Identification of consensus binding sites clarifies FMRP binding determinants

Abstract

Fragile X mental retardation protein (FMRP) is a multifunctional RNA-binding protein with crucial roles in neuronal development and function. Efforts aimed at elucidating how FMRP target mRNAs are selected have produced divergent sets of target mRNA and putative FMRP-bound motifs, and a clear understanding of FMRP's binding determinants has been lacking. To clarify FMRP's binding to its target mRNAs, we produced a shared dataset of FMRP consensus binding sequences (FCBS), which were reproducibly identified in two published FMRP CLIP sequencing datasets. This comparative dataset revealed that of the various sequence and structural motifs that have been proposed to specify FMRP binding, the short sequence motifs TGGA and GAC were corroborated, and a novel TAY motif was identified. In addition, the distribution of the FCBS set demonstrates that FMRP preferentially binds to the coding region of its targets but also revealed binding along 3' UTRs in a subset of target mRNAs. Beyond probing these putative motifs, the FCBS dataset of reproducibly identified FMRP binding sites is a valuable tool for investigating FMRP targets and function.

Figure 1.

Identification of FMRP consensus binding sequences (FCBS). (A) Workflow showing processing steps (orange) to generate FCBS set from parental CLIP datasets (gray). (B) Overlap of FMRP-bound CLIP tags from Darnell et al. and Ascano et al. datasets. (C) Genes containing sites bound in Darnell et al. and Ascano et al. datasets. (D) Location distribution of FMRP-bound sites in parental and FCBS sets.

Figure 2.

Previously described sequence motifs in FCBS set. Presence of sequence motifs (A) and frequency of motif occurrence (B) in FCBS set and in 1000 permutations of random 101 nt mRNA sequences.

Figure 3.

Ten most significantly enriched motifs in FCBS set as identified by DREME tool.

Figure 4.

Motifs with significant position bias in FCBS. (A) Ten motifs with the most significant bias in location in FCBS set as identified by CentriMo tool. For motifs originating from RNAcompete (9), the RNA binding protein which binds that motif is indicated. (B) Common patterns of position bias within FCBS set as identified by CentriMo tool.

Figure 5.

Normalized distribution of FCBS sites along all bound mRNAs, with the percentage of FCBS sites located in each region indicated.

Figure 6.

FMRP CLIP tags are not enriched near miRNA seed sites. Using the Ascano et al. FMRP CLIP dataset and a negative control CLIP dataset (c22orf28 (22)), the distance was calculated from each PAR peak to the nearest miRNA seed site for the 100 most highly expressed miRNAs in HEK cells.

Figure 7.

Codon usage in FCBS. For each FCBS site or in 1000 permutations of random positions within the same transcripts, the in-frame codon usage was calculated. Bar graph indicates percent of all in-frame codons. * indicates stop codons. For amino acids with ≥15% difference between FCBS and random positions, the fold change is indicated as well as corresponding motifs enriched or depleted (depletion indicated by parenthesis) in FCBS.