We have analyzed the coding regions from 10 different
mab-5 alleles using PCR cycle sequencing on amplified genomic and cDNA fragments. Eight alleles have the same phenotype, which appears to be the null phenotype based on genetic arguments. Four of these alleles contain nonsense mutations upstream of the homeobox, and are therefore indeed likely to be null. Three alleles affect a splice donor site, and two others affect splice acceptor sites. Finally, the semidominant hypomorphic allele
bx54 ,previously characterized in the Emmons laboratory, contains a missense mutation in the homeobox. First, the nonsense mutations:
e2011 contains an ochre mutation at glutamine 24 near the N terminus of the protein. Barring a secondary promoter internal to the gene, this allele is likely to be null.
e1936 ,
n1384 ,and
mu14 all contain the same opal mutation at tryptophan 106 just upstream of the homeobox. Since this residue is part of the aromatic region conserved in all Antennapedia class homeobox genes, these alleles are likely to be null. Several of the alleles contain mutations in splice donor or acceptor sites.
e2088 is a 140 bp deletion removing the splice donor from the first intron. In the one cDNA so far sequenced, out-of-frame splicing occurred using a cryptic donor four bases downstream of the deletion.
e1239 and
sy173 also affect the first splice donor, converting GTGTGTGTTT to ATGTGTGTTT This results in out-of-frame splicing two bases downstream of the normal location.
mu17 and
mu32 convert TTTCAG splice acceptors to mCAA. In the case of
mu17 ,in the second intron, this leads to efficient use of a cryptic splice acceptor, producing an out-of-frame junction. Consistent with this, the
mu17 phenotype appears similar to that of null alleles. In the case of
mu32 ,in the fourth intron, correct splicing occurs with low efficiency. This produces wild-type protein which can be detected in nuclei at low levels using polyclonal antibodies. Consistent with this,
mu32 is a hypomorphic allele with residual activity in the V cells.
bx54 ,a hypomorphic allele with a similar phenotype to
mu32 ,also has a dominant ray fusion phenotype (K.-L. Chow & S. Emmons, pers. comm.). Interestingly, this allele contains a missense mutation converting glutamate 135 (in the homeobox) to a lysine. (The homeodomain protein
egl-5 also contains lysine in this position.) We are now more carefully comparing the
bx54 and
mu32 phenotypes and trying to understand why they both have residual
mab-5 activity only in the V cells. One possibility is that the V cells are least sensitive to decreases in the level of
mab-5 . As might be expected,
bx54 expresses
mab-5 at near wild-type levels. The pattern of expression also appears normal with one interesting exception: Since
bx54 lacks
mab-5 activity in the QL descendants, they migrate anteriorly rather than posteriorly. As they do so, expression of
bx54 declines from near wild-type levels to below the limits of detection. This suggests that either the position of the Q descendants is important in maintaining
mab-5 expression, or
mab-5 itself is needed to maintain its own expression in the Q descendants. All alleles have been sequenced at least twice, in many cases from independant freezer stocks. Since several independant alleles contain identical base changes, there would appear to be mutational hotspots in the gene. [See Figure]