The MicroRNA Revolution: The 2012 Dr. Paul Janssen Award Symposium

Resources

Allison Abbott

Abbott AL, Alvarez-Saavedra E, Miska EA, et al. The let-7 MicroRNA family members mir-48, mir-84, and mir-241 function together to regulate developmental timing in Caenorhabditis elegans. Dev Cell. 2005;9(3):403-14.

Abbott AL. Uncovering new functions for microRNAs in Caenorhabditis elegans. Curr Biol. 2011;21(17):R668-71.

Brenner JL, Kemp BJ, Abbott AL. 2012. The mir-51 family of microRNAs functions in diverse regulatory pathways in Caenorhabditis elegans. PLoS One. 2012;7(5):e37185.

Brenner JL, Jasiewicz KL, Fahley AF, Kemp BJ, Abbott AL. Loss of individual microRNAs causes mutant phenotypes in sensitized genetic backgrounds in C. elegans. Curr Biol. 2010;20(14):1321.

Miska EA, Alvarez-Saavedra E, Abbott AL, et al. Most Caenorhabditis elegans microRNAs are individually not essential for development or viability. PLoS Genet. 2007;3(12):e215.

Victor Ambros

Chaudhuri AA, So AY, Mehta A, et al. Oncomir miR-125b regulates hematopoiesis by targeting the gene Lin28A. Proc Natl Acad Sci USA. 2012;109(11):4233-8.

Karp X, Ambros V. Dauer larva quiescence alters the circuitry of microRNA pathways regulating cell fate progression in C. elegans. Development. 2012;139(12):2177-86.

Lee RC, Feinbaum RL, Ambros V. The C. elegans heterochronic gene lin-4 encodes small RNAs with antisense complementarity to lin-14. Cell. 1993;75(5):843-54.

Mayr C, Hemann MT, Bartel DP. Disrupting the pairing between let-7 and Hmga2 enhances oncogenic transformation. Science. 2007;315(5818):1576-9.

Moss EG, Lee RC, Ambros V. The cold shock domain protein LIN-28 controls developmental timing in C. elegans and is regulated by the lin-4 RNA. Cell. 1997;88(5):637-46.

Gary Ruvkun

Grishok A, Pasquinelli AE, Conte D, et al. Genes and mechanisms related to RNA interference regulate expression of the small temporal RNAs that control C. elegans developmental timing. Cell. 2001;106(1):23-34.

Hayes GD, Ruvkun G. Misexpression of the Caenorhabditis elegans miRNA let-7 is sufficient to drive developmental programs. Cold Spring Harb. Symp. Quant. Biol. 2006;71:21-7.

Reinhart BJ, Slack FJ, Basson M, et al. The 21-nucleotide let-7 RNA regulates developmental timing in Caenorhabditis elegans. Nature. 200;403(6772):901-6.

Slack FJ, Basson M, Liu Z, Ambros V, Horvitz HR, Ruvkun G. The lin-41 RBCC gene acts in the C. elegans heterochronic pathway between the let-7 regulatory RNA and the LIN-29 transcription factor. Mol Cell. 2000;5(4):659-69.

Wightman B, Ha I, Ruvkun G. Posttranscriptional regulation of the heterochronic gene lin-14 by lin-4 mediates temporal pattern formation in C. elegans. Cell. 1993;75(5):855-62.

Oliver Hobert

Chang S, Johnston RJ Jr, Frøkjaer-Jensen C, Lockery S, Hobert O. MicroRNAs act sequentially and asymmetrically to control chemosensory laterality in the nematode. Nature. 2004;430(7001):785-9.

Cochella L, Hobert O. Diverse functions of microRNAs in nervous system development. Curr Top Dev Biol. 2012;99:115-43.

Johnston RJ Jr, Chang S, Etchberger JF, Ortiz CO, Hobert O. MicroRNAs acting in a double-negative feedback loop to control a neuronal cell fate decision. Proc Natl Acad Sci USA. 2005;102(35):12449-54.

Johnston RJ, Hobert O. A microRNA controlling left/right neuronal asymmetry in Caenorhabditis elegans. Nature. 2003;426(6968):845-9.

Poole RJ, Enkelejda Bashllari E, Cochella L, Flowers EB, Hobert O. 2011. A genome-wide RNAi screen for factors involved in neuronal specification in Caenorhabditis elegans. PLoS Genetics. 2011;7(6), e1002109.

Eric Lai

Bender W. MicroRNAs in the Drosophila bithorax complex. Genes Dev. 2008; 22(1):14-9.

Okamura K, Balla S, Martin R, Liu N, Lai EC. Two distinct mechanisms generate endogenous siRNAs from bidirectional transcription in Drosophila melanogaster. Nat Struct Mol Biol. 2008;15(6):581-90.

Okamura K, Liu N, Lai EC. Distinct mechanisms for microRNA strand selection by Drosophila Argonautes. Mol Cell. 2009;36(3):431-44.

Tyler DM, Okamura K, Chung WJ, et al. 2008. Functionally distinct regulatory RNAs generated by bidirectional transcription and processing of microRNA loci. Genes Dev. 2008;22(1):26-36.

Sponsors

This symposium was made possible with support from

Speakers:
Victor Ambros, University of Massachusetts Medical School
Gary Ruvkun, Massachusetts General Hospital and Harvard Medical School

Highlights

• The first microRNA, lin-4, and the transcription factor gene it regulates, lin-14, were among developmental-timing mutants found in genetic screens in the worm, C. elegans.
• The base-pairing mechanism for regulation, related to that in RNA interference, emerged from sequence comparisons of the 22-nucleotide RNA from lin-4 and regions in the 3′ untranslated region (UTR) of the lin-14 mRNA.
• The second known microRNA, let-7, is part of the same heterochronic pathway, and shows evolutionary conservation in a wide range of animals, including humans.
• Sequence conservation is particularly strong for the loops of RNA that are not involved in base pairing.
• Bioinformatic techniques have been generally ineffective in identifying targets for microRNAs.

Introduction

The identification of microRNAs (miRNAs) grew out of the efforts of a strong research community that exhaustively explored the worm, Caenorhabditis elegans, as model organism for development. H. Robert Horvitz at MIT, John Sulston at Cambridge University, and others had painstakingly traced the invariant development of this worm down to the level of individual cells, and had also developed extensive libraries of mutant animals that they shared with other researchers. "This animal presents us a great opportunity to learn about the precision of developmental-control mechanisms." Victor Ambros said, emphasizing that the almost "saturated" cataloguing of mutations was critical to revealing an unexpected regulatory mechanism.

The first microRNA, lin-4

"The discovery of miRNAs was accidental," said Ambros. "There was no theoretical foundation for predicting that there should be these small RNAs to regulate gene expression." As postdoctoral fellows in the Horvitz lab, Ambros and Ruvkun studied a set of mutants in heterochronic genes, which are involved in the timing of development and which can be phenotypically assessed by a number of morphological and functional features including egg-laying. Ambros analyzed a mutant called lin-4, which has a "retarded" phenotype: In contrast to the progression from first to second larval stage in a skin-cell lineage of a wild-type worm, the mutant repeats the first stage. "It happens over and over," Ambros said. "The animal is just basically repeating its larval stages. It's missing all kinds of parts that should have been made at later stages." In particular, these worms fail to develop a vulva and are unable to lay eggs, so the eggs accumulate within their bodies until they hatch and the progeny burst out of the parent.

At the same time, Gary Ruvkun focused on lin-14, which had the opposite phenotype in the same cell lineage. Instead of being stuck at an early larval stage, "progenitor cells don't even bother to do the first-stage program, they skip to the second," Ambros said. These mutants end up being very small with precocious development of some adult structures. When both genes are missing, only the lin-14 phenotype is manifested, supporting the idea that lin-4 regulates lin-14. Other cell lineages showed similar, opposing phenotypes for the two mutations.

The first microRNA, lin-4, and its target lin-14, were discovered in mutant worms whose development was either retarded or precocious. (Slide courtesy of Victor Ambros)

Ruvkun's group found that lin-14 codes for a transcription factor. Interestingly, they also found gain-of-function lin-14 mutants that were insensitive to regulation by lin-4. The genetics community tends to be suspicious of gain-of-function mutations, Ruvkun said, reasoning that the new functions might be irrelevant for normal organisms. But these gain-of-function mutations, which occur not in the protein-coding portion but in the untranslated region (UTR) of the 3′ tail of the mRNA, were very informative. "[The mutations] essentially remove the off switch from a gene," he said, and later experiments showed how this worked.

As Ambros recounts, "In parallel [with the Ruvkun lab's findings with lin-14], Rhonda Feinbaum and Rosalind Lee in my lab cloned lin-4 and found it encoded this little RNA, about 22 nucleotides. Gary and I shared sequences one day, and called each other up and said "Wow! Do you see it? They're complementary!' and started reading off the sequences. It was just an absolute wonderful, wonderful thing to do together."

The complementarity was conserved, so the interaction of the two had to occur by base pairing, Ambros explained. Conservation is usually an indication that the phenomenon, in this case complementarity and the resulting base pairing, is selected for throughout evolution because of its functional importance. Ruvkun went on to show that this interaction resulted in translational repression of the lin-14 mRNA.

In 1998, Craig Mello and Andy Fire discovered the RNA interference mechanism (RNAi), in which exogenous double-stranded RNA modifies the post-transcriptional activity of messenger RNAs containing complementary sequences. The endogenous microRNAs discovered by Ambros and Ruvkun then were viewed as part of a larger class of small RNAs that modify gene expression post-transcriptionally.

The early reception of developmental biologists to the base-pairing interactions of lin-4 and lin-14 was subdued, said Ruvkun. "It was originally thought, well, here's a funny little detail in an organism that you don't need to pay that much attention to." By contrast, the RNA community was enthusiastic and supportive. "They constantly chimed in about universality, because they live in a world of universals: the ribosomal RNA, how translation works, genetic codes," he said. "That sort of kept us going."

Mutations in the 3′ untranslated region (UTR) of lin-14 remove sequences with complementarity to the lin-4 RNA that regulates its translation. (Slide courtesy of Gary Ruvkun)

Ambros and Ruvkun looked at evolutionary conservation of the sequences for the microRNA and its targets. Such comparisons of related species had not been done much in developmental biology. "Even the parts in this [RNA-RNA binding] model that looped out [because they are not very complementary with other parts] are also well conserved," Ruvkun said. "It says it's not just base pairing," but also the non-pairing molecular folds that determine function. Those loops "are probably platforms for various auxiliary factors" that he continues to explore.

Comparing the changing levels of lin-14 mRNA with the levels of its protein, LIN-14, show that lin-4 is "regulating the production of this protein at a translational-control level." This mechanism contrasted with the then almost universal model of gene regulation in which transcription factors bind to DNA to modify transcription of a gene into mRNA. But the direct modification of translation rate also differs from another post-transcriptional mechanism, later demonstrated for RNA interference, in which complementary mRNAs is targeted for more rapid degradation.

The second microRNA, let-7

Ruvkun continued to explore the genetics of the heterochronic pathway. In 2000, he and his team, notably Frank Slack and Brenda Reinhart, described the second microRNA, let-7, which acts later in the pathway.

The discovery of human homologs to let-7 "is a case where writing a grant actually accomplished something," Ruvkun joked. He realized that he could query online the human sequences, which were then becoming available, to look for similarities to the 60-nucleotide precursor of let-7 that might flag a human homolog for this microRNA.

In a genetic screen for mutations that suppress the lethal phenotype of let-7, Slack identified lin-41. Suppression of a phenotype can occur when the suppressor gene (e.g., lin-41) encodes a product through which the first gene (lin-7) exerts its effects. But the complementarity between let-7 and lin-41 is not perfect, and the results of bioinformatic searches for microRNA targets have given many false matches. "It's quite a disaster for the microRNA field," Ruvkun said, because people "use these prediction programs that are generally misleading," and end up following up a lot of dead ends.

In spite of these challenges, it is clear that the microRNAs that Ambros and Ruvkun discovered are an ubiquitous feature of genetic regulation. For example, "of the 150 worm microRNAs, about a third are very well conserved across animals, some perfectly, like let-7," Ruvkun noted. (Plants appear to have evolved microRNAs independently.) Researchers will be exploring their effects for years to come.

Speakers:
Victor Ambros, University of Massachusetts Medical School
Gary Ruvkun, Massachusetts General Hospital and Harvard Medical School
Oliver Hobert, Columbia University and HHMI
Allison Abbott, Marquette University
Eric C. Lai, Memorial Sloan-Kettering Cancer Center

Highlights

• The let-7-family of microRNAs regulates the transcription factor hunchback during development in C. elegans, but after Dauer quiescence lin-4 takes a critical role in proper resumption of development.
• The processing of microRNAs shares key components with the RNA-interference pathway.
• The microRNA lsy-6 regulates the functional lateralization of chemosensory neurons, maintaining low expression and decompacted chromatin through six cell divisions after it is primed.
• Scores of microRNAs have been found in many species by cloning and sequencing, but in most cases their function is hard to discern and may be quite distinct from the crucial developmental roles of the first microRNAs.
• Although the segmentation program of the fly Drosophila is a classic example of genetic regulation through transcription factors, a microRNA plays an indispensable role.

Arrested development

In addition to reviewing the discoveries of microRNAs that earned them the Dr. Paul Janssen Award for Biomedical Research, Victor Ambros and Gary Ruvkun discussed some of their more recent research. "We're still studying lin-4" nearly 20 years after its discovery, said Ambros. He described how this microRNA helps progenitor cells maintain their identity during prolonged periods of quiescence, known as the Dauer state in C. elegans.

Although well-fed worms pass quickly between larval stage 2 and 3 (L2 and L3), when deprived of resources they stop developing for an indefinite time and enter the rugged Dauer phase where they can survive harsh environmental conditions. When resources become available, they seem to resume development where they left off, with no change in the programs of cell lineage for the progenitor cells.

How do progenitor cells reinforce their fate properly during periods of indefinite quiescence?

Ambros focused on a skin cell lineage in which cells change from symmetric division in L2 to asymmetric division in L3. The let-7 family of microRNAs helps enforce the transition between these larval stages by downregulating the key transcription factor hunchback. Individual let-7 family members can be knocked out without terrible consequences because they are redundant. "They're similar to one another, and they can share targets," Ambros noted. But mutations in three let-7 family microRNAs at once causes a worm to be "severely retarded, in the sense that it repeats this second larval stage over and over and it doesn't actually get to the adult stage."

Surprisingly, however, triple mutants developed normally if they entered L3 after Dauer quiescence. "These let-7-family microRNAs are really critical during continuous development, but during this optional, alternative, arrested development and resumption, they are essentially dispensable," Ambros said.

A genetic screen identified several genes involved in this "post-Dauer suppression" of the mutant phenotype, including the microRNA lin-4. This microRNA had been predicted to target hunchback mRNA by base pairing, but had not previously been thought to play a major role in this aspect of development. The results indicated "a reprogramming or a redeployment of the microRNA regulation of this gene hunchback, depending on life history, whether the animals arrested or not," Ambros said. "This illustrates that microRNAs could be incorporated into mechanisms of developmental robustness by being deployed in a situation-appropriate fashion."

Processing of microRNAs

The full significance of the early microRNA research became clearer with the discovery of RNA interference. In this phenomenon, extrinsic double-stranded RNAs are processed in the nucleus by complexes containing the Dicer protein. The resulting small interfering RNAs are incorporated in cytoplasmic complexes containing Argonaute-family proteins such as RDE to modulate translation of messenger RNAs that contain corresponding sequences. When early studies of RNA interference showed that small interfering RNAs are about 20–25 nucleotides long, "that sounded a lot like lin-4," said Gary Ruvkun, which led to important hints about the cofactors for microRNAs. Further research revealed the roles in the RNA interference (RNAi) pathway for Dicer and RDE-1, so it was natural to ask if they help process microRNA, too.

To address this question, Ruvkun collaborated with Craig Mello, using the delicate process of knocking down the RNAi components using RNAi itself. Mello's lab saw that when they used RNAi to knockout Dicer it caused a similar phenotype to let-7, Ruvkun said. RNAi of orthologs of RDE-1, notably the Argonaute proteins ALG-1 and ALG-2, caused the same phenotype. These proteins are carrying out both RNAi and miRNA activity, Ruvkun said. "One leads to translational repression, one leads to cleavage."

Dicer and the Argonaute proteins act in both RNAi and miRNA activity.

Ruvkun also described a powerful screen based on the expression of C. elegans genes in E. coli. They started with a weak allele of let-7 that doesn't cause the lethal bursting phenotype. In this sensitized genetic background, the researchers could look for genes that enhance the let-7 mutation, as evidenced by the bursting phenotype. They fed the worms bacteria that were engineered to express double-stranded DNA that, by means of RNAi, targeted a wide range of worm genes and in this way found about 60 "enhancer" genes for let-7. These genes are candidates for a role in the processing and translational control by microRNAs. "We do it on a whole organism basis," Ruvkun said. "It's completely different from a cellular screen."

Asymmetric brains

Turning to the discovery of microRNAs involved in other developmental processes, Oliver Hobert of Columbia University presented his studies of nervous system lateralization in C. elegans, where at least some of the answer lies in microRNAs. "Brains are, like the rest of our body, morphologically bilaterally symmetric," he began. Even at the molecular level, gene expression, as monitored by messenger RNA levels, appears to be largely the same on the left and right side of the brain. The paradox is that "for more than 150 years now we've know that brains are functionally asymmetric," he said. "How do you make the two sides different from one another if they look, at least superficially, so similar?"

The exhaustive cataloging of lineage in the worm shows that, just as in humans, the nervous system is largely symmetric, down to the cellular level. Most neuron types in the head of the worm come as bilaterally symmetric pairs. However, the main gustatory neurons, called ASE, form a pair with distinct chemosensory responses, explained Hobert.

Even for these neurons, hundreds of genes are expressed symmetrically in the right and left ASE neurons. The ASE-specific expression is induced by a master regulatory transcription factor called CHE-1, which is expressed in both ASE neurons.

Fluorescent labels show that that the microRNA lsy-6 is preferentially expressed in a cell on the left side of the C. elegans nervous system. (Slide courtesy of Oliver Hobert)

However, the transcriptome reveals a notable exception to the symmetry, specifically guanyl cyclase (gcy) receptor genes. "This left-right asymmetric expression of those chemoreceptors very nicely correlates with the functional left-right asymmetry of these two neuron types," Hobert said.

One key to understanding the system, Hobert stressed, is that "we have almost reached a saturation for screening this specific problem." He and his team identified phenotypes in which the usual molecular asymmetry was disrupted, some in which gene expression on both sides looked like that in the right cell, some like the left. This led them to a microRNA called lsy-6, which is expressed relatively late in embryonic development. It appears only in the left cell, where, through a direct target interaction, it represses the gene cog-1 which otherwise drives the cell to the right fate.

But this finding doesn't explain how the asymmetry arises. Hobert noted that the right and left ASE cells arise from two different cells, ABa and ABp, in the 4-cell blastocyst. The differing fates of these very early embryonic cells are driven by differential expression of the genes tbx37 and tbx38, members of a family of genes that contain a "T-box" sequence and often play roles in development. But their expression is long gone by the time CHE-1 drives differentiation of the ASE cells and the strong expression of lsy-6, six cell divisions later.

Genetic screens were surprisingly unhelpful in tracing the connection between these events. Hobert and his team engineered fluorescent markers to monitor the activity of the lsy-6 microRNA gene and to separately interrogate the regulatory regions upstream and downstream of the gene. "Only together do these two regulatory elements produce the correct expression pattern," Hobert stressed. The downstream element produced very low-level expression of lsy-6 in lineages where tbx37/38 had been expressed.

Differential microRNA expression could help impose functional lateralization in protein levels on a symmetric morphology.

This early exposure appears to "prime" the cells for the later activity of the master regulator, CHE-1, as manifested by continuing low-level expression of lsy-6. Visualizing lsy-6 expression within the cell suggested that the priming prevents compaction of the chromatin at this locus, allowing its later expression.

Hobert speculated that microRNA activity might induce the elusive molecular-level lateralization in the brain. Although messenger RNAs seem to be symmetrically expressed, microRNAs could enforce asymmetry in protein levels. A functional specialization could thus be imposed on an underlying symmetric morphology without endangering the precisely coordinated events that lead to the complex morphology. "I think we see this very nicely in C. elegans."

Where's the function?

The first microRNAs, lin-4 and let-7, were identified in genetic screens that found very strong, highly penetrant phenotypes, said Allison Abbott of Marquette University. "We can learn a lot with genetics when we have mutant phenotypes."

But as researchers gained the ability to perform RNA cloning and deep sequencing, the number of microRNAs exploded to more than 100 just in worms. "Now we're coming from a totally different perspective," Abbott said. "We have all these genes but we have no idea what they do." In work published in 2007, Abbott and Ambros, then at Dartmouth, teamed with the labs of Horvitz at MIT and Dave Bartel at the Whitehead Institute to look for phenotypes associated with mutations in these microRNA genes. They were encouraged by the observation that disruptions in the processing pathways for microRNAs are lethal, so the molecules, as a class, must be essential.

The results were disappointing. "We looked at deletion mutants that covered 87 microRNA genes and we plugged them through a whole bunch of assays, and they essentially all looked fine," Abbott said. The hard work uncovered no strong phenotypes like those of let-7 or lin-4 that would allow detailed study of the underlying mechanisms.

Since microRNA sequences are conserved and thus presumably functional, Abbott suggested three explanations for the lack of effect of individual mutations. First, the effects could be limited to specific tissues or cells, like the lsy-6 effect on chemosensory neurons described by Hobert. Second, the effect of these microRNAs might be a more subtle, quantitative fine tuning, rather than a binary on/off function. A third possibility is that the complex regulatory network may compensate for individual deletions, as in the redundant let-7-family microRNAs described by Ambros. That work showed that deleting multiple members of a microRNA family with related seed sequences revealed phenotypes that were not apparent with individual deletions. But "we can't account for everything just based on family-member redundancy," Abbott said.

To look for network interactions beyond family members, Abbott and her colleagues devised a broader approach to detect interactions. "We chose to take a genetic approach and use what we call a sensitized background," she said. They engineered worms in which the microRNA pathway was compromised by deleting one of the two Argonaute genes that are critical components of the complexes that implement post-transcriptional repression. Mutants missing alg-1 have weak defects in developmental timing, but still survive, and accumulate unprocessed miRNA precursors.

"In contrast to our initial studies, where we didn't really find phenotypes," Abbott said, "in this sensitized background, we were able to identify at least modest mutant phenotypes in the majority of microRNAs that we analyzed." They chose 31 microRNAs that were conserved or showed some interesting developmentally regulated expression pattern and found that 25 gave a phenotype when deleted.

They divided the microRNA phenotypes into three classes. Some deletions resulted in enhanced embryonic lethality, suggesting that these microRNAs play a role in embryogenesis. Some enhanced gonad migration defects. A third class had the surprising effect that the worms were healthier than the sensitized worms. The weak developmental-timing defects in the alg-1 mutant background were no longer obvious.

One possible reason for this effect is that these microRNAs act specifically in the developmental timing pathway, but antagonistically with genes (like let-7) whose effects are disrupted in the sensitized background. But Abbott and her colleagues focused on one, the miR-51 family, to explore the possibility that the effects reflect a broader function for these microRNAs. This highly conserved microRNA family is abundantly expressed in many cell types throughout development, so it seems unlikely to be specific to development like lin-4 and let-7. Instead, mutations in this family, especially loss of miR-52, are associated with suppression of several microRNA-dependent phenotypes.

"Broadly expressed microRNAs may have a different function than the highly specific microRNAs involved in development."

"This is a puzzle that we have not cracked yet. Perhaps it's just regulating multiple targets in multiple pathways," Abbott said. Alternatively, it may be that this abundant miRNA is clogging up the already compromised microRNA-processing pathways, so removing it allows other microRNAs to be processed normally. Rather than being specific to particular stages, pathways, or tissues, "these more broadly expressed microRNAs may have a different function," she said.

Finally, Abbott explored in detail one clear phenotype from the broad microRNA screen. Deletion of miR-786 disrupts the rhythmic defecation behavior of the worm and also causes sterility. Although the two behaviors seem unconnected, she said, "the connection is that they both involve calcium signaling, and they both involve the IP-3 receptor."

MicroRNAs help build a fly

The elucidation of the mechanism of body patterning, specifically the development of the segmented body pattern of the fly Drosophila, is a classic story of transcription-factor regulation of gene expression. As Eric Lai of Memorial Sloan-Kettering Cancer Center described, the sequential action of genes that limit and assign identity along the anterior/posterior axis sets up the expression of homeobox (Hox) genes in different domains. Hox genes encode transcription factors whose activity determines the identity of the different segments. They are expressed in the same position along the body axis from anterior to posterior as they appear on the chromosome, a phenomenon known as colinearity. Helping to enforce this expression pattern is the fact that the products of posterior genes transcriptionally repress the anterior genes.

The researchers studied a particular miRNA locus in the bithorax complex, a well-known group of genes that controls development of the posterior thorax. "This is a very complicated locus, because it's actually transcribed on both strands, and it makes microRNAs from both strands," Lai said. The top strand codes for iab-4 and the bottom strand for iab-8.

The homeobox region of Drosophila DNA contains hairpins on both strands that give rise to microRNAs involved in segmentation.

Following the colinearity principle, "it turns out these microRNAs actually repress anterior Hox genes," Lai said. For example, the iab-8 microRNA complements seven-nucleotide sequences in Ubx and abd-A, which are the two Hox genes immediately upstream. Misexpression of the microRNAs, especially iab-8, in patches of cells represses expression of both of the homeobox genes, suggesting that these are genuine targets of the microRNAs. Furthermore, Lai said, "If you just misexpress these microRNAs you can induce changes in the segment identities and the changes in body parts. We have been able to show for the first time that a microRNA is an essential input into the control of the segmental gene expression," said Lai. "But you have to look in the right place."

"If you misexpress these microRNAs you can induce changes in the segment identities."

However, Welcome Bender of Harvard Medical School showed that, surprisingly, knocking out the microRNA hairpin entirely, which removes iab-4 and iab-8, does not cause visible embryonic or adult segmental transformations, a result that is reminiscent of the missing phenotypes described by Abbott and that may arise for the same reasons she enumerated. But deletion of the iab-8 microRNA did have a strong phenotype: they are 100% sterile. The functional basis of this was unknown, but the observation supports the hypothesis that the microRNAs are important in specific locations. Indeed, using in situ markers for nascent transcripts before processing, Lai and his team traced this expression to specific nerve cells.

Removing the microRNA hairpin, they found haphazard spatial patterns of expression of Hox genes in the posterior segment of the ventral nerve cord, including "cells that express combinations of Hox genes they should never have" according to standard regulation models, Lai said. "Even though, in the early embryo the microRNAs are not sufficient to repress Hox genes, they have a non-redundant role in the ventral nerve cord in the central nervous system."

In view of the limitations of bioinformatic searches, Lai's lab also looked for genuine targets of the microRNAs in situ. They found expression of "hundreds of 3′ UTR extensions, specifically in the central nervous system," Lai said. "So they're not included in any gene predictions that people have used." For another gene, conserved sites are missed because of a failure of gene alignments. "Both of these couldn't be found by the traditional bioinformatic methods."

Lai traced the sterility phenotype to neurons that project to the oviduct and uterus, and which are required for egg-laying behavior. "They live exactly where we are seeing this phenotype in the ventral nerve cord with all the misexpression of Hox genes."

"Neurons are also very unusual cells," said Lai. "They have unusual morphology, and they live for a long time. It may be that it will turn out to be the case that microRNAs have very special functions in the nervous system."

Moderator: Gregory J. Hannon, Cold Spring Harbor Laboratories and HHMI

Panelists:
Victor Ambros, University of Massachusetts Medical School
Gary Ruvkun, Massachusetts General Hospital and Harvard Medical School
Oliver Hobert, Columbia University and HHMI
Allison Abbott, Marquette University
Eric C. Lai, Memorial Sloan-Kettering Cancer Center

Highlights

• Although microRNAs are expressed very specifically in particular tissues, treatments such as locked nucleic acids (LNAs) may be effective in inhibiting their activity.
• Study of microRNA pathways may reveal useful drug targets even if the drugs do not target the microRNAs themselves, and some current drugs may already do this.
• The poor understanding of which genes microRNAs target could be a barrier to their therapeutic use.
• Because nucleic acids are detectable in tiny quantities, microRNA may serve as sensitive diagnostics for cancer progression or other diseases.

MicroRNA therapeutics and diagnostics

The panel discussion at the Janssen Award Symposium addressed the potential of microRNAs for drug development. Although the five panelists do research in model organisms (four in worms and one in flies), the moderator, Gregory J. Hannon of Cold Spring Harbor Laboratories and the Howard Hughes Medical Institute, challenged them to speculate about prospects for translational applications of microRNAs in people.

Gary Ruvkun admitted that when he first heard of proposals to commercialize microRNAs, he thought the idea was misguided, since their expression is specific to particular cells. But subsequent work that demonstrated weeks-long knockdown of microRNAs using circulating locked RNAs (LRNAs) have caused him to reconsider. In addition to directly targeting natural microRNAs, understanding their regulatory pathways may help to identify new drug targets. "If microRNAs are a major regulatory axis of animals," he said, "then the natural world will have evolved drugs to target that." In fact, he speculated that some existing drugs, such as statins that target HMG-CoA reductase, may already be targeting microRNA pathways. "The mechanism of action of a wonder drug might be different than what people think it is."

Gary Ruvkun: "The mechanism of action of current wonder drugs may already involve microRNAs."

The frequent absence of any obvious phenotype for microRNA deletions could be a good thing for treatment, said Eric Lai. "It means that you're not going to die" if their expression is modified. But the subtle effects in particular tissues may be hard to find, especially by studying model organisms. "MicroRNAs may be a class of regulator which is more frequently mutated than we realize, and could lead to diseases," he said, in which case resupplying the microRNAs might be a useful treatment. "I'd be surprised if these things are not translatable on some level."

Off-target effects could be a big problem for therapies that manipulate microRNAs, said Allison Abbott. The disappointingly poor ability to predict target sequences makes it hard to know what unintended genes will be regulated, for example in an LNA knockdown. "We don't understand the rules of target repression and target recognition, so, moving forward, that seems to me to be the biggest challenge."

Oliver Hobert noted that the tissue-specific and cell specific expression of microRNAs could be an advantage for their application as diagnostics. "They could serve as wonderful markers for cancer." Victor Ambros agreed, adding that, because they are nucleic acids, microRNAs can be detected in very small samples such as biopsy samples.

Ambros also observed that, in contrast to the early prejudice that microRNAs are a peculiarity of C. elegans, many homologs have been found, especially for the let-7 family, in other organisms including humans. He noted that LNA knockdown, for example the knockdown of miR-122 for treating hepatitis C, could see therapeutic application fairly quickly. But all the panelists agreed that realizing the full translational potential of microRNAs will have to wait for a much better understanding of their roles. "We have to learn much more about the biology of the microRNAs, in the context of mammalian systems in particular, so we can start to think about how these microRNAs may be able to tell us about targetable pathways, said Ambros.