A Molecular Basis for Natural Selection at the timeless Locus in Drosophila melanogaster

Federica Sandrelli,^1* Eran Tauber,^2* Mirko Pegoraro,^1* Gabriella Mazzotta,¹ Paola Cisotto,¹ Johannes Landskron,³ Ralf Stanewsky,^3,4 Alberto Piccin,¹ Ezio Rosato,² Mauro Zordan,¹ Rodolfo Costa,¹ Charalambos P. Kyriacou²

^* These authors contributed equally to this work.

Deceased.

To whom correspondence should be addressed. E-mail: rodolfo.costa{at}unipd.it

Wild European populations of Drosophila melanogaster have two major alleles of the timeless (tim) gene, ls-tim and s-tim (1). These alleles differ in their use of two alternative translational starts to generate longer (L-TIM₁₄₂₁) and/or shorter (S-TIM₁₃₉₈) isoforms (2). The ls-tim allele is derived from the s-tim allele, and directional selection is thought to have created a latitudinal gradient of ls-tim frequency within the past 10,000 years, perhaps due to an enhanced fitness of ls-tim individuals in temperate environments (1). TIM is a cardinal component of the circadian clock (3), and its light sensitivity via its physical interaction with the circadian photoreceptor cryptochrome (CRY) (4) mediates the fly's circadian responses to light (5). This photoresponse can be quantified at the behavioral level by studying the fly's locomotor response to brief light pulses delivered at zeitgeber time 15 (ZT15), three hours into the night phase of a light/dark [12 hours of light alternating with 12 hours of darkness (LD12:12)] cycle that generates a phase delay of a few hours; the same light stimulus administered late at night (ZT21) generates a phase advance (6).

Flies homozygous for each natural tim allele (ls-tim and s-tim) were established from iso-female lines from natural populations in Italy, the Netherlands, and Russia (1, 7). We examined the two natural variants' locomotor phase response to 20-min saturating light pulses delivered at ZT15 and ZT21. Because we were interested in observing whether tim-mediated behavioral photoresponsiveness might be relevant to its latitudinal distribution, we initially used two temperatures, 18° and 24°C. For phase delays (ZT15 light pulse), analysis of variance (ANOVA) revealed significant genotype [F_(1,164) = 11.1, P = 0.001], temperature [F_(1,164) = 23.8, P < 0.001], and population [F_(2,164) = 4.47, P < 0.002] effects. For phase advances (ZT21 light pulse), significant genotype [F_(1,192) = 10.5, P < 0.0015], population [F_(2,192) = 3.17, P = 0.044], and temperature x population [F_(2,192) = 8.4, P < 0.0005] interactions were observed. In these tests, the s-tim variants clearly showed a larger phase response, as compared with that of ls-tim (Fig. 1A).

Fig. 1. Phase response to light pulses of tim genotypes. Mean phase responses [hours (h) ±SEM] to 20-min light pulses delivered at ZT15, giving phase delays (–, below), and ZT21, giving phase advances (+, above), are shown. (A) Natural lines: Bitetto (Bit), southern Italy; Houten (Hu), Netherlands; and Moscow (Mos), Russia at 18° and 24°C. Black bars, s-tim; white bars, ls-tim; h, hours. (B) Transformants at 18°, 24°, and 28°C. Black bars, P[S-tim]; gray bars, P[L-tim]; white bars, P[LS-tim]. [View Larger Version of this Image (18K GIF file)]

We also examined phase responses of flies transformed with the transgenes P[LS-tim], P[L-tim], and P[S-tim] (1), which are designed to generate both or each TIM length isoforms, respectively, in a tim⁰¹ mutant background at three temperatures (18°, 24°, and 28°C). ANOVA for delays gave highly significant effects for genotype [F_(2,311) = 28.7, P < 0.0001] and temperature [F_(2,311) = 3.52, P = 0.03], with P[S-tim] flies consistently showing larger delays than the other genotypes. Similarly for advances, ANOVA of the data for P[L-tim] and P[S-tim] transformants at all three temperatures gave only a significant genotype effect [F_(1,201) = 12.28, P = 0.0006]. A similar result was obtained for all three transformants at 18° and 28°C (P[LS-tim] data was not collected at 24°C), with a resulting significant genotype effect [F_(2,187) = 4.94, P = 0.008]; as with delays, the advances of P[LS-tim] were intermediate between those of P[S-tim] and P[L-tim].

We next examined whether the ls-tim variants would show the normal arrhythmic behavioral response to constant bright light (LL) (8). We placed the natural tim variants, as well as the P[tim] transformants, in LD12:12 for 3 days and then in LL for 7 days. Locomotor arrhythmicity in LL of all genotypes was high (90 to 100%), and no tim allele (natural or transgenic) significantly differed in the time it took for the line to reach arrhythmia (table S1). Thus, P[L-tim] and ls-tim are able to mediate normal behavior in LL and constant darkness (DD), though they are less responsive to short light pulses (1).

Our results imply that in natural ls-tim and transformant P[LS-tim] flies, the longer isoform is translated and has biological activity that leads to a reduction in the circadian response to light. To investigate whether this was indeed the case, Western blot analysis on fly heads was used to study whether ls-tim flies did express the longer L-TIM isoform, and if so, whether they also expressed S-TIM. The ls-tim allele putatively encodes a protein that is 23 residues longer than the s-tim allele (2), a relatively small difference between isoforms that are 1400 residues long. At ZT1 (the first hour of light), when TIM levels were low because of degradation by light, and at ZT13 (the first hour of darkness), when TIM levels began to rise (9), we detected two isoforms in ls-tim (Fig. 2). P[L-tim] transformants generated only the longer isoform, whereas ls-tim flies and the P[LS-tim] transformants produced both long and short TIM isoforms; s-tim flies and P[S-tim] transformants produced only the short isoform (Fig. 2). TIM bands from ls-tim flies also appeared more intense than did those from s-tim in flies from the same or different genetic backgrounds (Fig. 2). The higher–molecular weight band was maintained in ls-tim samples that had prior phosphatase treatment, in spite of a change in mobility for both ls-tim and s-tim genotypes that was consistent with dephosphorylated TIM isoforms (Fig. 2). The same phosphatase treatment had a more dramatic effect on the mobility of PER isoforms, given its more extensive levels of phosphorylation (10, 11). On the basis of these findings, we suggest that the L-TIM isoform mediates an enhanced diapause response (1) and is also responsible for the attenuated circadian light sensitivity of ls-tim flies.

Fig. 2. TIM Western blots reveal different TIM isoforms. Fly heads harvested at ZT1 or ZT 13 are shown. The upper row shows natural lines: columns 1 and 5, GAB (ls-tim); 2 and 6, ATDD (s-tim); 3 and 7, Nov75 (ls-tim); and 4 and 8, B16 (s-tim). The broader "doublet" TIM band in ls-tim genotypes is shown with arrows. The upper middle row shows CS, Canton-S (ls-tim); y w(s), yw (s-tim); T27L and T29L (P[L-TIM]); T28S and T30S (P[S-TIM]); and (P[LS-TIM]) from fly heads harvested at ZT1. There are doublets in CS and P[LS-tim] transformants, with single bands in (P[S-TIM]) and (P[L-TIM]) transformants. The lower middle row shows larger-scale figures of yw (s-tim), Canton-S (ls-tim) and their heterozygote s/ls [(s/ls-tim)] at ZT1. The bottom row shows the results of a phosphatase (P) treatment applied to ls-tim and s-tim samples at ZT1 and ZT13. Blots were performed with anti-TIM and anti-PER. ls-tim genotypes maintain the higher–molecular weight isoform after phosphatase treatment. [View Larger Version of this Image (54K GIF file)]

We then systematically performed Western blot analyses for a full day in LD12:12 (Fig. 3) and in the second cycle of DD (fig. S1). TIM levels were significantly elevated in the ls-tim genotype in LD12:12 at all points of the cycle, as compared with s-tim [ANOVA for genotype: F_(1,79) = 22.7, P < 0.0001; for time: F_(7,79) = 8.3, P < 0.0001] (Fig. 3A), suggesting that ls-tim generates a higher combined level of the two isoforms or has more stable products. This same pattern was also observed in DD, where ANOVA revealed a significant genotype effect [F_(1,73) = 10.2, P = 0.002], but the time (the oscillation began to damp by the second cycle) and genotype x time interactions were both insignificant (fig. S1). The expression of TIM in the four independent P[L-tim] and P[S-tim] transformant lines was also studied in LD12:12. A nested ANOVA revealed significant time [F_(7,42) = 5.21, P = 0.0003] and time x genotype [F_(7,42) = 4.53, P = 0.0008] interactions, reflecting the observation that P[L-TIM] levels were similar between day and night, as compared with P[S-TIM] (Fig. 3B). These results suggest a stability difference between the two TIM isoforms, rather than a difference in translational efficiency (12), and they are further supported by the mRNA profiles for the two natural variants, which are very similar [time: F_(5,36) = 91.02, P < 0.0001; genotype: F_(3,36) = 1.52, NS (not significant)] (fig. S2).

Fig. 3. Circadian TIM profiles in natural lines and transformants. (A) Natural variants. Left panels show mean ± SEM TIM/TUB ratios from Western blots of the Moscow line (23°C in LD12:12, n = 6 blots for each variant); right panels show examples of corresponding Western blots. (B) Transformants. Left panels show mean ± SEM TIM/TUB ratios for Western blots (right) of each of the P[S-tim] (n = 11) and P[L-tim] (n = 10) transformants. [View Larger Version of this Image (45K GIF file)]

The enhanced stability of L-TIM might therefore be expected to contribute to the higher levels of TIM observed in natural ls-tim flies and to reduced circadian photoresponsiveness. Circadian light responses in Drosophila are mediated both by the canonical visual pathway, which uses rhodopsins, and by CRY (13). After stimulation by light, CRY can physically interact with TIM and/or PERIOD in yeast, in Drosophila S2 cells, and invivo(4, 14–16). These PER/TIM/CRY interactions lead to TIM degradation (5, 15) and subsequent PER instability, which releases the negative autoregulation of PER on the per and tim genes (17). We therefore studied the physical interaction of the L-TIM and S-TIM isoforms with CRY in the yeast two-hybrid system (16). No interactions between TIM and CRY occurred in the dark, and the level of interaction between CRY and L-TIM in light was weaker than that between CRYand S-TIM in both plate and liquid assays (Fig. 4, A and B). As a control, we also examined the interaction of L-TIM and S-TIM with the large fragment of PER (residues 233 to 685) that is stable in yeast (16), but these PER/TIM interactions were not significantly different (Fig. 4, C and D). These results indicate that the differences in interaction between the two TIM isoforms and CRY are a specific effect due to the additional N-terminal 23 residues in L-TIM, which interfere with the light-dependent dimerization of CRY.

Fig. 4. TIM interactions with CRY and PER in the yeast two-hybrid system. In light, L-TIM shows a diminished interaction with CRY in (A) plate assays (pJG4-5, empty vector control) and (B) liquid assays (mean ± SD), as compared with S-TIM [F(1,16) = 141.4, P < 0.001] for at least nine cultures derived from at least eight independent clones are shown. (C and D) L-TIM and S-TIM show equally robust interactions with the PER(233-685) fragment in light or darkness in both plate and liquid assays [F(1, 20) = 0.04, NS]. [View Larger Version of this Image (25K GIF file)]

A reduced L-TIM/CRY interaction may explain the differences in the fly's circadian photoresponsiveness and the enhanced L-TIM stability. The observation that ls-tim females are more prone to diapause at any day length (1) is also consistent with the results presented here. As in the corresponding diapause profiles (1), the transformants conclusively reveal that the circadian photoresponsive phenotypes of natural tim variants are not due to linkage disequilibrium between tim and a nearby locus, but they are attributable to tim itself. Furthermore, the similarity in behavior of natural s-tim variants and P[S-TIM] transformants suggests that the residual putative truncated N-terminal 19-residue TIM product from the s-tim allele does not play any major role in the phenotypes we have studied (2).

It has been argued that the light sensitivity of the circadian clock needs to be abated in temperate zones because of the dramatic increase in summer day lengths in northern latitudes (18, 19). One mechanism for this process involves a reduced sensitivity to light-induced disturbance by having a higher pacemaker amplitude (18, 19). However, the amplitude of TIM cycling in DD was not significantly different between the two variants (fig. S1), nor were there any significant differences in amplitude or phase of the tim mRNA cycle between the s-tim and ls-tim genotypes (fig. S2). Another way to attenuate circadian photoresponsiveness in temperate zones may be by filtering light input into the clock. The molecular changes to the L-TIM protein may buffer the circadian response to light in ls-tim individuals, even in the presence of S-TIM, and may contribute to the positive Darwinian selection observed for ls-tim in the European seasonal environment (1).

References and Notes

The editors suggest the following Related Resources on Science sites:

In Science Magazine

PERSPECTIVES
EVOLUTION: Tantalizing timeless

William Bradshaw and Christina Holzapfel (29 June 2007)
Science 316 (5833), 1851. [DOI: 10.1126/science.1145053]
 |  Summary »  |  Full Text »  |  PDF »

REPORTS
Natural Selection Favors a Newly Derived timeless Allele in Drosophila melanogaster

Eran Tauber, Mauro Zordan, Federica Sandrelli, Mirko Pegoraro, Nicolò Osterwalder, Carlo Breda, Andrea Daga, Alessandro Selmin, Karen Monger, Clara Benna, Ezio Rosato, Charalambos P. Kyriacou, and Rodolfo Costa (29 June 2007)
Science 316 (5833), 1895. [DOI: 10.1126/science.1138412]
 |  Abstract »  |  Full Text »  |  PDF »  |  Supporting Online Material »

In Science Signaling

EDITORS' CHOICE
Evolution
Timeless Changes

Laura M. Zahn (3 July 2007)
Sci. STKE 2007 (393), tw236. [DOI: 10.1126/stke.3932007tw236]
 |  Abstract »

THIS ARTICLE HAS BEEN CITED BY OTHER ARTICLES:

ADVERTISEMENT