Identification of bantam/miR-58 targets by differential RIP-chip. (A) TAP::ALG-1 associated mRNAs were purified from synchronized L4 stage transgenic “wild-type” animals (WS4303; see Methods) and synchronized L4 stage transgenic “mir-58” mutants (WS5041; see Methods) and analyzed by one-color microarrays (Affymetrix). The average log₂ expression values of all the detected mRNAs are shown. The abundance of TAP::ALG-1-associated mRNAs correlated strongly between both samples (Pearson square correlation factor [R²] = 0.99). Potential bantam/miR-58 targets are indicated in red (perfect 7-mer binding site in the 3′ UTR) and light blue (7-mer, one mismatch allowed). These potential bantam/miR-58 target genes were significantly overrepresented among the genes depleted in the “mir-58” RIPs, as expected for potential bantam/miR-58 targets. (B) Genes containing potential bantam/miR-58 binding sites were overrepresented among genes that were significantly depleted in the “mir-58” RIPs (FDR < 5%, SAM), as expected for potential bantam/miR-58 target genes. In contrast, no such overrepresentation could be detected for genes that were significantly enriched in the “mir-58” RIPs (FDR < 5%, SAM). P-values were determined by hypergeometric distribution.

Targets identified by RIP-chip-SRM are enriched in evolutionarily conserved bantam/miR-58 seed binding sites. The same data sets as in Figure 3A (bantam/miR-58 targets identified by RIP-chip, targets predicted by TargetScan, and random control group) were tested for the presence (A,C) and frequency (B,D) of imperfect (one mismatch allowed, A,B) and perfect (C,D) bantam/miR-58 7-mer seed binding sites in their 3′ UTRs and the predicted 3′ UTRs of their corresponding C. briggsae homologs. The RIP-chip candidates are shown subdivided into the “RIP-chip (Protein ↑)” (log₂ protein_mir-58/wt > 0) and “RIP-chip (Protein ↓)” (log₂ protein_mir-58/wt < 0) groups (as in Fig. 3C). P-values for the presence of bantam/miR-58 seed binding sites (A,C) were determined by a Fisher’s exact test (one-sided, with the random group as the null hypothesis). P-values for the frequency of bantam/miR-58 seed binding sites (B,D) were determined by hypergeomtric distribution (using the frequency of binding sites in the whole 3′ UTR-ome as the null hypothesis). (*) P < 0.05, (**) P < 0.01, (***) P < 0.001.

TAP::ALG-1 RIP enriches for miRNA targets. (A) RIP-chip of TAP::ALG-1-associated mRNAs. The log₂ ratio profiles of the 4071 oligo probes (representing 3750 gene models, out of a total of 15,322 oligo probes, representing 12,890 gene models, detected) that were significantly changed (P-value < 0.05, two-sample t-test) in the TAP::ALG-1 RIP compared to the mock RIP are shown. Log₂ ratios (RIP mRNA/total mRNA) were calculated for each measured oligo probe and normalized to the corresponding mock average log₂ ratio for better visualization. A total of 99.7% of all features are enriched in the TAP::ALG-1 RIP. Three independent biological replicates were analyzed. Genotypes used: TAP::ALG1: alg-1(tm492); opIs205, wild type: N2 (= mock control). (B) The set of 3750 TAP::ALG-1-associated mRNAs identified here overlaps strongly (P-value < 3 × 10⁻⁵⁹; hypergeometric distribution) with a previously published set of ALG-1-associated mRNAs identified by Clip-seq (Zisoulis et al. 2010). Only Clip-seq mRNAs that had also been detected on our microarrays (2560 out of 3093) were included in the analysis. The total number of expressed genes (12,890) on our microarray was used for the statistical analysis. (C) TAP::ALG-1 RIP enriches for mRNAs with seed binding sites for abundant miRNAs. miRNA seeds were arranged in groups of five based on their miRNA microarray expression values (see Supplemental Table S4). TOP1 includes the five most highly expressed miRNA seeds (including bantam/miR-58), TOP2 includes the next five most highly expressed seeds, etc. The CTL1, CTL2, and CTL3 groups each contained five miRNA seeds that were below the detection limit on the miRNA microarray (all seed groups are listed in Supplemental Table S13). Relative seed binding site enrichment in the 3′ UTRs of TAP::ALG-1-associated mRNAs compared to nonassociated mRNAs was determined for each group as described in Methods. Two different P-values and three different signal-to-noise ratio (SNR) cutoffs were applied to define the TAP::ALG-1 associated mRNAs.

Bantam/miR-58 targets identified by RIP-chip are regulated mainly at the translational level. (A,B) Cumulative fraction plot of the log₂ protein (A) and mRNA (B) changes (“mir-58” to “wild type”) of potential bantam/miR-58 targets identified by RIP-chip or TargetScan prediction and of a random control group. (A, inset) Overlap in membership between the RIP-chip and TargetScan groups. (C) Log-log plot of the protein (x-axis) and mRNA (y-axis) changes shown in A and B for the RIP-chip group. The RIP-chip candidates were subdivided into “RIP-chip (Protein up)” (log₂ protein_mir-58/wt > 0; dark blue triangles) and “RIP-chip (Protein down)” (log₂ protein_mir-58/wt < 0; light blue circles) groups, based on whether their protein levels were elevated or reduced in “mir-58” mutants. The circled data points represent the general correlation trend of the majority of data points within both groups. (D) Comparison of the extent of change in abundance at the protein and mRNA levels (protein_mir-58/wt/mRNA_mir-58/wt) are depicted as box plots. The box plot marked with “*” differs significantly (P-value = 0.009, Kolmogorov-Smirnov test) from the random gene group. The RIP-chip candidates are also shown subdivided into the “RIP-chip (Protein ↑)” (log₂ protein_mir-58/wt > 0) and “RIP-chip (Protein ↓)” (log₂ protein_mir-58/wt < 0) groups. The bottom and the top of the boxes are the 25th and 75th percentile, respectively, while the whiskers represent the ninth and 91st percentile. The band within the box depicts the 50th percentile (median) and the “+” the mean value.