Identification of an Endogenous Dominant-Negative Short Isoform of Caspase-9 That Can Regulate Apoptosis¹

The cytochrome c-dependent apoptotic pathway is initiated by release of cytochrome c from the mitochondria in response to an apoptotic stimulus (reviewed in Refs. 1 and 2). Cytochrome c, in the presence of dATP or ATP, can form a complex with the human CED-4 homologue Apaf-1, enabling it to recruit the initiator procaspase-9 to the complex (3). The association of procaspase-9 and Apaf-1 is mediated by the interaction of the NH₂-terminal prodomain of procaspase-9 with the NH₂-terminal CARD⁴ of Apaf-1 (3). This results in activation of procaspase-9 by Apaf-1-mediated oligomerization and activation of the downstream executioner caspases-3 and -7 by the activated caspase-9 (4). Activation of caspase-3 and -7 is essential for dismantling important structural and regulatory components of the cell, leading to the final morphological changes of apoptosis, such as chromatin condensation (reviewed in Refs. 5, 6, 7, 8).

Activation of the cytochrome c/Apaf-1/caspase-9 pathway is crucial for most forms of apoptosis. As demonstrated with a dominant-negative caspase-9 mutant, caspase-9 activity is necessary for induction of apoptosis by the proapoptotic Bcl-2 family members, UV irradiation, and death receptors and their ligands in transfected MCF-7 cells (3, 4, 9, 10). Furthermore, in casp-9^−/− knockout mice, the loss of caspase-9 causes perinatal death due to a reduction in apoptosis at an earlier stage of brain development (11, 12). This is also associated with resistance of thymocytes, embryonic stem cells, and embryonic fibroblasts to several apoptotic stimuli (11, 12).

A potential role for alternative splicing in apoptosis was suggested by the discovery of functionally active alternatively spliced variants of major apoptosis regulators, such as Bcl-x, Ced-4, and Ich-1 (caspase-2; Refs. 13, 14, 15). In all these cases, the alternatively spliced isoforms play opposing roles in apoptosis. For example, the long isoforms of Bcl-x (Bcl-xL) and Ced-4 (Ced-4L) protect cells against apoptosis, whereas their short isoforms (Bcl-xS and Ced-4S, respectively) promote cell death (13, 15). On the other hand, the long isoform of Ich-1 (Ich-1L) induces apoptosis, whereas the short isoform (Ich-1S) inhibits it (14). Thus far, there is no known endogenous antiapoptotic molecule that has been clearly shown to disrupt the Apaf-1-procaspase-9 complex. Here, we show that an alternatively spliced isoform of procaspase-9 can interact with Apaf-1 CARD and block apoptosis in vivo and activation of procaspase-9 and -3 in vitro, by a dominant-negative mechanism. We predict that up-regulation of the expression of this isoform in certain instances may contribute to increased resistance of cells to most forms of apoptosis.

Poly(A)⁺ RNA from the human Jurkat T-cell line was reverse-transcribed using Moloney murine leukemia virus reverse transcriptase (Promega) with dT(17) primer. The reverse transcription products were then amplified by PCR using two caspase-9-specific primers, Mch6-ATG (ATGGACGAAGCGGATCGG) and Mch6-TAA (CCCTGGCCTTATGATGTT). The PCR products were subcloned into the pBluescript II SK(+) SmaI-digested vector and sequenced.

cDNAs encoding caspase-9b or caspase-9 prodomain were subcloned in the mammalian expression vectors pcDNA3 or pRSC-lacZ. Constructs encoding procaspase-9, caspase-9-Δpro, procaspase-3, Apaf-CARD, Bax, Bik, and Blk have been described previously (3, 4, 9, 16).

S100 cellular extracts (2 μg/μl, total protein) were prepared from 293 or MCF-7 cells transfected with empty construct or constructs encoding caspase-9b or caspase-9 prodomain, as described previously (3, 4). The extracts (5 μl) were incubated in a 10-μl total reaction volume with in vitro translated ³⁵S-labeled procaspase-9 (0.5 μl) or procaspase-3 (0.5 μl) in the presence of cytochrome c (5 ng/μl) and dATP (1 mm). SDS sample buffer was then added to each sample, and the samples were boiled and subjected to SDS-PAGE analysis, followed by autoradiography.

Transfection and Northern and Western analyses were performed as described previously (3, 16, 17).

MCF-7-Fas cells were transiently cotransfected with β-galactosidase reporter and test plasmids as described (3, 17). The percentage of round blue apoptotic cells (mean ± SD) was determined by phase contrast microscopy and then represented as a function of total blue cells under each condition (n ≥ 3).

To identify novel isoforms of caspase-9 that could participate in regulation of the cytochrome c/Apaf-1-dependent apoptotic pathway, we performed RT-PCR on Jurkat mRNA with two procaspase-9-specific primers spanning the initiation and stop codons of procaspase-9. In addition to the full-length procaspase-9 (1259 bp), we identified a short amplification product (809 bp) that represents ∼10% of the full-length product. Sequence analysis revealed that the short PCR product encodes an alternatively spliced 266-residue isoform of procaspase-9 (Fig. 1,A). Following the caspase nomenclature, this isoform was designated caspase-9b. Caspase-9b contains the first 139-residue prodomain fused inframe to the last 127-residue linker and small subunit caspase domain (Fig. 1, B and C); that is, it lacks the entire large subunit caspase domain (residues 140–289), which contains the active site pentapeptide QACGG (Fig. 1 B). Its predicted molecular mass is ∼30.1 kDa.

Northern blot analysis of mRNA from selected normal human tissues and tumor cell lines using a riboprobe complementary to caspase-9b (Fig. 2,A, left) detected two mRNA species (∼2.1 and ∼1.7 kb) in all these samples. However, using a riboprobe complementary to caspase-9 large subunit, which is missing in caspase-9b, detected only the 2.1-kb transcript, not the 1.7-kb transcript (Fig. 2,A, right). This suggests that the 1.7-kb transcript represents the caspase-9b mRNA, whereas the 2.1-kb transcript represents the full-length procaspase-9 mRNA because its size is consistent with the full-length cDNA clones isolated previously from the Jurkat cDNA library (Ref. 16; GenBank accession no. U60521). Highest expression of the two transcripts is seen in skeletal muscle and the A549 cell line. RT-PCR analysis detected varying amounts of caspase-9b species in several cell lines examined, confirming the Northern blot data (Fig. 2 B).

To investigate whether caspase-9b is expressed at the protein level, we analyzed cellular extracts from 293 and MCF-7 cells by immunoblot analysis with a caspase-9-specific monoclonal antibody directed against the prodomain of procaspase-9. We detected a M_r 46,000 band, representing the full-length procaspase-9, and a M_r 31,000 band in both cell lines (Fig. 2,C). The endogenous M_r 31,000 comigrated with the recombinant caspase-9b band, indicating that the M_r 31,000 band most probably represents caspase-9b. The M_r 31,000 band was also detectable in several other cell lines, suggesting that caspase-9b is expressed at the protein level in many cell types. Quantitative analysis of the RT-PCR products and the immunoreactive bands (Fig. 2, B and C) by scanning the caspase-9 and caspase-9b bands revealed that the ratio of caspase-9b/caspase-9 in most cell lines examined is ∼0.1.

To demonstrate that caspase-9b can interact with Apaf-1 CARD (residues 1–97), we incubated in vitro translated caspase-9b with Apaf-1 CARD (Fig. 3). Apaf-1 CARD was able to bind to the same extent both procaspase-9 and caspase-9b but not procaspase-9 without its prodomain, suggesting that both isoforms can be recruited by Apaf-1 (Fig. 3).

Because caspase-9b can actually bind to Apaf-1, it could interfere with recruitment of procaspase-9 to the cytochrome c-activated Apaf-1 complex and prevent its activation when it is overexpressed in cells. To test this possibility, we incubated S100 cellular extracts prepared from 293 or MCF-7 cells transiently transfected with caspase-9b or caspase-9 prodomain with cytochrome c and dATP. As shown in Fig. 4 A, caspase-9b or caspase-9 prodomain were able to completely block cytochrome c/dATP-induced activation of procaspase-9 and -3.

To determine the amount of caspase-9b required to inhibit processing of procaspase-3 in the in vitro cytochrome c-dependent assay, we incubated S100 extracts from 293 cells containing increasing amounts of caspase-9b with in vitro translated procaspase-3. As shown in Fig. 4 B, caspase-9b completely blocked activation of procaspase-3 at a ratio of 2.4:1 of caspase-9b to procaspase-9. Because procaspase-9 undergoes autoactivation by Apaf-1-mediated dimerization of two molecules of procaspase-9 (4), it is expected that when caspase-9b and procaspase-9 are present at equimolar concentration, only one-third of Apaf-1 complexes would contain functional procaspase-9 homodimers capable of autoactivation. The remaining two-thirds of Apaf-1 complexes would contain caspase-9b homodimers and caspase-9b/procaspase-9 heterodimers that would be nonfunctional. This was consistent with our results which shows ∼78% inhibition of procaspase-3 activation when the ratio of caspase-9b to procaspase-9 in the S100 extracts was ∼1.2:1. These results suggest that caspase-9b can equally compete with procaspase-9 for binding to Apaf-1 and interferes with Apaf-1-mediated activation of procaspase-9 by a dominant-negative mechanism.

The ability of caspase-9b to interfere with Apaf-1-mediated activation of procaspase-9 in vitro (Fig. 4,A) suggests that it could inhibit apoptosis in vivo. To test this possibility, we transiently transfected MCF-7 cells with a construct encoding caspase-9b. As shown in Fig. 4,C, caspase-9b transfected cells were substantially less sensitive than control vector-transfected cells to multiple apoptosis-inducing agents, such as the proapoptotic bcl-2 family members Bax, Bik, and Blk, and the death receptors ligands TRAIL, tumor necrosis factor, and Fas antibody. These cells were also less sensitive to UV-induced apoptosis. These results indicate that caspase-9b can, in fact, negatively regulate apoptosis induced by a wide variety of stimuli and support our earlier observation that the Apaf-1-caspase-9 pathway is a central apoptotic pathway on which most apoptotic signals converge (4). Inhibition of death receptor-induced apoptosis in MCF-7 cells by caspase-9b and Bcl-xL (Fig. 4 C; Ref. 4) suggests that activation of the cytochrome c/Apaf-1/caspase-9 pathway is necessary for induction of apoptosis by death receptor in these cells.

In conclusion, we have identified caspase-9b as an endogenous molecule that negatively regulates the cytochrome c-dependent apoptotic pathway by inhibiting Apaf-1-mediated activation of procaspase-9. Because of the importance of this pathway in many forms of apoptosis, the identification of this molecule may provide clues to the regulation of apoptosis in several cell types. Perhaps, caspase-9b may provide an inhibitory function similar to that of the antiapoptotic Bcl-2 family members in certain cell types that express high levels of caspase-9b. Furthermore, expression of caspase-9b in many cells may establish a threshold that regulates activation of caspase-9 by preventing spontaneous activation of the Apaf-1-caspase-9 complex. Because caspase-9b is generated by an alternative splicing mechanism, signaling pathways that regulate this mechanism may play a key role in apoptosis.