A Novel Enhancer of the Apaf1 Apoptosome Involved in Cytochrome c-dependent Caspase Activation and Apoptosis*

Apaf1/CED4 family members play central roles in apoptosis regulation as activators of caspase family cell death proteases. These proteins contain a nucleotide-binding (NB) self-oligomerization domain and a caspase recruitment domain (CARD). A novel human protein was identified, NAC, that contains anNB domain and CARD. The CARD of NAC interacts selectively with the CARD domain of Apaf1, a caspase-activating protein that couples mitochondria-released cytochrome c (cyt-c) to activation of cytosolic caspases. Cyt-c-mediated activation of caspases in cytosolic extracts and in cells is enhanced by overexpressing NAC and inhibited by reducing NAC using antisense/DNAzymes. Furthermore, association of NAC with Apaf1 is cyt c-inducible, resulting in a mega-complex (>1 MDa) containing both NAC and Apaf1 and correlating with enhanced recruitment and proteolytic processing of pro-caspase-9. NAC also collaborates with Apaf1 in inducing caspase activation and apoptosis in intact cells, whereas fragments of NAC representing only the CARD or NB domain suppress Apaf1-dependent apoptosis induction. NAC expressionin vivo is associated with terminal differentiation of short lived cells in epithelia and some other tissues. The ability of NAC to enhance Apaf1-apoptosome function reveals a novel paradigm for apoptosis regulation.

CED4 family proteins constitute a unique family of caspaseactivating molecules. The founding member of this family, CED4, was discovered in the nematode Caenorhabditis elegans in screens for genes that are essential for developmental programmed cell death (1). CED4 contains an N-terminal CARD 1 followed by an NB domain, the later containing classical Walker A and B box motifs recognized as important in binding nucleotide triphosphates. CED4 functions as an activator of the caspase, CED3, in vitro and in vivo (2,3). The NB domain of CED4 oligomerizes in an ATP-dependent manner (4,5), whereas the CARD binds a complementary N-terminal CARD found in the zymogen proform of CED3 (6). Protease activation is thought to result from the induced proximity of CED3 zymogens bound to oligomerized CED4, where the weak intrinsic protease activity of the proenzymes is sufficient for trans-proteolysis of closely juxtaposed pro-caspases (4,7). Proteolytic cleavage of pro-CED3 then produces the large and small subunits of the heterotetrameric, autonomously active enzyme.
The closest homologue of CED4 identified in humans and other mammals thus far is Apaf1 (apoptosis protease-activating factor-1) (8). Similar to CED4, the Apaf1 protein contains a CARD, followed by an NB domain that shares significant amino acid sequence identity with the NB domains of CED4 and a family of ATPases associated with pathogen resistance (R genes) in plants (3,5,9), thus constituting the NB-ARC (Apaf-1/R gene/CED4) domain family (also known as NACHT domain). Unlike CED4, however, the NB-ARC domain of Apaf1 is followed by multiple WD repeats. These WD domains participate in auto-repression of Apaf1, locking it into an inactive, unoligomerized state until bound by cyt-c. In response to multiple cell death stimuli, changes in mitochondrial membrane permeability result in release of cyt-c into the cytosol, where it binds and activates Apaf-1, thus coupling mitochondrial damage to a mechanism for caspase activation (10). Cyt-c, in conjunction with dATP or ATP, induces formation of a large Apaf1 oligomer (estimated to be an octamer), via its NB-ARC domain, and exposes the CARD of Apaf1 for interactions with a complementary CARD found in the N-terminal prodomain of procaspase-9 (11)(12)(13)(14). By the induced proximity method, juxtaposed pro-caspase-9 (pro-Casp9) zymogens then transproteolyze each other, generating the characteristic large and small subunits typical of activated caspases. Active caspase-9 bound to oligomerized Apaf1 then directly cleaves and activates pro-caspase-3, an effector caspase that is responsible both for cleavage and activation of additional downstream caspases and for direct cleavage of a variety of substrate proteins that commit the cell to an apoptotic demise. A close homologue of Apaf1 has recently been identified in the fly, Drosophila melanogaster, apparently operating as a caspase-activator via a similar cyt-c-inducible mechanism (15)(16)(17). Gene ablation studies in mice and flies indicate that Apaf1 plays a critical role in programmed cell death in certain tissues and in response to many types of cell death stimuli in vivo (18).
In this report, we describe the identification and initial functional characterization of a novel regulator of Apaf1, which we have termed NAC.

MATERIALS AND METHODS
cDNA Cloning and Plasmid Construction-The NAC cDNA sequence was found using PSI-BLAST and the CARD sequence of CARD4/Nod1 * This work was supported by CaPCURE, National Institutes of Health Grants GM61694 and NS36821, and the United States Department of Defense. 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 1 The abbreviations used are: CARD, caspase recruitment domain; LRR, leucine-rich repeat; EGFP, enhanced green fluorescent protein; PBS, phosphate-buffered saline; IP, immunoprecipitation; IVT, in vitro translated; PAGE, polyacrylamide gel electrophoresis; aa, amino acids; HA, hemagglutinin; cyt-c, cytochrome c; pro-Casp9, pro-caspase-9; ZVAD-fmk, benzyloxycarbonyl-Val-Ala-Asp (OMe)-fluoromethyl ketone; NB, nucleotide binding; Ac-DEVD-AFC, acetyl-Asp-Glu-Val-Asp-7-amino-4-trifluoromethyl-coumarin; DAPI, 4Ј,6-diamidino-2-phenylindole; GST, glutathione S-transferase. as query. This search revealed homology with a predicted protein of EST clone KIAA0926 in the Kazusa DNA Research Institute brain genomic data base. Jurkat total RNA was reverse-transcribed to cDNAs with Moloney murine leukemia virus reverse transcriptase (Stratagene) and random hexanucleotide primers. Three overlapping cDNA fragments of NAC were amplified using Turbo Pfu DNA polymerase (Stratagene) and three sets of oligonucleotide primers as follows: set 1, 5Ј-CCGAATTCACCATGGCTGGCGGAGCCTGGGGC-3Ј (forward) and 5Ј-CCGCTCGAGTCAACAGAGGGTTGTGGTGGTCTTG-3Ј (reverse); set 2, 5Ј-CCCGAATTCGAACCTCGCATAGTCATACTGC-3Ј (forward) and 5Ј-GTCCCACAACAGAATTCAATCTCAACGGTC-3Ј (reverse);, and set 3, 5Ј-TGTGATGAGAGAAGCGGTGAC-3Ј (forward) and 5Ј-CCGCT-CGAGCAAAGAAGGGTCAGCCAAAGC-3Ј (reverse). The resultant cDNA fragments were ligated into mammalian expression vector pcDNA3-Myc. From these overlapping cDNA fragments, full-length NAC cDNA was assembled in pcDNA3-Myc and pcDNA3-HA at EcoRI and XhoI cloning sites. The nucleotide sequence of the assembled fulllength NAC was confirmed by DNA sequencing analysis. The regions encoding the CARD or the NB domain (amino acids 329 -547) were polymerase chain reaction-amplified from Jurkat cDNA using primer set 3 and the primers 5Ј-CCCGAATTCGAACCTCGCATAGTCATACT-GC-3Ј (forward) and CCGCTCGAGTCAACAGAGGGTTGTGGTGGTC-TTG-3Ј (reverse), respectively. The resultant polymerase chain reaction fragments were digested with EcoRI and XhoI and ligated into pcDNA3-Myc and into vector pGEX-4T1 for GST fusion protein production.
Antibodies-Polyclonal antisera were generated in rabbits using keyhole limpet hemocyanin-and ovalbumin-conjugated (Pierce) synthetic peptides with sequences corresponding to residues aa 161-180 (Bur241) or aa 1058 -1077 (Bur242) of NAC. Mouse monoclonal antibody recognizing human APAF1 was purchased from R & D Systems (Minneapolis, MN). Epitope-specific antibodies for FLAG, HA, or Myc tag were obtained from Sigma, Roche Molecular Biochemicals, and Santa Cruz Biotechnology, respectively.
DNAzymes-An anti-NAC DNAzyme oligonucleotide was designed by the method of Joyce (19), targeting the translation initiation region of NAC mRNA and containing 2-O-methylnucleosides at the 5Ј-end and an inverted thymidine at the 3Ј-end for nuclease resistance. Immunohistochemistry-Normal human tissues for immunohistochemistry analysis were obtained from biopsy and autopsy specimens, fixed in Bouin's solution (Sigma), and embedded in paraffin. Tissue sections were immunostained using a diaminobenzidine-based detection method employing the Envision-Plus-horseradish peroxide system (Dako). Nuclei were counterstained with hematoxylin.
In Vitro Protein Interaction Assays-GST fusion proteins were expressed from pGEX-4T1 in XL-1-blue Escherichia coli cells (Stratagene) and affinity-purified using GSH-Sepharose. Purified GST fusion proteins (0.1-0.5 g) immobilized on 10 -15 l of GSH-Sepharose beads were incubated with 1 mg/ml bovine serum albumin in 100 l of buffer A for 30 min at 25°C. The beads were then incubated overnight at 4°C with 1 l of rabbit reticulocyte lysates (TnT-lysate; Promega) containing 35 S-labeled, IVT proteins in 100 l of buffer A supplemented with 0.5 mg/ml bovine serum albumin. Proteins on beads were washed four times in 500 l of buffer A, followed by boiling in 20 l of Laemmli-SDS sample buffer, SDS-PAGE and detection by fluorography.
Caspase Assays-Cytosolic extracts were prepared in hypotonic buffer B as described (21) and incubated (10 g) with various concentrations of cyt-c and 1 mM dATP in Caspase buffer (21) for 30 min at 30°C. Caspase substrate Ac-DEVD-AFC (100 M) (Calbiochem) was then added, and protease activity was measured continuously by monitoring the release of fluorigenic AFC at 37°C. Alternatively, transfected cells were directly lysed in Caspase Lysis buffer (10 mM HEPES (pH 7.4), 25 mM NaCl, 0.25% Triton X-100, and 1 mM EDTA), normalized for protein content, and monitored for cleavage of Ac-DEVD-AFC as described (21). Processing of IVT 35 S-labeled pro-caspase-9 in cytosolic extracts was monitored by SDS-PAGE (20).
Molecular Modeling-A three-dimensional model of the CARD domain of NAC was generated using the MODELLER program, essentially as described (22), based on the structures of the CARDs of Apaf1, pro-Casp9, and Raidd (23,24).

RESULTS AND DISCUSSION
NAC-encoding cDNAs were obtained by reverse transcriptase-polymerase chain reaction, revealing a continuous open reading frame encoding a 1473-amino acid protein (Fig.  1). The predicted NAC protein contains an NB domain, followed by leucine-rich repeats (LRR), and a CARD domain. Thus, unlike Apaf1/CED4 family proteins that also contain CARD and NB domains, the CARD domain of NAC is located at its C rather than N terminus. The NB domain of NAC contains classical Walker A and B boxes indicative of ATP-binding proteins and is most similar in amino acid sequence to the NB domain of Nod1/CARD4 (29%) (25,26), followed by human APAF1 (17%), the Drosophila Apaf1 homologue (12%), and the C. elegans CED4 protein (12%) (Fig. 1C). Moreover, recombinant NAC NB domain was observed to bind ATP and to selfassociate in an ATP-dependent manner in vitro (not shown). The CARD domain of NAC shares 21, 19, and 8% amino acid identity with the CARD domains of Nod1/CARD4, huApaf1, and CED4, respectively (Fig. 1D). The NAC CARD sequence was readily threaded onto the structures of other CARDs using the MODELLER program (Fig. 1E), suggesting conservation of the 6 ␣-helical fold typical of these domains (23). The LRRs of NAC are reminiscent of Nod1/CARD4 and plant stress-response (R) proteins, which also contain LRRs. In NB-containing plant R proteins, the LRRs function as interaction motifs for pathogen responses (9), suggesting a possible role of these structures in linking NAC to specific signaling pathways. Additional NAC cDNAs were obtained that presumably represent alternative mRNA splicing products that encode shorter proteins lacking 31-or 45-amino acid segments (or both) located between the LLR and CARD (Fig. 1).
NAC mRNAs were widely expressed in human tissues, with highest levels found in blood leukocytes, thymus, spleen, and heart ( Fig. 2A). Antisera were raised against synthetic NAC peptides and confirmed to bind specifically the NAC protein by immunoblotting and immunoprecipitation assays (Fig. 2B, and not shown). NAC protein was detected in several adult human tissues, with highest levels in kidney, brain, and epidermis among the tissues examined (Fig. 2C). A smaller anti-NAC immunoreactive band was detected in thymus lysates, which remains to be characterized (Fig. 2C). Differences in NAC mRNA and protein expression in some tissues suggest the possibility of post-transcriptional regulation (Fig. 2, A and C).
Immunohistochemical analysis of the in vivo patterns of NAC expression in adult human tissues demonstrated association with differentiation in stratified epithelia of the skin, esophagus, intestine, and cervix, as well as in the prostate gland where differentiated luminal secretory cells were NACimmunopositive and undifferentiated basal cells were immunonegative ( Fig. 2D and not shown). Differentiated macrophages and granulocytes were also strongly NACimmunopositive (3-4 intensity/on 0 -4 scale), whereas their bone marrow precursors were immunonegative (not shown). In the testis, NAC immunointensity also increased in concert with the differentiation program from spermatocytes (negative), to spermatids (0 -1 intensity/on 0 -4 scale), to spermatozoa (1-2 intensity), reaching highest intensity in the residual bodies that represent cellular remnants (2-3 intensity) (not shown). Thus, NAC expression is associated with differentiation of some types of short lived cells in vivo.
In vitro binding experiments were performed using a glutathione S-transferase (GST) fusion protein containing the CARD of NAC. The CARD of NAC bound efficiently to itself and also interacted selectively in vitro with the CARDs of Apaf1, Nod1, and CED4 but not with the CARDs of Bcl10, pro-Casp9, procaspase-1, pro-caspase-2, pro-caspase-11, Raidd, or Cardiak (RIP2) ( Fig. 2A and not shown). The ability of NAC to interact with itself, Apaf1, Nod1, and CED4 in cells was confirmed by coimmunoprecipitation of epitope-tagged proteins from transiently transfected HEK293 cells (Fig. 2B and not shown). The endogenous ϳ160-kDa NAC protein could also be coimmunoprecipitated with endogenous Apaf1 (but not Nod1) from cells that intrinsically express NAC, using anti-NAC antisera (Fig. 2D).
Interestingly, compared with full-length Apaf1, NAC coimmunoprecipitated more efficiently with a truncation mutant of Apaf1 which lacks the WD repeats that normally maintain this protein in an auto-repressed state (14), suggesting that "activated" Apaf1 interacts preferentially with NAC (Fig. 2B). Consistent with this observation, association of full-length Apaf1 with NAC was inducible by stimuli such as staurosporine, which trigger cyt-c release and result in Apaf1 activation (27). Before exposure to death stimuli, relatively little NAC was coimmunoprecipitated with Apaf1, but within 10 min NAC/ Apaf1 complexes were readily detected (Fig. 3E).
Since Apaf1 is known to form complexes with pro-Casp9 (10), we examined the effects of NAC on Apaf1/pro-Casp9 interactions by coimmunoprecipitation. Overexpression of NAC by transient transfection in HEK293 cells increased the relative amount of Apaf1 that coimmunoprecipitated with pro-Casp9 (Fig. 3C), suggesting that NAC enhances rather than inhibits interactions of these proteins. These coimmunoprecipitation experiments also hinted that NAC may exist in a complex simultaneously with Apaf1 and pro-Casp9, since NAC was recovered in anti-Casp9 immunoprecipitates when Apaf1 was coexpressed (Fig. 3C), even though NAC cannot directly bind pro-Casp9 ( Fig. 3A and not shown) and was not recovered in anti-Casp9 immune complexes when Apaf1 was not coexpressed (Fig. 3B).
Cyt-c induces formation of a large holoenzyme complex (apoptosome) containing multiple Apaf1 and Casp9 molecules in an estimated 8:8 stoichiometry (12). The apoptosome can be monitored by gel-sieve chromatography of cyt-c-stimulated cytosolic extracts and typically migrates at ϳ700 kDa (28). However, a larger apoptosome of Ͼ1 MDa has been reported in some types of cells, suggesting that other proteins may associate with the Apaf1-Casp9 complex (29). In cyt-c-stimulated extracts from HEK293 cells, which contain relative little endogenous NAC, as determined by immunoblotting (not shown), a single Apaf1-containing apoptosome of ϳ700 kDa was observed by gel-sieve chromatography (Fig. 2F). However, in extracts from NAC-transfected HEK293 cells, two Apaf1-containing apoptosomes were evident, including a large Ͼ1 MDa complex that contained both Apaf1 and NAC (as determined by coimmunoprecipitation analysis of the column fractions in which Apaf1 and NAC coeluted) (Fig. 2F). Similarly, in cells that contain relative high levels of endogenous NAC, such as Jurkat T-cells, the endogenous NAC and Apaf1 molecules coeluted in gel-sieve experiments, revealing two apoptosomes where greater amounts of NAC were present in the larger of these multiprotein complexes (Fig. 2F, top). In extracts lacking cyt-c treatment, Apaf1 eluted as a monomer, whereas NAC was spread over multiple fractions without a clear elution peak (not shown). Also, NAC did not coelute in gel-sieve chromatography experiments with Nod1 using untreated or cyt-c-stimulated extracts, 2 confirming that NAC associates selectively with the Apaf1 apoptosome.
To explore whether the Apaf1-containing apoptosome with which NAC associates displays caspase activity, endogenous NAC was immunoprecipitated from control-and cyt-c-stimulated Jurkat cell extracts, and associated caspase activity was measured based on cleavage of the fluorigenic caspase-sub-  1). B, full-length NAC forms complexes with Apaf1, CED4, and itself. Human 293T cells were cultured in 6-well plates and were transiently transfected with expression plasmids (pcDNA3, 1 g each) encoding epitope-tagged (HA or FLAG tag) full-length HA-NAC, full-length FLAG-Apaf1, HA-Apaf1 lacking the WD repeats (HA-Apaf1(⌬WD)), HA-CED4, or an inactive form of HA-tagged pro-Casp9 harboring a catalytic site mutation (C287A) (20) in the presence (ϩ) or absence (Ϫ) of pcDNA3 (1 g) encoding Myc-tagged full-length NAC. Immunoprecipitations IP) were performed 1 day later using either a mouse monoclonal antibody to Myc (lanes 4 and 6) or a control mouse IgG (Cntl) (lane 5). Immune complexes were resolved by SDS-PAGE and analyzed by immunoblotting using anti-HA or anti-FLAG antibodies. Lysates derived from each transfection (10% of immunoprecipitation input) were loaded directly in gels as controls (lanes 1-3). These interaction results were further confirmed reciprocally by probing Myc-NAC in immune complexes derived from the indicated target proteins (data not shown). C, NAC enhances Apaf1/pro-Casp9 interaction. 293T cells were transfected with FLAG-Apaf1 in combination with FLAG-pro-Casp9 (C287A) (lanes 2 and 3) (10 7 ) were treated with staurosporine for the indicated times before preparing lysates (250 g of total protein) and performing immunoprecipitations with monoclonal anti-Apaf1 antibody and protein G beads overnight at 4°C. Immune complexes were analyzed by SDS-PAGE/immunoblotting using anti-NAC (Bur241) (top) and anti-Apaf1 (bottom) antibodies. Immunoblot analysis of lysates (lanes 1-4) confirmed that staurosporine treatment had no effect on total NAC protein levels under these conditions (not shown). F, NAC associates with Apaf1 apoptosome. Top panel, cytoplasmic extracts (1.5 mg) prepared from Jurkat cells were treated with cyt-c and dATP for 10 min at 30°C, and then 100 M ZVAD-fmk was added, and the samples were placed on ice before fractionation on a Superose-6 gel filtration column. Column fractions were assayed by SDS-PAGE/immunoblotting for Apaf1 and NAC, using specific antibodies. The positions of molecular mass markers and the void volume fraction are indicated. Bottom panels, lysates prepared from 293T cells transfected (Transf.) with plasmids encoding FLAG-Apaf1 and Myc-NAC or FLAGApaf1 alone were analyzed by gel-seive chromatography, as above. Column fractions were analyzed by SDS-PAGE/immunoblotting using anti-FLAG and anti-Myc antibodies. Apoptosome complexes derived from epitopically expressed Apaf1 or NAC were probed using epitope-specific antibodies (FLAG or Myc). Fractions in which Apaf1 and NAC coeluted were pooled and immunoprecipitated with anti-FLAG antibodies to recover Apaf1 or with an IgG control antibody (Cntl), and the resulting immune complexes were immunoblotted for NAC using anti-Myc epitope antibody.
strate Ac-DEVD-AFC (21). These experiments revealed that NAC is associated with active caspases after but not before cyt-c stimulation (Fig. 4A). Control immunoprecipitates lacked significant caspase activity, confirming the specificity of these results. Although activities of different size apoptosomes can vary depending on salt concentrations, use of detergents, and the presence of endogenous inhibitors (X-chromosone-linked inhibitor of apoptosis protein and second mitochondrial activator of caspases) (29), 2 these findings argue that the larger apoptosome containing NAC is active.
In HEK293 cells, which contain little endogenous NAC, overexpression of NAC by transient transfection promotes formation predominantly of the larger Ͼ1-MDa apoptosome upon cyt-c stimulation, with relatively little of the smaller ϳ700-kDa apoptosome present (Fig. 3F). To contrast the function of Apaf1 in the presence and absence of NAC, therefore, we compared cyt-c-induced proteolytic processing of pro-Casp9 and activation of downstream caspases (Ac-DEVD-AFC cleavage) in extracts prepared from control-and NAC-transfected HEK293 cells (Fig. 4B). Extracts containing elevated NAC displayed increased processing of pro-Casp9 and greater cyt-c-induced activation of downstream caspases, suggesting that NAC enhances Apaf1 activity. Conversely, reducing NAC protein levels using antisense expression plasmids (not shown) or antisense/ DNAzyme oligonucleotides decreased the ability of cyt-c to activate caspases in cell extracts in vitro. Fig. 4C, for example, shows experiments performed using Jurkat cells, which contain relatively higher levels of endogenous NAC, demonstrating DNAzyme-mediated ablation of NAC protein without concomitant changes in the levels of Apaf1. Extracts prepared from anti-NAC DNAzyme-treated Jurkat cells were less sensitive to cyt-c compared with control oligonucleotide-treated Cytoplasmic extracts were prepared from 293T at 1 day following transfection with plasmids encoding the following: (a) pro-Casp9 (50 ng DNA), Apaf1 (50 ng), and NAC (2 g); (b) pro-Casp9 and Apaf1; or (c) pro-Casp9 alone (all transfections normalized for 2.5 g of total DNA using pcDNA3) and assayed for caspase activity by addition of 100 M Ac-DEVD-AFC to cell lysates and monitoring AFC release (RFU/g lysate) over time.
Although not shown, negligible caspase activity was detected in lysates of cells transfected individually with Apaf1 or NAC or with the combination of NAC and Apaf1 (in the absence of pro-Casp9). E, NAC enhances apoptosis induced by staurosporine and by Apaf1. 293T cells were transfected with pEGFP (0.1 g) and plasmids encoding pro-Casp9 (0.05 g), Apaf1 (0.05 or 2.0 g), or NAC (0.5 or 2 g), as indicated. Total DNA input was normalized with empty pcDNA3. Transfected cells were cultured in media containing 0.1% fetal bovine serum for 1.5 days (1st to 7th columns) or treated with staurosporine (STS) at the indicated concentration for 3 h (8th to 13th columns) prior to fixing and staining with DAPI (black bars ϭ with NAC; white bars ϭ without NAC). The % GFP-positive apoptotic cells (exhibiting nuclear fragmentation and chromatin condensation) was determined by fluorescence microscopy (mean Ϯ S.E., n ϭ 3). Immunoblot analysis of replicate transfections confirmed production of plasmidencoded proteins and demonstrated that NAC and Apaf1 did not affect each other's expression (not shown). F, dominant-negative mutants of NAC inhibit apoptosis induction by Apaf1 and staurosporine (STS) but not Fas. 293T cells were transfected with 0.1 g of pEGFP DNA and either 0.5 or 2 g of plasmids encoding the CARD or NB domain of NAC, together with plasmids encoding pro-Casp9 (50 ng) and Apaf1 (1 g) (left) or Fas (0.3 g) (right). Alternatively, cells were treated with 1 M staurosporine for 5 h (middle). Cells were fixed and stained with DAPI, enumerating the percentage of apoptotic GFP-positive cells (mean Ϯ S.E., n ϭ 3). Immunoblot analysis confirmed production of the CARD and NB domains of NAC and demonstrated no affect on levels of Apaf1 or pro-Casp9 (not shown). cells, in terms of caspase activation. In contrast, sensitivity to granzyme B-mediated caspase activation was not affected, confirming a specific defect in the cyt-c pathway which depends on Apaf1 for caspase activation. Further evidence that NAC regulates the cyt-c/Apaf1-dependent activation of caspases in cell extracts was obtained by affinity preabsorption of NAC from extracts using a GST fusion protein containing the CARD of NAC and by expression in cells of a dominant-negative fragment of NAC consisting of only the CARD domain (not shown).
Consistent with experiments involving cell extracts, NAC also collaborated with Apaf1 in inducing caspase activation and apoptosis in intact cells. In transient transfection experiments using HEK293 (Fig. 4, D and E) or other cell lines (not shown), overexpression of NAC by itself (not shown) or in combination with pro-Casp9 had little effect on caspase activation or apoptosis. In contrast, overexpressing NAC together with Apaf1 and pro-Casp9 resulted in synergistic increases in activation of caspases and induction of apoptosis, as determined from cotransfections that employed suboptimal amounts of Apaf1-encoding plasmid (Fig. 4, D and E). Overexpression of NAC also sensitized cells to suboptimal concentrations of apoptosis inducers such as staurosporine, which triggers apoptosis through an Apaf1-dependent mechanism (30), but not by anti-Fas antibody which utilizes an Apaf1-independent pathway ( Fig. 4E and not shown). Immunoblotting experiments demonstrated that the enhanced sensitivity of NAC-transfected cells to staurosporine was not due to differences in the levels of Apaf1 or pro-Casp9 proteins produced in cells (not shown).
Whereas full-length NAC enhanced apoptosis and caspase activation induced by overexpressing Apaf1 or by treatment of cells with Apaf1-dependent apoptotic stimuli (Fig. 4, D and E), fragments of NAC containing only the CARD or NB domain had the opposite effect, interfering with apoptosis induced by coexpression of Apaf1/pro-Casp9 and by Apaf1-dependent stimuli such as staurosporine without affecting Apaf1-independent pathways activated by Fas (Fig. 4F). Again, immunoblotting experiments demonstrated that these reductions in Apaf1-dependent apoptosis caused by these dominant-negative fragments of NAC were not secondary to effects on levels of the Apaf1 or pro-Casp9 proteins (not shown).
Cells of various tissues vary in their sensitivity to cyt-cinduced activation of caspases, a finding that cannot be accounted for by differences in the levels of Apaf1 protein or downstream caspases (33,34). The discovery of NAC suggests a mechanism for fine-tuning Apaf1 function, based on whether NAC is expressed and perhaps on whether NAC interacts with unidentified proteins via its LRR or other domains. Apaf1/ CED4 family proteins directly bind CARD-containing caspases, promoting protease activation upon oligomerization by bringing the suboptimally active pro-enzymes into close proximity, allowing them to trans-process each other (4,(12)(13)(14). Although NAC enhances cyt-c-mediated pro-Casp9 processing, it does not directly bind this caspase (nor caspases-1, -2, -6, -7, -8, -10, or -11). 2 Rather, interactions of NAC with Apaf1 facilitate Apaf1-mediated activation of pro-Casp9, thus revealing a new paradigm for apoptosis regulation. The precise mechanism by which NAC enhances Apaf1 function remains to be elucidated.