BIOLOGICAL OVERVIEW

The central dogma of molecular biology maintains that there is a one-to-one correspondence between the genetic code held in the DNA and the encoded sequence found in protein. RNA is the mediator of this correspondence. The phenomenon of RNA editing was first reported in trypanosomes 14 years ago with evidence of the insertion of four ribonucleotides into a frameshifted mitochondrial transcript (Benne, 1986). The insertion restored an open reading frame that was not encoded by the trypanosome mitochondrial genome. Since then, several distinct classes of RNA editing have been found in both plants and animals (reviewed by Simpson, 1999). ADARs (adenosine deaminase acting on RNA) are capable of the site-specific conversion of A-to-I in precursor messenger RNAs (pre-mRNAs). The examples of site-specific editing are limited to transcripts found in the nervous system of vertebrate and invertebrate animals. Most of these transcripts encode ligand- or voltage-gated ion channels and G protein-coupled receptors (reviewed in O'Connell, 1997; Rueter, 1998; Keller, 1999). Because inosine has basepairing properties like that of guanosine (G), the translation machinery interprets I as G. Thus, A-to-I conversion in mRNA has the potential to recode genomic information and alter protein function (Palladino, 2000b and references therein).

Despite advances in understanding of the mechanisms that underlie RNA editing, a very basic question remains; what is the biological function of this process? To address this question, null mutations have been generated in a gene responsible for an entire class of RNA editing in Drosophila, which possesses a single Adar gene. Correspondingly, all known Drosophila site-specific RNA editing (25 sites in three ion channel transcripts) is abolished. Adult flies lacking Adar are morphologically wild-type but exhibit extreme behavioral deficits including temperature-sensitive paralysis, locomotor uncoordination, and tremors that increase in severity with age. Neurodegeneration accompanies the increase in phenotypic severity. Surprisingly, Adar mutants are not short-lived. Thus, A-to-I editing of pre-mRNAs in Drosophila acts predominantly through nervous system targets to affect adult nervous system function, integrity, and behavior (Palladino, 2000b).

ADAR enzymes convert adenosine (A) to inosine (I) in double-strand (ds)RNA and were first discovered in Xenopus oocytes (Bass, 1988). ADARs have been purified and cloned from many diverse sources and act via hydrolytic deamination of A-to-I. In mammals, there are three known ADAR enzymes; ADAR1, ADAR2, and RED2. They contain two or three dsRNA binding domains and a catalytic deaminase domain. The deaminase domain is distantly related to cytidine deaminases, and in particular to another editing enzyme, APOBEC-1, that catalyzes the conversion of C to U in transcripts of mammalian apolipoprotein B. The deaminase domain is also related to the ADATs (adenosine deaminases acting on tRNA) that convert A-to-I in tRNA (Palladino, 2000b and references therein).

ADARs are defined by their ability to convert A-to-I nonselectively in extended perfect dsRNA duplexes. Such activities have been detected in extracts from lower invertebrates through vertebrates, and hence, the enzymes for converting A-to-I in mRNA are of ancient origin. The exact biological role for this promiscuous activity is not known but the activity has been proposed to play roles in viral defense, viral life cycles, and gene regulation. Recently, edited mRNAs were identified in C. elegans on the basis that they contain inosine (Morse, 1999). These mRNAs have been shown to possess extensive modification of 3' untranslated regions within extended hairpin structures, and candidate ADARs have been identified in C. elegans (Hough, 1999). Likewise, a Drosophila Adar has been identified and shown to possess promiscuous A-to-I activity (Palladino, 2000a).

The best characterized examples of A-to-I RNA editing are found in the glutamate receptor subunit genes (GluRs) of the mammalian nervous system (reviewed by Seeburg, 1998). One site in GluR-B transcripts, the glutamine (Q)/arginine (R) site, undergoes editing that changes a genomically encoded Q codon (CAG) to an R codon (CGG) (Sommer, 1991). The functional consequences of editing at this site are striking: Ca²⁺ permeability of GluR channels containing GluR-B subunits is determined largely by the editing status of this site (Verdoorn, 1991; Kohler, 1993). Mechanistically, RNA editing of the Q/R site is dependent upon cis-acting sequences in the intron downstream of and complementary to the editing site (Higuchi, 1993). This editing site complementary sequence (ECS) has been shown to base-pair with the region surrounding the Q/R site and form a double-stranded (ds) RNA secondary structure. Transgenic mice missing the ECS fail to edit the Q/R site. Mice expressing such GluR-B Q/R editing-incompetent alleles in various heterozygous combinations exhibit seizures and early death (Brusa, 1995; Feldmeyer, 1999). Thus, site-specific A-to-I RNA editing profoundly affects protein function in vivo (Palladino, 2000b and references therein).

Further examples of specific A-to-I editing in mammals include other GluR editing sites, several sites in the serotonin receptor (5HT_2CR) (Burns, 1997; Niswender, 1999), and the ADAR2 gene (Rueter, 1999). Although the functional significance has been investigated for some of these sites, their in vivo significance remains largely unknown. Interestingly, while the best studied example of mammalian A-to-I editing, the GluR-B Q/R site, is edited at a very high frequency in vivo (>99% of transcripts are edited), all of the remaining examples of editing occur at lower frequencies. Such intermediate levels of editing suggest a mixture of edited and unedited proteins in neurons or tissue specificity of editing. This is particularly intriguing considering a recent proposal (Seeburg, 2000) that RNA editing in the nervous system evolved to fine-tune the function of the nervous system through subtle functional effects on ion channels (Palladino, 2000b and references therein).

In support of such a proposal, A-to-I RNA editing as a posttranscriptional regulatory mechanism for generating protein diversity in the nervous system appears to have ancient origins, as evidenced by several invertebrate examples. For instance, the squid voltage-gated potassium channel sqKv2 has been shown to undergo RNA editing at numerous positions (Patton, 1997). Transcripts of the major action-potential Na⁺ channel in Drosophila, the product of the paralytic (para) locus, have been shown to undergo RNA editing proceeding through a mechanism similar to that of the mammalian GluR Q/R sites (Hanrahan, 2000; Reenan, 2000). Other reported substrates for editing in Drosophila also encode nervous system signaling components, namely, the cacophony (cac) voltage-gated Ca²⁺ channel (Peixoto, 1997; Smith, 1998b) and the glutamate-gated Cl^- channel, DrosGluCl-alpha (Semenov, 1999) (Palladino, 2000b and references therein).

Mutant Adar^1F1 larvae appear normal for locomotion, response to stimuli, and are also normal in a specific behavioral assay. The phototactic behavior of Adar^1F1 versus WT larvae was examined and no measurable difference was observed. Adar mutant adults, however, exhibit severe neurobehavioral phenotypes. Motor deficits, manifesting as slow uncoordinated locomotion, occasional tremors and varying degrees of abnormal body posture are observed immediately upon eclosion. Also apparent soon after eclosion, Adar^- animals spend an inordinate amount of time grooming (5 ± 0.5% for WT, 27 ± 6% for Adar^1F1). This obsessive cleaning is apparent throughout the lifetime of mutant animals. Adar mutants are capable of flying and jumping but do so only when repeatedly provoked and then only rarely. Not unexpectedly, flight in Adar mutants is erratic. Adar^- males exhibit an extreme mating defect. When presented with WT or Adar^- virgin females, males were not observed to initiate any displays of courtship (Palladino, 2000b).

Since the targets of Adar activity are ion channel transcripts of the nervous system, two paradigms that have been found to be extremely successful in detecting ion channel mutations in Adar mutants are temperature-sensitive (ts) paralysis and ether-induced leg shaking. While Adar mutants recover more slowly from ether anesthesia, they do not exhibit the leg-shaking characteristic of mutations that increase membrane excitability. However, Adar mutant flies exhibit a strong temperature-dependent enhancement of behavioral defects at the restrictive temperature (37.5°C) resulting in bouts of paralysis and extreme motor uncoordination. Mutations in a number of genes affecting ion channels or the process of neurotransmission confer similar phenotypes. The lack of a significant developmental phenotype, outwardly normal adult morphology, varied behavioral deficits, and ts phenotype are all consistent with the primary role of Adar being the editing of transcripts encoding signaling components of the adult nervous system (Palladino, 2000b).

Although Adar mutants are severely compromised for many aspects of adult behavior, they perform the behaviors and functions necessary to sustain life (eating, respiration, metabolism). It was of interest to determine whether Adar mutants are short-lived as a result of their neurobehavioral enfeeblement. Life span analysis was performed on Adar^1F1 male populations under optimal environmental conditions (low population density). As controls, the life spans of populations of Adar^1F1 males bearing Dup(X;Y)901 and FM7G males were determined. The life span of Adar^1F1 males under these ideal conditions parallels that of control animals. Importantly, the maximal life span of Adar^1F1 is not different from controls. The oldest flies obtained from Adar^1F1 and the corresponding Dup(X;Y)901 rescued male controls were 112 and 115 days, respectively. Similar results were obtained for Adar^1F4 and controls. Therefore, most Adar^- individuals, though extraordinarily compromised neurologically, can survive as long as WT animals (Palladino, 2000b).

A stark contrast to the apparent vitality of Adar mutants is their genesis in a screen for lethal mutations. Given the normalcy of development and lack of inviability during larval development and morphogenesis, one question was whether Adar mutants would be compromised under conditions of competition with WT animals. Simply, populations were composed of flies of a given mixture of genotypes. Females in the populations were phenotypically WT and heterozygous for Adar^1F1 and FM7G. Males were of two genotypes; half were Adar^1F1 and the remaining half were Adar^1F1 bearing Dup(X;Y)901. Survivorship for each of the genotypes in these populations was determined for one generation (14 days). That is, no progeny from the populations in question were allowed to eclose and thus alter the size of the population. Even under relatively low population density, Adar mutants exhibit a high mortality rate with respect to WT animals. Thus, Adar mutants are at an extreme selective disadvantage even early in their life span, and this disadvantage is rescued by a translocation that provides Adar⁺ function and behavioral rescue (Palladino, 2000b).

The behavioral phenotypes of Adar^- animals become more severe with age and some new phenotypes appear. Tremors increase dramatically such that locomotion is severely compromised in animals beyond day 50. Animals fall over and become increasingly inefficient at righting themselves; sometimes leading to periods of inactivity resting on their backs. Many animals beyond day 30 exhibit circling behavior that varies from wide circling to circling while standing in place. A majority of animals beyond day 50 exhibit a persistent upheld wing phenotype. Marked asymmetries appear in animals, manifesting as one upheld wing or leg, extension of one or both back legs, and more severe asymmetries in posture. The progressive nature of these defects prompted a determination of the gross nervous system morphology of Adar mutants. Frontal sections through adult heads of male Adar^1F1 and Adar^1F4 were stained with hematoxylin and eosin. One gross difference between Adar mutants and controls is the organization of the retina. The photoreceptors in Adar^- animals appear disorganized and extend longitudinally projecting further to reach the laminar layer in mutants versus the more compact, organized retinal structure seen in WT controls. This retinal abnormality is seen in all sections through mutant heads of all ages tested (Palladino, 2000b).

Other than structural abnormalities in the retina, young Adar mutant brains (1–3 day) appear grossly normal. However, by day 30, lesions appear in the brains of Adar mutants and are distributed randomly in the central brain, optic lobes, and retina. The lesions appear as vacuolated regions, are distributed throughout the sectioned brains, and are most prevalent in the retina, lamina, and optic lobes. Vacuoles appear to increase in size and number with age and, by day 50, animals can be found with extensive brain degeneration. Control animals do not demonstrate brain degeneration and seldom have even one small vacuole per head (Palladino, 2000b).

What are the global roles of ADARs in gene regulation and organismal function? A hint at the answer may be the prevalence of nervous system targets of ADARs in both invertebrates and mammals. From data presented here, it is concluded that the process of A-to-I RNA editing in Drosophila is, primarily, a process subserving the execution of adult behaviors through effects on numerous and varied nervous system targets of ADAR activity. Moreover, it is proposed that the process of A-to-I editing is, by its very nature, ideally suited for modifying activity in the nervous system (Palladino, 2000b).

The developmental time course of Adar^- animals is essentially the same as that of WT controls. Thus Adar activity does not appear to be necessary maternally or zygotically for normal morphology of adult animals. Additionally, lack of Adar does not shorten the life span of mutants under optimal conditions. The mutant phenotypes associated with loss of Adar activity are entirely consistent with perturbations of electrical and chemical signaling mechanisms in the nervous system and proximal effects on behavior. In fact, mutations have been generated in some of the ion channel targets of Adar (described in this paper) which confer behavioral phenotypes. For instance, cac mutants exhibit specific defects in male courtship song, visual defects, and temperature-dependent convulsions and uncoordination. Mutations of para also confer a wide range of behavioral phenotypes including recessive and dominant ts-paralysis, defects in learning and olfaction, cold-sensitive lethality, genetic suppression of certain mutations in the Shaker K⁺ channel gene, and resistance to insecticides. Glutamate-gated Cl^- channels, like DrosGluCl-alpha, have only been reported in invertebrates. They are a target of the avermectin class of insecticides and are found in several neuronal and muscle preparations. Though no mutations in DrosGluCl-alpha have been reported, the presence of glutamate-gated Cl^- channels on the soma of neurons and their similarity to the mammalian inhibitory glycine receptor, mutations that cause inherited startle disorders to unexpected stimuli, are also consistent with the idea that RNA editing of DrosGluCl-alpha modulates neuronal membrane excitability (Palladino, 2000b).

The extensive and varied behavioral defects in Adar mutants are consistent with affects on target genes such as para, cac, and DrosGluCl-alpha; ion channels of the nervous system. Like all other examples of site-specific A-to-I editing though, these 25 sites in three Drosophila genes were discovered serendipitously and other unidentified targets of Adar activity surely exist. Thus, it appears unlikely that the nervous system impairment seen in Adar mutants is due solely to any one particular Adar target site. A more likely scenario, given the potential for hundreds of Adar editing sites in scores of nervous system genes, is that the Adar^- phenotype is complex and results from effects on the posttranscriptional processing of many transcripts in the nervous system (Palladino, 2000b).

Since most A-to-I editing sites have been discovered in signaling components of the nervous system, one role of editing that has been proposed is in the fine-tuning of neurophysiological processes (Seeburg, 2000). Site-specific mRNA editing can have profound biological significance for individuals. The only examples, and these are extreme, are the in vivo consequences of failure to edit the GluR-B Q/R site in mice. Editing at this site occurs at high frequency (>99%) and mutant mice expressing an editing-incompetent allele of GluR-B die postnatally from neurological dysfunction (Brusa, 1995). However, most editing sites are not edited at such high levels and must confer altered signaling properties in a cell-specific manner or alter neuronal signaling properties through dominant effects on protein function (Palladino, 2000b and references therein).

The singular role of Adar in Drosophila pre-mRNA editing along with the adult neurological phenotypes of Adar null mutants suggest a novel evolutionary role for Adar in supplying genetic diversity. Genetic variation in large, diploid, randomly mating populations appears through mutation and is discrete in nature. That is, an individual animal can have zero, one or two alleles for a given nucleotide position and stochastic or selective processes act to determine the final frequency of that allele in a population over time. Cells can usually express, at most, two alleles differing at a single amino acid in a 1:1 ratio (Palladino, 2000b).

An important example of selection acting on variation in nature exists for the para gene. In natural insect pest populations, such as Heliothis (tobacco horn worm) or Blatella (cockroach), resistance to pyrethroid insecticides is prevalent and confers a strong selective advantage on animals carrying knock-down resistant (Kdr) alleles in para orthologs. In the laboratory as well, certain Drosophila para mutations that are recessive ts-paralytic mutations confer dominant resistance to pyrethroid insecticides and DDT in heterozygotes. In fact, certain doubly heterozygous mutant combinations confer even higher levels of resistance. Thus, the advantage of modifying certain amino acid positions in ion channels is not without precedence and demonstrates the selective value of possessing two different alleles of a particular locus (Palladino, 2000b and references therein).

A-to-I RNA editing via ADARs is different from the usual mutational gene-altering mutational paradigm. Changes introduced by editing are not discrete, such as naturally occurring genetic variation, but appear as a continuum. Within the 25 examples given in this report, the level of editing at particular sites ranges from ~5% to 90%. Though little is known about what determines the efficiency of deamination of pre-mRNAs in vivo, the substrate requirements of ADARs must play a large role. Specific pre-mRNA editing via ADARs has been shown, in mammals, to proceed through dsRNA intermediates in which intronic editing-site complementary sequences (ECSs) base-pair with regions around the edited adenosine (Higuchi, 1993; Herb, 1996). This mechanism has been conserved in Drosophila para editing (Reenan, 2000). Thus, the evolution of such RNA editing sites likely proceeds through the mutation of noncoding sequences. Many mutational changes to the genome may be required in intronic sequences to generate an initial, weakly complementary ECS. This nascent ECS would then direct modification of a particular nucleotide position at a low efficiency in nearby coding sequences resulting in an amino acid change in a small percent of messages. Certain de novo editing sites, though edited at low levels, would be selectively advantageous through dominant effects on protein function and resultant changes in behavior. Such a model provides an ideal substrate for evolution, the intron, for which mutation can provide a large sequence space of variants most of which would be selectively neutral. Larger introns would perforce generate more sequence variants through time. Editing introduced at a low level at a particular site within a protein would introduce subtle variants within a population that could benefit from a selective advantage in a niche-specific manner altering behavior to suit environment. Future enhancements of editing, through improvements of the ECS-editing site interaction to generate a better ADAR substrate, could titrate in the edited form of the protein to maximally advantageous levels -- all via changes to noncoding sequences. In effect, RNA editing provides the possibility to make use of amino acid changes that would be selectively disadvantageous at 50% (as a mutation appearing in coding sequence) offering a continuum of expression of two residues at a given amino acid position. Further diversity could be obtained in cases where the process of RNA editing became spatially regulated (Palladino, 2000b).

A-to-I RNA editing, while providing the potential for tailored mixtures of proteins differing at a single amino acid position, seems to have supplied this diversity primarily to the nervous system -- a tissue whose hierarchical function, complex structure and direct influence on organismal behavior ideally suit it to take advantage of such continuous and subtle genetic variation. The future challenge of studying Adar mutants will be in addressing the potential multitude of unknown pre-mRNA targets, the role of different Adar isoforms in the editing of targets, the association of specific editing sites with specific modifications of behavior, and the natural history of ADARs and their substrates in the course of evolution (Palladino, 2000b).

GENE STRUCTURE

Transcripts of the Drosophila Adar locus originate from two regulated promoters. In addition, alternative splicing generates at least four major Adar isoforms that differ at their amino-termini as well as altering the spacing between their dsRNA binding motifs (Palladino, 2000a).

cDNA clone length - 10164

Bases in 5' UTR - 991

Exons - 10

PROTEIN STRUCTURE

Amino Acids - 632 and 669

Structural Domains

A homolog of the ADAR (adenosine deaminases that act on RNA) class of RNA editases from Drosophila, Adar, has been identified. Drosophila Adar is most homologous to the mammalian RNA editing enzyme ADAR2, the enzyme that specifically edits the Q/R site in the pre-mRNA encoding the glutamate receptor subunit GluR-B. Analysis of the completed Drosophila genome sequence supports the data since Adar and dADAT1 (a tRNA adenosine deaminase: Keegan, 2000) are the only predicted genes in the Drosophila genomic sequence that have ADAR-type deaminase motifs (Palladino, 2000a).

date revised: 3 September 2000

Home page: The Interactive Fly © 1995, 1996 Thomas B. Brody, Ph.D.

The Interactive Fly resides on the
Society for Developmental Biology's Web server.