In WTs, Pax6 and Cdk6 were correlated inversely: Pax6

levels were significantly higher and Cdk6 levels were significantly lower in rostral than in caudal cortex (p < 0.0001 and p < 0.0009, respectively; Figure 3E, black lines). In PAX77 embryos, Pax6 levels were elevated both rostrally and caudally and there was a significant reduction of Cdk6 levels caudally. Whereas elevated Pax6 levels repressed Cdk6 expression caudally, repression was not detected rostrally. It is possible that rostral increases might have little effect on Cdk6 expression if Pax6-mediated repression of Cdk6 in WTs is already relatively close to maximum in this region. As a next step toward defining a biochemical pathway through which selleck chemicals Pax6 might regulate cortical progenitor cell cycles, we used bioinformatics to identify potential Pax6 binding sites in genomic regions surrounding some of the cell-cycle genes regulated by Pax6. Using a 21 bp consensus binding motif that is known from previous work to be recognized by the Pax6 paired domain (P6CON; Epstein et al., 1994a, 1994b), we identified a particularly large number of putative Pax6 binding sites within

10 kb on either side of the Cdk6 coding see more sequence and focused further work on testing for direct regulation of this gene by Pax6. We used two position weight matrix (PWM) databases, TRANSFAC and JASPAR, and identified 14 putative Pax6 binding sites within the 10 kb genomic regions immediately upstream and downstream of the Cdk6 coding region. These candidate sites were filtered by conservation analysis across five vertebrate species (mouse, rat, dog, chimpanzee, and human) using the Mulan program. Five putative binding sites (BS1–BS5; Figure S6A) with at least 85% similarity to P6CON and >90% identity

across the five vertebrate species were defined for experimental testing. The murine Cdk6 promoter has not been characterized, but the human CDK6 promoter has ( Cram et al., 2001). By aligning its sequence against the mouse genome (NCBI Build 37, UCSC mm9) we identified Terminal deoxynucleotidyl transferase a 2.3 kb region flanking the 5′ end of the murine Cdk6 gene with ∼80% homology to the human CDK6 promoter sequence. This region, which probably contains the murine Cdk6 promoter, also contains one of the five putative Pax6 binding sites (BS1; Figure 5C). We tested whether BS1–BS5 (Figure 4A) could specifically bind Pax6, using Pax6 protein generated by in vitro translation (Figure 4B). Electrophoretic mobility shift assays (EMSAs) are shown in Figures 4C–4G. The migration of radioactively labeled oligonucleotides containing each predicted Pax6 binding site (BS1–BS5) was retarded by binding to Pax6 protein (shift) and retarded further upon addition of an anti-Pax6 antibody (supershift). Radiolabeled oligonucleotides containing mutations that abolish the Pax6 consensus binding site in BS1–BS5 (Figure S6A) showed reduced or no retardation (compare lanes 1 and 2).