Contributions of mRNA abundance, ribosome loading, and post- or peri-translational effects to temporal repression of C. elegans heterochronic miRNA targets

Abstract

miRNAs are post-transcriptional regulators of gene activity that reduce protein accumulation from target mRNAs. Elucidating precise molecular effects that animal miRNAs have on target transcripts has proven complex, with varied evidence indicating that miRNA regulation may produce different molecular outcomes in different species, systems, and/or physiological conditions. Here we use high-throughput ribosome profiling to analyze detailed translational parameters for five well-studied targets of miRNAs that regulate C. elegans developmental timing. For two targets of the miRNA lin-4 (lin-14 and lin-28), functional down-regulation was associated with decreases in both overall mRNA abundance and ribosome loading; however, these changes were of substantially smaller magnitude than corresponding changes observed in protein abundance. For three functional targets of the let-7 miRNA family for which down-regulation is critical in temporal progression of the animal (daf-12, hbl-1, and lin-41), we observed only modest changes in mRNA abundance and ribosome loading. lin-41 provides a striking example in that populations of ribosome-protected fragments from this gene remained essentially unchanged during the L3-L4 time interval when lin-41 activity is substantially down-regulated by let-7. Spectra of ribosomal positions were also examined for the five lin-4 and let-7 target mRNAs as a function of developmental time, with no indication of miRNA-induced ribosomal drop-off or significant pauses in translation. These data are consistent with models in which physiological regulation by this set of C. elegans miRNAs derives from combinatorial effects including suppressed recruitment/activation of translational machinery, compromised stability of target messages, and post- or peri-translational effects on lifetimes of polypeptide products.

Figure 1.

miRNAs in the C. elegans heterochronic pathway. (A) Five genes of the heterochronic pathway are known to be targeted by miRNAs during C. elegans larval development. (B) Overview of the high-throughput ribosome profiling procedure. (NGS) Next generation sequencing.

Figure 2.

Changes in mRNA levels of miRNA targets measured by mRNA-seq. (A) Steady-state mRNA levels measured by mRNA-seq (solid lines) and protein levels (dashed lines) measured by immunoblotting for lin-4 targets (lin-14 and lin-28). mRNA and protein levels were normalized to L1 levels, and relative levels at subsequent larval stages were plotted on the same scale with a logarithmic vertical axis. (B) mRNA levels measured by mRNA-seq for let-7 and miR-48/miR-84/miR-241 targets (lin-41, daf-12, and hbl-1) for each larval stage, with nonlogarithmic vertical axis. Counts were normalized using the TMM method in the EdgeR package (Robinson and Oshlack 2010) to account for both library size and composition differences. (Data points) Average normalized tag count for all replicates at each stage; (error bars) standard error of the mean. Significance of differential mRNA levels across larval development (L1–L4) was determined by a two-sided t-test: lin-14 P = 0.07, lin-28 P = 0.05, lin-41 P = 0.21, hbl-1 P = 0.001, daf-12 P = 0.02. Additional normalizations are described in Supplemental Figure S2.

Figure 3.

lin-14 mRNA abundance in animals heterozygous for a 3′ UTR deletion. (A) Schematic of wild-type and mutant lin-14 alleles in the heteroallelic configuration. The n536 deletion is confined to the 3′ UTR and eliminates five lin-4 binding sites. The n539 point mutation is used to distinguish mutant from wild-type transcript; reverse transcription and PCR primers were selected using regions around the point mutation that are identical in both alleles, generating products of identical length and differing only at a single internal position. (B) Abundance ratios for wild-type and mutant alleles in L1 and L4 larvae, as measured by Illumina sequencing. Individual ratios were measured as the simple ratio of mutant to wild-type reads, error bars represent standard error of the mean, P-value is from a two-sided t-test. Equivalent data derived using Sanger sequencing is shown in Supplemental Figure S3.

Figure 4.

Changes in ribosome-protected fragment abundance and ribosome loading for heterochronic miRNA targets. Total RPF levels and ribosome loading at each larval stage are shown for lin-4 targets lin-14 and lin-28 (A–B) and let-7 family targets hbl-1, daf-12, and lin-41 (C–D). Count normalization and axis scaling are as in Figure 2, and protein levels are plotted for LIN-14 and LIN-28 as a reference (protein data are the same as in Fig. 2). Ribosome loading is defined as the log₂ ratio of normalized counts from RPF libraries to corresponding mRNA-seq libraries. The significance of changes in RPF abundance across larval development (L1–L4) was determined by a two-sided t-test: lin-14 P = 0.003, lin-28 P = 0.01, lin-41 P = 0.29, hbl-1 P = 0.01, daf-12 P = 0.002. (Error bars) Standard error of the mean.

Figure 5.

Ribosome occupancy profiles of miRNA targets. Stage-specific ribosome occupancy profiles are shown for targets of lin-4 (A), miR-48/miR-84/miR-241 (B), and let-7 (C). Occupancy profiles are generated by first assigning counts to each codon position in a transcript based on the number of RPF reads whose P-site falls on that codon (RPFs contain 12 nt 5′ to the P-site, see Ingolia et al. 2009; Stadler and Fire 2011). Occupancy values are then normalized within each data set by dividing the counts value at each position by the total counts for that transcript, then multiplying by the transcript length, such that the average codon value is one. Finally, individual profiles are smoothed using kernel density estimation. Data integrated from all replicate data sets are shown: The top of the gray region represents the minimum value observed at that position among all replicates, while the top of the colored region represents the maximum value observed. Equivalent non-smoothed and non-normalized plots are shown in Supplemental Figure S7.

Figure 6.

Ribosome occupancy profiles: gross changes. Each gene was divided into eight bins of equal length, and the relative occupancy of RPFs within each bin was calculated as a proportion of the total reads mapping to the gene within the sample that map to the bin. Occupancy for each bin was then compared between miRNA-unregulated and -regulated stages for miRNA targets (L1–L2 for lin-14 and lin-28, L2–L4 for hbl-1, daf-12, and L3–L4 for lin-41). The change for a given bin is the simple difference (regulated − unregulated), such that a positive score indicates the bin had greater occupancy in the miRNA-regulated stage. (Gray bars) Standard deviation of bin changes for mRNAs of similar coverage, with similar coverage defined as those genes with reads-per-codon values within 20% of the mRNA of interest.