MicroRNAs (miRNAs) are a class of short (∼22 nucleotides) noncoding RNAs involved in the degradation and translational regulation of specific target RNAs. Since the discovery of the let-7 and lin-4 miRNAs in Caenorhabditis elegans (reviewed in ref. 1), several hundred such RNAs have been found in both animals and plants2. Although the functions of these miRNAs have been well documented in some cases, their presence in vertebrates has yet to be convincingly associated with particular developmental mechanisms, partly due to the difficulty of visualizing their spatial and temporal distributions in the embryo. On page 1079 of this issue, Jennifer Mansfield and colleagues3 report a first step in meeting this difficult challenge. Their results suggest that miRNAs have a role in fine-tuning specific Hox mRNA expression patterns during mouse development.

Invisible no more

Presumptive miRNA genes are widely distributed in mammals (http://www.sanger.ac.uk/Software/Rfam/mirna/). Techniques for monitoring the expression of these genes has been mostly limited to northern blots and microarray analyses4,5. Determining their precise spatial distribution in the developing embryo requires a technique for visualizing these molecules in situ, analogous to the well-developed in situ hybridization methods for localizing mRNAs. But despite recent methodological improvements in plants6, direct visualization of miRNAs in situ remains difficult to implement owing to their very small size. An alternative approach is to use an indirect assay in which the action of a given miRNA on a target reporter system, rather than the miRNA itself, is monitored. Mansfield et al. adapted this technology, originally pioneered by Cohen and colleagues studying the bantam miRNA in Drosophila melanogaster7, to developing mice. They found that it is a valuable tool for evaluating the potential effects of selected miRNAs in time and space.

The system uses a transgene 'sensor' composed of a constitutively expressed lacZ gene carrying a sequence complementary to a given miRNA in its 3′ untranslated region (UTR) (Fig. 1). Transgenic mice carrying the same transgene without the target sequence show widespread blue staining after the X-gal reaction (Fig. 1a), but transgenic mice carrying the sensor lack staining wherever the lacZ RNA is degraded or impaired as a result of its modified UTR (Fig. 1b). Consequently, much like a photographic negative, areas where the miRNA is expressed (Fig. 1c) show weak or absent staining. The authors used several miRNA target sequences to validate this approach and showed that it could be successfully applied to developing mouse embryos.

Figure 1: Detection of miRNAs in developing mouse embryos using 'sensor' transgenes.
figure 1

Figure courtesy of Jennifer Mansfield

(a) Transgenic embryo expressing the lacZ reporter gene (blue box) ubiquitously. After X-gal staining, the embryo is blue. (b) Transgenic embryo expressing the same transgene carrying a target sequence complementary to a particular miRNA in its 3′ UTR (red). The blue staining disappears wherever such miRNAs are present. (c) Deduced distribution of the miRNA (red). The example shown here corresponds to a Hox-like pattern, such as that reported by Mansfield et al.3

Full size image

The miR-196 and miR-10 miRNAs located in the Hox gene clusters are a good example of the application of the technology. The well-documented role of Hox genes in specifying regional identities along the body's main axes requires the precise temporal and spatial activation of Hox gene transcription. This choreography depends to a large degree on the genomic organization of the four Hox clusters (HoxA–HoxD), such that neighboring genes are activated in temporal and spatial sequences in accordance with their positions along the chromosome8. The molecular processes underlying some of the unique features of the Hox clusters are beginning to be unraveled, but they remain largely unknown.

Yekta et al.9 reported a possible function for the miR-196 miRNAs, whose genes are evolutionarily conserved at the same relative positions in the HoxA, HoxB and HoxC clusters. In particular, they described the presence of target UTR sequences in the closely related Hoxb8, Hoxd8, Hoxc8 and Hoxa7 transcripts and showed, in the case of Hoxb8, that miR-196 could direct cleavage of this transcript. While their results imply that miRNAs may have a role in the colinear distribution of Hox transcripts, neither the presence of miR-196 nor its effect on the Hoxb8 mRNA was determined in the context of the developing embryo.

Profiting from colinearity

Using transgenic sensors, Mansfield et al. now show that both miR-196a and miR-10a are distributed in Hox-like patterns in the embryo (Fig. 1). This illustrates the advantage of having these miRNA genes embedded in the Hox complexes; only in this location can they be assured of having a similar colinear activation as their neighboring Hox genes. Notably, this phenomenon was also shown in previous studies in which promoters introduced into Hox clusters tend to adopt colinear regulation. In the case of miR-10a, the sensor technique showed that the anterior limits of its expression and that of its genomic neighbor Hoxb4 are the same. In contrast, maximal expression of miR-196a, as judged by weaker lacZ staining, was observed posteriorly, precisely where Hoxb8 mRNA is less abundant, suggesting that miRNAs could participate in the fine regulation of at least some Hox expression patterns. Because miR-196a target UTR sequences are found in all group 8 Hox transcripts, miR-196a may negatively regulate either the steady-state levels or the translational efficiency of all group 8 Hox transcripts in the caudal part of the embryo. This possibility is supported by in vitro work showing that miR-196a can direct processing of the Hoxb8 transcript3,9.

Although these observations might bring new players into the complex game of colinearity, it is not yet obvious how this element will contribute to the puzzle's solution. First, the presence of miRNAs and related UTR target sequences seems to be restricted to only a few Hox genes. Therefore, it seems unlikely that this mechanism is a global determinant of Hox expression patterns. Second, transcripts of many Hox/lacZ transgenes are localized quite faithfully even without their corresponding Hox 3′ UTRs. Third, the functional relevance of miR-196 has yet to be documented genetically. In worms and flies, loss-of-function mutations have clearly shown that specific miRNAs have a role in development7,10. Engineering similar mutations in mice (e.g., by microdeletion) will be a difficult task given the presence of related miRNAs on three of the four Hox clusters. An alternative approach would be to upregulate Hoxb8 in the posterior portion of the embryo (e.g., by expressing a transgene carrying a deletion of the target UTR sequence) to assess the effect, if any, of high steady-state levels of Hoxb8 posteriorly. Our current knowledge of the system suggests that Hox gain of function posteriorly should not markedly alter morphologies, but such experiments may uncover some surprises.

So far, the only target of the miR-196 or miR-10 RNAs that has been demonstrated in vivo is the miR-196a-mediated degradation of Hoxb8. This makes it tempting to assume that the only targets of these miRNAs are the Hox transcripts. As miRNAs generally have multiple targets, however, the miRNAs in the Hox complexes could synergistically influence Hox function by targeting other mRNAs.

miRNAs are now entering the field of vertebrate development. It is particularly notable that they do so in a genetic system in which the control of timing is crucial8, given that the founding members of this class of molecules are involved in tightly time-controlled processes in Caenorhabditis elegans1. Further analyses will determine whether these small RNA molecules have a related function in the precise temporal regulation of Hox gene transcripts.