Report

MicroRNA-Directed Cleavage of HOXB8 mRNA

See allHide authors and affiliations

Science  23 Apr 2004:
Vol. 304, Issue 5670, pp. 594-596
DOI: 10.1126/science.1097434

Abstract

MicroRNAs (miRNAs) are endogenous ∼22-nucleotide RNAs, some of which are known to play important regulatory roles in animals by targeting the messages of protein-coding genes for translational repression. We find that miR-196, a miRNA encoded at three paralogous locations in the A, B, and C mammalian HOX clusters, has extensive, evolutionarily conserved complementarity to messages of HOXB8, HOXC8, and HOXD8. RNA fragments diagnostic of miR-196–directed cleavage of HOXB8 were detected in mouse embryos. Cell culture experiments demonstrated down-regulation of HOXB8, HOXC8, HOXD8, and HOXA7 and supported the cleavage mechanism for miR-196–directed repression of HOXB8. These results point to a miRNA-mediated mechanism for the posttranscriptional restriction of HOX gene expression during vertebrate development and demonstrate that metazoan miRNAs can repress expression of their natural targets through mRNA cleavage in addition to inhibiting productive translation.

Nearly 1% of the predicted mammalian genes encode microRNAs (miRNAs) (13). As previously reported for miR-10 (1), genes for miR-196 (1, 2) map to homeobox (HOX) clusters (Fig. 1 and table S1). HOX clusters are groups of related transcription factor genes crucial for numerous developmental programs in animals (4). Mammals have four HOX clusters (HOX A to D) containing a total of 39 genes organized into 13 paralogous subgroups (Fig. 1) (4).

Fig. 1.

Genomic organization of HOX clusters. Colored arrows indicate HOX genes representing 13 paralogous groups; black arrowheads depict miRNA genes. Repression supported by bioinformatic evidence only (dotted red line), cell-culture and bioinformatic evidence (dashed line), or in vivo, cell culture, and bioinformatic evidence (solid line) are indicated. The vertical red line indicates that miRNAs from any of the three loci could repress the targets.

The miR-196 miRNAs have intriguing complementarity to sites in the 3′ untranslated regions (3′ UTRs) of HOX genes representing each cluster. With the exception of a single G:U wobble, pairing between miR-196a and the human HOXB8 3′ UTR is perfect. The functional importance of this miR-196 complementary site is supported by its conservation in the fish and frog HOXB8 3′ UTRs, despite the divergence of surrounding UTR sequences (Fig. 2A). This conserved, near-perfect pairing suggested that, like natural miRNA targets in plants (57) or engineered miRNA targets in animals (8, 9), HOXB8 mRNA is targeted for cleavage. MicroRNA-directed cleavage can be detected by using a modified form of 5′-RACE (rapid amplification of cDNA ends) because the 3′ product of this cleavage has two unusual properties: (i) a 5′ terminus that is a suitable substrate, without further modification, for ligation to an RNA adaptor using T4 RNA ligase and (ii) a 5′ terminus that maps precisely to the nucleotide that pairs with the tenth nucleotide of the miRNA (5, 7). Accordingly, we used this method to examine whether HOXB8 mRNA was a natural target of miR-196–directed cleavage. Having determined that miR-196 is expressed during mouse embryogenesis starting at or before day 7 (10) (fig. S1), total RNA from day 15 to day 17 mouse embryos was chosen for this analysis. Of the eight 5′-RACE clones that ended in the vicinity of the miR-196 complementary site [i.e., within the 150-nucleotide (nt) segment centering on the complementary site], seven terminated precisely at the position diagnostic of miR-196–directed cleavage (Fig. 2A, the eighth terminated 41 nt downstream).

Fig. 2.

MiR-196–directed cleavage of HOXB8 mRNA. (A) Sequence alignment of the miR-196 complementary site in the 3′ UTRs of HOXB8 genes (table S2). Absolutely conserved nucleotides are highlighted in black; those of the complementary site are either red (Watson-Crick pairing to miR-196a) or pink (G:U wobble). The vertical arrowhead indicates the 5′ end of cleavage products detected by 5′-RACE, with the frequency of clones noted. Hs, human; Mm, mouse; Rn, rat; Xl, frog; Dr, zebrafish; Fr, pufferfish. (B) Response to miR-196a in HeLa cells (16). Firefly luciferase reporters containing either the miR-196 complementary site from mammalian HOXB8, or the perfect antisense sequence of miR-196a or miR-1d (HOXB8, miR-196a-as, and miR-1d-as, respectively) were cotransfected with the indicated amount of cognate or noncognate miRNA (10). Firefly luciferase activity was normalized to that of the Renilla transfection control, then activity with cognate miRNA (solid) was normalized to the median activity with noncognate miRNA at the same concentration (open). Each box represents the distribution of activity measured for each reporter (n = 9; bars define 10th and 90th percentiles; box spans 25th and 75th percentiles; line and number indicate median normalized activity). In each case, the expression observed for the cognate miRNA significantly differed from that for the noncognate miRNA (P < 0.001, Mann-Whitney test). (C) Reporter mRNA levels monitored with an ribonuclease protection assay (10, 25). Protected fragments of each probe are indicated (arrowheads), and were absent when cells did not receive reporter plasmid (Ø). Intensity of firefly relative to Renilla bands (Firefly/Renilla) is shown and normalized to that of the noncognate miRNA. In three independent repetitions of this experiment, normalized mRNA levels were 0.35 ± 0.03, 0.53 ± 0.02, and 0.52 ± 0.03 for miR-1d-as, miR-196a-as, and HOXB8, respectively (±SD).

By analyzing RNA isolated from the mouse, the HOXB8 5′-RACE results validated this miRNA-target interaction in the animal, whereas previous experimental support for predicted mammalian miRNA targets has come from reporter assays in cultured cells (11). Moreover, this experiment detected miRNA-mediated cleavage of the targeted message, whereas, in the previously examined metazoan examples, translational inhibition had been the mechanism of endogenous miRNA-mediated repression (12, 13). Nevertheless, the HOXB8 5′-RACE results do not rule out the possibility that the predominant mode of silencing is translational inhibition, as illustrated for miR172 regulation of Arabidopsis APATELA2 (7, 14, 15). To explore the mechanism of HOXB8 repression, the miR-196 complementary site from HOXB8 or control complementary sites were placed into the 3′ UTR of the firefly luciferase reporter gene, and the reporter plasmid was cotransfected into HeLa cells together with a transfection control and either cognate or noncognate miRNAs. As expected, miR-196a inhibited luciferase expression from the construct with the complementary site from HOXB8 mRNA (Fig. 2B). Inhibition was essentially the same as that observed for a reporter with perfect antisense complementarity to the miRNA (miR-196a-as), indicating that the conserved G:U wobble involving U5 of miR-196 (Fig. 2A) does not substantially decrease miRNA-directed inhibition. Accompanying this inhibition was a substantial decrease in the amount of reporter mRNA, again similar to that seen with perfect antisense complementarity (Fig. 2C), indicating that a large fraction of the miR-196–directed repression occurred through mRNA degradation.

Genes from each of the other clusters also appear to be miR-196 targets (Fig. 1). The HOXA7 3′ UTR has multiple conserved matches to residues 2 to 8 of miR-196, called “seed matches,” which previously identified it as a likely miR-196 target (Fig. 3A) (11). HOXC8 and HOXD8 UTRs have both seed matches and more extensive complementary sites (Fig. 3A), although none of these sites resemble the HOXB8 site in having perfect pairing at their center (fig. S2). Segments from all three UTRs imparted miR-196–dependent repression to the luciferase reporter without a substantial decrease in reporter mRNA, indicating predominantly translational inhibition (Fig. 3, B and C). To the extent that these experiments in cell culture reflect regulation in animals, miR-196 represses its targets by two posttranscriptional mechanisms. As reported for engineered targets (8, 9, 16, 17), the choice of mechanism appears to depend on the nature of the complementary sites, with translational inhibition apparently requiring more complementarity sites but less pairing within each site.

Fig. 3.

Additional HOX genes predicted to be regulatory targets of miR-196. (A) miR-196 complementary sites in HOXC8, HOXD8, and HOXA7 3′ UTRs (table S2). Wide red bars indicate extensive complementarity, narrow bars indicate seed matches, and asterisks indicate slightly relaxed seed matches that permit a G:U wobble involving U8 of miR-196 (fig. S2). Abbreviations as in Fig. 1A, plus Gg, chicken, and Hf, hornshark. (B) Experimental support for predicted targets. Experiments were analogous to those of Fig. 2B, except the firefly reporters contained fragments of human HOXC8, HOXD8, and HOXA7 UTRs bracketed in (A) (table S3). Red boxes denote statistically significant repression (P < 0.001, Mann-Whitney test). (C) Reporter mRNA levels monitored as in Fig. 2C. In three independent repetitions, normalized mRNA levels were 1.01 ± 0.08, 1.00 ± 0.09, and 0.95 ± 0.06 for HOXC8, HOXD8, and HOXA7, respectively (±SD).

For some HOX genes, posttranscriptional processes combine with complex transcriptional regulation to generate the pattern of HOX expression seen in early development (18, 19). Our results indicate that miR-196 delimits or dampens the expression of HOXB8 and probably also HOXC8, HOXD8, and HOXA7, thus pointing to additional posttranscriptional control of genes in the HOX clusters. The temporal and spatial order of HOX gene expression along the anterior-posterior axis is initially colinear with the physical position of the HOX genes on their chromosomes (4, 20). If the same is true for mir-196 genes, their locations suggest that they would initially be transcribed slightly later than their targets and could help define the posterior boundary of target gene expression. Close orthologs of miR-196 have not been found in invertebrates, although miR-196 is a distant homolog of the let-7 miRNAs (21). Nonetheless, an unrelated miRNA, miR-iab-4, maps to the corresponding region in the Bithorax complex of insect HOX genes (22), and this miRNA is predicted to target Ubx (23), an insect counterpart of the HOX 6 to 8 paralogous groups, suggesting that analogous regulation might extend to insects (Fig. 1).

The observation of endogenous miRNA-directed mRNA cleavage in animals raises the question of how many other metazoan miRNA targets might be down-regulated by cleavage. Few other human messages have such extensive complementarity to the known miRNAs. Nonetheless, analysis of off-target siRNA-mediated regulation indicates that extensive complementarity is not always required (24), leaving open the possibility that a large fraction of metazoan miRNA regulation might be achieved through mRNA cleavage.

Supporting Online Material

www.sciencemag.org/cgi/content/full/304/5670/594/DC1

Materials and Methods

Figs. S1 and S2

Tables S1 to S3

References and Notes

View Abstract

Navigate This Article