Loss-of-function mutations in
ced-4 prevent programmed cell death in C. elegans. The genomic sequences of
ced-4 genes from C. briggsae and C. vulgaris revealed that the 3' end of the third intron of these genes is conserved with the respective intron in C. elegans. We suspected that this region might contain an alternatively-used exon, such that a splice acceptor 72 bp upstream of the previously identified splice acceptor might be used to generate an in-frame insertion of 24 amino acids in the translated product. To test this idea, we hybridized a probe derived from the conserved region to northern blots of wild-type RNA. We observed a band slightly larger than the previously known
ced-4 transcript of 2.2 kb. In addition, we used RT-PCR to amplify cDNAs containing the alternatively spliced message from wild-type RNA and confirmed their identities by sequence determination. The abundance of the alternative transcript (designated
ced-4L, for long) is approximately 10-30 fold less than that of the more common transcript (designated
ced-4S, for short). We have previously shown that overexpression of
ced-4S can kill cells (WBG, October, 1993). To test the function of
ced-4L, we established transgenic lines containing a cDNA corresponding to this message under the control of the worm heat-shock promoters. Transgenic animals treated with a heat shock had many extra cells, suggesting that normal programmed cell deaths had been inhibited. We also introduced constructs containing
ced-4L under the control of the constitutive
dpy-30 promoter into
ced-9(lf) animals, which die because of massive ectopic programmed cell death. Transgenic animals were rescued from the
ced-9(lf)-associated lethality, suggesting that overexpression of
ced-4L can inhibit ectopic as well as normal cell deaths and supporting the idea that
ced-4L normally protects against programmed cell death. Genetic evidence supports the notion that
ced-4 encodes two transcripts of opposite functions. Ellis and Horvitz (Cell 44, 817-829, 1986) showed that while
egl-1/+ animals are ~60% Egl (because of the ectopic deaths of the HSN neurons),
ced-4(null);
egl-1/+ animals are non-Egl, providing one line of the original evidence that
ced-4 encodes a killing function. However,
egl-1/+;
ced-4(null)/+ animals are ~85% Egl, suggesting that loss of
ced-4 function can increase cell death. This finding is consistent with the hypothesis that
ced-4 encodes a protective function as well. Further genetic evidence of a dual function for
ced-4 is provided by the
ced-4 allele
n2273, which contains a mutation in the conserved G of the
ced-4S-specific splice acceptor of intron 3. This mutation weakly prevents cell death, yet synergizes with a weak
ced-9 allele to enhance
ced-9-associated lethality (M. Hengartner, personal communication). This enhancement is
ced-3-dependent, suggesting that
n2273 enhances killing by programmed cell death. Thus, this mutation affects both the protective and killing functions of
ced-4. Sequence analysis of
ced-4 transcripts from
n2273 animals obtained by RT-PCR suggests that both
ced-4S and
ced-4L transcripts are defective, supporting the notion that the opposing genetic functions encoded by
ced-4 correspond to the alternative transcripts. Genetic analysis of
n2273 has also suggested that the protective function of
ced-4 is negatively regulated by
ced-9, just as is the killing function of
ced-4, consistent with the observation of Hengartner and Horvitz (Nature 369, 318-320, 1994) that
ced-9 can act to induce, rather than prevent, cell death in certain genetic backgrounds. These observation have suggested to us a model for the function of the
ced-9(
n1950) dominant gain-of-function mutation, which prevents cell death. We propose that
ced-9(
n1950) is incapable of negatively regulating the protective (
ced-4L) function of
ced-4, yet is still capable of negatively regulating the killing (
ced-4S) function of
ced-4, thus resulting in extra cell survival as a result of increased
ced-4L activity. We suggest, therefore, that
ced-9(
n1950) causes a loss of a specific
ced-9 function rather than an activation of
ced-9 function in general.