Figure 3

HuR binds U-rich motifs in antisense Alu elements. (A) CLIP-Seq alignments from three different studies demonstrated HuR binding within an antisense AluSx in the 3′ UTR of ITPRIPL2, (B) which generalized to genome-wide binding to two poly-U tracts in the antisense Alu consensus, shown here as alignment coverage normalized by dataset to sum to one across the element. (C) In a formaldehyde RNA immunoprecipitation (fRIP)-Seq of HuR, genes containing exonic antisense Alu elements (Alu- RNAs) were strongly enriched in the RIP over input compared with genes devoid of antisense Alu (dAlu- RNAs). Genes targeted via intronic antisense Alu elements also showed strong evidence of binding (Figure S6 in Additional file 1). (D) Plotting all HuR TE-specific motifs by their nucleotide composition revealed a diversity of motif compositions but a strong tendency towards Us. The x-axis specifies the expected number of As and Us in the motif model, and the y-axis specifies the expected number of As and Cs. Point size is proportional to the log2 of the number of CLIP-Seq alignments overlapped by the motif. (E,F) To better reveal the relationships between the strongest motifs, we collapsed highly redundant motifs into 12 representatives and hierarchically clustered them using information coverage Euclidean distance (see Materials and methods) [75].