Apoptosis is regulated by interaction of viral and cellular BCL-2 family antiapoptotic proteins with various pro-apoptotic proteins, several of which are also members of the BCL-2 family. Cellular protein BNIP3 is a BCL-2 family proapoptotic protein that interacts with viral antiapoptosis proteins such as adenoviruses E1B-19K and EBV-BHRF1 and cellular antiapoptosis proteins such as BCL-2 and BCL-xL. Database searches indicate that the human genome encodes an open reading frame for a protein, BNIP3α, that shares substantial homology with BNIP3. The BNIP3α open reading frame encodes a protein of 219 amino acids that contains a conserved BH3 domain and a COOH-terminal trans-membrane domain, characteristic of several BCL-2 family proapoptotic proteins. BNIP3α interacts with viral antiapoptosis protein E1B-19K and cellular antiapoptosis proteins BCL-2 and BCL-xL. Overexpression of BNIP3α in transfected cells results in apoptosis and suppresses the antiapoptosis activity of E1B-19K and BCL-xL. Like BNIP3, BNIP3α seems to be predominantly localized in mitochondria. These results suggest that BNIP3α is a structural and functional homologue of BNIP3. BNIP3 and BNIP3α seem to be the first examples of homologues among the various human proapoptotic proteins. Northern blot analysis reveals that BNIP3α is expressed ubiquitously in most human tissues. In contrast, BNIP3 is expressed well in several human tissues and less abundantly in certain tissues such as placenta and lung. These results suggest that although BNIP3 and BNIP3α may promote apoptosis simultaneously in most human tissues, BNIP3α may play a more universal role.
Apoptosis is a critical physiological process of cell death, which is essential for tissue remodeling and homeostasis in multicellular organisms. The process of apoptosis is also initiated as a defensive mechanism in cells infected by pathogenic agents, such as viruses. Several cellular and viral proteins related to the BCL-2 proto-onco-protein are efficient inhibitors of apoptosis. The mechanism by which these antiapoptosis proteins promote cell survival remains to be elucidated. The BCL-2 family antiapoptosis proteins have been shown to complex with a number of cellular proteins (reviewed in Ref. 1). Some of these interacting proteins themselves are also members of the BCL-2 family. These latter BCL-2 family proteins generally promote apoptosis when ectopically over-expressed. The BCL-2 family proapoptotic proteins share one or more conserved domains with BCL-2 and related antiapoptosis proteins. All of the proapoptotic proteins share a common death effector domain designated BH34 (2, 3). The BH3 domain of BCL-2 family proapoptotic proteins is essential for the cell death activity and for heterodimerization with antiapoptosis proteins (2, 3). Although proapoptotic proteins such as BAX (4), BAK (5, 6, 7), and BOK (8) have more extensive homology with BCL-2, several other pro-apoptotic proteins such as BIK (3), BID (9), HRK (10), BAD (11), and BIM (12) share only the BH3 domain with BCL-2, suggesting that the BH3 domain of these proteins is an important determinant for their proapoptotic activity. We have recently shown that one of the BCL-2/adenovirus E1B-19K interacting proteins, BNIP3, identified in our laboratory (13), is also a member of the family of BH3-containing proapoptotic proteins (14).
Although most BH3-containing proapoptotic proteins induce rapid cell death when overexpressed, BNIP3 exhibits delayed proapoptotic activity (14, 15). The proapoptotic activity of BNIP3 seems to be only partially dependent on the BH3 domain (2, 3). This observation has raised the possibility that BNIP3-mediated apoptosis may involve additional mechanisms. Because structural homologues may facilitate unraveling of the mechanisms by which a given protein functions, we undertook a search for homologues of BNIP3. Such functional homologues may exhibit similar or opposing activities. By searching the GenBank dbEST, we have identified a human homologue of BNIP3. In this study, we report the characterization of this homologue designated BNIP3α. BNIP3 and BNIP3α seem to be the first examples of highly homologous human proapoptotic proteins.
Materials and Methods
Plasmid pcDNA3-HA-BNIP3 has been described (14). Plasmid pcDNA3-HA-BNIP3α was constructed by cloning the BNIP3α ORF, generated by digestion of cDNA clone h16979 (American Type Culture Collection, Manassas, VA) with PvuII and NotI, in vector pcDNA3-HA. A plasmid expressing the COOH-terminal truncation mutant of BNIP3α (pcDNA3-HA-BNIP3αΔC) was constructed by deleting the DNA sequences (by PCR) coding for amino acids 188–219.
In Vitro Protein Binding.
GST fusion proteins were expressed in Escherichia coli and purified by affinity chromatography. 35S-labeled proteins were prepared by in vitro transcription and translation using a commercial kit (Promega). In vitro protein-protein interaction studies were carried out, essentially as described in Ref. 2. Briefly, the labeled proteins were precleared by incubation with glutathione-sepharose beads. The cleared extracts were incubated either with GST or GST fusion protein at 4°C for 1 h. Protein complexes were recovered by incubation at 4°C for 1 h with the addition of GSH-agarose beads. After extensive washing, the interacting proteins were eluted from the beads by incubation in SDS sample buffer at 100°C for 5 min and analyzed by 4–20% gradient SDS-PAGE.
MitoTracker Red Fluorescence and Indirect Immunofluorescence.
COS-7 cells (plated on coverslips in 35-mm dishes) were transfected with 5 μg of pcDNA3 empty vector or pcDNA3-HA-BNIP3α. For mitochondrial staining, 25 nm MitoTracker Red (Molecular Probes, Inc.) were added to the media for 30 min at 24 h after transfection. Cells were immediately fixed with 3.7% formaldehyde in PBS and permeabilized with ice-cold methanol. They were stained with monoclonal anti-HA (12CA5) antibody, followed by FITC-conjugated secondary goat antimouse IgG (Pierce Chemical Co.). Cells were photographed using a 495-nm filter to visualize HA-fluorescence and a 546-nm filter to visualize mitochondria (MitoTracker Red fluorescence).
Cell Survival and Apoptosis Assays.
Cell survival assays were carried out using a BRK cell line (BRK/p53val135-E1A) transformed by cotransfection of p53val135 (16) and adenovirus E1A (17). To determine the effect of pHA-BNIP3α on the antiapoptotic activity of E1B-19K or BCL-xL, the BRK/p53val135-E1A cell line was transfected with pRcCMV-19K or pcDNA3-BCL-xL (neo) and pBabe-BNIP3α (puro) or pBabe-BNIP3 (puro), and the transfected cells were maintained at 32.5°C for 5 days. The surviving cells were then allowed to form visible colonies by growing at 37.5°C in the presence of 100 μg/ml G418 and 1 μg/ml puromycin, stained with Giemsa, and counted. Transient apoptosis assays were carried out in MCF-7 cells transiently transfected with pcDNA3 empty vector or pHA-BNIP3 or pHA-BNIP3α along with the reporter plasmid pCMV-βgal. Forty-eight hours after transfection, apoptosis was quantitated, essentially as described earlier (14).
Northern Blot Analysis.
Human multiple tissue poly(A)+ RNA blot was purchased from Clontech. The blot was first hybridized with a random primed 32P-labeled 561 bp BNIP3α probe, corresponding to amino acids 1–187 at 68°C overnight. The membrane was washed twice in 2 × SSC/0.05%SDS for 30 min each at room temperature, followed by twice in 0.1 × SSC/0.1%SDS for 40 min each at 50°C. The membrane was stripped and then hybridized at 68°C overnight with a 582-bp BNIP3 probe corresponding to amino acids 1–194 (full-length), and the membrane was then washed as above and autoradiographed.
Results and Discussion
Identification of BNIP3α.
To identify the potential homologues of BNIP3 (HSU15174), we searched the GenBank dbEST for homologous cDNA sequences, using the BLASTN algorithm (18), and for cDNA sequences that encode amino acid sequences homologous to the BNIP3 protein, using the TBLASTN algorithm (18). The BLASTN search revealed 113 homologous human sequences in the database; the TBLASTN search revealed 100 homologous human sequences (P ≤1.7 × 10−5). Combining these two sets revealed that 117 human sequences with a high degree of homology to BNIP3 are present in dbEST.
These sequences could be divided into two subsets: those with complete identity to the published BNIP3 sequence; and those with more limited homology. The homologous, but nonidentical sequences were compared. Overlapping sequences were identified and used to create a complete assemblage. Searching the database revealed 84 identical human sequences could be fit into this assemblage. The completed assemblage could be represented using three of the longer overlapping sequences: H16979, H18031, and H25924 (Fig. 1,A). The assembled cDNA sequence contains an ORF entirely encoded within cDNA clone H16979. The DNA sequence encompassing this ORF was confirmed by further DNA sequence analysis. This ORF codes for a protein of 219 amino acids, which shares 63% identity and 69% similarity with BNIP3 (Fig. 1 B). We have designated this protein BNIP3α. The highest homology between these two proteins cluster in two regions: one located between BNIP3 residues 32 and 67 and one located between residues 92 and 194 at the COOH-terminal half. The homologous COOH-terminal region includes a BH3 domain and a trans-membrane domain. The sequences corresponding to the BNIP3α ORF were cloned in an expression vector (pcDNA3-HA-BNIP3α) and used for further functional characterization.
In Vitro Binding of BNIP3α with E1B-19K, BCL-2, and BCL-xL.
Because BNIP-3 binds to various viral and cellular BCL-2 family antiapoptosis proteins, we examined BNIP3α for a similar binding property. S-labeled BNIP3α wt or a mutant (BNIP3αΔC) lacking the COOH-terminal trans-membrane domain (31 amino acids) was prepared by in vitro transcription and translation. Binding to GST fusion proteins containing E1B-19K or BCL-2 or BCL-xL was determined (Fig. 2,A). As expected, both BNIP3α wt and BNIP3αΔC did not bind significantly to the control GST protein, whereas they specifically bound to GST-BCl-xL, GST-BCL-2, and GST-E1B-19K. However, the level of binding of BNIP3α to BCL-xL was lower than that of BAK binding to BCL-xL (data not shown). In addition, under the same binding conditions, BNIP3α did not significantly bind to the GST fusion protein containing the proapoptotic protein BAK (Fig. 2 B). These results suggest that BNIP3α binds specifically and directly with various BCL-2 family antiapoptosis proteins.
Proapoptotic Activity of BNIP3α.
Because structurally related proteins can exhibit opposing activities, we then examined BNIP3α for proapoptotic activity. MCF-7 cells were transfected with the pcDNA3 empty vector or plasmids expressing either BNIP3 or BNIP3α and examined 48 h after transfection for protein expression (Fig. 3) and for apoptosis (Fig. 3 B). About 40% of cells transfected with BNIP3α exhibited apoptosis. This level of apoptosis is similar to that induced by BNIP3. These results suggest that BNIP3α is also a proapoptotic protein, like BNIP3.
We then determined if BNIP3α could antagonize the activity of antiapoptosis proteins E1B-19K and BCL-xL. For this purpose, an apoptosis indicator cell line (BRK/p53val135-E1A) that expresses a p53 ts mutant allele and adenovirus E1A was used. This cell line, which grows normally at 37.5°C, undergoes rapid and total apoptosis when shifted to 32.5°C (mutant p53 assumes wt conformation at 32.5°C). BRK/p53val135-E1A cells were transfected with plasmids expressing either BNIP3 or BNIP3α (cloned in vectors expressing the puro selection marker) or empty (puro) vector. These cells were also cotransfected with plasmids expressing either E1B-19K or pBCL-xL (cloned in vectors expressing the neo selection marker). Transfected cells were maintained at 37.5°C for 24 h and shifted to 32.5°C for 5 days. The surviving cells were then allowed to proliferate (in the presence of G418 and puromycin), form visible colonies at 37.5°C, and were quantitated. In the absence of any antiapoptosis protein, the BRK/p53val135-E1A cells undergo total apoptosis and do not form any detectable colonies (data not shown; Ref. 14), whereas transfection of the antiapoptosis protein E1B-19K or BCL-xL resulted in significant colony formation. Coexpression of E1B-19K or BCL-xL either with BNIP3 or BNIP3α significantly reduced the antiapoptotic activity of both E1B-19K and BCL-xL, as evidenced by reduced levels of colony formation (Fig. 3 C). These results suggest that BNIP3α may directly or indirectly antagonize the activity of BCL-2 family antiapoptosis proteins, similar to BNIP3.
Previous indirect immunofluorescence analysis revealed that BNIP3 localizes primarily in mitochondria (3, 15). This exclusive mitochondrial localization of BNIP3 is unique among various BCL-2 family proteins because other members exhibit both mitochondrial and nuclear envelope localization. To examine if BNIP3α also exhibits mitochondrial localization, COS-7 cells were either transfected with empty pcDNA3 vector or pHA-BNIP3α (which expresses BNIP3α tagged with the HA epitope), and the transfected cells were double-stained with MitoTracker Red (Molecular Probes, Inc.), a mitochondrial specific dye (Fig. 4, A and C) or HA monoclonal antibody for indirect immunofluorescence (Fig. 4, B and D). In cells transfected with the pcDNA3 vector, there was no detectable HA-specific fluorescence (Fig. 4,A), whereas in cells expressing HA-BNIP3α, punctate cytoplasmic HA-specific fluorescecence was observed (Fig. 4 C). The MitoTracker fluorescence of cells expressing BNIP3α coincides with HA-specific immunofluorescence. Because BNIP3α, like BNIP3, localizes exclusively in the mitochondria, this suggests that both BNIP3α and BNIP3 function in mitochondria.
Expression of BNIP3α in Human Tissues.
Considering the similarities between BNIP3 and BNIP3α, it was of interest to determine the patterns of expression for these proapoptotic genes in various human tissues. Because the mRNA sequences of BNIP3 and BNIP3α share about 50% nucleotide sequence homology, we reasoned that both BNIP3 and BNIP3α mRNA can be differentially detected in Northern blots under stringent hybridization conditions. Hybridization of a human multiple blot with BNIP3α probe detected two transcripts of 1.6 kb and 3.9 kb (Fig. 5,A). These two transcripts were found to be ubiquitously expressed in all tissues examined, albeit at lower levels in the liver, skeletal muscle, and the pancreas. The identification of two transcripts suggests that the BNIP3α mRNA may be expressed as two differentially spliced forms. Alternatively, one of the transcripts may be derived from a closely related gene. When the same multiple tissue blot was rehybridized with the BNIP3 probe, a prominent transcript of 1.9 kb was observed in the heart and brain, whereas two transcripts of 1.9 and 1.5 kb were observed in other tissues (Fig. 5 B). Interestingly, BNIP3 expression was less in placenta and lung compared with expression of BNIP3α. Although cross-hybridization cannot be ruled out, differential expression and different sizes of the transcripts detected by the two respective probes suggest that the probes are specific. These results suggest that at least in certain human tissues these functionally related genes are differentially expressed and, thus, may contribute to apoptosis with some degree of specificity more selectively. These differences indicate that BNIP3 and BNIP3α may play different roles in development and cellular homeostasis.
We have identified and functionally characterized a human homologue of the BCL-2 family proapoptotic protein BNIP3. Like BNIP3, the homologue BNIPα is also a proapoptotic protein and induces apoptosis in a delayed fashion in transfected cells. BNIP3α seems to bind directly with antiapoptosis proteins E1B-19K and BCL-xL and suppresses their antiapoptosis activity. Among the various proapoptotic proteins that have been identified thus far, BNIP3 and BNIP3α seem to be the first examples of human pro-apoptotic proteins that are homologues, underscoring the fact that these two highly homologous proteins may play a concerted role in the human apoptosis pathways. BNIP3 and BNIP3α share two regions of high amino acid homology, suggesting that these regions of the huBNIP3 family may be functionally important. We have also observed that a 19-aa region of BNIP3 located between residue 113 and 131 is highly conserved in a putative protein coded by an ORF, C14.F, identified by the Caenorhabditis elegans genome sequencing center (data not shown). Interestingly, this region is also conserved in BNIP3α. However, the function of this highly conserved domain remains unclear because deletion of this domain in BNIP3 only partially relieves its apoptotic activity.5 Human BNIP3 and BNIP3α are rich in Ser/Thr residues, raising the possibility that the activities of these proteins may be regulated by phosphorylation in response to apoptotic signals.
Both BNIP3 and BNIP3α are primarily localized in mitochondria. Mitochondrial dysfunction seems to be an important early event of apoptosis in mammalian cells (reviewed in Ref. 19). Because both BNIP3 proteins are ubiquitously expressed in several human tissues, it is possible that these proteins may be important for the onset of apoptosis in multiple human tissues by contributing to mitochondrial dysfunction. BNIP3 seems to multimerize in mammalian cells (15).6 The COOH-terminal trans-membrane domain has been implicated in BNIP3 homodimerization (15). Because the trans-membrane domain of BNIP3 and BNIPα seem to be significantly homologous, the possibility that these proteins function as heterodimers remains to be investigated.
It is interesting that both BNIP3 and BNIP3α induce delayed cell death, whereas most other BH3-containing proapoptotic proteins promote rapid cell death in transfected mammalian cells. The mechanism of delayed cell death induced by these proteins remains to be investigated. Previous mutational studies of BNIP3 have revealed that deletion of BH3 domain only partially relieves the proapoptotic activity of BNIP3 (14). It is possible that the BNIP3 and BNIP3α may promote apoptosis by both BH3-dependent and independent mechanisms.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Supported by National Cancer Institute research Grants CA-33616 and CA-73803 and American Cancer Society Grant VM-174.
The abbreviations used are: BH3, Bcl-2 homology domain 3; GST, glutathione S-transferase; BRK, baby rat kidney; ORF, open reading frame; dbEST, Expressed Sequence Tag database; wt, wild type.
M. Yasuda and G. Chinnadurai, unpublished observations.
X-L. Gong and G. Chinnadurai, unpublished observations.
We thank Bob Lutz for various GST fusion proteins. We also thank Bob Lutz, Tom Chittenden, and Walter Blättler for discussion and comments on the manuscript.