An Extensive Class of Small RNAs in Caenorhabditis elegans

See allHide authors and affiliations

Science  26 Oct 2001:
Vol. 294, Issue 5543, pp. 862-864
DOI: 10.1126/science.1065329


The lin-4 and let-7 antisense RNAs are temporal regulators that control the timing of developmental events inCaenorhabditis elegans by inhibiting translation of target mRNAs. let-7 RNA is conserved among bilaterian animals, suggesting that this class of small RNAs [microRNAs (miRNAs)] is evolutionarily ancient. Using bioinformatics and cDNA cloning, we found 15 new miRNA genes in C. elegans. Several of these genes express small transcripts that vary in abundance during C. elegans larval development, and three of them have apparent homologs in mammals and/or insects. Small noncoding RNAs of the miRNA class appear to be numerous and diverse.

Small RNAs perform diverse functions within cells, including the regulation of gene expression (1–4). One class of regulatory RNA includes the small temporal RNA (stRNA) products of the geneslin-4 and let-7 in Caenorhabditis elegans. The lin-4 and let-7 RNAs are ∼22 nucleotides (nt) in length, and are expressed stage-specifically, controlling key developmental transitions in worm larvae by acting as antisense translational repressors (2–4).

lin-4 and let-7 were identified by their mutant phenotypes (2, 3) and, until recently, were the only known RNAs of their class. However, the phylogenetic conservation oflet-7 RNA sequence and developmental expression (5), and the overlap between the stRNA and RNA interference (RNAi) pathways (6, 7), suggested that stRNAs are part of an ancient regulatory mechanism involving ∼22-nt antisense RNA molecules (8).

To identify more small regulatory RNAs of the lin-4/let-7class in C. elegans, we used informatics and cDNA cloning to select C. elegans genomic sequences that exhibited four characteristics of lin-4 and let-7: (i) expression of a mature RNA of ∼22 nt in length; (ii) location in intergenic (non–protein-coding) sequences; (iii) high DNA sequence similarity between orthologs in C. elegans and a related species, Caenorhabditis briggsae; and (iv) processing of the ∼22-nt mature RNA from a stem-loop precursor transcript of ∼65 nt (2, 3).

In an informatics approach to identifying candidate small regulatory RNAs, predicted C. elegans intergenic sequences that were also highly conserved in C. briggsae (9,10) were analyzed using the RNA folding program “mfold” (11–13). Forty sequences were predicted by mfold to form a stem-loop similar in size and structure to lin-4and let-7. Probes complementary to these sequences were tested against Northern blots of total worm RNA (13), and three of them detected small RNA transcripts (Table 1 and Fig. 1A).

Figure 1

Northern blots of small RNA transcripts. (A through C) Total RNA from C. elegans larvae (stages L1 through L4) or from mixed stage (M) populations were blotted and probed with oligonucleotides complementary to either the 5′ or 3′ half of the indicated transcript (13). U6 = the same filters were probed with probe to U6 snRNA as a loading control. (A)mir-60 5′ probe detects a transcript of ∼65 nt. The ratio of L1 to L4 mir-60 signal, normalized to U6, is about 5:1. The mir-60 3′ probe (not shown) detects a similar-sized species with a similar developmental profile. (B) mir-80 3′ probe detects a ∼22-nt RNA expressed uniformly at all stages. (C) mir-52 5′ probe. The normalized mir-52 signal is threefold greater in the L1 versus the L3. (D) mir-1 3′ probe detects a transcript of ∼22 nt in total RNA from mouse (Mm) 17-day embryos, mixed-stage C. elegans (Ce), Drosophila melanogaster (Dm) mixture of embryo-larvae-pupae, and in a sample of human heart (ht) tissue. Other human tissue samples were brain (br), liver (li), kidney (ki), and lung (lu). (E)mir-1 and mir-58 probes to total RNA from mixed populations of wild-type (+) and dcr-1(ok247) (–) animals. An increase in the proportion of unprocessed ∼65-nt precursor is observed in the dcr-1 RNA.

Table 1

MicroRNA gene products in Caenorhabditis elegans.

View this table:

In a second approach, a cDNA library (about 1.6 × 106 independent lambda clones) was prepared from a size-selected (∼22-nt) fraction of C. elegans total RNA (14) and sequence was obtained for 5025 independent inserts, representing 3627 distinct sequences (13). Some 386 of these sequences were represented by multiple (from 2 to 129) clones. Each of these multiple-hit cDNA sequences was compared using BLAST (15) to the NCBI database, and to approximately 800,000 raw sequence traces of C. briggsae genomic sequence (16). Single-copy cDNA sequences that corresponded to no previously known (or previously predicted) transcripts (17), and that were conserved in the C. briggsae genome, were analyzed using mfold for a predicted stem-loop structure. A total of 38 novel cDNA sequences fit these criteria, of which 13 were tested for expression by Northern hybridization; in all 13 cases, small transcripts (∼22 nt and/or ∼65 nt) were detected (Table 1 and Fig. 1). (The other 25 sequences have not been tested for expression.)

These 13 new genes identified by cDNA cloning, together with two additional genes from the informatics screen, were namedmir, for microRNA (18, 19). All 15 of these miRNA genes appear to produce ∼65-nt stem-loop transcripts (Fig. 2) that may be processed to ∼22-nt forms by the same DCR-1/ALG-1/ALG-2 system involved withlin-4 and let-7 processing (6,7). For the two RNAs that we tested (mir-1 andmir-58), dcr-1 activity was required for normal processing of the ∼65-nt precursor (Fig. 1E). So in some cases, such as lin-4 and let-7, the ∼22-nt form is processed from the 5′ part of the stem (6, 7), and in other cases, such as mir-1 and mir-58, from the 3′ part (Fig. 2), suggesting gene-specificity of miRNA processing and/or stabilization. For the three miRNA genes identified in our informatics screen (mir-60, mir-88, andmir-89), the longer stem-loop transcripts were detected by Northern blot, but ∼22-nt forms were not detected, suggesting that their processing is inefficient, or is sharply restricted developmentally. For mir-60, 20-nt cDNA clones were identified, suggesting that mir-60 is processed, but the mature form accumulates at levels below threshold for detection by Northern blot.

Figure 2

Predicted secondary structures of stem-loop precursors of selected C. elegans miRNAs. Sequences of the ∼22-nt mature small RNA are red, and were inferred from cDNA sequence, northern blots, and/or C. elegans::C. briggsae homology (Table 1). Phylogenetically conserved nucleotides are bold. The 5′ end is to the upper left.

At least 10 of the 15 miRNAs vary in abundance during C. elegans larval development, perhaps reflecting roles for these particular genes in developmental timing (Table 1 and Fig. 1).mir-1, mir-2, and mir-87 have apparent orthologs in mammals and/or insects (Table 1 and Fig. 1).mir-1 is expressed tissue-specifically in humans (heart), and stage-specifically in mouse embryogenesis (Fig. 1D). An evolutionarily conserved miRNA such as mir-1 may have coevolved with its mRNA targets, and hence, could retain a similar developmental or physiological role in diverse taxa (5).

lin-4, let-7, and the 15 new miRNA genes described here are members of a gene family that could number in the hundreds in C. elegans (18) and other animals (19). To date, approximately 100 miRNA genes have been identified in worms, flies, and human cells (18, 19), and it is very likely that the screens conducted so far have not reached saturation. Therefore, additional C. elegans miRNAs can be identified using cDNA library sequencing. Also, continued application of whole-genome sequence alignment should identify additional new miRNAs, because this informatics approach complements the cDNA cloning. For example, using only a sample of the worm genome forC. elegans/C. briggsae alignment, we found two miRNAs (mir-88 and mir-89) that were not represented in the size-selected cDNA library (perhaps due to absent or inefficient processing to the ∼22-nt form).

This collection of new miRNAs exhibits a diversity in sequence, structure, abundance, and expression profile. If miRNA genes are as numerous and diverse as they appear to be, they likely occupy a wide variety of regulatory niches, and exert profound and complex effects on gene expression, development, and behavior. The challenge now is to determine the functions of the miRNAs, to identify potential antisense target mRNAs, and to characterize the consequences of their regulatory interactions.

  • * To whom correspondence should be addressed. E-mail: vambros{at}


Stay Connected to Science

Navigate This Article