In the last newsletter, we reported that DNase I footprints showed
unc-86 protein binding to three conserved sites in the upstream regions of the
mec-3 gene of C. elegans and C. briggsae (CS1, CS2 and CS3). In addition gel shifts demonstrated that
mec-3 protein binds to CS3. We have extended our work to search for more
unc-86 binding sites using footprint analysis. In addition to the three sites previously found,
unc-86 binds to three other sites (nCS1, nCS2 and nCS3). These sites are upstream of the concensus sites and are not conserved between the two species (see figure below). The
mec-3 footprint using E. coli derived protein on the same region reveals four binding sites that overlap with four of the
unc-86 binding sites (CS1, CS2, CS3 and nCS1, we don't know whether
mec-3 binds to nCS3 and nCS2). The
unc-86 and
mec-3 footprint sequences are co-centered in CS3 and nCS1, but are skewed in CS2 and CS3. To test whether
unc-86 and
mec-3 interact with each other, we performed gel- shift analysis with CS2 and CS3 oligos. The
mec-3 protein binds both oligos and shows one retarded species. In contrast, the
unc-86 protein binds to both oligos with multiple retarded species, supposedly in monomer, dimer and mutimeric forms. There are three retarded species for CS2 and four species for CS3 (CS2 and CS3 do not share sequence homology; CS2 has POU binding motifs, while CS3 is AT- rich). When both proteins are added together in the presence of excess probe, the gel-shift pattern does not change with the CS2 oligo. However, the CS3 gel-shift pattern is changed. Though the
mec-3 band, which has faster mobility than all
unc-86 bands, and the second fastest
unc-86 band appear unchanged, the remaining
unc-86 bands have disappeared. A new band appears at a position between the two fastest original
unc-86 bands. These results suggest that the interaction of
mec-3 and
unc-86 proteins inhibits the formation of the multimeric complexes by
unc-86 protein. Also the loss of the fastest
unc-86 species (supposedly the monomeric form) and the appearance of a new band at a slightly higher position (supposedly a heterodimeric form) by the addition of
mec-3 protein in the presence of excess probe suggest that the binding of the
unc-86 and
mec-3 monomer to CS3 may be cooperative. We don't know the in vivo significance of this phenomenon, but are testing its importance by in vitro mutagenesis. We are also using in vitro mutagenesis to knock out the
unc-86 and
mec-3 binding sites. Changing five nucleotides in CS3 completely abolishes the
unc-86 and
mec-3 binding to this region as verified by
unc-86 footprinting and
mec-3 gel-shift. When this mutant is introduced into wild-type animals as a
mec-3-lacZ construct, the staining is almost completely abolished, suggesting that CS3 is essential for
mec-3 expression. Three nucleotide changes in CS1 abolish the
unc-86 binding at this region (these changes are not in the
mec-3 footprint region). This mutant seems to have no effect on staining in ALMUR and PLMUR, but greatly reduces the frequency of staining in FLPL/R, and PVDUR (we are not quite sure about AVM/PVM), suggesting that this CS is important for the expression of
mec-3 in FLP and PVD cells (non-touch-cell). Preliminary results of similar CS2 mutants suggest that they affect ALM and PLM staining but not PVD and FLP (ALM staining is abolished; PLM staining is greatly reduced). Thus, CS1 and CS2 may be important for cell-specific
mec-3 expression. Deletion of a region which contain the three non-conserved
unc-86 binding sites leads to the appearance of additional staining cells, suggesting that this region contains negative control elements. Perhaps
unc-86 binding (or
mec-3 binding ) at this region could repress
mec-3 expression in these other cells. Moreover, these results suggest a caution in relying too heavily on conservation as the sole method of identifying important cis acting elements. [See Figure 1]