The alternative reading frame (ARF) mRNA encodes two pro-death proteins, the nucleolar p19ARF and a shorter mitochondrial isoform, named smARF (hsmARF in human). While p19ARF can inhibit cell growth by causing cell cycle arrest or type I apoptotic cell death, smARF is able to induce type II autophagic cell death. Inappropriate proliferative signals generated by proto-oncogenes, such as c-Myc and E2F1, can elevate both p19ARF and smARF proteins. Here, we reveal a novel means of regulation of smARF protein steady state levels through its interactions with the mitochondrial p32. The p32 protein physically interacts with both human and murine smARF, and colocalizes with these short isoforms to the mitochondria. Remarkably, knocking down p32 protein levels significantly reduced the steady state levels of smARF by increasing its turn over. As a consequence, the ability of ectopically expressed smARF to induce autophagy and to cause mitochondrial membrane dissipation was significantly reduced. In contrast, the protein levels of full-length p19ARF, which mainly resides in the nucleolus, were not influenced by p32 depletion, suggesting that p32 exclusively stabilizes the mitochondrial smARF protein. Thus the interaction with p32 provides a means of specifically regulating the expression of the recently identified autophagic inducer, smARF, and adds yet another layer of complexity to the multifaceted regulation of the ARF gene.
The mouse p19 alternative reading frame (ARF) tumor suppressor (p14ARF in human) is localized to the nucleolus, and can inhibit cell growth in a p53-dependent or -independent manner (Sherr, 2006). Recently, we found that ARF mRNA also encodes an additional shorter mitochondrial isoform, named smARF (hsmARF in human), that is translated from an internal methionine at position 45 and therefore lacks all the nucleolar functions of the full-length protein. smARF is a short-lived protein that is rapidly degraded by the proteasome, but accumulates after inappropriate proliferative signals generated by oncogenes. Overexpression of this isoform results in damage to the structure of the mitochondria, dissipation of the mitochondrial membrane potential and autophagic cell death (Reef et al., 2006). A search for novel p19ARF interacting proteins, based on immunopurification of ectopically expressed p19ARF from total cell lysates and identification of co-immunoprecipitated bands by mass spectrometry, yielded several candidate proteins. One of them was the p32 protein (Figure 1a). The p32 protein (named also gC1qR, SF2-associated binding protein) is a doughnut-shaped homotrimeric protein (Jiang et al., 1999), which localizes predominantly to the mitochondrial matrix (Muta et al., 1997; Dedio et al., 1998; Matthews and Russell, 1998; Seytter et al., 1998). p32 has been shown to bind many cellular (Ghebrehiwet et al., 1994; Simos and Georgatos, 1994; Yu et al., 1995a; Deb and Datta, 1996; Herwald et al., 1996; Lim et al., 1996) and viral proteins (Desai et al., 1991; Luo et al., 1994; Fridell et al., 1995; Yu et al., 1995b; Bruni and Roizman, 1996; Tange et al., 1996; Wang et al., 1997; Matthews and Russell, 1998). Despite abundant biochemical data, many aspects in p32s function are still unknown, although it has been suggested that it has a role in the maintenance of mitochondrial oxidative phosphorylation in yeast (Muta et al., 1997), and that it can regulate Hrk-mediated apoptosis in mammalian cells (Sunayama et al., 2004). We found that the human p14ARF was also capable to pull down the p32 protein (data not shown). Moreover, the interaction between ARF protein and p32 was direct, since 35S-labeled p32, produced by in vitro translation in rabbit reticulocyte lysate, was pulled-down by glutathione S-transferase (GST)-p14ARF, but not by GST alone, GST-Hdm2 or GST-p53 (Figure 1b). Thus, although the mouse and human ARF proteins share a relatively low degree of identity over the region of overlap (45%), the ability to bind p32 is conserved, implying the significance of this interaction.
Since p32 is a mitochondrial protein, it was appealing to examine whether smARF, the mitochondrial ARF short isoform, can also interact with p32. To this end, 293T cells were transfected with Flag-tagged smARF (ΔNp19- see the scheme in Figure 1e) or hsmARF (ΔNp14- the corresponding human protein lacking the first 47 amino acids). Flag-tagged α-actinin was used as a non-relevant negative control. The proteins were immunoprecipitated using anti-Flag antibodies. Flag-tagged smARF pulled down endogenous p32, similar to the full-length p19ARF, while Flag-tagged α-actinin could not (Figure 1c). Similarly, in the reciprocal immunoprecipitation, endogenous p32 specifically pulled down Flag-smARF (Figure 1d). Owing to the low expression levels of endogenous smARF, and the background appearing at the expected size with the available anti-p19ARF antibodies, we were not able to detect specific co-IP with the endogenous smARF. In addition, similar to smARF, Flag-tagged hsmARF also co-immunoprecipitated with p32 (Figure 1f). The fact that both smARF and hsmARF are able to interact with p32 indicates that p19/p14ARFs N terminus (aa 1–44, or 1–47, respectively), which drives the nucleolar functions of the full-length protein, is not essential for the interaction with p32. Notably, further truncation of a stretch of 35 amino acids from the N-terminal region of smARF (ΔNp19Δ1−35 see the scheme in Figure 1e) abrogated the binding to p32 (Figure 1g), suggesting, that the first 35 amino acids of smARF are necessary for the interaction. The N-terminal truncation of smARF also caused further destabilization of the protein (Figure 1h), and therefore, to reach comparable levels of the two proteins, the pull down experiment in Figure 1g was carried out in the presence of the proteasome inhibitor MG132.
The p32 protein was shown to be expressed mainly in the soluble matrix of the mitochondria (Muta et al., 1997; Dedio et al., 1998; Matthews and Russell, 1998; Seytter et al., 1998). However, several studies have reported that under certain conditions, a small fraction of p32 can be found also in other cellular compartments, such as the nucleus and the plasma membrane (Matthews and Russell, 1998; Soltys et al., 2000; Brokstad et al., 2001). Therefore, the localization of p32 was examined in HeLa cells by staining with specific anti-p32 antibodies and compared by co-staining to Flag-tagged smARF or hsmARF. In all of the smARF/hsmARF transfected cells, endogenous p32 was found colocalized with the short ARF isoform to punctate structures (Figure 2a, rows I and III). These structures stain positively for the mitochondrial marker cytochrome c (Figure 2a, row II), and are in fact the fragmented, damaged mitochondria generated by smARF expression (Reef et al., 2006). Unlike the smARF/hsmARF-expressing cells, the adjacent non-transfected cells when stained with anti-p32 antibodies, always showed the typical tubular staining of intact mitochondria (Figure 2a, row I and III), which overlapped with the cytochrome c staining (Figure 2a, row II). The colocalization of p32 and smARF to the mitochondria was also confirmed in 293T cells (data not shown). Notably, in a fraction of cells expressing ectopic smARF or hsmARF, nucleolar staining was also observed. However, even in those cells, endogenous p32 was detected only in the mitochondria, and not in the nucleolus (data not shown), indicating that the ARF short isoform interacts with p32 in the mitochondria and not in the nucleolus. Of note, the truncated mutant ΔNp19Δ1−35, which could not interact efficiently with p32, was no longer exclusively mitochondrial, but could be found also in the cytoplasm and/or the nucleus (Figure 2a, row IV). Thus the degree of co-immunostaining of p32 and the truncated mutant was severely reduced. This suggests that the first 35 amino acids of smARF are also necessary for its appropriate mitochondrial localization. To confirm further the mitochondrial colocalization of smARF and p32, the mitochondria were fractionated from 293T cells which were transfected with ΔNp19ARF (smARF) or luciferase (Cont.) constructs. The purified mitochondrial fraction clearly contained both p32 and smARF proteins (Figure 2b). Additionally, p32 co-immunoprecipitated with Flag-tagged smARF specifically from the mitochondrial fraction (Figure 2c). Thus, the co-immunostaining of smARF and hsmARF with the mitochondrial p32, together with the physical interaction between smARF and p32 when extracted from purified mitochondria, suggest that the mitochondrial compartment is the main cellular milieu where smARF and hsmARF interact with p32 within cells. We therefore assume that the binding of the full-length nucleolar p19ARF to p32 shown in Figure 1a and c probably occurred after cell lysis in the crude extracts.
To examine whether smARFs function is dependent on p32, an shRNA plasmid was generated to knock down endogenous p32. The levels of p32 protein were partially reduced upon transfection of 293T cells with the shRNA construct (Figure 3a). Remarkably, the knock down of p32 significantly reduced the steady-state levels of ectopically expressed smARF when the latter was expressed below saturation levels. In contrast, the steady-state levels of full-length p19ARF were not affected upon similar transfection with p19M45A (mutated at Met45, which initiates smARF expression (Reef et al., 2006)), or with the wild type construct, which express mainly the full-length p19ARF (Figure 3a). Thus, p32 is capable of regulating smARFs protein expression levels, and does not affect the steady state levels of the full-length p19ARF. Next, the effect of p32 depletion on the previously reported smARF-induced dissipation of the mitochondrial membrane potential was tested. To this end, 293T cells in which p32 had been knocked down, were transfected with control or ΔNp19ARF expression vectors. After 24 h, the transfected cells were stained with a fluorescent mitochondrial dye, MitoTracker Red, which accumulates only in actively respiring mitochondria that have an intact mitochondrial membrane potential. The cells were then fixed and immunostained with anti-Flag antibodies. Expression of smARF in the p32 knock down cells led to smaller reductions in the mitochondrial membrane potential compared to cells transfected with control shRNA (Figure 3b), as would be expected considering that lower levels of smARF are expressed in these cells. Similar result was obtained by using flow cytometric analysis of transfected cells stained with the mitochondrial probe DiOC6. The loss of mitochondrial membrane potential caused by smARF was attenuated from 45.5% in cells expressing the normal levels of p32 (control shRNA), to only 30% in p32 knock down cells (Figure 3c). The final cellular outcome of smARF overexpression, which involves induction of autophagy, was also examined after p32 depletion. To that end, 293T cells in which p32 had been knocked down were co-transfected with green fluorescent protein (GFP)-LC3 and ΔNp19ARF. The extent of autophagy induction, that is, of smARF-transfected cells displaying punctate LC3 staining, was reduced by knocking down p32 (Figure 3d). As expected, the truncated mutant ΔNp19Δ1−35, which could not interact efficiently with p32, and which was not exclusively mitochondrial, and expressed to lower levels, was less efficient both in reducing the mitochondrial membrane potential (Figure 3e), and in inducing autophagic vesicle formation (Figure 3f). Notably, the effect of p32 on smARF was concentration-dependent; above a certain threshold of smARF expression, p32 levels no longer influenced its steady-state levels, and therefore, knock down of p32 did not affect smARFs ability to dissipate membrane potential or induce autophagy (data not shown). This implies that p32s direct role is to regulate the steady-state expression of smARF, rather than its function.
To examine the effect of p32 protein on smARF steady state levels under physiological conditions, 35-8 cells (immortalized p53-null MEFs) were electroporated with siRNA against murine p32, and the expression levels of p19ARF isoforms were examined. Notably, in cells in which p32 was knocked down, the expression levels of endogenous smARF were significantly reduced, but the expression levels of full-length p19ARF were not affected (Figure 4). To examine whether the reduction of smARF protein levels in p32 knockdown cells was due to differences in protein turnover, the stability of smARF protein following administration of cycloheximide was measured in the 35-8 cells. Notably, the turnover of smARF in p32 knockdown cells was faster than in control cells (Figure 4), suggesting that depletion of p32 enhances the degradation rate of smARF. As mentioned previously, the N-terminal truncated mutant (ΔNp19Δ1−35), which fails to bind to p32 (Figure 1g), was significantly more unstable than ΔNp19 (compare Figure 1h and g), consistent with the conclusion that p32 stabilizes smARF. Although p32 was also able to bind to ectopically expressed full-length p19ARF in cell lysates, it could not regulate its turnover, underscoring the specificity of the regulation of smARF by p32 protein and further suggesting that smARF is the ultimate physiological partner of p32 in intact cells. Thus, unlike oncogene expression, which enhances the expression of both isoforms (Reef et al., 2006), probably by common mechanisms which operate at the transcriptional level, different proteins control the turnover of the nucleolar and mitochondrial isoforms. p19ARF is stabilized through interaction with nucleolar NPM/B23 protein (Kuo et al., 2004), and smARF is specifically stabilized by the mitochondrial p32 protein. These findings reveal a novel function for p32, which, although shown to interact with many proteins, the full repertoire of its cellular functions was not ascribed yet. A future challenge will be to reveal the mechanism by which p32 regulates the stabilization of smARF. As previously suggested by using the MG132 inhibitor, smARF is kept under tight regulation by proteasome-mediated degradation (Reef et al., 2006). p32 may protect smARF from such degradation by binding it and sequestering it to the mitochondria. Alternatively, smARF degradation may occur within the mitochondria via ATP-dependent mitochondrial proteases that may be also sensitive to the proteasome inhibitor MG132 (Granot et al., 2003). In such a scenario, p32 may protect smARF from these proteases within the mitochondria. Additional experiments are required to gain further insight into this novel pathway.
In summary, the interaction with p32 provides a means of specifically regulating the expression of the recently identified autophagic inducer, smARF.
We thank WC Russell for providing the polyclonal antibody against p32. We thank S Bialik for critical reading of the article, E Zalckvar and G Tarcic for help and advice. This work was supported by grants from the European Union (LSHB-CT-2004-511983) (to AK) and by the Center of Excellence grant from the Flight Attendant Medical Research Institute (FAMRI) to AK and MO. AK is the incumbent of Helena Rubinstein Chair of Cancer Research.
About this article
Journal of Controlled Release (2016)