Noxa, a BH3-Only Member of the Bcl-2 Family and Candidate Mediator of p53-Induced Apoptosis

See allHide authors and affiliations

Science  12 May 2000:
Vol. 288, Issue 5468, pp. 1053-1058
DOI: 10.1126/science.288.5468.1053


A critical function of tumor suppressor p53 is the induction of apoptosis in cells exposed to noxious stresses. We report a previously unidentified pro-apoptotic gene, Noxa. Expression of Noxa induction in primary mouse cells exposed to x-ray irradiation was dependent on p53. Noxa encodes a Bcl-2 homology 3 (BH3)–only member of the Bcl-2 family of proteins; this member contains the BH3 region but not other BH domains. When ectopically expressed, Noxa underwent BH3 motif–dependent localization to mitochondria and interacted with anti-apoptotic Bcl-2 family members, resulting in the activation of caspase-9. We also demonstrate that blocking the endogenous Noxa induction results in the suppression of apoptosis. Noxa may thus represent a mediator of p53-dependent apoptosis.

The mechanism of p53-induced apoptosis has been extensively studied in the context of tumor suppression (1). p53-dependent apoptosis is regulated, at least in part, by transcriptional activation of its target genes (1), and this process is dependent on the Apaf-1/caspase-9 activation pathway (2). Among the identified target genes of p53,Bax encodes a pro-apoptotic Bcl-2 family of proteins that can activate this pathway (3). However, inBax-deficient mice, DNA damage–induced apoptosis occurs normally in thymocytes, and apoptosis induced by treatment with anticancer drugs is only partly inhibited in mouse embryo fibroblasts (MEFs) expressing the adenovirus oncoprotein E1A (4). Furthermore, thymocytes fromp53-deficient mice with a Bax transgene nevertheless showed resistance similar to that of thymocytes without the transgene to DNA damage–induced apoptosis (5). Therefore, the existence of at least one other target gene appears to be necessary to explain the full p53-dependent apoptotic response.

p53 and interferon regulatory factor–1 (IRF-1), a critical transcription factor in the interferon response, cooperate in tumor suppression and in regulation of the cell cycle and apoptosis (6). These observations prompted us to search for target genes of IRF-1, p53, or both that show increased transcription in response to DNA damage. We used a mRNA differential display method (7) to isolate cDNAs whose mRNA expression profiles differed between the x-ray–irradiated wild-type andIRF-1/p53 doubly deficient MEFs. We identified a gene termed Noxa (for damage). Its cDNA encodes a 103–amino acid protein (Fig. 1A), and it lacks any known motif except for two mutually related 9–amino acid sequences (A and B) characteristic to the Bcl-2 homology 3 (BH3) motif of the Bcl-2 family of proteins (Fig. 1B) (8).

Figure 1

Primary sequence of Noxa and its expression. (A) Predicted amino acid sequence of Noxa (28). Two putative BH3 motifs are underlined (regions A and B). (B) Alignment of the Noxa BH3 motifs with the BH3 domains of Bcl- 2 family proteins; anti-apoptotic Bcl-2 subfamily proteins (human Bcl-2, GenBank accession numberM14745; human Bcl-Xl, GenBank accession number Z23115; human Mcl-1, GenBank accession number Q07820), pro-apoptotic Bax subfamily proteins (human Bax, GenBank accession number L22473; human Bak, GenBank accession number U23765), and pro-apoptotic BH3-only subfamily proteins (mouse Bad, GenBank accession number L37296; human Bik, GenBank accession number U34584; mouse Bid, GenBank accession number U75506). Amino acids identical to Noxa's BH3 motifs are shaded. (C) Expression of Noxa mRNA following x-ray irradiation in MEFs. Noxa mRNA was analyzed by RNA blotting with RNAs (5 μg in each lane) isolated from wild-type (WT) andp53-deficient (p53–/–) MEFs following x-ray irradiation [20 grays (Gy)]; time after radiation is indicated in hours. The same filter was probed with MDM2 or β-actin cDNA. (D) Expression of Noxa mRNA in p53–/– MEFs following p53 expression. Cells were infected with p53-expressing adenovirus (Ad-p53) for the indicated time periods, and RNA blotting was performed as in (C). The same filter was probed with β-actin cDNA. (E) Expression of Noxa mRNA in thymocytes following x-ray irradiation. Noxa mRNA was analyzed by RNA blotting with RNAs (5 μg in each lane) isolated from WT and p53–/– thymocytes following x-ray irradiation (5 Gy). The same filter was probed with β-actin cDNA. (F) Expression of Noxa protein. Noxa protein was determined by immunoblot analysis with antibody to Noxa in WT thymocytes following x-ray irradiation (5 Gy); time after radiation is indicated.

Noxa mRNA was constitutively expressed in small amounts in the brain, thymus, spleen, lung, kidney, and testis of adult mice (9). X-ray irradiation of wild-type MEFs increased expression of Noxa mRNA about fivefold (Fig. 1C), with kinetics similar to those of the p53-dependent gene MDM2 (1). In contrast, expression of the Noxa gene was totally abolished in p53-deficient MEFs but not in IRF-1–deficient MEFs (Fig. 1C) (9, 10). Moreover, ectopic expression of p53 resulted in increased expression of Noxa mRNA inp53-deficient MEFs (Fig. 1D). Thymocytes undergo DNA damage–induced apoptosis in a p53-dependent manner (11). Increased expression of Noxa mRNA in response to x-ray irradiation also occurred in wild-type thymocytes (fivefold increase) but not inp53-deficient thymocytes (Fig. 1E). Noxa protein also accumulated in wild-type thymocytes after x-ray irradiation (Fig. 1F) (12).

To determine whether the p53-dependent expression of theNoxa gene involves direct activation of its promoter, we isolated and characterized the mouse Noxa gene (13). This gene contains three exons in which the BH3 motifs A and B are encoded by exons 2 and 3, respectively. The transcription initiation site was determined to be 158 base pairs upstream from the initiator ATG by polymerase chain reaction (PCR)–based primer extension, and one potential p53-recognition sequence, located at −155 to −174, was found in the promoter region (Fig. 2A). The contribution of p53 to the activation of the Noxa promoter was examined by a transient cotransfection assay using a luciferase reporter gene linked to Noxa promoter (Noxa-luc in Fig. 2A) (14). The promoter was activated (on average, sevenfold) by coexpressed p53 in p53-deficient MEFs. In contrast, reporter genes containing a deletion (Noxa-Δluc) or point mutations (Noxa-mt-luc) in the putative p53-recognition sequence were not activated by p53 (Fig. 2B). Collectively, these results lend support to the idea that expression of the Noxa gene in x-ray–irradiated cells involves direct activation of its promoter by p53.

Figure 2

Activation of the Noxa promoter by p53. (A) The Noxa promoter and luciferase reporter gene constructs. Putative p53-recognition sequence and p53-consensus binding sequence (p53 CBS) are shown. The following reporter plasmids using this assay are also indicated: Noxa-luc containing Noxa promoter and putative p53-recognition sequence, Noxa-Δluc lacking putative p53-recognition sequence, and Noxa-mt-luc in which four critical nucleotide residues for p53 binding were altered (indicated by lowercase letters). (B) Transient cotransfection analysis of p53. p53 expression vector (pEF-p53) (0.05 μg) was transfected into 8 × 104p53–/– MEFs with 0.2 μg of each reporter plasmid. Luciferase activity was measured 24 hours after transfection. RGC-luc (29) containing synthetic p53-binding sequences was used as a positive control. Histogram shows the mean of three independent experiments, and error bars show standard deviations. The assay was repeated three times, and the results were reproducible.

On the basis of its retention of various BH domains, the Bcl-2 family can be divided into three classes: the anti-apoptotic Bcl-2, the pro-apoptotic Bax, and BH3-only subfamilies (8). The BH3 domain of the pro-apoptotic members is critical for association with other Bcl-2 family proteins in the promotion of apoptosis (8). Unlike Bax, whose expression is also regulated by p53 (3), Noxa contains BH3 but not other BH motifs (BH1, BH2, and BH4) or a transmembrane domain; hence, it appears be a previously unknown member of the BH3-only subfamily. Ectopic expression of Noxa in HeLa cells with an adenovirus-mediated gene expression system caused apoptosis in >90% of the cells 24 hours after virus infection (Fig. 3A) (15). This Noxa-induced apoptosis was also observed in other human cancer cell lines independently of their p53 status (9). Because substitution of the NH2-terminal leucine to alanine in the BH3 domain of another BH3-only member, Bad, is known to cause a loss of pro-apoptotic activity (16), we generated mutant Noxa cDNAs in which one amino acid substitution was similarly introduced in either or both of the leucine residues within the BH3 motifs (15). Noxa mutants carrying one substitution (A mt and B mt) had lower pro-apoptotic activities than wild-type Noxa, and the mutant carrying both substitutions (AB mt) was totally inactive (Fig. 3A). Thus, the BH3 motifs are central to Noxa's pro-apoptotic activity. Noxa-induced apoptosis was suppressed by coexpression of the anti-apoptotic members of the Bcl-2 family, Bcl-Xl or Bcl-2 (9).

Figure 3

Functional characterization of Noxa. (A) Induction of apoptosis by Noxa and the effect of mutations in its BH3 motifs. HeLa cells were infected with control adenovirus (control) (solid circles), adenovirus expressing Noxa (Noxa) (open circles), and adenoviruses expressing a mutant form of Noxa: A mt (open triangles), B mt (solid triangles), and AB mt (squares). Percentage of TUNEL-negative cells was determined at the indicated times in hours (left). One of two similar results is shown. The expression of Noxa and its mutant proteins in the same cells was determined by immunoblot analysis with antibody to HA at the indicated time (right). (B) Subcellular localization of Noxa protein. Twelve hours after infection in HeLa cells by adenovirus expressing Noxa, the HA-tagged Noxa protein was stained with antibody to HA (green), and mitochondria were stained with CMTMRos (red) (Molecular Probes); the two images were overlaid (Noxa + mitochondria) by confocal laser microscopy (μRadiance, Bio-Rad). (C) Subcellular distribution of the Noxa protein. HeLa cells infected with adenovirus expressing Noxa (Ad-Noxa) for 12 hours were separated into cytosol (soluble fraction, S), plasma membrane (light membrane fraction, LM), mitochondria-rich (heavy membrane fraction, HM), and nuclear (containing nuclei and some mitochondria) (low-speed pellet, P1) fractions. All fractions were adjusted to the same volume and analyzed by immunoblotting with antibodies to HA (Noxa), cytochrome oxidase subunit IV (Cyt. Oxidase) (as a mitochondrial marker), and Bax. Ad-control, control adenovirus; Ad-AB mt, adenovirus expressing mutant forms of Noxa. (D) Noxa protein is associated with Bcl-Xland Bcl-2. HeLa cells were transfected with Bcl-Xl or Bcl-2 expression vectors (pEF-Bcl-Xlor pEF-Bcl-2, respectively). After 36 hours, these cells infected with Ad-Noxa or Ad-AB mt for 12 hours. Cell extracts were immunoprecipitated (IP) with antibodies to Bcl-Xl (α-Bcl-Xl) or Bcl-2 (α-Bcl-2) and subjected to immunoblot (IB) analysis with antibodies to HA (α-HA), Bcl-Xl, or Bcl-2. Essentially identical results were obtained when the same extracts were analyzed by IB with α-Bcl-Xl followed by IP with α-HA. Ectopically expressed protein is indicated as +. (E) Cytochrome c release and caspase-9 activation by Noxa. Cytosolic extracts were prepared from HeLa cells that were mock-infected (−), or they were prepared 12 hours after infection with the indicated adenoviruses. Cytosolic extracts (cytosol) and extracts from the residual pelleted fraction (pellet) were subjected to immunoblot analysis using antibody cytochrome c (Cyt. c) (left). Caspase-9 activation was determined with extracts from HeLa cells at the indicated time after infection with Ad-Noxa or Ad-AB mt by immunoblot analysis with antibody to caspase-9 (right). The unprocessed caspase-9 precursor (procasp 9) and the cleaved 37-kD product of active caspase-9 (p37/casp 9) are indicated. (F) Reduction in ΔΨm by Noxa. ΔΨm was measured by fluorescence of the cationic lipophilic dye CMTMRos with a flow cytometer at the indicated time after infection with Ad-control, Ad-Noxa, and Ad-AB mt. A reduction in ΔΨm is observed as “ΔΨm low.” The reduction was inhibited by the addition of a caspase inhibitor, z-VAD fmk, suggesting that the permeability change may be a postcaspase event (9).

We examined subcellular localization of epitope-tagged Noxa by immunohistochemical analysis in HeLa cells. Noxa was colocalized with a mitochondrial marker, CMTMRos (Fig. 3B) (17). Immunoblot analysis of subcellular fractions also showed that most of the ectopically expressed Noxa protein was located in the mitochondria-rich heavy membrane fraction and a small amount was detected in the low-speed pellet, which contains residual mitochondria together with nuclei (Fig. 3C) (18). On the other hand, the Noxa mutant lacking functional BH3 motifs (AB mt) was found in all fractions (Fig. 3C), indicating that the selective localization of Noxa to mitochondria is contingent on its functional BH3 motifs. Bax is known to accumulate in mitochondria in response to death signals (18, 19). Endogenous Bax protein was dispersed in all fractions, even after ectopic expression of Noxa (Fig. 3C), and Noxa failed to bind Bax (9). Therefore, the function of Noxa is likely to be independent of that of Bax. In fact, BH3-only subfamily members are known to induce apoptosis by association with anti-apoptotic Bcl-2 family members or by stimulating other apoptosis-promoting factors (8). Noxa indeed coimmunoprecipitated with coexpressed Bcl-Xl or Bcl-2, and this coimmunoprecipitation was dependent on the BH3 motifs of Noxa (Fig. 3D). We also found that endogenous Noxa, induced in irradiated thymocytes, also coimmunoprecipitated with Bcl-Xl (9). Such an interaction was also observed with another Bcl-2 member, Mcl-1 (20), collectively suggesting the selective interaction of Noxa with the anti-apoptotic Bcl-2 subfamily of proteins.

Because p53-dependent apoptosis is dependent on the activation of Apaf-1 and caspase-9 (2), we also examined whether Noxa affected these events. Cytochrome c release, which induces Apaf-1 activation (21), and caspase-9 activation were also observed in these cells (Fig. 3E). The mitochondrial permeability change is also induced during the process of p53-dependent apoptosis (22). A decrease in mitochondrial membrane potential (ΔΨm), which is mediated by the opening of the mitochondrial permeability transition pore (17, 23), was detected 12 hours after infection of HeLa cells with the Noxa-expressing adenovirus (Fig. 3F).

To examine the involvement of Noxa in p53-induced apoptosis, we used human Saos2 cells, which lack p53 expression (24). We screened for a human homolog of Noxa cDNA and found that the cloned cDNA showing the highest degree of similarity is identical to the previously identified human gene APR(25). However, the function of this gene is not known, and its regulation by p53 has not been demonstrated. Human Noxa, or APR, encodes 54 amino acids, containing only one BH3 motif at amino acids 29 to 37 (Fig. 4A). This motif probably corresponds to motif B of mouse Noxa, and the humanNoxa gene lacks a DNA segment corresponding to the second exon of mouse Noxa (9). Human Noxa also induced apoptosis in various cells, including Saos2 cells in a BH3 motif–dependent manner (9). The promoter region of the human Noxa gene indeed contains one p53-response element (9), and increased expression of Noxa mRNA was observed in Saos2 cells infected with adenovirus encoding p53 (Fig. 4B). When an antisense oligonucleotide to Noxa was exposed in Saos2 cells, the increased expression of endogenous Noxa in response to p53 was inhibited, whereas control oligonucleotide had no effect (Fig. 4C) (26). Introduction of the antisense oligonucleotide also inhibited p53-induced apoptosis (Fig. 4D). Radiation-induced apoptosis in a hematopoietic cell line, BAF-3, is known to be dependent on p53 (27). Introduction of the antisense oligonucleotide to Noxa also inhibited the induction of Noxa expression and apoptosis (Fig. 4E). These results also support the notion that Noxa is a mediator of p53-induced apoptosis, at least in these assay systems.

Figure 4

Role of Noxa in p53-induced apoptosis. (A) Comparison of the amino acid sequences of human and mouse Noxa (28). Human Noxa is identical to APR (25). Identical amino acids are indicated with an asterisk. BH3 motifs are also indicated. (B) Induction of Noxa mRNA in Saos2 cells by p53. Cells were infected with adenovirus expressing p53 (Ad-p53) for the indicated time periods in hours, and RNA blotting was performed. The same filter was probed with β-actin cDNA. (C) Reduction of endogenous Noxa protein by transfection with Noxa antisense oligonucleotide. Noxa protein was determined by immunoblot analysis with antibody to human Noxa in Saos2 cells 20 hours after infection with Ad-p53 following transfection with 4 μM antisense (Ad-p53 + antisense) or control (Ad-p53 + control) oligonucleotide for 4 hours. Noxa protein was also determined before (−) and after infection with Ad-p53 (Ad-p53) for 20 hours without oligonucleotides. (D) Effect of Noxa antisense oligonucleotide in p53-induced apoptosis. Saos2 cells were transfected with antisense (open circles) or control (solid circles) oligonucleotide as in (C) and infected with Ad-p53 for the indicated times. Viable cells were determined by trypan blue exclusion and calculated as the percentage of survival in relation to the number at the start of trial. Error bars represent standard deviations from two independent samples. We tested two other control oligonucleotides and confirmed that these oligonucleotides had no effect. (E) Effect of Noxa antisense oligonucleotide in x-ray irradiation–induced apoptosis of BAF-3 cells. BAF-3 cells were incubated with 10 μM antisense or control oligonucleotide for 12 hours and were subjected to x-ray irradiation (4 Gy). Noxa protein was determined by immunoblot analysis with antibody to mouse Noxa 16 hours after x-ray irradiation (+) (−, before infection) in the absence or presence of indicated oligonucleotides (left). Viable cells were determined after irradiation in the presence of antisense (open circles) or control (solid circles) oligonucleotides by trypan blue exclusion and calculated as percentage of survival (right). Error bars represent standard deviations from two independent samples.

Noxa may be an attractive candidate mediator of p53-mediated apoptotic response (1). It is likely that Noxa, and other p53 target genes, functionally cooperate with each other for the efficient induction of apoptosis in various cell types.

  • * These authors contributed equally to this report.


View Abstract

Navigate This Article