrA family of serine proteases and consists of two tandem HtrA repeats, each completed with two PDZ domains (Clausen et al., 2002; Pallen and Wren, 1997; Ponting, 1997). Whereas the N-terminal HtrA repeat harbours a complete catalytic triad characteristic for serine proteases, the second repeat lacks two of the three active-site residues (Pallen and Wren, 1997). To test whether the N-terminal HtrA repeat is required and sufficient to promote cell death, we generated two truncations of Nma111p that respectively expressed the N-terminal and the C-terminal HtrA repeat fused to protein A. The resulting plasmids, pNOPPATA1L–Nma111p-N (residues 2-449) and pNOPPATA1L–Nma111p-C (residues 450-997), were transformed into Δnma111 cells and indirect immunofluorescence microscopy revealed that ProtA–Nma111p-N was predominantly nuclear with some cytosolic staining, whereas ProtA–Nma111p-C showed no nuclear accumulation (supplementary material Fig. S4). Next, ProtA–Nma111p-N and ProtA–Nma111p-C cells were incubated with 0.4 mM H₂O₂ for 4 hours and analysed for apoptotic markers. ProtA–Nma111p-N cells produced ROS as determined by DHE staining (Fig. 6A), showed single-strand DNA breaks as detected by TUNEL staining (supplementary material Fig. S5A) and were annexin-V positive (supplementary material Fig. S5B), whereas ProtA–Nma111p-C cells showed no apoptotic markers. Quantification of DHE staining revealed that about 9.5% of ProtA–Nma111p-N cells were ROS positive, as compared with only 2.5% of the ProtA–Nma111p-C cells (Fig. 6B).

Next, ProtA–Nma111p-N and ProtA–Nma111p-C cells were tested for cell survival after H₂O₂ treatment. Whereas about 20% (19.6±6.9%) of ProtA–Nma111p-N cells survived, ∼36% (35.6±14.4%) of ProtA–Nma111p-C cells were resistant to H₂O₂ (Fig. 6C).

We have previously shown that serine 235 (S235) is required for the death-promoting activity of Nma111p (Fahrenkrog et al., 2004). To confirm that S235 is necessary for Nma111p activity and not simply localisation, we mutated this serine residue to a cysteine by oligonucleotide site-directed mutagenesis in the plasmid pNOPPATA1L–Nma111p-N. The resulting pNOPPATA1L–Nma111p-N(S235C) was transformed into Δnma111 cells and the subcellular localisation of the protein was found to be predominantly nuclear, based on indirect immunofluorescence microscopy (supplementary material Fig. S4). ProtA–Nma111p-N(S235C) cells that were treated with 0.4 mM H₂O₂ for 4 hours showed no production of ROS (Fig. 6A) and were TUNEL and annexin-V negative (supplementary material Fig. S5), similar to Δnma111 cells. Quantification of DHE staining revealed ∼3.5% of the ProtA–Nma111p-N(S235C) and the nma111-disrupted cells to be ROS positive (Fig. 6B). Consistent with these results, the clonogenicity assay revealed survival rates of about 40% (40.3±8.6%) for ProtA–Nma111p-N(S235C) cells and ∼30% (28.7±6.0%) for Δnma111 cells (Fig. 6C).

To elucidate the role of the two HtrA repeats of Nma111p under more natural conditions, we performed chronological ageing assays over 14 days and found that expression of ProtA–Nma111p-N led to increased cell death as compared with cells expressing ProtA–Nma111p-C or ProtA–Nma111p-N(S235C), and with Δnma111 cells (Fig. 7A and supplementary material Fig. S6B). When these yeast cells were tested for apoptotic markers after 7 days in culture, ProtA–Nma111p-N cells showed the production of ROS and showed DNA condensation, whereas ProtA–Nma111p-C, ProtA–Nma111p-N(S235C) and Δnma111 cells lacked apoptotic markers (Fig. 7B,C and supplementary material Fig. S6A).

Taken together, the N-terminal HtrA repeat of Nma111p is sufficient to promote apoptosis and the death-promoting activity of Nma111p is reduced by a mutation of S235 to cysteine.

The HtrA-like serine protease Nma111p is a nuclear protein that is able to promote apoptosis in yeast in a serine-protease-dependent manner. We show here that Nma111p harbours a classical bipartite NLS within the first 35 N-terminal amino acids and identified Kap95p as its nuclear-import receptor. Also, we show that Nma111p is a non-shuttling protein that, even under oxidative-stress conditions, remains nuclear. The proapoptotic activity of Nma111p requires its nuclear localisation, because mutations of any critical residue in the bipartite NLS reduced the ability of Nma111p to mediate apoptosis. We show further that the death-promoting activity of this protein is restricted to its N-terminal HtrA repeat, which harbours the NLSs and the active catalytic site of the protein.

Nma111p is a nuclear protein under steady-state conditions in the absence or presence of H₂O₂ added to induce apoptosis. By contrast, a nuclear reporter protein (i.e. cNLS-GFP) is primarily nuclear in the absence of H₂O₂, but cNLS-GFP becomes more cytosolic after exposure to the oxidising agent (Fig. 1A). The release of cNLS-GFP from the nucleus upon H₂O₂ stress is most probably simply due to increased size of the yeast NPC permeability barrier (Mason et al., 2005). Alternatively, Nma111p might be `anchored' inside the nucleus owing to an interaction with a thus-far-unknown binding partner.

Nuclear localisation of a protein typically requires the presence of an NLS. Nma111p is an exclusively nuclear protein and we have mapped a bipartite basic NLS within the first 35 amino acids of the protein. Such basic NLS sequences are most often recognised by the Kap60p-Kap95p import-receptor complex and consequently the expression of Nma111p^1-83-GFP in a kap95-3 mutant leads to cytoplasmic redistribution of the fusion protein (Fig. 2C). Surprisingly, full-length Nma111p is still present in the nucleus in this kap95 mutant. Similarly, ProtA-Nma111p is nuclear, whereas ProtA–Nma111p-N is cytoplasmic in the kap95-3 mutant (supplementary material Figs S1 and S2). These data suggest that Nma111p gets trapped inside the nucleus owing to an interaction with a thus-far-unknown binding partner and that the binding to this partner protein is mediated by the C-terminal HtrA repeat. This binding partner is unlikely to be Bir1p, because Bir1p interacts with the N-terminal HtrA repeat of Nma111p (Walter et al., 2006). Alternatively, more than one Kap protein could mediate the nuclear import of the full-length protein, as is known for histones and ribosomal proteins (Mosammaparast et al., 2002; Mosammaparast et al., 2001; Rout et al., 1997). Post-translational modifications can also affect protein localisation and, for example, mask NLSs (Poon and Jans, 2005). The human homologue of Nma111p, Omi, is regulated by phosphorylation (Plun-Favreau et al., 2007) and Nma111p is phosphorylated at S989 in the C-terminal HtrA repeat (http://www.yeastgenome.org/). It will be interesting to see whether this phosphorylation site in fact is implicated in the regulation of Nma111p nuclear import and/or retention.

Our previous studies revealed that Nma111p and its only known substrate, the IAP Bir1p, are both nuclear proteins (Fahrenkrog et al., 2004; Walter et al., 2006), which indicated that the nucleus seems to play a significant role in yeast apoptosis. This is further supported by our data presented here, which revealed that Nma111p is not shuttling between the nucleus and the cytoplasm under normal and oxidative-stress conditions (Fig. 1). Moreover, disruption of any of the two basic stretches that act as NLSs for Nma111p affects its ability to promote cell death in response to oxidative stress and during chronological aging (Figs 4 and 5; supplementary material Figs S2 and S3). Therefore, Nma111p function in yeast apoptosis seems to not be linked to mitochondria, which is in clear contrast to its metazoan homologue Omi, which is predominantly localised to mitochondria and is released into the cytosol under apoptotic conditions. This, in turn, allows its interaction with and degradation of the IAP XIAP, leading to the activation of executioner caspases (Challa et al., 2007; Hegde et al., 2002; Khan et al., 2008; Martins et al., 2002; Suzuki et al., 2001; Suzuki et al., 2004). It still remains to be seen which proteins are the downstream targets of Nma111p that lead to the execution of apoptosis. Bir1p is one such target and degradation of Bir1p by Nma111p induces apoptosis, but Bir1p does not directly inhibit the yeast caspase Yca1p (Walter et al., 2006). Therefore, the bridging factor to the caspase or an unknown executioner of apoptosis in yeast remains to be elucidated, but it is most likely a nuclear protein.

Nma111p consists of two tandem HtrA repeats with the N-terminal HtrA repeat harbouring a complete catalytic triad characteristic for serine proteases, whereas the second repeat lacks two of the three active-site residues (Pallen and Wren, 1997). Consequently, we show here that the N-terminal HtrA repeat of Nma111p is sufficient and required to induce apoptosis in response to oxidative stress and during chronological ageing (Figs 6 and 7; supplementary material Figs S5 and S6). The active catalytic serine of trypsin-like serine proteases is typically embedded in a sequence motif GNSGG as consensus (Clausen et al., 2002; Pallen and Wren, 1997), which, in Nma111p, is 234GSSGS238, accordingly. We have previously shown that a mutation in S235 inhibits the ability of Nma111p to promote cell death (Fahrenkrog et al., 2004), whereas others have recently shown that mutation of S236 impairs its chaperone activity (Padmanabhan et al., 2009). Consistent with our previous data, we found here that cells that express ProtA–Nma111p-N with a cysteine mutation at S235 are less sensitive to oxidative-stress- and ageing-induced apoptosis than cells expressing a wild-type ProtA–Nma111p-N (Figs 6 and 7; supplementary material Figs S5 and S6). Therefore, S235 is unambiguously important for the proapoptotic activity of Nma111p. The controversy between our data and that of Padmanabhan et al. (Padmanabhan et al., 2009) most probably arises from the different strain background used. Whereas in our background, i.e. BMA41/BMA64/W303 (Fahrenkrog et al., 2004), and in BY4741 wild-type cells (Tong et al., 2006) deletion of nma111 causes no defects in growth or morphology, it does in the YB332 background used by Padmanabhan et al. (Tong et al., 2006), indicating some strain-specific features of the YB332 derivatives. Alternatively, S235 and S236 are both crucial for the catalytic activity of Nma111p; S235 primarily for its proapoptotic activity and S236 primarily for its chaperone activity. Future studies are required to address this issue more systematically.

In summary, we have identified and characterised the NLS and nuclear-import receptor of Nma111p. Moreover, Nma111p is a non-shuttling protein that remains in the nucleus under steady-state as well as apoptotic conditions. This inability of Nma111p to exit the nucleus in essence excludes a role for Nma111p in the mitochondria-dependent apoptotic pathway in yeast. This is further supported by our data that indicate that Nma111p exerts its apoptotic activity in the nucleus and that nuclear localisation is crucial for Nma111p-dependent cell death. Therefore, nuclear signalling cascades seem to be of utmost significance for the execution of apoptosis in yeast.

Enzymes for molecular biology were purchased from New England Biolabs (Beverly, MA) and Sigma-Aldrich (St Louis, MO), and were used as per the manufacturer's instructions. Yeast transformations were performed as described (Woods and Gietz, 2001), as were genetic manipulations, yeast cell culture and media preparation (Guthrie, 1991). Plasmids pGAL::CCA-GFP and pGAL::H2B-GFP were generous gifts from Anita Hopper (Penn State, Hershey, PA). Yeast strains and plasmids used are indicated in Tables 1 and 2, respectively.

Plasmids pKBB282 (Nma111^1-35-GFP) and pKBB280 (Nma111p^1-83-GFP) were constructed by amplifying the DNA encoding the NMA111 promoter and the first 35 or 83 amino acids of Nma111p, respectively, and inserting the resulting DNA into pLDB350 (CEN URA3 GFP) by homologous recombination. For pKBB282, NMA111 DNA was amplified using primers KOL 155 (5′-GTACCGGGCCCCCCCTCGAGGTCGACGGTATCGATAAGCTTGATATCGAATTCCTGCAGGCATCAGCATCAGCAAGATCC-3′) and KOL157 (5′-TCCAACAAGAATTGGGACAACTCCAGTGAAGAGTTCTTCTCCTTTGCTGGCGCTTTCCAACTGTTTCCTTTTTACAAGCG-3′). For pKBB280, amplification was performed using KOL155 and KOL156 (5′-ATCTAATTCAACAAGAATTGGGACAACTCCAGTGAAGAGTTCTTCTCCTTTGCTCACTGATTTAACAACGTTGGAGATGG-3′). pLDB350 was digested with EcoRI prior to co-transformation into yeast with the NMA111 PCR products to stimulate homologous recombination. pKBB428 (Nma111p^1-83-NLS1Δ–GFP) was generated by amplifying NMA111 by PCR using KOL156 and mutagenic primer KOL207 (5′-CAGTAAAGGTTTTTTAGATCTACTAATGACCATATCGTTGAGCAATATAGCTGCTGCTGACCATTCTAAAATTTCCG-3′), and inserting the resulting product into pLDB352 as described above. pKBB430 (Nma111p^1-83-NLS2Δ–GFP) was similarly constructed using KOL155 and KOL209 (5′-TCATGGTCTGTATATTCTTCCTCTTGATCTCCGGTGGCGCTTTCCAACTGAGCAGCAGCTACAAGCGATGATTCACCAG-3′). pKBB394 (Nma111p^1-83-NLS1ΔNLS2Δ–GFP) was generated using KOL209 and KOL207 as primers for amplification. pKBB460 (Nma111p-GFP) was constructed by PCR-amplifying the entire NMA111 gene sequence, including 500 nucleotides of the upstream promoter, using oligonucleotides KOL155 and KOL278 (5′-CATCACCATCTAATTCAACAAGAATTGGGACAACTCCAGTGAAGAGTTCTTCTCCTTTGCTAGCTTTTTCACTTTGGCTGTTGCC-3′) and integrating the resulting DNA into pLDB351 using homologous recombination as described above. pKBB465 (Nma111p-NLS1Δ–GFP) was generated by site-directed mutagenesis of pKBB460 using U.S.E. mutagenesis (Pharmacia, NY) and mutagenic oligonucleotide KOL288 (5′-CCATATCGTTGAGCAATATAGCGGCAGCAGACCATTCTAAAATTTCCGATGG-3′) as per the manufacturer's instructions. pKBB467 (Nma111p-NLS2Δ–GFP) was similarly generated using KOL289 (5′-GGTGAATCATCGCTTGTAGCAGCGGCACAGTTGGAAAGCGCCACCGG-3′) and pKBB466. Nma111p-NLS1ΔNLS2Δ–GFP was made by site-directed mutagenesis using both KOL288 and KOL289. All plasmids were rescued from yeast using glass-bead lysis (Hoffman and Winston, 1987) and transformed into Escherichia coli. pKBB439 (2 μ URA3 GAL::NMA111-GFP) was constructed by PCR-amplifying the entire coding region of NMA111 using oligonucleotides KOL277 (5′-CAACAAAAAATTGTTAATATACCTCTATACTTTAACGTCAAGGAGAAAAAACTATAATGACCATATCGTTGAGC-3′) and KOL278, and inserting the resulting DNA into pLDB352 by homologous recombination. Expression of Nma111p-GFP in media containing 2% galactose and the lack of expression in the presence of 2% dextrose was confirmed by western blotting using anti-GFP antibodies (Roche Pharmaceuticals, Basel). The complete nucleotide sequence of all NMA111 fusions was confirmed using dideoxynucleotide sequencing on an AbiPrism 310 Genetic Analyzer (Applied Biosystems, Foster City, CA).

Nucleocytoplasmic shuttling of Nma111p was performed by the method of Feng and Hopper (Feng and Hopper, 2002) with the following modifications. Briefly, pKBB439 (2 μ URA3 GAL::Nma111p-GFP), pGAL::Cca1-GFP and pGAL::H2B-GFP were transformed into BY4741. Transformants were grown overnight in SD –Ura + 2% raffinose to A₆₀₀ of 0.05-0.2, then supplemented with galactose to 2% and grown for 2.5 hours at 30°C. Cells were harvested by centrifugation and resuspended in SD –Ura containing 2% glucose and incubated for 2 hours at 30°C. Cells were then mixed with an equivalent number of MS739 (kar1-1) cells, centrifuged and resuspended in 50 ml SD –Ura. A slurry of cells was spotted on a YPD plate and incubated at 30°C. Samples collected 3 hours and 8 hours after mating were fixed in 70% EtOH at 4°C and DAPI stained for DNA visualisation.

To examine Nma111p protein localisation, plasmids pKBB282 (Nma111p^1-35-GFP), pKBB280 (Nma111p^1-83-GFP), pKBB428 (Nma111p^1-83-NLS1Δ–GFP), pKBB430 (Nma111p^1-83-NLS2Δ–GFP), pKBB394 (Nma111p^1-83-NLS1ΔNLS2Δ–GFP), pKBB460 (Nma111p-GFP), pKBB465 (Nma111p-NLS1Δ–GFP), pKBB467 (Nma111p-NLS2Δ–GFP), and pKBB466 (Nma111p-NLS1ΔNLS2Δ-GFP) were transformed into strains W303 and BY4741, grown in SD –Ura to A₆₀₀ 0.1-0.4, and observed by direct fluorescence microscopy using a Nikon E600 epifluorescence microscope. Localisation in karyopherin mutants was performed by transforming plasmids into PSY1102 (kap95^ts), EY10609 (msn5Δ) and LDY1008 (crm1^ts). Cells were grown overnight at 25°C in SD –Ura to early log phase, then shifted to 37°C for 2-4 hours and observed by direct immunofluorescence microscopy. Images were captured using SPOT cameras and software (Diagnostic Instruments, Sterling Heights, MI) and final images were produced in Adobe Photoshop CS (Adobe Systems, San Jose, CA).

Indirect immunofluorescence microscopy was performed as described (Fahrenkrog et al., 2004). Primary antibodies were anti-protein-A (Sigma, St Louis, MO) diluted 1:1000. Secondary antibodies were Alexa-Fluor-488-labelled anti-rabbit-IgG antibody (Molecular Probes, Eugene, OR) diluted 1:1000. Images were recorded using a confocal laser scanning microscope (Leica TCS NT/SP1, Leica, Vienna, Austria) and were analysed using NIH Image and Adobe Photoshop CS (Adobe Systems, San Jose, CA).

For DHE staining, 1×10⁷ cells were harvested by centrifugation, resuspended in 1 ml of 2.5 μg/ml DHE in PBS and incubated for 15 minutes in the dark. Cells were washed with 1 ml PBS and analysed by fluorescence microscopy. DNA was stained using Mowiol containing 1 μg/ml DAPI as mounting medium. TUNEL assay and survival platings/clonogenicity assays were performed as described (Fahrenkrog et al., 2004; Walter et al., 2006). Chronological ageing assays were performed as described elsewhere (Walter et al., 2006).

Exposed phosphatidylserine was detected by reaction with FITC-coupled annexin V (Annexin-V-FITC Apoptosis detection Kit I, BD Bioscience). Yeast cells were washed in sorbitol buffer (1.2 M sorbitol, 0.1 mM KPP, pH 7.4) and digested with zymoylase in sorbitol buffer for 30 minutes at 30°C. Next, cells were washed in incubation buffer (10 mM HEPES, 140 mM NaCl, 5 mM CaCl₂, 0.6 mM sorbitol), resuspended in 30 μl incubation buffer containing 6 μl PI (50 μg/ml) and 3 μl annexin-V–FITC and incubated for 20 minutes at room temperature. The cells were harvested, resuspended in sorbitol buffer, applied to a microscopic slide and imaged by confocal microscopy.