The SR protein B52/SRp55 regulates splicing of the period thermosensitive intron and mid-day siesta in Drosophila

Abstract

Similar to many diurnal animals, Drosophila melanogaster exhibits a mid-day siesta that is more robust as temperature increases, an adaptive response that aims to minimize the deleterious effects from exposure to heat. This temperature-dependent plasticity in mid-day sleep levels is partly based on the thermal sensitive splicing of an intron in the 3' untranslated region (UTR) of the circadian clock gene termed period (per). In this study, we evaluated a possible role for the serine/arginine-rich (SR) splicing factors in the regulation of dmpi8 splicing efficiency and mid-day siesta. Using a Drosophila cell culture assay we show that B52/SRp55 increases dmpi8 splicing efficiency, whereas other SR proteins have little to no effect. The magnitude of the stimulatory effect of B52 on dmpi8 splicing efficiency is modulated by natural variation in single nucleotide polymorphisms (SNPs) in the per 3' UTR that correlate with B52 binding levels. Down-regulating B52 expression in clock neurons increases mid-day siesta and reduces dmpi8 splicing efficiency. Our results establish a novel role for SR proteins in sleep and suggest that polymorphisms in the per 3' UTR contribute to natural variation in sleep behavior by modulating the binding efficiencies of SR proteins.

Figure 1

B52 stimulates dmpi8 splicing in Drosophila cultured cells. (a) Schematic diagram of the VT1.1 and VT1.2 haplotypes for the dper 3′ UTR showing the different SNP variants (red) and a previously identified B52 cross-linking site (X; Bradley et al.). (b–d) Drosophila S2 cells were transfected with either the pAct-Luc-VT1.1 (VT1.1) or pAct-Luc-VT1.2 (VT1.2) plasmid and grown at the indicated temperatures (12°, 22° or 25 °C). Cells were either mock-treated (control) or treated with double stranded RNA-mediated RNAi directed against the shown SR protein. RNA was purified from cell extracts and dmpi8 splicing efficiency calculated. Each experiment was done at least three times and values averaged. Values for dmpi8 splicing efficiency (% spliced) were significantly different between mock-treated (control) and RNAi-treated cells; *p < 0.05; **p < 0.01; two-tailed t-test. The following p values were determined (two-tailed t-test): [Panel (b); (1) 12 °C; VT1.1 control vs VT1.2 control, 4.2 × 10⁻⁶; VT1.1 control vs VT1.1 RNAi-B52, 3.6 × 10⁻⁴; VT1.2 control vs VT1.2 RNAi-B52, 0.045; (2) 22 °C; VT1.1 control vs VT1.2 control, 7.1 × 10⁻⁴; VT1.1 control vs VT1.1 RNAi-B52, 1.2 × 10⁻³; VT1.2 control vs VT1.2 RNAi-B52, 0.010]. [Panel (c); for VT1.1 control vs RNAi, B52, 0.012; SC35, 0.44; SF2, 0.89; Rbp1, 0.95; RSF1, 0.61; XL6, 0.97; for VT1.2 control vs RNAi, B52, 0.012; SC35, 0.65; SF2, 0.78; Rbp1, 0.10; RSF1, 0.63; XL6, 0.59]. [Panel (d); for VT1.1 control vs RNAi, B52, 0.006; SC35, 0.026; SF2, 0.10; Rbp1, 0.44; RSF1, 0.98; XL6, 0.58; for VT1.2 control vs RNAi, B52, 0.049; SC35, 0.33; SF2, 0.094; Rbp1, 0.040; RSF1, 0.62; XL6, 0.072].

Figure 2

Enhanced binding of B52 to transcripts containing the VT1.1 haplotype compared to the VT1.2 version. Drosophila S2 cells were transfected with either the the pAct-Luc-VT1.1 (VT1.1) or pAct-Luc-VT1.2 (VT1.2) plasmid, in the presence (+) or absence (−) of a plasmid expressing FLAG-B52, as indicated. After UV irradiation (+UV) or mock-treatment (−UV), cells were homogenized in 350 μl of lysis buffer. A fraction of the cell extract (100 μl) was used to measure the relative levels of Luc-dper transcripts (Input; top left) and FLAG-B52 using immunoblotting (Input; bottom left, lanes 1–5). The remainder of the cell extract was subjected to immunoprecipitation (IP) in the presence of anti-FLAG antibodies. Following IP, an aliquot was used for immunoblotting of FLAG-B52 (IP; bottom right, lanes 7–11), and the remainder used to measure the relative levels of Luc-dper transcripts bound to B52 (IP; top right). For the immunoblots shown, the left panel (lanes 1–5) and the right panel (lanes 7–11) come from two different gels (for images of the full-length blots, see figure S1). Note that for the set of immunoblots shown we did not include extracts prepared from the control samples expressing pAct-Luc-VT1.2 in the absence of FLAG-B52 (lanes 6 and 12). For quantitation of transcript levels, results from three experiments were averaged. The values for Luc-dper mRNA bound to B52, either with or without cross-linking, was significantly higher for VT1.1 compared to VT1.2, even though the starting amount of VT1.2 RNA was generally higher compared to VT1.1 (Input); the following p values were determined for IP samples comparing VT1.1 and VT1.2 values: +UV, 0.0035; −UV, 0.043. Also, the levels of Luc-dper transcripts present in immune complexes for cells expressing FLAG-B52 were significantly higher compared to negative control samples derived from cells not expressing FLAG-B52. The following p values were obtained comparing either Luc-VT1.1 or Luc-VT1.2 transcript levels in immune complexes expressing FLAG-B52 (lanes 7–10) compared to control samples not expressing FLAG-B52 (lanes 11 and 12); VT1.1/+UV vs VT1.1/No B52, 0.003; VT1.1/−UV vs VT1.1/No B52, 0.01; VT1.2/+UV vs VT1.2/No B52, 0.005; VT1.2/−UV vs VT1.2/No B52, 0.0004. *p < 0.05; **p < 0.01; two-tailed t-test.

Figure 3

Knock down of B52 in clock cells preferentially modulates daytime sleep during a daily light-dark cycle. (a–c) Young male progeny from the indicated cross (top of panels) were kept at 18 °C and entrained for 5 days in 12 hr:12 hr light/dark (LD) cycles [where Zeitgeber time (ZT) 0 is lights-on] followed by several days in constant darkness (DD; see Fig. 4). Daily wake sleep levels (a), total daytime sleep (b) and sleep latency (c) were measured, and average values during the last 3 days of LD are shown. For each genotype, data from 16 individual flies was used. Daily sleep levels (a) and sleep latency (c) for TUG > RNAi-B52(V101740) and control crosses are shown. Similar results were obtained with other RNAi-B52 lines (data not shown). (b) Total daytime sleep levels (ZT0–12) for several independent lines of RNAi-B52 (V38862, V101740, T37519) and appropriate control crosses. The following p values were determined: [Panel (a), ANOVA, TUG > V101740 compared to both controls; for daytime values from ZT0-12, 3.4 × 10⁻⁶; for nighttime values from ZT12-24, 0.057]. [Panel (b), two-tailed t-test; TUG > V38862 vs V38862 x w¹¹¹⁸, 5.0 × 10⁻¹⁰; TUG > V38862 vs TUG x w¹¹¹⁸, 1.2 × 10⁻⁸; TUG > V101740 vs V101740 x w¹¹¹⁸, 1.68 × 10⁻²⁰; TUG > V101740 vs TUG x w¹¹¹⁸, 1.96 × 10⁻¹⁸; TUG > T37519 vs T37519 x w¹¹¹⁸, 2.19 × 10⁻⁶; TUG > T37519 vs TUG x w¹¹¹⁸, 1.06 × 10⁻⁸]. [Panel (c), two-tailed t-test; for day values, TUG > V101740 vs V101740 x w¹¹¹⁸, 9.56 × 10⁻⁸; TUG > V101740 vs TUG x w¹¹¹⁸, 4.8 × 10⁻¹⁰; for night values, TUG > V101740 vs V101740 x w¹¹¹⁸, 0.081; TUG > V101740 vs TUG x w¹¹¹⁸, 0.00025]. (b,c) Values for TUG > RNAi-B52 were significantly different compared to one or more control crosses; *p < 0.05; **p < 0.01; two-tailed t-test. The corresponding daily activity profile for panel (a) is shown in Figure S2a.

Figure 4

The effect of B52 on daily sleep levels is strongly reduced in constant dark conditions. (a–c) Young male progeny from the indicated cross (top of panels) were kept at 18 °C and entrained for 5 days in 12 hr:12 hr light/dark (LD) cycles followed by several days in constant darkness [where circadian time (CT) 0 is the beginning of the subjective day; i.e., herein defined as equivalent to ZT0 in LD]. Figures are derived from the same data shown in Fig. 3, except that data are from the first day of constant darkness based on averaging values obtained from 16 individual flies for each genotype. The following p values were determined: [Panel (a), ANOVA, TUG > V101740 compared to both controls, 0.13]. [Panel (b), two-tailed t-test; TUG > V38862 vs V38862 x w¹¹¹⁸, 4.3 × 10⁻³; TUG > V38862 vs TUG x w¹¹¹⁸, 0.11; TUG > V101740 vs V101740 x w¹¹¹⁸, 0.022; TUG > V101740 vs TUG x w¹¹¹⁸, 9.38 × 10⁻⁵; TUG > T37519 vs T37519 x w¹¹¹⁸, 0.13; TUG > T37519 vs TUG x w¹¹¹⁸, 0.079]. [Panel (c), two-tailed t-test; for day values, TUG > V101740 vs V101740 x w¹¹¹⁸, 0.12; TUG > V101740 vs TUG x w¹¹¹⁸, 0.16; for night values, TUG > V101740 vs V101740 x w¹¹¹⁸, 0.15; TUG > V101740 vs TUG x w¹¹¹⁸, 0.65]. The corresponding daily activity profile for panel (a) is shown in figure S2b.

Figure 5

Modulation of daily sleep levels is not a shared feature of all SR proteins. (a–e) Young male progeny from the indicated crosses (top of panels) were kept at 18 °C and entrained for 5 days in 12 hr:12 hr light/dark (LD) cycles followed by several days in constant darkness (DD). Shown are the daily sleep levels (averaged over the last 3 days of LD). For each genotype, data from 16 individual flies was used to generate the graphs shown in each panel. The driver and SR protein targeted by RNAi are indicated. For RNAi-B52, the V101740 line is shown. The following p values (ANOVA) were determined for driver x RNAi (red) compared to both control crosses (blue and green): [daytime values from ZT0-12; panel (a), 0.0040; panel (b), 0.00039; panel (c), 0.92; panel (d), 0.48; panel (e), 0.28; nighttime values from ZT12–24; panel (a), 0.36; panel (b), 0.015; panel (c), 0.26; panel (d), 0.021; panel (e), 0.13]. The corresponding daily activity profiles are shown in Figure S4.

Figure 6

B52 stimulates dmpi8 splicing in flies. (a,b) Crosses were set up between the TUG driver and independent lines of RNAi-B52 (V38862, V101740 and T37519). For each RNAi line, we set up two crosses; in one set we used male TUG and female virgin RNAi lines and in the other set we did a reciprocal cross. For each cross, young adult progeny were placed in 6 vials (each vial had ~40 flies), and entrained for 5 days in 12 hr:12 hr light/dark (LD) cycles at 25 °C. In addition, the same was done for the parental controls (TUG, V38862, V101740 and T37519). On the last day of LD, at the indicated times, 1 vial for each cross was collected by freezing. For each RNAi line, flies from both crosses were pooled. Total RNA was extracted from fly heads and the splicing efficiency of dmpi8 (a) and total levels of dper (b) were calculated. Because results were highly similar for the different RNAi-B52 line and their parental controls (Fig. S5), we averaged all three TUG > RNAi-B52 crosses, and pooled the data from the parental controls to yield the group averages shown. Values shown for dmpi8 splicing efficiency (shown as fraction, where 1.0 is equal to 100% splicing of dmpi8) and dper RNA levels are from the average of two independent experiments. Note that the values for ZT0 were re-plotted for ZT24. (a) The daily dmpi8 splicing efficiency for flies expressing RNAi-B52 in tim-expressing cells (TUG > RNAi-B52) was significantly different compared to the parental controls (ANOVA, p = 0.010); in addition, for each time point we determined p values (two-tailed t-test); ZT0/24, 0.027; ZT4, 0.0054; ZT8, 0.45; ZT12, 0.0051; ZT16, 0.031; ZT20, 0.026. (b) For dper RNA levels we determined p values for each time point (two-tailed t-test); ZT0/24, 0.23; ZT4, 0.15; ZT8, 0.009; ZT12, 0.004; ZT16, 0.23; ZT20, 0.60. *p < 0.05; **p < 0.01; two-tailed t-test. The data clearly show that knock down of B52 in tim-expressing cells reduces the daily splicing efficiency of dmpi8.