Adapted from an original article by Jennifer Altman in the September-October issue of Alzheimer Actualités, a newsletter published in French by the Ipsen Foundation. The Alzforum editors acknowledge the Foundation’s generosity in making this summary freely available in English.
Until recently, the control of protein synthesis seemed straightforward and of little concern to most neuroscientists. However, the simple story of gene transcription into messenger RNA (mRNA) and subsequent translation into a protein has recently become considerably more complicated. Small RNA molecules have been discovered that can determine when and if the mRNA for a particular protein will be translated. This higher level of control is proving to be crucial in many aspects of cell biology, and small RNAs are now also being implicated in many diseases, including cancer and heart disease. For neuroscientists, they not only provide sophisticated tools for dissecting intracellular signaling and control pathways, but microRNAs, a particular type of small RNA, are also proving to be of considerable functional and pathological importance: in neuron differentiation and synaptic plasticity, neurodegenerative diseases, and mental illness (Kosik, 2006; Fiore et al., 2008; Hébert and De Strooper, 2009). Twelve leading researchers discussed the actions and implications of microRNAs in the nervous system at the Fondation IPSEN’s 17th annual Neuroscience Colloquium held in Paris on April 20. The meeting was organized by Bart de Strooper of the Flanders Interuniversity Institute for Biotechnology, Leuven, Belgium, and Yves Christen, Fondation IPSEN, Paris, France.
Every cell in an organism contains the same complement of genes, derived from the DNA of the fertilized oocyte. What distinguishes skin cells from muscle or neuron and so on is the set of genes in each type of cell that is transcribed into mRNA to produce that cell’s specific complement of proteins. This protein complement also changes through the life of a cell as the cell gains its identity, matures, and dies; in the mature cell, the output of certain gene—mRNA—protein pathways may alter from time to time depending on demand and the challenges they face. A highly stimulated neuron, for example, will need to increase production of all the proteins involved in synaptic transmission. This active set of genes has come to be called the transcriptome, and the proteins produced from the active genes are now termed the proteome, explained Kenneth Kosik of the University of California, Santa Barbara. These new words evoke a sense of the integrated, networked regulation of protein production in a cell.
The picture of gene transcription has become increasingly complex, as more levels of control have been discovered linking the cell’s outer and inner environments to the expression of genes in its nucleus. According to the standard story of protein synthesis, mRNA reflects a given gene’s DNA template. Transferring to the cytoplasm, the mRNA attaches to ribosomes and is translated into the sequence of amino acids that make up the protein. Decisions about whether a protein is made or not, and in what quantity, are regulated at the transcription stage to a large extent by proteins known as transcription factors, which operate on parts of each gene that are not transcribed into mRNA: the promoter and suppressor regions. The discovery of small RNA molecules in the past 15 years has introduced the concept that regulation can also take place at the second stage of protein synthesis—the translation of mRNA into protein (Bartel, 2004).
The small RNA molecules were once called “junk,” as they were thought to have no significant role, but they are now considered to control protein synthesis at an executive level. Acting late in the process of protein production, they both switch off protein synthesis and fine-tune the level of protein production, said Nikolaus Rajewsky, Max-Delbrück-Centrum, Berlin, Germany, as well as directly repress the translation of hundreds of genes (Selbach et al., 2008). As the functions of the small RNAs are probed in greater detail, it is becoming apparent that they are crucial to many decisions in the life of cells, and so they are attracting attention as possible targets for a new type of therapeutic intervention.
In contrast to the several hundred nucleotide units in a typical mRNA, small RNAs contain about 20 units. Two types have been identified: microRNA and small interfering RNA (siRNA). Both prevent protein translation by binding to mRNAs, but their modes of action differ somewhat. Synthetic small RNAs, but particularly siRNAs, have become powerful tools for dissecting intracellular signaling pathways by experimentally preventing the synthesis of a selected protein. So far, only a few hundred naturally occurring microRNAs have been identified, compared to about 25,000 proteins, and most are known prosaically by “miR” and a number. A few exceptions are named according to the conventions used in the nematode worm and fruit fly where they were discovered.
Because microRNAs are so small, they are difficult to detect accurately. Rajewsky’s group has described the development of highly sophisticated techniques for detection, sequencing, and computational analysis that are underpinning the rapid expansion of this field (Friedlander et al., 2008; Selbach et al., 2008). Once microRNAs are identified and located, the more conventional techniques of molecular biology, genetics, and cell biology are being employed to determine their functions and dysfunctions.
Each microRNA can interact with many mRNAs, so overall, microRNAs form a powerful network regulating protein production. Many are restricted to particular cell types, where they appear to inhibit completely or partially the level of mRNA (Sood et al., 2006). Through a small change in the sequence of its nucleotides, a microRNA may switch its affinity to a completely different set of mRNAs. One particularly potent target, which Kosik pointed out, is the mRNAs coding for the transcription factors that determine when a gene is transcribed, an interaction that creates feedback loops regulating the rate at which specific proteins are synthesized (Mercer et al., 2008).
Like proteins, microRNAs are produced from genes, which are transcribed into primary RNA (Pri-miRNA) chains of about 1,000 nucleotides. Still in the nucleus, these are shortened into microRNA precursor (Pre-miRNA) molecules of about 70 nucleotides by an enzyme, Drosha, in a complex with a protein termed DGCRB (aka Pasha). Another protein, Exportin 5, carries the Pre-miRNAs into the cytoplasm, where they are further cut by the enzyme Dicer, producing the mature ~19-23 nucleotide microRNAs. These then associate with a group of protein molecules to form the RNA-induced silencing complex that binds to the mRNA (Hébert and De Strooper, 2009).
The microRNA genes and the genes coding for the various production enzymes and helper proteins are, of course, all susceptible to mutation, with disruptive consequences for the regulation of protein production. Mutations disabling Dicer, the second cutting enzyme, are particularly serious, as they stop production of all mature microRNAs. Mutations in the genes coding for microRNAs and the microRNA processing proteins are increasingly being found to contribute to a range of diseases. One particular interest is that they may possibly be involved in so-called sporadic forms of diseases, in particular in neurodegenerative conditions such as Alzheimer’s and Parkinson’s, and mental illness. Most patients with these diagnoses have no family history of the disease, or any known mutations in identified disease genes, such as presenilin in Alzheimer disease or α-synuclein in Parkinson’s.
MicroRNAs in Development
A key question in embryonic development is when and how a cell becomes committed to become a particular cell type. In stem-cell biology, the microRNA network is now known to be involved in determining whether a newly divided cell remains an undifferentiated stem cell or commits to become a particular cell type (see an overview). The nematode worm, Caenorhabditis elegans, offers a unique opportunity for investigating the dynamics of all the known microRNAs and their effects on protein synthesis in a whole animal as it develops from egg to larva, a project now underway in Rajewsky’s laboratory. This worm is beloved by embryologists because every cell in the adult body can be individually identified and its origins traced back to early development. The distribution of conserved microRNA targets has also been mapped throughout the worm’s genome (Lall et al., 2006). This type of investigation will take the regulation of embryonic development to a sub-microscopic level.
Another model animal, the zebrafish Danio rerio, is small enough to map the occurrence of microRNAs in the whole nervous system. Marika Kapsimali, INSERM U784, Ecole Normale Supérieure, Paris, France, showed the first anatomical maps of microRNA distribution at various embryonic stages and in the adult (Kapsimali et al., 2007). Some microRNAs are expressed in particular developmental stages, e.g., miR-92b is found in stem cells and neuronal precursors, while miR-124 is present in cells that are differentiating into neurons. Others are restricted to particular brain areas, such as miR-222 in the telencephalon; yet others occur only in a specific cell type, e.g., miR-218a in motor neurons. Knowing the time and place where particular microRNAs are expressed allows predictions to be made about their functions. These maps will also enable researchers to focus on questions such as how microRNAs interact with proteins already known to determine the organization of the developing nervous system.
While most neurons in the adult mammalian brain are not replaced if they die, one known exception is in a part of the hippocampus, an area involved in memory storage, where a nest of neural stem cells gives rise to mature neurons and glial cells. Peng Jin of Emory University School of Medicine, Atlanta, Georgia, demonstrated that, as in embryonic stem cells, at least one microRNA helps to keep the adult neural stem cells in their dividing state, rather than differentiating into mature neurons and glial cells. The molecule here is miR-137. It is regulated by a feedback loop: when active it suppresses the production of a protein known as Ezh2 which, in turn, modifies the gene coding for miR-137. By reducing production of the microRNA, Ezh2 promotes the differentiation of the adult neural stem cells and stops them proliferating. Two other proteins also act directly on this gene to reduce miR-137 production; mutations in the gene coding for one of these, MeCP2, are found in children with Rett syndrome, a neurodevelopmental disorder that causes mental retardation in girls, implicating an overproduction of miR-137 as a factor in the cause of this condition (Smrt et al., 2007; also see ARF related news story).
Some microRNAs support differentiation rather than proliferation. One such is miR-133b, which helps maintain the health of the dopamine-secreting neurons in the midbrain, reported Asa Abeliovich, Columbia University, New York (Kim et al., 2007; see also ARF related news story). These neurons are essential for fine control of voluntary movements and their loss is central to Parkinson disease. The microRNA interacts with specific proteins, including one known as Pitx3 that itself promotes maturation of the dopamine neurons. This is another example of a feedback loop: Pitx3 also increases production of miR-133b, which, as its level rises, reduces production of Pitx3 to maintain the neuron population in a stable state. Identification of such control loops will help in the understanding of why dopamine neurons die in Parkinson disease, as well as improve the manipulation of stem cells in vitro for possible transplantation into Parkinson disease patients.
MicroRNAs and Synaptic Plasticity
As the nervous system develops, young neurons grow processes, or neurites, that in time become richly branched dendrites. These receive information from the axon terminals of other neurons through synaptic contacts, located on small protrusions known as spines. To begin with, a dense network of connections is established, followed by a pruning phase when synapses and branches that are regularly used become stronger and little-used ones disappear. A similar process occurs throughout life as new connections are established and reinforced by repetition or lost through lack of use. Given their roles in development, it is hardly surprising that microRNAs are involved in these growth processes. Mice lacking one of the proteins necessary for processing the microRNA-precursor molecules have stunted dendritic bushes, finds Maria Karayiorgou, Columbia University, New York (Mukai et al., 2008), and several microRNAs have been identified in dendrites and synapses that regulate the protein synthesis required for growth, as both Kosik and Gerhard Schratt, Universität Heidelberg, Germany, reported.
Schratt showed that dendritic growth occurs when the stimulation of synapses activates a transcription factor, MEF2, that in turn switches on the gene coding for miR-134 (Fiore et al., 2009). The mature form of miR-134 is then transported to the stimulated dendrite, where it takes the brakes off new protein synthesis by shutting off a molecule that prevents local mRNA translation. Through similar actions, miR-132 also regulates the growth of dendritic spines, while miR-138 promotes the contraction of under-used spines.
The discovery of locally regulated microRNA action is also helping to answer a long-debated question in synapse biology. Repetitive activation of a synapse requires newly synthesized proteins, many of which are made locally from mRNAs produced in the nucleus. How do these mRNAs get delivered specifically to the synapses that need them, when one dendrite may receive inputs from several sources, not all of which are necessarily active at the same time? The answer, according to Florence Rage of the Institut de Génétique moléculaire de Montpellier, UMR 5535, France, seems to lie, at least in part, in the packets in which both mRNAs and microRNAs are carried from the cell body to the spines. Called dendritic-like P-bodies, these small granules contain microRNA bound to mRNA embedded in a complex of molecules that maintain the mRNA in a suppressed state. Their transport along the dendrites is accelerated when neurotransmitter receptors on the spine are stimulated (Cougot et al., 2008). In the spine, both Kosik and Rage described how a local enzyme, switched on by the synaptic activity, removes the microRNA and packages it for degradation, so that the mRNA can be released from the P-body and translated into protein.
A second long-debated question concerns the communication between the dendritic spine and the presynaptic terminal on the axon. It has been known for many years that information flows both ways across the synapse, with the post-synaptic site reporting back on its state of activation—a process thought to be important in establishing the changes in synaptic function associated with memory formation, as well as maintaining healthy synapses. David Simon, Harvard Medical School, has been looking at the synapses between motor neurons and muscles in the nematode worm, which is providing a useful model for investigating the nature of these retrograde signals. A microRNA commonly found in muscle, miR-1, seems to act on two levels. It controls the sensitivity of the synaptic receptors for acetylcholine, the neurotransmitter in the nerve-muscle junction, and it regulates the presynaptic release of the transmitter by suppressing the post-synaptic release of the transcription factor MEF2. MEF2 is activated directly when one type of acetylcholine receptor is stimulated and in its turn inhibits a presynaptic molecule involved in mobilizing synaptic vesicles for release (Simon et al., 2008; see also ARF related news story). Here again, the microRNA is a component of a complex regulatory loop, in this case dampening down synaptic transmission to prevent runaway excitation.
MicroRNAs and Neuropathology
With microRNAs’ multiple roles in development, maintenance, and function of synaptic connectivity, their involvement in brain pathologies is to be expected. The implication of miR-137 in Rett syndrome, a neurodevelopmental disorder, has been mentioned already (see Jin, above); similar regulatory disruptions are also being found in other inherited developmental conditions, such as Fragile-X syndrome, and may also be implicated in autism. In adult nervous systems, microRNAs are being identified in both neurodegenerative and psychiatric illnesses.
Employing simpler organisms, such as the fruit fly Drosophila melanogaster, to model aspects of human disease is paying off in identifying and analyzing the contributions of microRNAs to neurodegeneration, as Stephen Cohen, of the Temasek Life Sciences Laboratory, Singapore (see also ARF Interview with Stephen Cohen), and Nancy Bonini, Howard Hughes Medical Institute, Philadelphia, Pennsylvania, both demonstrated. So far, Cohen reported, 148 genes coding for microRNAs have been validated in the fly, of which 128 have related equivalents in other species—in other words, they are evolutionarily conserved. One-third of the microRNAs produced from these genes are found in the brain (Ruby et al., 2007), but so far the functions of only a few of them are known.
One advantage of working with flies is that mutations can quickly be generated and identified through their effects on the animal’s behavior and/or appearance. For example, Cohen showed how a mutation in the miR-8 gene produces flies with poorly coordinated movements and an excessive number of dying neurons (Karres et al., 2007). One function of this microRNA is to limit the production of atrophin, a protein that is defective in a human neurodegenerative condition called dentatorubral-pallidoluysian atrophy; this is one of a group of genetic diseases that includes Huntington disease and spinocerebellar atrophy, all of which are characterized by proteins that are toxic because they contain abnormally long strings of the amino acid glutamine. Mapping experiments are now under way to match up many more microRNA mutants with their effects on fly behavior and then to identify the pathway that each of these microRNAs is regulating.
Drosophila is providing Bonini with a useful way of exploring what makes these proteins with abnormal numbers of glutamines toxic, with a focus on spinocerebellar atrophy type 3 (SCA3). When the gene for the abnormal SCA3 protein, also known as ataxin-3, is transplanted into flies, its effects can easily be monitored by changes in the structure of the eyes. The protein accumulates because the extra glutamines make it less soluble, but cells turn out to have a multilayered protection against its toxicity with at least three other factors identified. First, a protein belonging to the heat-shock protein family is engaged; these molecules, known as chaperones, help to remove damaged proteins. The second protection is through microRNAs: increasing the level of a microRNA called bantam, which is not found in mammals, reduces eye degeneration and other microRNAs are implicated as well (Bilen et al., 2006). The third level of protection involves the rather surprising finding that the mRNA coding for the abnormal protein itself seems to have toxic effects—and possibly microRNAs help by inactivating it (Li et al., 2008).
Identifying changes in microRNA levels in the brains of humans with neurodegenerative conditions such as Alzheimer or Parkinson diseases is problematic: the work has to be done postmortem or on very limited biopsy material, and finding adequate control material is difficult. However, there are indications of lower miR-29a/b levels in Alzheimer patients with high levels of the β-secretase enzyme BACE1, which is responsible for producing the pathogenic amyloid-β protein, said Sébastien Hébert, Centre de Recherche du CHUQ (CHUL), Québec, Canada (see ARF related news story). In Parkinson disease, Abeliovich described a decrease in the level of miR-133b, mentioned above as part of the feedback loop maintaining mature dopamine neurons (see ARF related news story).
Molecular analysis is proving more fruitful: in Alzheimer disease, binding sites for microRNAs have been identified on the BACE1 mRNA, as well as on the mRNAs of the amyloid precursor protein (APP) and presenilin-1, two key proteins in the pathway that produces amyloid-β. Work in cell culture indicates that members of the miR-106 family can regulate APP production, and miR-106b, which is relatively abundant in human brain, is decreased in sporadic Alzheimer patients. It seems likely that loss of specific microRNAs and decline in the overall fine-tuning of the microRNA network may be one of several factors that contribute to neurodegeneration.
Another target of microRNA regulation seems to be the tau protein, the abnormally phosphorylated forms of which cause neurofibrillary tangles and neuron death. In flies, Bonini showed, the phosphorylation and toxicity of tau increase when the production of mature microRNAs is prevented by deleting the gene for the processing enzyme Dicer (but see Hébert and De Strooper, 2009, for problems with interpretation of Dicer deletion). Because microRNAs have significant roles in the development and maintenance of synapses, they are probably also implicated in the loss of synapses that is a characteristic of the neurodegenerative diseases.
The subtle and complex ways in which the microRNA networks are likely to participate in brain health and disease were illustrated by Karayiorgou’s discussion of recent studies on schizophrenia. Although schizophrenia is known to run in families, attempts to find a common genetic background have been repeatedly thwarted and 60 percent of sufferers have no affected relatives. However, schizophrenia is one of several conditions associated with a small deletion of genetic material at locus q11 on chromosome 22, and Karayiorgou reported that a gene located in this region codes for the molecule known as DGCR8 that is involved in the processing of the microRNA precursor molecules. Deleting DGCR8 has a similar effect to disabling the Dicer gene: no mature microRNAs are made (Stark et al., 2008; see DGCR8 on SZGene).
Mice bred with a similar deletion on chromosome 16, the equivalent to human chromosome 22, are providing a test system for tracking down how the missing microRNAs act. When the deletion is on only one of the paired chromosomes, leaving the other intact, the mice display several characteristics of schizophrenia, including restlessness, agitation, and compromised memory (Stark et al., 2008). Compared with normal mice, those with the deletion have fewer dendritic spines in the hippocampus and prefrontal cortex (Mukai et al., 2008), areas in humans where connectivity problems seem to contribute to schizophrenic symptoms. A clue to the targets of the missing microRNAs is that production of several proteins involved in synaptic plasticity increases in these mice, indicating that they are no longer tightly regulated; similar spine defects are seen in other mice missing just one or another of these proteins. Curiously, the genes for these proteins are also on the deleted part of the chromosome, so the deletion seems to upset the balance between the microRNAs and the proteins they regulate. In what Karayiorgou described as the “genetically complex architecture” of schizophrenia, changes in microRNA production seem to disturb the orchestration of many small contributions, resulting in a devastating mental illness.
Clearly, only the surface of microRNA biology is being scratched at the moment, and in the coming years we can expect more clarity as the breadth and depth of the microRNA regulatory networks are revealed. A sign of the youth of this field is that many questions in the discussions were answered by, “We have yet to look at that.” A burning topic is, of course, how useful microRNAs will be as therapeutic targets. Although the temptation is to manipulate the microRNAs involved in particular diseases, the general opinion was that much caution is required because microRNAs have such widespread and complex actions. So little is known about how their production is regulated in different tissues, let alone in different types of neurons, that interfering with one microRNA could have widespread unintended consequences. But as more precise information emerges, microRNAs may yet prove potent additions to the therapeutic toolbox.—Jennifer Altman.
Jennifer Altman is a science writer in Todmorden, U.K.