Ndi1 mediates apoptosis induced by various stresses and is functionally independent of yeast Z-VAD-fmk–binding activity
We previously showed that Ndi1 overexpression induces yeast apoptosis and is involved in aging-related apoptosis. To determine whether NDI1
is involved in apoptosis induced by other external stress factors, NDI1
-overexpressing or deletion strains were subjected to H2
(Madeo et al., 1999
) and Mn (Liang and Zhou, 2007
). Low doses of H2
and Mn are excellent yeast apoptosis inducers. The viability test results showed that the strain overexpressing NDI1
survived worse than the corresponding vector control (). In contrast, the deletion mutant of NDI1
was resistant to Mn- and H2
-induced apoptosis compared with wild-type yeast (). In addition, we found that Ndi1 is involved in apoptosis induced by acetic acid (Ludovico et al., 2001
), another commonly used apoptosis inducer (unpublished data).
FIGURE 1: NDI1 participates in apoptosis induced by various stimuli. (A, B) NDI1 is involved in Mn-induced apoptosis. NDI1 overexpression and deletion resulted respectively in reduced and increased viabilities under Mn stress, accompanied by changes of caspase-like (more ...)
To determine whether the yeast cells undergoing NDI1-mediated apoptosis display increased endogenous metacaspase activity, we assayed the VAD-binding activity (an assay for potential metacaspase activity) in the NDI1-overexpressing and deletion mutant strains. After treatment by various stresses, yeast cells were incubated with a fluorescein isothiocyanate (FITC)–conjugated caspase inhibitor (FITC-VAD-fmk) probe, which can bind and abrogate activated metazoan caspases irreversibly, and then analyzed by flow cytometry. Of interest, strains overexpressing NDI1 accumulated more FITC-VAD-fmk (), whereas ndi1 mutants were labeled less ().
As a homologue of AMID, which is known to participate in apoptosis in a caspase-independent manner, it is very important to establish whether the Ndi1-mediated apoptosis is caspase dependent, in other words, whether the increased VAD-binding signal associated with Ndi1-induced apoptosis is functional. To address this question, we analyzed the survival rates of the NDI1
-overexpressing strain with or without preincubation with the pan-caspase inhibitor Z-VAD-fmk. The addition of Z-VAD-fmk did not suppress the apoptotic phenotype due to Ndi1 overexpression (). As a control to show that Z-VAD-fmk is working, we stressed wild-type yeast with Mn with or without Z-VAD-fmk. Mn-induced apoptosis was previously shown to be partially dependent on Yca1 and associated with increased VAD-binding signal (Liang and Zhou, 2007
). Very clearly, Z-VAD-fmk could significantly abrogate cell death caused by Mn ().
Because Yca1 is the only known yeast metacaspase, to demonstrate genetically that Ndi1 functions independently of Yca1, we overexpressed NDI1 in yca1 mutant background. Consistently, NDI1 overexpression exacerbated cell death in the absence of Yca1 ().
The results show that Ndi1 plays an important role in Mn- and H2O2-induced apoptosis. Although accompanied by increase in VAD-binding signal, Ndi1-induced apoptosis appears to be functionally independent of the potential yeast metacaspase activity.
Proapoptotic activity of Ndi1 is separable from its ubiquinone oxidoreductase activity and independent of electron transport chain
The mitochondrial electron transport chain (ETC) assimilates electrons from substrates such as NADH and polarizes the mitochondrial membrane potential for the ATP formation while serving as the major source of reactive oxygen species (ROS) production. In S. cerevisiae
, Ndi1, together with Nde1—both mitochondrial ubiquinone oxidoreductases (internal or matrix side, and external or cytosolic side, respectively; Luttik et al., 1998
; Small and McAlister-Henn, 1998
)—acts as the major entry point of electrons into the respiratory chain. Previously we showed that when NDI1
, but not NDE1
, is overexpressed, yeast growth is strongly inhibited and apoptotic cell death is induced (Li et al., 2006
). Conversely, ndi1
-mutant cells have reduced ETC activity and are very resistant to oxidative stress. To determine whether the proapoptotic activity of Ndi1 is dependent on its ubiquinone oxidoreductase function, we made a series of point-mutated variants of Ndi1 (), which may have disrupted NADH dehydrogenase activity, based on structural and functional information (Feng et al., 2013
), and tested the proapoptotic activities of these mutant forms. These Ndi1 variants were cloned into an ADH promoter-driven vector, transformed into an ndi1
-deletion mutant, and analyzed for the loss of respiration activity. Some of these mutants indeed had lost the ubiquinone oxidoreductase function, as attested by their drastically reduced ability to grow on media with glycerol as the sole carbon source (). Spotting assays for several of these mutants, including Ndi1S83W
, and Ndi1L217D,F510D
, which had undetectable ubiquinone oxidoreductase function, and Ndi1L217D,F258D
, which retained most ubiquinone oxidoreductase function, were further carried out to test their proapoptotic activity. All these mutants retained almost the same growth-inhibitory activity as the wild-type Ndi1 (). Several representative mutants were analyzed further for their protein expression and localization. All were shown to be expressed at roughly comparable levels and free in the cytosol, indicating that they are neither unstable nor mistargeted ().
FIGURE 2: Proapoptotic activity of Ndi1 is independent of its ubiquinone oxidoreductase activity. (A) Loss of ubiquinone oxidoreductase activity is not associated with change of Ndi1 apoptotic activity. Shown here are spotting assays of ndi1-mutant strain transformed (more ...)
Having shown Ndi1-mediated apoptosis is independent of its NADH dehydrogenase function, we next explored whether it is independent of ETC activity in general. To determine whether a change of ETC would affect NDI1
-mediated apoptosis, we tested the survival rates of wild-type, ndi1-
-deletion mutant, and petite strains treated by H2
is important for optimal cellular growth with a number of nonfermentable carbon sources, in particular ethanol, whereas petite yeast strain is defective in mitochondrial DNA and respiration. The nde1
and petite yeast strains almost had the same viability rates as the wild-type control. However, only ndi1
-disrupted cells survived better, indicating that the Ndi1 proapoptotic property is not acting through ETC (). Furthermore, various ETC mutations in critical respiration genes such as cor1
, and cox17
did not abrogate the growth deficiency caused by NDI1
overexpression, again supporting the notion that Ndi1-mediated apoptosis is independent of ETC (). We previously reported that a majority but not a small minority of ETC mutants could affect NDI1
-induced toxicity (Li et al., 2006
). This discrepancy appears to arise from the instability of some of these mutant strains during strain storage and propagation. Aided by freshly generated mutants, we confirmed that the growth deficiency caused by NDI1
overexpression is independent of all ETC mutants tested. We therefore finally conclude that the proapoptotic activity of Ndi1 is distinguishable from its ubiquinone oxidoreductase function and is independent of mitochondrial ETC activity.
Ndi1-induced apoptosis is at least partially mediated by ROS production (Li et al., 2006
). We next examined whether apoptotic cell death associated with Ndi1 mutants lacking oxidoreductase function is regulated by ROS. Using dihydrorhodamine 123 (DHR123) as an indicator, we showed that the overexpression of these Ndi1 mutants was still associated with ROS production (). To assess whether the ROS is functional to Ndi1-induced apoptosis, we transformed these mutant NDI1
into a sod2
-mutant strain. Whereas normal-sized colonies were observed for the control vector, no colonies appeared on Ndi1 and Ndi1S83W
plates 3 d after transformation, and only several small colonies appeared on plates transformed with Ndi1Q253A,D254A,R479A
. This result indicates that ROS plays an important role both in the apoptotic cell death associated with Ndi1 and Ndi1 mutants lacking oxidoreductase function.
Mitochondrial Ndi1 translocates to the cytoplasm during apoptosis
In apoptosis, the mitochondria, cytoplasm, and nucleus closely interact with each other. The past 20 years of research has found that upon induction of apoptosis some mitochondrial factors escape. AIF, for example, translocates from the mitochondria into the nucleus. Conversely, some proteins translocate from the nucleus into the mitochondria or cytoplasm. Human Rad21 is cleaved and moves from the nucleus to the cytoplasm and acts as a nuclear signal for apoptosis (Chen et al., 2002
; Pati et al., 2002
), and Mcd1 is cleaved and translocated from the nucleus to the mitochondria to promote apoptosis (Yang et al., 2008
Considering the translocating behaviors of AIF during apoptosis, we examined Ndi1 localizations in the process of apoptosis. To determine whether the subcellular localization of Ndi1 may change when undergoing apoptosis, we used confocal microscopy to track the localization of Ndi1 in the cells. An NDI1-RFP fusion gene was constructed. By spotting assay, this red fluorescent protein (RFP)-tagged Ndi1 was shown to be properly functional; the ndi1-mutant strain overexpressing NDI1-RFP exhibited almost the same growth inhibition as that of the ndi1-mutant strain overexpressing wild-type NDI1, suggesting that NDI1-RFP retained its apoptotic activity ().
FIGURE 3: Ndi1 translocates from mitochondria to cytoplasm during apoptosis. Ndi1-RFP is overexpressed, whereas Ndi1-TAP is under the endogenous regulation. (A) Ndi1-RFP maintains the proapoptotic function of Ndi1. (B) Ndi1-RFP is localized in the mitochondria (more ...)
The NDI1-RFP construct was then transformed into an Om45-GFP (BY4742) strain in which the mitochondrial Om45 (a major constituent of the mitochondrial outer membrane, located on the cytoplasmic face of the outer membrane) was fused with green fluorescent protein (GFP) to mark the mitochondria. The RFP and GFP signals can therefore be used to locate Ndi1 and mitochondria, respectively, when yeast is insulted by different apoptosis triggers. We used Mn as the apoptosis inducer. When treated with Mn, Ndi1-RFP relocalized from the mitochondria to the cytoplasm, whereas under no Mn stress Ndi1-RFP and Om45-GFP were perfectly colocalized (). This suggests that Ndi1 translocates from the mitochondria to the cytoplasm when stressed by apoptotic inducers.
A concern here is whether the big RFP tag or the Ndi1 overexpression might affect Ndi1’s behavior or subcellular localization—for example, whether these might alter the mitochondrial inner membrane residence of Ndi1. Because our homemade Ndi1 antibody does not react well with the N-truncated Ndi1, we tried a tandem affinity purification (TAP)–tagged, endogenously expressed NDI1-TAP strain in which Ndi1 is C-terminal fused to a TAP in its original chromosomal state (under endogenous promoter control). The fusion protein Ndi1-TAP is expressed in the mitochondria, as shown by both immunolocalization and Western blot (). Although none of these experiments can pinpoint the inner membrane localization of Ndi1-TAP (like the normal Ndi1), its ability to grow on yeast extract/peptone/glycerol suggests that it is indeed correctly located (). This is because Ndi1 and Nde1 catalyze the internal and external NADH dehydrogenation, respectively. Loss of Ndi1 function results in poor growth in glycerol media, although growth is fine in ethanol media. The ability to use glycerol well indicates that Ndi1-TAP resides in the matrix side instead of the outer membrane. In addition, the ability to induce apoptosis is not affected by this TAP tagging (). All these demonstrate that Ndi1-TAP is functionally equivalent to native Ndi1. Using this NDI1-TAP strain, we also saw an obvious release of the protein to the cytoplasm (see )
FIGURE 4: Ndi1 is cleaved in mitochondria during apoptosis. (A) Ndi1 is cleaved when treated with Mn. NDI1-TAP yeast was treated or not treated with 8 mM Mn for 12 h, and the extracts were immunoblotted and probed with a TAP antibody. (B) Ndi1 cleavage occurs in (more ...)
Ndi1 is N-terminal cleaved before its translocation out of the mitochondria
In the huge machinery network of apoptosis, many components are cleaved to be activated. A prominent case is the caspase family. Another crucial family is the B-cell lymphoma-2 family (Taylor et al., 2008
). In both yeast and mammalian cells, some mitochondrial proteins involved in apoptosis were cleaved and then released from mitochondria. For example, during the process of apoptosis, AIF undergoes N-terminal proteolysis at the intermembrane space to form an apoptogenic fragment that can translocate into the nucleus (Otera et al., 2005
). Considering that Ndi1 is a protein loosely bound to the mitochondrial inner membrane, we suspected that Ndi1 might also undergo some conformational or other kind of change to facilitate the translocation from mitochondria into cytoplasm.
To facilitate the detection of the potential degradation or cleavage of Ndi1, we used the NDI1-TAP strain. Indeed, besides the full-length Ndi1-TAP band, an additional, smaller band was also detected from the protein extract of yeast cells undergoing Mn-induced apoptosis, whereas under the normal growth condition this shorter band was not detectable ().
To check whether this cleavage occurs before or after the release from mitochondria, we used density gradient centrifugation to fractionate cell extracts into the mitochondrial and the cytoplasmic fractions, confirmed by respective antibodies against their endogenous marker proteins. It is obvious that during apoptosis, Ndi1 is cleaved before it moves out of the mitochondria, as evidenced by the observation that the postmitochondrial fraction only contained the shorter form of Ndi1 and the mitochondrial fraction contained both the full-length and shorter ones (). Another question is whether or not Ndi1 release from the matrix side is specific. It is possible that the entire inner membrane is permeabilized during Ndi1 release. As a control, we monitored the release of the tricarboxylic acid enzyme mitochondrial NAD(+)-dependent isocitrate dehydrogenase Idh1 during Ndi1 release (Cupp and McAlister-Henn, 1992
). No Idh1 escaping to the cytoplasm was detected. Obvious truncation of Idh1 was not observed either (). Thus it appeared that the mitochondrial release of Ndi1 from the matrix side is at least relatively specific.
Further experiments showed that Ndi1 cleavage induced by Mn occurred in a time- and concentration-dependent manner (). At very high levels of Mn (12 or 16 mM), an even greater percentage of yeast cells died (), but less Ndi1 protein was converted into the shorter form, consistent with our previous finding that at high Mn levels yeast undergoes mostly necrosis rather than apoptosis. It appears that the cleavage of Ndi1 was specific in apoptosis.
Taken together, these data suggested that Ndi1 in its full length might be a target of some proteases that are activated during apoptosis induced by Mn, consequently generating a shorter fragment of Ndi1 to be released to the cytoplasm. Because the TAP signal is at the C-terminal of Ndi1, it is reasonable to consider that the smaller band observed in the Western analysis is Ndi1 with the very N-terminal removed.
The mitochondrial release of apoptotic proteins so far reported is normally from the intermembrane space. Considering the inner membrane location of Ndi1, it is not known how this translocation could occur. YBH3
was recently reported to be involved in mitochondrial outer membrane permeabilization (Büttner et al.
, 2011). We therefore tested whether mitochondrial release of Ndi1 and its apoptotic action would be affected in ybh3
-deficient background. Neither Ndi1’s apoptotic function nor its release is obviously dependent on Ybh3 ().
N-terminal cleavage is essential for Ndi1’s apoptotic action
Is N-terminal removal functionally necessary for the proapoptotic activity of Ndi1? To address this question, we tested different forms of Ndi1 for their apoptotic activities. First, the mature full-length Ndi1 form without the mitochondrial signal peptide (cytosol-Ndi1 or Ndi1ΔN1-27) was expressed in the cytosol. Cytosol-Ndi1 showed much better growth than the yeast with the normal mitochondrial form of Ndi1. In other words, when the full-length mature form of Ndi1 was expressed in the cytosol, a much-reduced toxicity was observed than with the expression of its native mitochondrial form (). To exclude the possibility that this phenotype was caused by attenuated protein expression, we analyzed the protein level by Western blotting using a homemade murine Ndi1 antibody. No decrease of protein expression was observed (). It appears therefore that the cytoplasmic full-length Ndi1 would not be nearly as active in terms of executing Ndi1-mediated apoptosis.
FIGURE 5: The N-terminal likely serves normally a protective role. (A) Cytosol-NDI1 (Ndi1ΔN1-27) displays much reduced toxicity as compared with normal NDI1. (B) A Western blotting control showing that cytosol-NDI1 was expressed similarly. (C) The N-terminal (more ...)
Next we constructed a few N- and C-terminal–truncated Ndi1 variants () and tested their apoptogenic activity. The truncated variants were designed according to potential protease cleavage sites, as well as functional and structural predictions. Several N-terminal–truncated Ndi1 variants were expressed in the cytoplasm to reproduce the growth-inhibitory activity of the native mitochondrial Ndi1. These variants include Ndi1ΔN1-37, Ndi1ΔN1-42, Ndi1ΔN1-47, Ndi1ΔN1-72, and Ndi1ΔN1-191. In particular, Ndi1ΔN1-72, which lacks the first 72 amino acids (or the potential mitochondrial localization sequence plus 45 amino acids in the N-terminal of the mature Ndi1), displayed almost the same inhibiting phenotype as the wild-type Ndi1 (). This form, which is predicted to be 4.6-kDa smaller than the wild-type mature Ndi1, closely matches the in vivo molecular weight of the shortened protein product detected by Western blotting during apoptosis. In contrast, removal of a few C-terminal residues reduced the cell growth–inhibition activity (). The proapoptotic activities of various Ndi1 forms are summarized in .
These data suggest that the N-terminal–cleaved Ndi1 is the activated form for apoptosis induction, and this cleavage involves about the first 40–50 amino acids of the N-terminal of the mature mitochondrial Ndi1.