Isolation and Analysis of Six timeless Alleles That Cause Short- or Long-Period Circadian Rhythms in Drosophila

Corresponding author: Michael W. Young, Rockefeller University, 1230 York Ave., New York, NY 10021., young{at}rockvax.rockefeller.edu (E-mail)

Communicating editor: J. J. LOROS

	ABSTRACT

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

In genetic screens for Drosophila mutations affecting circadian locomotion rhythms, we have isolated six new alleles of the timeless (tim) gene. Two of these mutations cause short-period rhythms of 21–22 hr in constant darkness, and four result in long-period cycles of 26–28 hr. All alleles are semidominant. Studies of the genetic interactions of some of the tim alleles with period-altering period (per) mutations indicate that these interactions are close to multiplicative; a given allele changes the period length of the genetic background by a fixed percentage, rather than by a fixed number of hours. The tim^L1 allele was studied in molecular detail. The long behavioral period of tim^L1 is reflected in a lengthened molecular oscillation of per and tim RNA and protein levels. The lengthened period is partly caused by delayed nuclear translocation of TIM^L1 protein, shown directly by immunocytochemistry and indirectly by an analysis of the phase response curve of tim^L1 flies.

CIRCADIAN rhythms have been studied genetically in species ranging from prokaryotes to mammals. The organisms in which mutations have led to significant insights into the working of the daily clock include Synechococcus, Neurospora, Drosophila, and mouse (reviewed in DUNLAP 1999 ). All such clocks share a common fundamental mechanism of negative autoregulation, and even the genes involved in the fly and mouse clock are conserved.

The first mutations of a single gene affecting the daily locomotion rhythm were found in the fruit fly Drosophila melanogaster. Three alleles of the period (per) gene were isolated that resulted in short- and long-period rhythms of locomotion and eclosion, or complete arrhythmia (KONOPKA and BENZER 1971 ). Since then, other alleles of per have been isolated that include ultrashort- and other long-period rhythms (HAMBLEN et al. 1998 ). The second central clock gene isolated in flies was called timeless (tim). As with per, a null mutation at tim results in arrhythmic individuals, indicating that tim is necessary to generate rhythmic behavior (SEHGAL et al. 1994 ). Another allele of tim was found as a specific suppressor of the per^L mutation, but by itself, tim^SL has a very subtle phenotype (RUTILA et al. 1996 ). A third allele, tim^rit, from a natural population, shows long-period phenotypes that deteriorate into arrhythmicity at 30° (MATSUMOTO et al. 1999 ).

Three more clock genes were found in forward genetic fly screens: dClock (dClk) and cycle (cyc) mutations both result in arrhythmia (ALLADA et al. 1998 ; RUTILA et al. 1998 ). The dClk gene is the fly homologue of the mouse Clock gene, which was found by a mutation leading to long-period and arrhythmic mice (VITATERNA et al. 1994 ). The double-time (dbt) gene was found via period-altering alleles and was shown to be required for molecular rhythms by studying a pupal lethal allele (KLOSS et al. 1998 ; PRICE et al. 1998 ).

The continued investigation of these genes has led to the following model of the fly clock (reviewed in HARDIN 1998 ): PER and TIM are negative autoregulators, and their RNA levels cycle, peaking at dusk. The protein levels also oscillate and PER and TIM proteins accumulate from dusk onwards in the cytoplasm of pacemaker cells to form a heterodimer that translocates into the nucleus later at night. There, the PER/TIM complex interferes with the transcriptional activators dCLK and CYC and represses transcription of per and tim. After the turnover of PER and TIM, the cycle restarts with per and tim RNA accumulation in the morning. Proper turnover of PER requires phosphorylation and the activity of DBT, a casein kinase I family member. Additional genes, like vrille (BLAU and YOUNG 1999 ), modulate this central oscillation.

The TIM protein also functions as an important sensor for light-induced phase shifts and entrainment (YOUNG 1998 ). Light causes TIM degradation via the proteasome pathway (NAIDOO et al. 1999 ). However, little is known about the domains in TIM important for degradation or repression, and TIM has no homology to other known proteins. The only defined domain in TIM is the cytoplasmic localization domain (CLD), deletion of which allows monomeric TIM to enter the nucleus without PER in cultured cells (SAEZ and YOUNG 1996 ). However, there is normally a mutual requirement between PER and TIM for nuclear localization, and two binding domains have been defined with respect to TIM (SAEZ and YOUNG 1996 ).

One common property of period-altering mutations is that they show semidominant phenotypes (see DUNLAP 1993 , DUNLAP 1996 ; HALL 1998 ). We made use of this fact in a screen for mutations that would either dominantly or semidominantly alter circadian locomotion rhythms and report the isolation of four new alleles of tim from this screen. Two further mutations of tim were isolated in a prior screen for homozygous mutations on the second chromosome affecting rhythmic behavior. These alleles show phenotypes ranging from 21- to 28-hr rhythms and may define important portions of TIM for the maintenance of a 24-hr day.

	MATERIALS AND METHODS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Fly methods and mutational screens:
Flies were raised on standard cornmeal/agar medium. EMS mutagenesis was done according to LEWIS and BACHER 1968 . The crossing scheme depicted in Fig 1A was followed for the locomotion screen for homozygous mutations on the second chromosome. Four to six homozygous males per stock (if viable, otherwise heterozygotes) were screened for their locomotion rhythms. [Note: Df(2R)XTE11 was included to find lethal genes in the interval 51D3-52A10, notably a mutation in the igloo gene].

View larger version (23K): In this window In a new window Download PPT slide   Figure 1. Crossing schemes followed to screen for novel mutations affecting circadian locomotion rhythms (see MATERIALS AND METHODS for details).

For the screen for dominant mutations on the second or third chromosome the scheme in Fig 1B was employed. Primary males (each unique in their genotype) were assayed for their rhythms and were crossed to two balancer strains only if they showed an interesting phenotype. Depending on whether a reproducible phenotype was observed segregating with a mutant second or third chromosome, a balanced stock was then established with that chromosome.

Behavioral assays:
For eclosion, flies were entrained for at least 3 days in cycles of 12 hr light followed by 12 hr dark (LD) and then released into constant darkness (DD). Every 2 hr, the freshly emerged adults were collected under a Kodak GBX-2 red safe light and then counted. Locomotion assays (at 25° unless otherwise noted) and chi-square periodogram analysis were performed as described in PRICE et al. 1998 . Anchored phase response curve: flies were entrained to LD for 3 days. Groups of flies (12–16) were exposed to 10-min light pulses (3000 lux) at various times after the last lights off, and their locomotion activity was assayed for 5–7 days. Their average activity offset was determined and compared to the average offset of unpulsed flies to calculate a phase shift. The offset was determined as the point where the activity fell below the mean activity.

Molecular biology and biochemistry:
RNase protection assays and Western blotting were performed as described (PRICE et al. 1998 ). Sectioning and immunocytochemistry was done according to MYERS et al. 1996 , using a 1:1500 dilution of rat anti-TIM antibody. For sequencing mutant and parental strains, total RNA was isolated and reverse-transcribed with random hexamers, or genomic DNA was isolated. Gene-specific primers for tim were used for PCR, and DNA sequencing was performed using a cycle-sequencing kit (Perkin Elmer, Branchburg, NJ).

	RESULTS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Isolation of novel tim alleles:
In two genetic screens for mutations changing circadian locomotion rhythms (see MATERIALS AND METHODS), we isolated new alleles of the timeless gene. The tim^L1 and tim^L2 alleles (Fig 2) were found as homozygous mutants, while tim^L3, tim^L4, tim^S1, and tim^S2 (Fig 3) were isolated due to their semidominant phenotypes. The tim^L1 and tim^L2 mutations showed no recombination with the original tim⁰¹ allele (415 and 143 potentially recombinant chromosomes were assayed, respectively; data not shown), indicating very tight linkage of these mutations. Sequencing of the tim loci from tim^L1 and tim^L2 mutant strains revealed the presence of single missense mutations that mapped to the PER-TIM interaction domains (Fig 4), confirming the genetically derived conclusion that they are alleles of tim. The tim^S2 mutation was also genetically mapped to the tim genomic region, with 114 recombinants using Sp and dp markers (data not shown). The other mutations were determined to be alleles of tim due to their noncomplementation with amorphic tim⁰ alleles (Table 1). Since the tim gene is dosage insensitive (see below), i.e., tim⁰/tim⁺ has no phenotype itself, complementation tests are straightforward (in contrast to some per complementation tests; see HALL 1998 ). Nevertheless, in the absence of sequencing data it is still formally possible that the three unmapped mutations are not tim alleles and that they show an uncommon intergenic noncomplementation with tim⁰ alleles.

View larger version (38K): In this window In a new window Download PPT slide   Figure 2. Locomotion and eclosion profiles of wild-type, timL1, and timL2 flies. (A) Locomotion activity was measured for 4 days in LD (represented by bars on top). The flies were then released into DD (indicated by an asterisk) and assayed for 7.5 days. The genotypes and individual period lengths of representative flies are indicated. (B) Eclosion profiles of flies emerging in DD. Strains were previously entrained to LD cycles (prior lights-on are indicated by stippled lines). The genotype and most likely eclosion period of the fly populations are shown. The periods are ±0.5 hr for wild type and ±1 hr for timL1 and timL2.

View larger version (32K): In this window In a new window Download PPT slide   Figure 3. Locomotion actograms of two further timL alleles and short-period tim mutants. Activity was measured in DD after LD entrainment and the genotype and individual period length are indicated. Note that the chromosome carrying timL3 also harbors an unlinked mutation resulting in homozygous lethality.

View larger version (10K): In this window In a new window Download PPT slide   Figure 4. Schematic of PER and TIM proteins. The timL1 and timL2 mutations map to the PER-TIM interaction domains. The interaction domains of the two proteins are connected by dashed lines in the middle: PER 233-365 interacts with TIM 514-587, and PER 452-512 with TIM 724-923 (SAEZ and YOUNG 1996 ). Mutations that fall into these domains are shown along with their amino acid change and position. They include PERL and the two TIML variants sequenced. NLS, nuclear localization signal; CLD, cytoplasmic localization domain; and PAS, PER-ARNT-SIM homology.

Five of the six alleles show the same fully penetrant phenotype when homozygous or hemizygous over tim⁰ (while one, tim^L3, is on a homozygous lethal chromosome). In addition, all six alleles show a semidominant phenotype where the tim^mutant/tim⁺ period length falls between the mutant and wild-type homozygous phenotypes. The two alleles tested, tim^L1 and tim^L2, also showed an eclosion period similar to the periods of their locomotion rhythms (Fig 2B).

The tim gene is dosage insensitive:
Most of the central clock genes isolated are dosage sensitive. Flies hemizygous for per or cyc have 25-hr period lengths (KONOPKA and BENZER 1971 ; RUTILA et al. 1998 ). Dosage sensitivity for dClk is somewhat tentative, since the point mutant dClk^Jrk probably produces a dominant-negative protein rather than a null mutation, and dosage sensitivity is inferred from a large deficiency removing the dClk locus, resulting in 25-hr flies (ALLADA et al. 1998 ). Amorphic vri mutations result in period-shortening of about half an hour in hemizygous flies (BLAU and YOUNG 1999 ). So far, the only Drosophila clock gene isolated that was shown not to be dosage sensitive is dbt (PRICE et al. 1998 ). We tested the dosage sensitivity of the tim gene by assaying hemizygous tim^- flies and flies carrying a duplication of tim⁺, Dp(2;3)DTD33. Flies harboring one to three copies of tim⁺ all showed indistinguishable locomotion period lengths (Table 2), showing that tim, in contrast to its molecular partner per, is dosage insensitive. The molecular basis for this difference in dosage sensitivity may be related to the fact that PER is destabilized by the activity of the kinase DOUBLE-TIME, but stabilized by physical interaction with TIM. Such a relationship tends to promote higher levels of TIM than PER protein throughout most of the circadian cycle (PRICE et al. 1998 ). Indeed, higher levels of TIM than PER have been measured directly in heads from wild-type Drosophila (ZENG et al. 1996 ; SURI et al. 1999 ).

Genetic interactions among tim alleles:
Since tim is dosage insensitive, but all the period-altering tim alleles are semidominant, these alleles cannot be hypomorphs. Nor can they be hypermorphs since tim^mutant/tim⁰ flies show a more severe phenotype than tim^mutant/tim⁺ flies. We characterized some alleles more carefully, utilizing the tim⁺-carrying duplication Dp(2;3)DTD33. We assayed tim^L1, tim^L2, and tim^UL [described more closely in ROTHENFLUH et al. 2000 ]. For all three alleles tested, the period length of locomotion rhythms was dependent on the ratio of mutant to wild-type tim allele (Table 3).

Tendency toward multiplicative interaction of period-altering alleles:
We also tested the genetic interactions of period-altering tim alleles with period-altering per alleles. Table 4 shows the resulting period length of various double-mutant combinations. The phenotypes generally reflect the action of alleles at both loci in such combinations (see below), and no period-altering mutation is epistatic to another. Thus, these mutant alleles seem to independently affect circadian rhythmicity.

The absolute number of hours a given tim^mutant allele changes the period length of the genetic background is rather variable, and higher, the longer the period of the genetic background. However, the percentages by which a tim^mutant alters the period length of the different per genetic backgrounds are generally similar. The genetics of period-altering alleles is perhaps best described by a multiplicative interaction, where each allele changes the period of the genetic background by a fixed percentage. This would be in contrast to an additive interaction, where each allele changes the period by a fixed number of hours.

Temperature compensation of tim alleles:
One characteristic of the biological clock is that it runs at the same speed over a broad span of temperatures, i.e., it is temperature compensated. Many period-altering mutations are defective in temperature compensation: the per^S and per^L alleles all show stronger deviations from wild type at higher temperatures (KONOPKA et al. 1989 ), while the period of per^SLIH is closer to wild type at higher temperatures (HAMBLEN et al. 1998 ). We tested a number of other mutant alleles for their temperature compensation. Table 5 shows that the three long-period alleles of tim tested are still temperature compensated. This contrasts with tim^rit flies, whose period is progressively longer at higher temperature, and at 30° most flies are arrhythmic (MATSUMOTO et al. 1999 ; see also Table 5 legend). The period length of tim^S2 flies increases slightly at higher temperatures, but the effect is subtle, with a deviation of <1 hr over a span of 10°. The period of dbt^L flies shows an improvement toward wild type at higher temperatures, indicating a defect in temperature compensation that is similar to the response observed in per^SLIH flies (see also HAMBLEN et al. 1998 ). Taken together, these data show that in Drosophila, a period-altering mutation does not necessarily disrupt temperature compensation or show temperature sensitivity. This is most clearly seen for tim^UL flies: even a 33-hr clock can run accurately over a span of 10°. Similarly, in Neurospora, certain period-altering mutations do not affect temperature compensation (cf. LOROS et al. 1986 ; DUNLAP 1993 ).

Long-period molecular oscillations in tim^L1:
To test whether a change in the behavioral periods (see Fig 2) is also reflected in a change of the underlying molecular oscillation, we decided to study tim^L1 mutant flies in molecular detail. Heads from tim^L1 flies were collected for 1 day in LD and two subsequent days in DD. RNA and protein were isolated and tested for levels of per and tim (Fig 5). In LD, a phase delay of 2 hr is observed in the accumulation of per and tim RNA and protein in tim^L1. In DD, the first peak of per and tim RNA is delayed by 4 hr and while two full RNA cycles are completed in 2 days of DD in wild type, only one and a half cycles are observed over the same time span in tim^L1, reflecting the lengthened behavioral period in tim^L1 flies. A delay can also be seen in the PER protein oscillation in tim^L1 in DD, especially the shift from hyper- to hypophosphorylated forms of PER, an indication for newly accumulating PER (EDERY et al. 1994 ). In wild type these shifts occur between CT 6 and 10 on day 1 (CT, circadian time, reflects time measured in DD with CT 6 = subjective noon and CT 18 = subjective midnight) and then between CT 2 and 6 on day 2. In tim^L1 they are delayed to around CT 14 of day 1 and then between CT 14 and 18 of day 2 in DD (Fig 5A), again consistent with a long-period cycle.

View larger version (52K): In this window In a new window Download PPT slide   Figure 5. per and tim protein and RNA oscillations in timL1. Three days were measured, the first one in LD and two subsequent ones in DD. Anti-PER (A) and anti-TIM (B) Western blots from wild-type (wt) and timL1 head extracts. Equal amounts of protein were loaded (controlled with a nonspecific cross-reacting band on the PER blots, not shown). (C and E) Wild-type and (D and F) timL1 RNase protections and their corresponding quantitations. Quantitations in E and F can be compared directly, since timL1 gels always included wild-type samples as well. Three experiments yielded similar results.

Delayed nuclear translocation in tim^L1:
The missense mutations in tim^L1 and per^L fall into the PER-TIM interaction domains (see Fig 4), thus opening the possibility that the similar behavioral phenotypes of these two mutations might result from similar molecular defects. The long-period phenotype of per^L was correlated with delayed nuclear translocation of PER^L protein by (CURTIN et al. 1995 ). The entry of PER^L into the nuclei of lateral neurons was further delayed by increasing the assay temperature, thereby correlating the per^L behavioral period lengthening at higher temperatures with a molecular phenotype. Since this temperature effect was also correlated with a decrease in PER^L/TIM interaction in yeast cells at increased temperatures (GEKAKIS et al. 1995 ), the following model of the per^L molecular defect has emerged: PER^L is an amino acid substitution in one of the two PER/TIM interaction domains that leads to decreased affinity of the proteins. This delays heterodimer formation in the cytoplasm of clock cells, and thus PER/TIM nuclear translocation occurs later than in wild type.

If a similar molecular mechanism is at least partially responsible for the tim^L1 phenotype, then a delay in nuclear translocation could also be expected in this genotype. We found that nuclear staining of TIM protein in the photoreceptor cells is delayed in tim^L1 relative to wild type (Fig 6). At early time points, diffuse cytoplasmic staining with little nuclear TIM was observed (Fig 6A and Fig D). Then the outer photoreceptors (1–7) showed nuclear staining (solid arrows), while the R8 photoreceptors (open arrows) continued to show little nuclear TIM (Fig 6B and Fig E). Still later, nuclear TIM staining predominated in both photoreceptor populations (Fig 6C and Fig F). Note the different phases of nuclear staining in wild type and tim^L1 (e.g., no R8 staining could be seen at ZT 17 in tim^L1, Fig 6D, while in wild type, nuclear staining was observed in all photoreceptors at ZT 17, Fig 6B; ZT, Zeitgeber time, reflects time measured in LD where ZT 0 = lights on and ZT 12 = lights off).

View larger version (90K): In this window In a new window Download PPT slide   Figure 6. TIM nuclear translocation in wild-type and timL1 photoreceptor cells. Frontal head sections from flies kept in LD were stained with anti-TIM antibody. Wild- type sections are (A) ZT 16, (B) ZT 17, and (C) ZT 18. timL1 sections are (D) ZT 17, (E) ZT 18, and (F) ZT 20. Nuclear staining in R1-7 photoreceptors (black ar- rows). Nuclear staining of R8 photoreceptors (open arrows). On a subjective scale from 1 (predominantly cytoplasmic staining) to 3 (exclusive nuclear staining) we obtained the following staining intensities for wild-type sections (with standard error and number of sections in parentheses): ZT 14, 1.0 ± 0.0 (4); ZT 15, 1.4 ± 0.3 (5); ZT 16, 1.9 ± 0.2 (8); ZT 17, 2.5 ± 0.2 (10); ZT 18, 2.8 ± 0.2 (8). For timL1 we obtained the following: ZT 15, 1.2 ± 0.2 (6); ZT 16, 1.4 ± 0.2 (9); ZT 17, 1.6 ± 0.3 (5); ZT 18, 1.8 ± 0.2 (10); ZT 19, 2.2 ± 0.2 (10); ZT 20, 2.6 ± 0.2 (9). Note that in photoreceptors nuclear translocation is 2–3 hr advanced compared to the lateral neurons (cf. CURTIN et al. 1995 ).

Yet in contrast to per^L, where a delay of 5 hr in nuclear staining closely reflects the behavioral period lengthening (CURTIN et al. 1995 ), nuclear entry in tim^L1 is only delayed by 2 hr, while tim^L1 flies show a behavioral period lengthening of 4 hr. The delay in nuclear translocation of TIM^L1 protein presumably contributes to the behavioral phenotype, but does not account for the full 4-hr-longer period.

To test delayed nuclear translocation in tim^L1 in an indirect way, the phase response curve (PRC) for tim^L1 was established (Fig 7). Light pulses at different times in DD result in different phase changes, depending on the time of the pulse. Three time domains can be defined in a PRC: (i) a phase delay domain in the early subjective night, when light pulses reset the fly clock to late afternoon; (ii) a phase advance domain, late at night, where light pulses advance flies into the early morning; and (iii) a domain of relative insensitivity to light during the subjective day. Each of these domains has a molecular correlate: the delay domain corresponds to the time of PER and TIM cytoplasmic accumulation, while during the advance domain PER and TIM are in the nucleus. The transition point (where delay changes to advance) occurs at the time of PER/TIM nuclear translocation and is delayed in per^L flies (RUTILA et al. 1997 ). Thus the PRC transition point can be regarded as an indirect behavioral means of measuring the timing of nuclear translocation. Fig 7 shows that in wild type the transition occurs shortly after ZT 18, while in tim^L1 it is delayed to about ZT 19.5. Note that the difference is only 1.5 hr, again indicating that delayed nuclear translocation contributes to the 4-hr period lengthening observed in tim^L1 flies, but does not fully explain it.

View larger version (24K): In this window In a new window Download PPT slide   Figure 7. Anchored phase response curve of wild-type, timL1, and dbtL flies. The phase shift for each time point was assayed two to five times, and standard error bars are shown (unless they are smaller than the plot symbol). At CT 19, there is a significant difference (P < 0.01, Student's t-test) between timL1 and wild type or dbtL, showing that the transition point is delayed specifically in timL1 flies.

To ensure that the delayed transition point is not just a reflection of a long period and concomitant phase delay in LD conditions (cf. Fig 5), we established a PRC for dbt^L. The long-period phenotype of dbt^L (27 hr; see Table 5) is the result of delayed PER phosphorylation and turnover late at night in the nucleus and includes a phase delay of PER and TIM protein accumulation in LD similar to tim^L1 (cf. Fig 5; PRICE et al. 1998 ). The transition point of the dbt^L PRC is very close to wild type, at about ZT 18.5, indicating that an 2-hr phase delay in PER and TIM protein accumulation does not automatically lead to a delay of the PRC transition point. At ZT 19, a light pulse results in an 1.5-hr phase advance in dbt^L flies, while tim^L1 flies are still phase delayed by 3.5 hr at this time point. The late transition point and delayed nuclear translocation of the PER/TIM complex seen in tim^L1 is therefore a specific phenotype of tim^L1 flies.

The amplitude of the advance domain in the tim^L1 PRC is more than double that seen with wild-type flies. This is similar to the phenotype observed in tim^UL flies, where it has been correlated with prolonged stability and nuclear localization of the PER/TIM^UL complex (ROTHENFLUH et al. 2000 ). Thus a secondary defect, possibly similar to the one seen in tim^UL flies late at night, is likely to occur as well as delayed nuclear translocation, and together these defects lead to the 4-hr period lengthening observed in the tim^L1 strain.

Since delayed nuclear translocation in per^L was correlated with a decreased interaction between PER^L and TIM, we tested the PER/TIM^L1 interaction in the yeast two-hybrid system. Our results were inconclusive. While PER 233-685 showed a reproducible decrease in interaction with TIM^L1 457-1014 to 65% PER/TIM⁺ interaction, a TIM^L1 300-1237 construct showed no difference in interaction with PER 233-685 compared to wild-type TIM (data not shown). Therefore, we do not know whether the delayed nuclear translocation is a result of a diminished interaction between PER and TIM^L1, analogous to that seen with PER^L/TIM.

	DISCUSSION

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

New timeless alleles:
In this report, we describe the isolation and characterization of six new alleles of the timeless gene. All alleles result in period alterations of locomotion rhythms with full penetrance, and the two alleles tested for eclosion, tim^L1 and tim^L2, also show a comparably altered period. These two mutations were isolated in a screen for homozygous mutants affecting circadian rhythms. Since they are both semidominant, and previously isolated period-altering alleles of per also show semidominance (KONOPKA and BENZER 1971 ), we changed our behavioral screen and assayed unique F₁ males for (semi)dominant locomotion phenotypes (see MATERIALS AND METHODS). Using this new strategy, we increased our throughput of mutant chromosomes tested by 8- to 10-fold (and even more per person-hour), and we isolated five additional tim alleles, one of which, tim^UL, is described in ROTHENFLUH et al. 2000 . Two of these alleles, tim^L3 and tim^L4, are similar to tim^L1. The other two, tim^S1 and tim^S2, are short-period mutations, the first reported for tim. There are now 10 tim alleles reported that include arrhythmic, short, long, and ultralong phenotypes (SEHGAL et al. 1994 ; MATSUMOTO et al. 1999 ; ROTHENFLUH et al. 2000 ) and an allele-specific suppressor of per^L, which by itself has a very small phenotype (RUTILA et al. 1996 ). Clearly, the tim gene can be altered in many ways to result in abnormal circadian rhythmicity.

In contrast to most period-altering per mutations (HAMBLEN et al. 1998 ), the tim alleles we isolated are well temperature compensated. Little is known about the molecular mechanism of temperature compensation, but our data suggest that the tim gene plays a lesser role in this mechanism than the per gene. That loss of temperature compensation in tim^rit can be rescued by introducing an additional dose of the per gene (MATSUMOTO et al. 1999 ) supports this notion that per is central to the mechanism of temperature compensation.

The genetics of period-altering mutations:
As mentioned above, all period-altering tim alleles described here are semidominant, and we found that, for those alleles tested, the ratio of mutant to wild-type product determines the severity of the phenotype. KING et al. 1997A found that the mouse Clock mutation, which causes an internal deletion of 51 amino acids (KING et al. 1997B ), is semidominant and behaves as an antimorph, or dominant negative mutation (MULLER 1932 ). Wild-type CLOCK and BMAL-1 heterodimers can transactivate transcription from mper-1-derived E-boxes in mammalian tissue culture (GEKAKIS et al. 1998 ), yet mutant CLOCK and BMAL-1 cannot activate, although they are still capable of binding one another and of binding to target DNA. Thus, mutant CLOCK acts in a dominant-negative fashion, titrating out BMAL-1 protein and CLOCK/BMAL-1 DNA binding sites. This results in reduced activation of mper-1 in vivo (JIN et al. 1999 ) and in long-period behavioral rhythms that deteriorate to arrhythmia (VITATERNA et al. 1994 ). Thus in the instance of Clock, dominant-negative antimorphic behavior can be explained in a straightforward manner.

Can this dominant-negative interpretation apply for the tim^L mutations? In contrast to mutant CLOCK protein, TIM^L proteins clearly retain most of their function, since tim^L mutants are all highly rhythmic. Thus they are incompletely dominant and not particularly negative. We have previously interpreted the semidominant nature of dbt mutations by proposing the formation of distinct pools of stably interacting proteins (KLOSS et al. 1998 ). In the case here of semidominant tim alleles this could also apply: in a cell where TIM^L and TIM⁺ proteins are present, both proteins bind PER and two distinct pools of PER/TIM⁺ and PER/TIM^L are generated. Both pools can perform their function, albeit at different rates, and the two pools may equally contribute at the molecular level. In such a model, since both pools feed back onto the transcriptional regulation of both tim alleles present, an integration of the independent reactions results, which ultimately contributes to an intermediate period length.

A nearly multiplicative interaction:
Period-altering mutations are generally thought to affect a specific step in the circadian cycle: per^L delays cytoplasmic PER^L/TIM complex formation (see above), per^S increases nuclear turnover of PER^S (ZERR et al. 1990 ; MARRUS et al. 1996 ), and tim^UL results in a prolonged nuclear PER/TIM^UL complex (ROTHENFLUH et al. 2000 ). These changes seem to be achieved independently, and it is therefore not surprising that double-mutant combinations give phenotypes that reflect the contribution of each mutant gene. On further inspection though, the genetic interaction was found not to be additive. Rather, as previously seen with some period-altering mutations in Neurospora (LAKIN-THOMAS and BRODY 1985 ), interactions among mutant alleles tend to be multiplicative or nearly so. That tim^UL, for example, lengthens the period of per^S flies by 6 hr, but lengthens the per^L rhythm by almost 13 hr seems surprising if indeed different steps are affected by these mutations. One explanation is that a mutant phenotype that affects a single step in the cycle has consequences throughout the cycle. tim^L1 affects nuclear translocation of PER/TIM^L1, but the net result of this effect is that both per and tim RNA oscillations are affected in timing and in amplitude (Fig 5). Similarly, tim^rit and tim^UL affect both per and tim RNA oscillations and levels (MATSUMOTO et al. 1999 ; ROTHENFLUH et al. 2000 ). Thus the cause of a mutant phenotype may be one step in the cycle, but the consequences reach over the whole cycle. This altered cycle may now be more (if lengthened) or less (if shortened) susceptible to further perturbations by introducing additional period-altering mutations, which themselves may have a unique cause but affect the whole molecular cycle.

The molecular defect in tim^L1:
Because the missense mutation in tim^L1 maps to one of the PER/TIM interaction domains, it was an attractive hypothesis that TIM^L1 would interact less strongly with PER and result in delayed formation and nuclear translocation of the PER/TIM^L1 complex, analogous to the model of the per^L defect (CURTIN et al. 1995 ; GEKAKIS et al. 1995 ). We were unable to observe a reproducible difference in PER/TIM^L1 interaction compared to the PER/TIM interaction in the yeast two-hybrid system. However, we did observe delayed nuclear translocation of TIM^L1 protein in photoreceptor cells. This was also reflected by a delayed transition point from phase delay to phase advance in a tim^L1 phase response curve. In both cases, the difference between wild type and tim^L1 was only 2 hr. The cause for the delayed nuclear translocation we observed may be a subtle PER/TIM^L1 interaction difference that emerged with some two-hybrid constructs, but not others.

The 2-hr phase delay we detected in PER and TIM^L1 protein accumulation in an LD cycle could delay complex formation and thus nuclear translocation. The same steady-state phase delay of PER and TIM accumulation is also seen in dbt^L flies (PRICE et al. 1998 ), yet according to their PRCs there is some difference in nuclear entry of PER/TIM in tim^L1 compared to dbt^L, suggesting specificity of delayed nuclear translocation to the tim^L1 mutation. Even if delayed nuclear entry is a cause for the tim^L1 period lengthening, the delay observed does not fully reflect the 4-hr behavioral period lengthening seen in tim^L1 flies. The differences in wild-type and tim^L1 PRCs lead us to propose that there is an additional defect in tim^L1 mutant flies when PER/TIM^L1 complexes are nuclear.

Conclusion:
We have isolated six new alleles of the timeless gene. They are associated with long- and short-period rhythms, showing that tim mutant phenotypes for altered period lengths can be as diverse as those found for tim's partner, per. The genetics of these mutations suggests that there is a finely tuned inter- and even intragenetic interplay between clock genes that is easily disturbed by subtle changes in their encoded proteins. Further analysis of these mutants will allow more insight into the various steps that are involved in generating a precise 24-hr internal clock, as well as identifying domains of TIM protein important for its proper function.

	FOOTNOTES

¹ Present address: Department of Physiology, University of California, San Francisco, CA 94143.
² Present address: School of Biological Sciences, University of Missouri, Kansas City, MO 64110.

	ACKNOWLEDGMENTS

We thank Liz Alcamo, Terry Everett, Amy Kiger, Yvonne deLotto, and Evelyn Icasiano for their help during the screens, and the Bloomington stock center for fly strains. We are grateful to Justin Blau for many discussions and critical reading of the manuscript. A.R. was supported by the Arnold and Mabel Beckman Foundation. This work was supported by the National Institutes of Health-GM (54339) and by the National Science Foundation Science and Technology Center for Biological Timing.

Manuscript received February 3, 2000; Accepted for publication May 31, 2000.

	LITERATURE CITED

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

ALLADA, R., N. E. WHITE, W. V. SO, J. C. HALL, and M. ROSBASH, 1998  A mutant Drosophila homolog of mammalian Clock disrupts circadian rhythms and transcription of period and timeless.. Cell 93:791-804[Medline].

BLAU, J. and M. W. YOUNG, 1999  Cycling vrille expression is required for a functional Drosophila clock. Cell 99:661-671[Medline].

CURTIN, K. D., Z. J. HUANG, and M. ROSBASH, 1995  Temporally regulated nuclear entry of the Drosophila period protein contributes to the circadian clock. Neuron 14:365-372[Medline].

DUNLAP, J. C., 1993  Genetic analysis of circadian clocks. Annu. Rev. Physiol. 55:683-728[Medline].

DUNLAP, J. C., 1996  Genetic and molecular analysis of circadian rhythms. Annu. Rev. Genet. 30:579-601[Medline].

DUNLAP, J. C., 1999  Molecular bases for circadian clocks. Cell 96:271-290[Medline].

EDERY, I., L. J. ZWIEBEL, M. E. DEMBINSKA, and M. ROSBASH, 1994  Temporal phosphorylation of the Drosophila period protein. Proc. Natl. Acad. Sci. USA 91:2260-2264[Abstract/Free Full Text].

GEKAKIS, N., L. SAEZ, A.-M. DELAHAYE-BROWN, M. P. MYERS, and A. SEHGAL et al., 1995  Isolation of timeless by PER protein interaction: defective interaction between timeless protein and long-period mutant PER^L. Science 270:811-815[Abstract/Free Full Text].

GEKAKIS, N., D. STAKNIS, H. B. NGUYEN, F. C. DAVIS, and L. D. WILSBACHER et al., 1998  Role of the CLOCK protein in the mammalian circadian mechanism. Science 280:1564-1569[Abstract/Free Full Text].

HALL, J. C., 1998  Genetics of biological rhythms in Drosophila.. Adv. Genet. 38:135-184[Medline].

HAMBLEN, M. J., N. E. WHITE, P. T. J. EMERY, K. KAISER, and J. C. HALL, 1998  Molecular and behavioral analysis of four period mutants in Drosophila melanogaster encompassing extreme short, novel long, and unorthodox arrhythmic types. Genetics 149:165-178[Abstract/Free Full Text].

HARDIN, P., 1998  Activating inhibitors and inhibiting activators: a day in the life of a fly. Curr. Opin. Neurobiol. 8:642-647[Medline].

JIN, X., L. SHEARMAN, D. WEAVER, M. ZYLKA, and G. DE VRIES et al., 1999  A molecular mechanism regulating output from the suprachiasmatic circadian clock. Cell 96:57-68[Medline].

KING, D. P., M. H. VITATERNA, A. M. CHANG, W. F. DOVE, and L. H. PINTO et al., 1997a  The mouse Clock mutation behaves as an antimorph and maps within the W19H deletion, distal of Kit.. Genetics 146:1049-1060[Abstract].

KING, D. P., Y. ZHAO, A. M. SANGORAM, L. D. WILSBACHER, and M. TANAKA et al., 1997b  Positional cloning of the mouse circadian Clock gene. Cell 89:641-653[Medline].

KLOSS, B., J. L. PRICE, L. SAEZ, J. BLAU, and A. ROTHENFLUH et al., 1998  The Drosophila clock gene double-time encodes a protein closely related to human casein kinase I. Cell 94:97-107[Medline].

KONOPKA, R., C. PITTENDRIGH, and D. ORR, 1989  Reciprocal behaviour associated with altered homeostasis and photosensitivity of Drosophila clock mutants. J. Neurogenet. 6:1-10[Medline].

KONOPKA, R. J. and S. BENZER, 1971  Clock mutants of Drosophila melanogaster.. Proc. Natl. Acad. Sci. USA 68:2112-2116[Abstract/Free Full Text].

LAKIN-THOMAS, P. L. and S. BRODY, 1985  Circadian rhythm mutations in Neurospora crassa: interactions between clock mutations. Genetics 109:49-66[Abstract/Free Full Text].

LEWIS, E. B. and F. BACHER, 1968  Method for feeding ethyl methane-sulfonate (EMS) to Drosophila males. Dros. Inf. Serv. 43:193.

LOROS, J. J., A. RICHMAN, and J. F. FELDMAN, 1986  A recessive circadian clock mutation at the frq locus of Neurospora crassa.. Genetics 114:1095-1110[Abstract/Free Full Text].

MARRUS, S. B., H. ZENG, and M. ROSBASH, 1996  Effect of constant light and circadian entrainment of per^S flies: evidence for light-mediated delay of the negative feedback loop in Drosophila.. EMBO J. 15:6877-6886[Medline].

MATSUMOTO, A., K. TOMIOKA, Y. CHIBA, and T. TANIMURA, 1999  tim^rit lengthens circadian period in a temperature-dependent manner through suppression of PERIOD protein cycling and nuclear localization. Mol. Cell. Biol. 19:4343-4354[Abstract/Free Full Text].

MULLER, H. J., 1932  Further studies on the nature and causes of gene mutations. Proc. Sixth Int. Congr. Genet. 1:213.

MYERS, M. P., K. WAGER-SMITH, C. S. WESLEY, M. W. YOUNG, and A. SEHGAL, 1995  Positional cloning and sequence analysis of the Drosophila clock gene, timeless.. Science 270:805-808[Abstract/Free Full Text].

MYERS, M. P., K. WAGER-SMITH, A. ROTHENFLUH-HILFIKER, and M. W. YOUNG, 1996  Light-induced degradation of TIMELESS and entrainment of the Drosophila circadian clock. Science 271:1736-1740[Abstract].

NAIDOO, N., W. SONG, M. HUNTER-ENSOR, and A. SEHGAL, 1999  A role for the proteasome in the light response of the Timeless clock protein. Science 285:1737-1741[Abstract/Free Full Text].

PRICE, J. L., J. BLAU, A. ROTHENFLUH, M. ABODEELY, and B. KLOSS et al., 1998  double-time is a novel Drosophila clock gene that regulates PERIOD protein accumulation. Cell 94:83-95[Medline].

ROTHENFLUH, A., M. W. YOUNG, and L. SAEZ, 2000  A TIMELESS-independent function for PERIOD proteins in the Drosophila clock. Neuron 26:505-514[Medline].

RUTILA, J. E., H. ZENG, M. LE, K. D. CURTIN, and J. C. HALL et al., 1996  The tim^SL mutant of the Drosophila rhythm gene timeless manifests allele-specific interactions with period gene mutants. Neuron 17:921-929[Medline].

RUTILA, J. E., O. MALTSEVA, and M. ROSBASH, 1997  The tim^SL mutant affects a restricted portion of the Drosophila melanogaster circadian cycle. J. Biol. Rhythms 13:380-392.

RUTILA, J. E., V. SURI, M. LE, W. V. SO, and M. ROSBASH et al., 1998  CYCLE is a second bHLH-PAS clock protein essential for circadian rhythmicity and transcription of Drosophila period and timeless.. Cell 93:805-814[Medline].

SAEZ, L. and M. W. YOUNG, 1996  Regulation of nuclear entry of the Drosophila clock proteins Period and Timeless. Neuron 17:911-920[Medline].

SEHGAL, A., J. L. PRICE, B. MAN, and M. W. YOUNG, 1994  Loss of circadian behavioral rhythms and per RNA oscillations in the Drosophila mutant timeless.. Science 263:1603-1606[Abstract/Free Full Text].

SURI, V., A. LANJUIN, and M. ROSBASH, 1999  TIMELESS-dependent positive and negative autoregulation in the Drosophila circadian clock. EMBO J. 18:675-686[Medline].

VITATERNA, M. H., D. P. KING, A.-M. CHANG, J. M. KORNHAUSER, and P. L. LOWREY et al., 1994  Mutagenesis and mapping of a mouse gene, Clock, essential for circadian behavior. Science 264:719-725[Abstract/Free Full Text].

YOUNG, M. W., 1998  The molecular control of circadian behavioral rhythms and their entrainment in Drosophila.. Annu. Rev. Biochem. 67:135-152[Medline].

ZENG, H., Z. QIAN, M. P. MYERS, and M. ROSBASH, 1996  A light-entrainment mechanism for the Drosophila circadian clock. Nature 380:129-135[Medline].

ZERR, D. M., J. C. HALL, M. ROSBASH, and K. K. SIWICKI, 1990  Circadian fluctuations of period protein immunoreactivity in the CNS and the visual system of Drosophila.. J. Neurosci. 10:2749-2762[Abstract].

This article has been cited by other articles:

				Z. Liu, X. Wang, Y. Yu, X. Li, T. Wang, H. Jiang, Q. Ren, Y. Jiao, A. Sawa, T. Moran, et al. A Drosophila model for LRRK2-linked parkinsonism PNAS, February 19, 2008; 105(7): 2693 - 2698. [Abstract] [Full Text] [PDF]

				R. Gunawan and F. J. Doyle III Phase Sensitivity Analysis of Circadian Rhythm Entrainment J Biol Rhythms, April 1, 2007; 22(2): 180 - 194. [Abstract] [PDF]

				C. I. Hong, E. D. Conrad, and J. J. Tyson A proposal for robust temperature compensation of circadian rhythms PNAS, January 23, 2007; 104(4): 1195 - 1200. [Abstract] [Full Text] [PDF]

				J. C. Hall, D. C. Chang, and E. Dolezelova Principles and Problems Revolving Round Rhythm-related Genetic Variants Cold Spring Harb Symp Quant Biol, January 1, 2007; 72(0): 215 - 232. [Abstract] [PDF]

				S. A. Cyran, G. Yiannoulos, A. M. Buchsbaum, L. Saez, M. W. Young, and J. Blau The Double-Time Protein Kinase Regulates the Subcellular Localization of the Drosophila Clock Protein Period J. Neurosci., June 1, 2005; 25(22): 5430 - 5437. [Abstract] [Full Text] [PDF]

				K. D. Edwards, J. R. Lynn, P. Gyula, F. Nagy, and A. J. Millar Natural Allelic Variation in the Temperature-Compensation Mechanisms of the Arabidopsis thaliana Circadian Clock Genetics, May 1, 2005; 170(1): 387 - 400. [Abstract] [Full Text] [PDF]

				C. Wulbeck, G. Szabo, O. T. Shafer, C. Helfrich-Forster, and R. Stanewsky The Novel Drosophila timblind Mutation Affects Behavioral Rhythms but Not Periodic Eclosion Genetics, February 1, 2005; 169(2): 751 - 766. [Abstract] [Full Text] [PDF]

				J. S. Takahashi Finding New Clock Components: Past and Future J Biol Rhythms, October 1, 2004; 19(5): 339 - 347. [Abstract] [PDF]

				L. J. Ashmore, S. Sathyanarayanan, D. W. Silvestre, M. M. Emerson, P. Schotland, and A. Sehgal Novel Insights into the Regulation of the Timeless Protein J. Neurosci., August 27, 2003; 23(21): 7810 - 7819. [Abstract] [Full Text] [PDF]

				T. Stempfl, M. Vogel, G. Szabo, C. Wulbeck, J. Liu, J. C. Hall, and R. Stanewsky Identification of Circadian-Clock-Regulated Enhancers and Genes of Drosophila melanogaster by Transposon Mobilization and Luciferase Reporting of Cyclical Gene Expression Genetics, February 1, 2002; 160(2): 571 - 593. [Abstract] [Full Text] [PDF]

				S. Bao, J. Rihel, E. Bjes, J.-Y. Fan, and J. L. Price The Drosophila double-timeS Mutation Delays the Nuclear Accumulation of period Protein and Affects the Feedback Regulation of period mRNA J. Neurosci., September 15, 2001; 21(18): 7117 - 7126. [Abstract] [Full Text] [PDF]