Dynamic response of RNA editing to temperature in Drosophila

Abstract

Background: Adenosine-to-inosine RNA editing is a highly conserved process that post-transcriptionally modifies mRNA, generating proteomic diversity, particularly within the nervous system of metazoans. Transcripts encoding proteins involved in neurotransmission predominate as targets of such modifications. Previous reports suggest that RNA editing is responsive to environmental inputs in the form of temperature alterations. However, the molecular determinants underlying temperature-dependent RNA editing responses are not well understood.

Results: Using the poikilotherm Drosophila, we show that acute temperature alterations within a normal physiological range result in substantial changes in RNA editing levels. Our examination of particular sites reveals diversity in the patterns with which editing responds to temperature, and these patterns are conserved across five species of Drosophilidae representing over 10 million years of divergence. In addition, we show that expression of the editing enzyme, ADAR (adenosine deaminase acting on RNA), is dramatically decreased at elevated temperatures, partially, but not fully, explaining some target responses to temperature. Interestingly, this reduction in editing enzyme levels at elevated temperature is only partially reversed by a return to lower temperatures. Lastly, we show that engineered structural variants of the most temperature-sensitive editing site, in a sodium channel transcript, perturb thermal responsiveness in RNA editing profile for a particular RNA structure.

Conclusions: Our results suggest that the RNA editing process responds to temperature alterations via two distinct molecular mechanisms: through intrinsic thermo-sensitivity of the RNA structures that direct editing, and due to temperature sensitive expression or stability of the RNA editing enzyme. Environmental cues, in this case temperature, rapidly reprogram the Drosophila transcriptome through RNA editing, presumably resulting in altered proteomic ratios of edited and unedited proteins.

Figure 1

Overall editing responses to temperature. (A) The temperature response of fifty-four editing sites in 14 transcripts. Editing (total guanosine trace to (total adenosine + guanosine traces)) at 10°C (blue), 20°C (black) and 30°C (red) is presented for each site. All sites are ranked by editing level at 20°C. Bars represent standard error. Notable sites are labeled by gene abbreviation and adenosine site as annotated in Savva et al. [9]. (B) Editing of synaptotagmin-1 sites 2 to 4 is largely insensitive to temperature. (C) Editing of sites 1 to 3 in the paralytic transcript decreases at 30°C. (D) Editing of shab site 6 increases with increasing temperature, while site 7, edited to 100%, is temperature-insensitive. All editing sites are indicated within chromatograms by black carrots above Sanger traces.

Figure 2

Conservation of editing responses across Drosophilidae . (A) Editing at sites 2 to 4 in the synaptotagmin-1 transcript. (B) Editing at sites 1 to 3 in the paralytic transcript. (C) Editing at sites 6 and 7 in the shab transcript. The slopes of the species-specific editing response curves, rather than the absolute editing, were statistically compared to those of D. melanogaster (black) for each site. P <0.0001: **, P <0.05: *. Bars in A through C represent standard error. (D) Phylogeny of the five Drosophilidae species used in this study. D. ananassae represents an outgroup species. Colors are as in A through C. Branch lengths represent evolutionary time, based on Tamura et al. [30].

Figure 3

ADAR protein level is sensitive to temperature. (A) Western blot analysis of the HA-tagged dADAR protein, from the loxP control allele as well as the dadar hypomorphic allele, both generated through homologous recombination [32,33]. β-actin is presented as a loading control. Wild type dADAR, which lacks the HA tag is presented as a negative control (−HA). (B) Quantification of western blot analysis. dADAR-HA signal is normalized to β-actin signal from each lane. Bars represent standard deviation. (C) Relative editing at sites that respond to dADAR dosage and/or to temperature. Editing at 20°C (black) is aligned along the black line. Green sites are unresponsive to temperature, but sensitive in the hypomorph (gray), suggesting editing is more dependent on RNA structure than on dADAR level. Purple sites are sensitive to both temperature and dADAR levels, while orange sites are more responsive to temperature than to dADAR level, suggesting at these sites that an additional factor, such as RNA structure, is partially responsible for the response of editing to temperature.

Figure 4

Reversibility of temperature-induced dADAR and editing levels. (A) Western blot analysis of the HA-tagged dADAR after three days at 10°C, 20°C and 30°C and after being shifted from 30°C to 20°C for 24, 48, or 72 hours. β-actin is presented as a loading control. Wild type dADAR, which lacks the HA tag is presented as a negative control (−HA). (B) Quantification of western blot analysis. dADAR-HA signal is normalized to the β-actin signal from each lane. Bars represent standard deviation. (C) Editing at specific sites after temperature shift. Editing after 72 hours at 30°C is depicted in red. After animals were held at 30°C for 72 hours they were shifted to 20°C for 24 (white), 48 (gray) and 72 (dark gray) hours. Animals held at 20°C for 72 hours but not shifted are depicted in black. Editing of synaptotagmin-1 sites 2 to 4 is unresponsive to temperature (Figure 1B) yet editing at sites 2 and 3 significantly decreases after 24 hours at 20°C. These sites recover to near-20°C levels after 48 and 72 hours. Editing of paralytic sites 1 to 3 begins to recover from 30°C levels after just 24 hours. Editing at shab site 6 increases at elevated temperature (Figure 1C) and after 72 hours recovers to near-20°C levels. Shab site 7, edited to 100%, is unresponsive to temperature (Figure 1D) and recovery. Bars represent standard error. P <0.0001: **, P < .05: *, not significantly different: NS.

Figure 5

Relative temperature stability of dADAR isoforms. (A) Western blot analysis of HA-tagged hardwired S and G dADAR isoforms [9]. The hypomorphic allele at 20°C is presented for comparison. β-actin is presented as a loading control. Wild type dADAR, which lacks the HA tag is presented as a negative control (−HA). (B) Quantification of western blot analysis. dADAR-HA signal from S (white) and G (black) alleles is normalized to the β-actin signal from each lane. Bars represent standard deviation.

Figure 6

Effect of RNA structural mutations on temperature-sensitivity. (A) Paralytic editing sites 1 to 3 (red), within an exon (blue), are encompassed within a complex tertiary structure involving three intronic (black) sequences: the editing site complementary sequence (ECS), the donor site complementary sequence (DCS) and a hairpin (HP), the loop of which forms a tertiary psueoknot with a docking site 3’ to the ECS. The DCS region is boxed for comparison with the knock in structural mutations. (B) In the ‘DCS delete’ mutation, the DCS region of the intron is excised (gray dotted line), resulting in a loss of secondary structure. (C) In the ‘DCS zip’ mutation, the DCS secondary structure is extended by nine base pairs, due to the insertion of seven nucleotides (green) within the intronic sequence. (D) Because these mutations overall decrease (DCS delete) or increase (DCS zip) editing at all three sites [8], the slopes of the editing response curves, rather than the absolute editing, was compared to loxP. The DCS delete (light green) and DCS zip mutation (dark green) show a different temperature-sensitive response pattern than that of the loxP control (gray). P <0.0001: **, P <0.05: *.

Figure 7

Molecular model of the effect of temperature on RNA editing. (A) At increased temperature, some RNA structures that direct RNA editing, formed between sequences in an exon (blue) and intron (black), melt, resulting in a decrease in editing at 30°C. (B) dADAR protein (pink) level decreases at elevated temperatures, leading to a decrease in editing at the RNA structures that still form at 30°C. (C) Because some RNA structures melt (A), dADAR protein, although present at lower concentrations, is free to edit remaining highly stable RNA structures, leading to an increase in editing at certain adenosines.