A genomic survey uses innovative genetics to make neurons susceptible to RNA-mediated gene inactivation. The results implicate many genes in communication at the synapse between neurons and muscle.
RNA interference (RNAi) is a spectacularly useful technique for selectively reducing gene activity — and thereby garnering clues about gene function1. In this issue, a collaborative group of geneticists, genomicists and neuroscientists report the first large-scale RNAi screen in neuroscience (Sieburth et al., page 510)2. They identify more than 100 novel genes that have specific functions in the transmission of signals across the junction between neurons and muscle cells.
The nematode worm Caenorhabditis elegans was the subject of the first genome-wide RNAi screens, aimed at finding the genes involved in development3,4. But neurons in C. elegans have proved inexplicably resistant to the technique. The Ruvkun lab5,6 has taken the geneticist's gambit to solve this problem: if you want something but have no idea how to get it, just find the right mutant. Their clever two-step screen to look for such mutants yielded a strain, designated eri-1;lin-15B, in which neuronal RNAi is possible.
The first gene of this combination, eri-1, encodes a ribonuclease enzyme that may suppress RNAi by degrading small interfering RNAs (siRNAs)5 — the intermediates in the RNAi process that mark out specific gene products for destruction. The second gene, lin-15B, is the C. elegans equivalent of the mammalian retinoblastoma tumour-suppressor gene (Rb), involved in gene regulation. In this issue, Ruvkun and colleagues (page 593)6 report that in lin-15B mutants, neurons and other mature cells are partly transformed into immature germline cells. Germline cells are responsive to RNAi, and this transformation also sensitizes neurons to RNAi. The eri-1;lin-15B strain provided the key element that allowed Sieburth et al.2 to find neural genes using RNAi.
Synaptic transmission is the crucial process by which neurons communicate with their neuronal or non-neuronal targets (Fig. 1). Acetylcholine is the neurotransmitter at the neuromuscular synapse of C. elegans, as it is in vertebrates. The neuron releases acetylcholine from bubble-like vesicles that burst into the synaptic space. The neurotransmitter binds to receptors on the target cell to pass the signal across the synapse, and any excess neurotransmitter is broken down by enzymes called acetylcholinesterases, so that the synapse is primed for the next signal.
In an early application of chemical genetics, Nguyen et al.7 found that drugs that inhibited acetylcholinesterases kill normal worms, but are less poisonous to worms with reduced neuromuscular transmission. The researchers screened a series of mutant worms for resistance to the inhibitors, and found many proteins that are involved in the cycle of synaptic-vesicle formation and breakdown (for example, UNC-13 and UNC-18). But much clearly remained to be discovered. In their work, Sieburth et al.2 applied RNAi together with this highly selective drug screen in the eri-1;lin-15B strain and in a second eri-1;lin-15B strain sensitized by another mutation in neurotransmission. Even with these sensitized and specific assays, the experiments were difficult, so they screened only 2,072 of the roughly 19,500 C. elegans genes, cherry-picking those with relatives in other species or possessing particular structural domains that hinted at neural functions. This sub-screen yielded 185 genes, 132 of which had not previously been implicated in synaptic transmission in any animal.
The next hurdle was to characterize this huge set of 185 genes further. Following an approach first used on genes involved in fat metabolism8, the authors analysed 60 RNAi candidates in the presence of various drugs and in different mutant-gene combinations. Genes with related functions should behave similarly in these assays. By clustering the patterns of defects seen in new RNAi genes with the defects caused by known genes, the authors assigned some of the novel genes to various steps of synaptic neurotransmission (Fig. 1).
But they didn't stop there. They examined the subcellular distribution of 100 of the proteins encoded by the genes by fusing them with green fluorescent protein (GFP). This work was facilitated by the ‘ORFeome’ project, led by one of the authors, Marc Vidal, in which full-length clones of all C. elegans genes are being isolated. Of the 100 proteins, 26 were located in presynaptic regions, some in association with synaptic vesicles, others associated with the plasma membrane at or near the sites of synaptic release. Ten proteins seemed to be in postsynaptic regions. Given that so many of the proteins reside in or near the synapse, and that many had associations with known synaptic components such as the kinesin UNC-104 (KIF-1A), this confirmed that the screens yielded exactly the desired molecules.
For many of the candidate genes from the screens, mutants were available in other gene collections created by classical genetic methods or from gene-knockout screens. When these traditional mutants were examined, 28 out of 34 mutants had characteristics suggesting functions at the synapse. Finally, the authors examined synapses in live animals using tagged synaptic-vesicle proteins to subdivide the mutants into even more specialized functional groups.
Most of the genes identified have close relatives in other animals, and detailed mechanistic studies should elucidate their exact molecular functions. For the present, the framework provided by Sieburth et al. is an impressive beginning. The work raises the standard for future experiments in the field: it's not just a gene list from a screen, but a thorough and thoughtful combination of genetics, molecular biology and data analysis.
RNAi is subject to false-negative results from inefficiencies in the system, and to false-positive results from effects on unintended targets, but Sieburth et al. were successful largely because they used highly selective screens in sensitized genetic backgrounds. Moreover, they could validate their methods using existing mutations (19 of 21 known cholinesterase-resistance genes were confirmed by at least one of their screens), and were able to verify their results by reference to classical genetic mutations — tools that are not available in every system.
The considerable efforts under way to define the enzymes involved in RNAi should make it an even more powerful and reliable tool. It has the potential to bring genetics to an exciting array of unsolved questions in neuroscience. Armed with the data from genome projects, we can look forward to bold and imaginative RNAi screens in previously intractable, but highly desirable, experimental systems — perhaps songbirds, honeybees, cichlid fish or even the primate brain.
Fire, A. Nature 391, 806–811 (1998).
Sieburth, D. et al. Nature 436, 510–517 (2005).
Fraser, A. G. et al. Nature 408, 325–330 (2000).
Gönczy, P. et al. Nature 408, 331–336 (2000).
Kennedy, S., Wang, D. & Ruvkun, G. Nature 427, 645–649 (2004).
Wang, D. et al. Nature 436, 593–597 (2005).
Nguyen, M. et al. Genetics 140, 527–535 (1995).
Ashrafi, K. et al. Nature 421, 268–272 (2003).
About this article
Progress in Neurobiology (2007)