Catalytic residues of microRNA Argonautes play a modest role in microRNA star strand destabilization in C. elegans

Abstract

Many microRNA (miRNA)-guided Argonaute proteins can cleave RNA ('slicing'), even though miRNA-mediated target repression is generally cleavage-independent. Here we use Caenorhabditis elegans to examine the role of catalytic residues of miRNA Argonautes in organismal development. In contrast to previous work, mutations in presumed catalytic residues did not interfere with development when introduced by CRISPR. We find that unwinding and decay of miRNA star strands is weakly defective in the catalytic residue mutants, with the largest effect observed in embryos. Argonaute-Like Gene 2 (ALG-2) is more dependent on catalytic residues for unwinding than ALG-1. The miRNAs that displayed the greatest (albeit minor) dependence on catalytic residues for unwinding tend to form stable duplexes with their star strand, and in some cases, lowering duplex stability alleviates dependence on catalytic residues. While a few miRNA guide strands are reduced in the mutant background, the basis of this is unclear since changes were not dependent on EBAX-1, an effector of Target-Directed miRNA Degradation (TDMD). Overall, this work defines a role for the catalytic residues of miRNA Argonautes in star strand decay; future work should examine whether this role contributes to the selection pressure to conserve catalytic activity of miRNA Argonautes across the metazoan phylogeny.

Graphical Abstract

Figure 1.

Mutation of catalytic tetrad in both miRNA Argonautes does not impact physiology. (A) Schematic of miRNA Argonaute genes in C. elegans with positions encoding residues of catalytic tetrad in ALG-1 and ALG-2 highlighted. ALG-5 residues in equivalent positions are not conserved. For ALG-1/2, positions in short isoform are denoted. Yellow position highlights aspartate that is mutated to alanine in ‘AEDH’ mutants. (B) Fecundity assay. Each dot represents one brood; four replicates conducted shown side by side for each genotype. (Only three replicates were done for alg-1(null).) All mutants were compared to wild type by one-way ANOVA with Dunnett's test. Only alg-1(null) shows significant difference from wild type. ****P< 0.0001. Assay was performed at 25°C.

Figure 2.

Some miRNA star strands are slightly elevated in catalytic residue mutants. (A) MA plot showing average abundance on X-axis and log₂(fold change) on Y-axis. miRNAs showing significant changes in alg-1(AEDH); alg-2(AEDH) versus wild type are shown in blue (miRNA guide strands) or pink (star strands) (DESeq2 FDR < 0.1). Three (L4) or four (embryo and young adult) biological replicates of each genotype were analyzed. (B) MA plot as in (A), with all guide strands shown in black and all star strands in pink. (C) Summary of log₂(fold change) values for all guides or star strands in each stage in alg-1(AEDH); alg-2(AEDH) versus wild type. Log₂(fold change) of star strands were compared to those of guide strands in the same sample by one-way ANOVA followed by Sidak's multiple comparisons test. ****P< 0.0001.

Figure 3.

Catalytic tetrad of ALG-2 plays a greater role in star strand destabilization than that of ALG-1. (A) MA plot showing average abundance on X-axis and log₂(fold change) on Y-axis. miRNAs showing significant changes in alg-1(AEDH) or alg-2(AEDH) versus wild type embryos are shown in blue (miRNA guide strands) or pink (star strands) (DESeq2 FDR < 0.1). Four biological replicates of each genotype were analyzed. (B) MA plot as in (A), with all guide strands shown in black and all star strands in pink.

Figure 4.

High duplex stability favors a greater role for catalytic residues in unwinding. (A) Percent trimming of guide strand and duplex stability (minimum free energy) of each miRNA duplex. Frequency of star strand misregulation in alg-1(AEDH); alg-2(AEDH) among three stages (embryo, L4, adult) is color-coded. (B) Histogram of duplex stabilities with bins containing mir-63 and mir-72 and their mutant variants indicated. (A, B) Only miRNAs with sufficient abundance for trimming analysis in (133) are shown. (C, D) Predicted structure of mir-63 and mir-72 and engineered mutations in each duplex. (E, F) Absolute quantification by qPCR of mir-63 and mir-72 guide and star strands in L4 samples (three biological replicates for each genotype). (G, H) Ratio of guide to star strand calculated from absolute quantification. (E–H) Unpaired t-tests, **P-value < 0.01, *P-value < 0.05.

Figure 5.

Changes in star strand abundance result in subtle changes in target repression. (A) MA plots showing average abundance (of all genes >10 reads per million) on X-axis and log2(fold change) on Y-axis in alg-1(AEDH); alg-2(AEDH) versus wild type. Only 1, 0 and 2 genes are significantly altered in expression in embryo, L4 and adult, respectively (DESeq2 P_adj< 0.05). (B) Empirical cumulative distribution function for log₂(fold change) values of predicted targets for mir-63, mir-72 or neither (non-targets). Low abundance transcripts (baseMean < 10) were excluded from the analysis. P-values < 0.05 are shown for Kolmogorov-Smirnoff test (comparison to non-targets). (A, B) Three (L4) or four (embryo and young adult) biological replicates of each genotype were analyzed.

Figure 6.

Down-regulation of select guide strands in the catalytic residue mutants is EBAX-1-independent. (A) Schematic of model being tested. Highly complementary targets that would otherwise be sliced may instead induce TDMD when slicing is inactivated. (B) Possible outcomes that would support the model proposed in (A) (top) or support an alternative mechanism (bottom). (C, D, G) MA plots showing average abundance on X-axis and log₂(fold change) on Y-axis. miRNAs showing significant changes are shown in light blue or pink (DESeq2 FDR < 0.1). Genotypes being compared are indicated in graph title. (E) Correlation of changes induced by alg-1(AEDH); alg-2(AEDH) in wild type (X-axis) or ebax-1(null) background (Y-axis). (F) Level of mir-243 in each sample after default DESeq2 normalization in pairwise comparisons shown. One-way ANOVA was performed, and P-values are from post hoc Sidak's multiple comparisons test. (E, H) Only miRNAs with baseMean > 50 in wild type experiment 1 are included. (C–H) All samples are young adults. Four biological replicates of each genotype were analyzed.

Figure 7.

Model of observed effects on miRNA star strands in catalytic residue mutants. A small portion of miRISC maturation depends on catalytic residues. In alg-1(AEDH);alg-2(AEDH) mutant backgrounds, this small portion of miRISC retains the star strand (leading to elevated star strand levels). The major portion of miRISC is matured in a slicing-independent manner, thus leaving overall miRISC function largely intact in the mutant background. Dependence of unwinding on catalytic residues is most apparent in embryos in which ALG-2 is highly expressed and for certain thermodynamically-stable miRNA:star duplexes like mir-63 and mir-72. For most miRNAs in L4 and adult samples, dependence of unwinding on catalytic residues is negligible.