Apoptosis is a kind of programmed cell suicide crucial for health, homeostasis, and embryonic development. Its important role in different diseases such as cancer, neurodegenerative disorders, or stroke, and its very complex regulatory network were discovered in model organisms such as Drosophila melanogaster or Caenorhabditis elegans. Recent studies support the notion that apoptosis also occurs in Saccharomyces cerevisiae (Frohlich and Madeo, 2000 ; Madeo et al., 2002a ). Apoptosis in yeast has been demonstrated in very different cases, e.g., in cdc48 mutants (Madeo et al., 1997 ); in cells expressing Bax, the mammalian apoptosis inducer gene (Ligr et al., 1998 ); during perturbation of the vesicular trafficking (Levine et al., 2001 ) and salt stress (Huh et al., 2002 ); and after cell treatment with osmotin, an antifungal protein from tobacco implicated in host-plant defense (Narasimhan et al., 2001 ). Very recently, evidence for the existence of a caspase-related protease regulating apoptosis has been reported in yeast cells (Madeo et al., 2002b ). Most of these scenarios were related to oxygen stress, suggesting that reactive oxygen species (ROS) are key regulators of yeast apoptosis (Madeo et al., 1999 ).

Interestingly, it has also been reported that altered pre-mRNA splicing or mRNA stability are involved in mammalian apoptotic-linked diseases (Guhaniyogi and Brewer, 2001 ; Nissim-Rafinia and Kerem, 2002 ).

In previous work (Mazzoni and Falcone, 2001 ), we demonstrated that the Kluyveromyces lactis LSM4 gene (KlLSM4), and a truncated form (Kllsm4Δ1) still containing the Sm-like domains, could restore viability in a S. cerevisiae strain not expressing the endogenous Lsm4p, a subunit of the Lsm complex that is involved in mRNA decapping and splicing (Cooper et al., 1995 ; Tharun et al., 2000 ). We also reported that cells expressing KlLsm4Δ1p showed an increased loss of viability when reaching the stationary phase and, therefore, we wondered whether the lack of the C-terminal region of Lsm4p could lead to apoptotic death in yeast.

Because most of the reports concerning yeast apoptosis have been done with S. cerevisiae, we looked for the cytological markers of apoptosis in an S. cerevisiae strain deprived of the endogenous LSM4 gene and expressing the truncated form of the K. lactis gene.

Our results demonstrated that all the hallmarks of apoptosis are present in this strain and indicated that this phenomenon is due an increase in mRNA stability as also confirmed by the analysis of S. cerevisiae strains lacking specific components of the mRNA turnover machinery (i.e., LSM1, DCP1, DCP2, DHH1, and PAT1).

Table 1 Yeast strains used in this study

Cells were grown in YP (1% yeast extract, 2% peptone) supplemented with 2% glucose (YPD), 2% galactose (YPGal), or glycerol (YPGly) at 28°C (unless indicated). Solid media were supplemented with 2% bactoagar (Difco, Detroit, MI).

Induction of [rho°] was obtained by growing the MCY4/KlLSM4 and MCY4/Kllsm4'441 strains for 24 h in the presence of 25 μg/ml ethidium bromide. The [rho°] mutants were selected by their inability to grow on respiratory medium (YPGly) followed by observation of 4,6-diamidino-2-phenylindole (DAPI) staining.

The presence of free intracellular radicals or strongly oxidizing molecules (ROS) were detected with dihydrorhodamine (DHR) (D1054; Sigma-Aldrich, St. Louis, MO) as described previously (Madeo et al., 1999 ).

Free 3′-OH was detected by terminal deoxynucleotidyl transferase dUTP nick-end labeling (TUNEL) (Madeo et al., 1997 ) by using the Roche Diagnostics (Mannheim, Germany) in situ cell detection kit peroxidase and observed with an Axioscope microscope (Carl Zeiss, Jena, Germany).

Kaiser and Schekman, 1990

The samples were infiltrated overnight at 4°C in a 1:1 mixture of ethanol with Epon 812 embedding medium. The mixture was replaced with pure Epon 812, and the samples were allowed to polymerize at 60°C for 24 h. Ultrathin sections were stained with lead citrate before examination at electron microscope.

Schmitt et al., 1990

Sambrook et al., 1989

Saavedra et al., 1997

To test splicing efficency, 15 μg of total RNA was run onto 6% acrylamide/urea gel and elettroblotted on Hybond-N⁺ for 1 h at 60 V in Tris buffer 0.5×, UV cross-linked, and then hybridized by random priming labeled U3 probe, obtained by polymerase chain reaction amplification reaction from two oligonucleotides (5′-CCAACTTGGTTGATGAGTCC-3′; 5′-GGATGGGTCAAGATCATCGC-3′) complementary to the exon2 region of SNR17 gene (Hughes et al., 1987 ).

The hybridization was performed at 42°C in 50% formammide, 5× SSC, 10× Denhardt's, 0.3% SDS and 150 μg of single-stranded DNA. After hybridization the membrane was washed for 30 min at 37°C in 2× SSC and 0.2% SDS and then for 2 h at 48°C in 0.1× SSC and 0.4% SDS. As molecular marker, we used pBR322 digested with MspI.

Lazebnik et al., 1993

Figure 1 Chromatin fragmentation. DAPI staining and phase contrast of S. cerevisiae strain MCY4 containing plasmids pRS416/KlLSM4 (a and b), pRS416/Kllsm4Δ1 (c–f), and the corresponding [rho°] derivatives (KlLSM4, g and (more ...)

To be sure that the bright spots revealed with DAPI staining were actually derived from chromatin fragmentation and not due to mitochondrial DNA, we obtained by ethidium bromide treatment [rho°] strains from both the KlLSM4- and Kllsm4Δ1-expressing cells (see MATERIALS AND METHODS). As shown in Figure 1, also in the [rho°] strains, the expression of the entire gene KlLSM4 led to a normal morphology (Figure 1h) and DAPI staining (Figure 1g), whereas, in the case of the truncated gene, the morphology of cells was abnormal (Figure 1, j and l) and DAPI staining still showed chromatin fragmentation (Figure 1, i and k), indicating that the fluorescent DNA fragments observed had nuclear and not mitochondrial origin.

To have a closer look at these nuclear apoptotic effects, we undertook electron microscopy observations. As already detected by DAPI staining, MCY4 cells expressing the truncated KlLsm4p showed chromatin condensation and margination (Figure 2, arrows in b and c) along the inner part of the nuclear envelope, typical of cells undergoing apoptosis and, in some cases, we also observed nuclear fragments (Figure 2d). On the contrary, normal nuclear and cellular morphology was observed when we used as a control the MCY4 strain transformed with the entire KlLSM4 gene (Figure 2a) or expressing the endogenous LSM4 gene (our unpublished data).

Figure 2 Electron micrographs of strain MCY4-containing plasmids pRS416/KlLSM4 (a) and pRS416/Kllsm4Δ1 (b–d) grown on YPD. The arrows indicate the condensed and marginated chromatin. N, nucleus; M, mitochondria; V, vacuole. Bar, 1 μm.

DNA fragmentation is another hallmark of apoptosis and this phenomenon can be detected in situ by TUNEL staining, which reveals free 3′-OH ends originated by DNA breaks (Gavrieli et al., 1992 ).

The TUNEL test performed in cells expressing the wild-type LSM4 and the KlLSM4 genes (Figure 3, a and b) showed no or only slightly stained nuclei, whereas cells expressing the truncated form of the protein showed an intense staining, indicating that most of cells contained DNA strand breaks (Figure 3c).

Figure 3 DNA strand breakage detection. To express the endogenous LSM4p, MCY4 cells were grown exponentially on galactose (a), whereas MCY4 cells containing plasmids pRS416/KlLSM4 (b) and pRS416/KllsmΔ1 (c) were grown on YPD. After cell fixing and digestion (more ...)

Electrophoretic analysis of isolated chromosomal DNA from exponential and stationary phase cultures did not revealed the DNA ladder that has been observed in most, but not all, apoptotic systems (Cornillon et al., 1994 ) (our unpublished data). This observation, also reported by other authors studying yeast apoptosis, can be explained by the fact that S. cerevisiae chromatin structure has short or no linker DNA between nucleosomes (Madeo et al., 1999 ).

Madeo et al., 1999

Laun et al., 2001

Schulz et al., 1996

Substantially, <1% of cells expressing the entire KlLSM4 gene showed a marginal fluorescence as described previously (Madeo et al., 1999 ), whereas ~40% of cells expressing the truncated form of the gene showed an intense intracellular staining with DHR (Figure 4, b and c).

Figure 4 MCY4 expressing Kllsm4Δ1 accumulates ROS. Cell fluorescence (a–d) and phase contrast displays of cells (e–h) after 2 h incubation with dihydrorhodamine 123 are shown. Wild-type control (a–e), MCY4/Kllsm4Δ1 (b, c, (more ...)

H₂O₂ treatment still increased ROS production in both strains in that 80% of cells transformed either with the entire or the truncated KlLSM4 gene (Figure 4d) became highly fluorescent.

He and Parker, 2000

To investigate whether the mRNA degradation was altered in the presence of the truncated form of KlLsm4p, we followed the stability of the heat shock gene SSA4. This gene is induced at the transcriptional level by high temperatures and, after backshift to the normal temperature, its transcription is shut off and the amount of the mRNA decreases to the basal level within 2 h (Boorstein and Craig, 1990 ). We looked at the SSA4 mRNA levels in strains expressing the LSM4 endogenous gene, the KlLSM4 gene, and its truncated form Kllsm4Δ1.

As shown in Figure 5A, incubating wild-type or lsm4/pKlLSM4 cells for 15 min at 45°C induced the SSA4 gene (lanes 2, 8, and 14); the transcript level reached a maximum 15 min after the backshift to 24°C (lanes 3, 9, and 15) and then declined back to the basal levels by 2 h. mRNA degradation followed very similar kinetics for strains expressing LSM4 (lanes 3–6) and KlLSM4 (lanes 9–12). In the case of Kllsm4Δ1p, we observed a significant increase in mRNA stability (lanes 16 and 17) in that only a slight decrease in SSA4 mRNA was seen after shift back to 24°C (lane 18). mRNA stabilization also occurs in K. lactis cells expressing Kllsm4Δ1p, which showed an increased stability of the ethanol-repressible gene KlADH3 (Mazzoni, Mancini, Madeo, and Falcone; unpublished data).

Figure 5 MCY4 expressing Kllsm4Δ1 shows an increased mRNA stability. Strain MCY4 was grown on YP-galactose to induce the GAL1-LSM4 gene (lanes 1–6), whereas MCY4/KlLSM4 (lanes 7–12) and MCY4/Kllsm4Δ1 (lanes 13–18) were grown (more ...)

As already mentioned, in S. cerevisiae the depletion of Lsm4p causes defects in the excision of introns from pre-mRNAs, indicating a role for this protein also in mRNA maturation (Cooper et al., 1995 ). To investigate a possible effect of Kllsm4Δ1p on pre-mRNA splicing, total RNA was extracted from strains MCY4 (grown on galactose) and MCY4/KlLSM4 and MCY4/Kllsm4Δ1 (grown on glucose), and splicing efficiency was determined by probing Northern blots with the yeast SNR17 gene (U3 snoRNA; see MATERIALS AND METHODS).

After prolonged autoradiographic exposure, as shown in Figure 5B, we could detect in cells expressing Kllsm4Δ1p the presence of the U3 precursor, which accumulated at a lower level than that observed in cells not expressing LSM4 (Mayes et al., 1999 ).

Lsm1p is a specific component of the Lsm1p-7p complex and is not present in the Lsm2p-8p complex involved in mRNA splicing (Bouveret et al., 2000 ; Tharun et al., 2000 ). As can be seen in Figure 6, DAPI staining and electron microscopy (EM) analysis of cells lacking Lsm1p showed evident chromatin fragmentation (Figure 6a), chromatin condensation along the nuclear membrane (Figure 6c) and the presence of multiple nuclei (Figure 6d), indicating that apoptosis occurred also in this lsm mutant.

Figure 6 DAPI staining (a and b) and EM analysis (c and d) of S. cerevisiae mutant strain lsm1. N, nucleus; M, mitochondria; V, vacuole. Bar, 1 μm.

We also investigated the onset of apoptosis in strains lacking other factors that are known to interact with Lsm1p-7p: Dcp1p and Dcp2p (Beelman et al., 1996 ; LaGrandeur and Parker, 1998 ; Dunckley and Parker, 1999 ), which are two subunits of the decapping holoenzyme (Roy Parker, personal communication); Dhh1p, which is involved in decapping (Coller et al., 2001 ; Fischer and Weis, 2002 ); and Pat1p. The Lsm1p-7p/Pat1p complex, independently of its role in promoting decapping (Hatfield et al., 1996 ; Bonnerot et al., 2000 ; Bouveret et al., 2000 ; Tharun et al., 2000 ), is also involved in the protection of the 3′ end of the mRNA from trimming (He and Parker, 2001 ).

Strains lacking each of these factors were analyzed by DAPI, TUNEL, and DHR staining. As shown in Figure 7, dcp1 and dcp2 mutants showed evident DNA damages as well as increased ROS production, and these phenotypes could be also detected, although at a lower level, in dhh1 and pat1 mutants.

Figure 7 DAPI staining and its phase contrast, TUNEL test, DHR staining of S. cerevisiae mutants dcp1, dcp2, dhh1, and pat1 (see text and MATERIALS AND METHODS for details).

EM analysis confirmed that apoptotic phenotypes were more severe in mutants dcp1 and dcp2 (Figure 8, a and b) compared with mutants dhh1 and pat1 (Figure 8, c and d).

Figure 8 Electron micrographs of strains dcp1 (a), dcp2 (b), dhh1 (c), and pat1 (d). The arrows indicate the condensed and marginated chromatin. N, nucleus; M, mitochondria; V, vacuole; ER, endoplasmic reticulum. Bar, 1 μm.

In a previous work, we reported the construction of a viable K. lactis mutant carrying a truncated form of the essential gene KlLSM4 and we showed that this mutated allele could also rescue viability in an S. cerevisiae strain deprived of its endogenous Lsm4p. Nevertheless, in both yeasts K. lactis and S. cerevisiae, cells expressing this mutated protein showed a consistent loss of viability as soon as they reached the stationary phase (Mazzoni and Falcone, 2001 ).

In this work, we report that yeast cells expressing this truncated protein show cytological markers of apoptosis such as nuclei fragmentation and chromatin condensation.

The question as to whether mitochondria are necessary for mammalian apoptosis is very much under discussion (Reed and Green, 2002 ). A similar debate concerning the role of mitochondria in yeast apoptosis is also ongoing, with some authors suggesting an important role for mitochondria, whereas others report that this process does not necessarily require mitochondrial function (Gross et al., 2000 ; Kissova et al., 2000 ). We observed that chromatin fragmentation occurred in KlLsm4Δ1 [rho°] cells lacking mitochondrial DNA, indicating that functional mitochondria are not required for this particular pathway of yeast apoptosis.

Almost all scenarios described about apoptosis in yeast so far were somehow connected with oxygen radicals (Madeo et al., 1999 ; Laun et al.,2001 ). Also, in our case, dihydrorhodamine staining revealed that a high percentage of cells expressing the truncated KlLsm4p accumulated oxidizing molecules (ROS) in the absence of external oxidative stress.

What has been clearly demonstrated in yeast is that Lsm4p plays an important role in mRNA decapping and splicing when assembled in different complexes with other proteins of the Lsm family.

The analysis of cells expressing Kllsm4Δ1 showed a slight defect in RNA splicing, whereas RNA degradation was significantly delayed in both S. cerevisiae and K. lactis (our unpublished data).

The onset of apoptosis is not a specific case due to the truncation of Lsm4p in that we demonstrated that the absence of other components of the RNA decay machinery result in the same phenotypes. In fact, very strong effects have also been observed in strains lacking Lsm6p (our unpublished data) and Lsm1p, a protein that differently from Lsm2p-Lsm7p, which are involved both in splicing and mRNA decay, is specific for the latter activity.

Very interestingly, among the other factors that interact with the Lsm1p-7p complex, we found that in the absence of the decapping mRNA enzymes, Dcp1p and Dcp2p, cells show the same phenotypes observed in the absence of Lsm proteins, indicating the decapping is a crucial step in triggering apoptosis.

In the case of mutants in DHH1, the gene encoding a DEAD-box RNA helicase involved in mRNA decapping together with Lsm1p-7p (Coller et al., 2001 ; Fischer and Weis, 2002 ), the apoptotic markers resulted less evident compared with dcp1, dcp2, and lsm mutants.

A particular case has been observed in the pat1 mutant that showed a very similar picture to the wild-type strain for DAPI and DHR staining, whereas a significant percentage of cell resulted positive to TUNEL test and EM observation. Although we cannot exclude a relationship between the mRNA 3′ protection and apoptosis, the less severe effects observed in the pat1 mutant, which has defects in both decapping and trimming protection, together with the existence of a normal mRNA 3′ protection in the dcp1 mutant (He and Parker 2001 ), indicated that defects in the latter process does not have a strong effect as observed for decapping.

It is well known that mammalian genes involved in cell cycle and apoptosis are tightly regulated at the transcriptional and posttranscriptional level (Guhaniyogi and Brewer, 2001 ;. Lam et al., 2001 ). Our results in yeast suggest that the stabilization of mRNAs could lead to the abnormal production of some protein(s), which in turn trigger apoptosis. Related to this, it is interesting to remark that Dhh1p and Pat1p play also a role in mRNA translation and that their absence result in a less efficient protein production (Coller et al., 2001 ; Wyers et al., 2000 ; Tharun and Parker, 2001 ).

One can hypothesize that in these mutants the increased stability of transcription products could be counteracted by a less efficient translation of the accumulated mRNAs, which result in a less severe apoptotic phenotypes. Alternatively, the defects in translation would retard the synthesis of an apoptosis-promoting factor, as suggested by the resistance of H₂O₂-treated cells to entering in apoptosis in the presence of the protein synthesis inhibitor cycloheximide (Madeo et al., 1999 ; Ligr et al., 2001 ). Further work has to be done to understand whether the stabilization of total mRNAs or some mRNAs encoding specific proteins is the signal that trigger cellular apoptosis.

We thank Prof. Jean D. Beggs for kindly providing the S. cerevisiae strains MCY4, AEMY24, AEMY14, and BMA38. We are also grateful to Dr. Roy Parker for giving the strains YRP840, YRP1069, YRP1346, YRP1372, and YRP1560. We thank Giuseppe Lucania for excellent technical assistance. This work was supported by a grant “Cofin 2000” protocol MM05C63814.