Outline
Graphics
The frequency (frq) locus of Neurospora crassa was originally identified in searches for loci encoding components of the circadian clock.The frq gene is now shown to encode a central component in a molecular feedback loop in which the product of frq negatively regulated its own transcript, which resulted in a daily oscillation in the amount of frq transcript. Rhythmic messenger RNA expression was essential for overt rhythmicity in the organism and no amount of constitutive expression rescued normal rhythmicity in frq loss-of-function mutants. Step reductions in the amount of FRQ-encoding transcript set the clock to a specific and predicted phase. These results establish frq as encoding a central component in a circadian oscillator.
Circadian rhythms are widespread phylogenetically [1] and are essential for proper temporal regulation of a variety of phenomena, ranging from microbial growth and development [2], to plant and animal reproduction, and to human sleep-wake cycles and cognitive function [3]. The cellular machinery that generates this capability is known as the biological clock. Such clocks are intrinsic to the cell [2,4,5], are endogenous and self-sustaining, and can be reset in response to time cues such as daily light-dark or temperature cycles [1].
The molecular mechanisms that constitute clocks are not well described. Therefore, investigators have searched for clock components in genetically and biochemically manipulable microbial systems. Much effort has focused on dissection of the circadian system of Neurospora crassa, a filamentous fungus in which the clock regulates a clearly expressed overt rhythm in developmental potential and morphology [2,6]. This rhythmicity is manifest through the cyclical production of asexual spores known as conidia. Genetic approaches to the identification of components of the clock in N. crassa have identified seven loci, one of which, frequency (frq), was repeatedly identified. The frq gene is a likely candidate for encoding a central component of the oscillator: strains carrying different frq alleles are differentially altered in multiple clock properties, including period length and temperature compensation (the ability to maintain period length over a wide range of physiological temperatures), but are otherwise generally morphologically and developmentally normal. Cloning of the locus [7] allowed identification of a long open reading frame [8,9] that is phylogenetically conserved [8] and that encodes the protein FRQ, the key transacting product of the frq gene [9,10]. A frameshift mutation within the open reading frame mimics the deletion of frq and results in complete loss of stable rhythmicity and compensation [9]. These observations are compatible with the inclusion of frq in the feedback loop that comprises the clock.
A tenet of chronobiology is that feedback is central to the generation of circadian rhythms [11]. However, although some biochemical and genetic feedback oscillator will likely be a part of the clock, the simple description of a feedback loop does not constitute proof of its involvement in the clock [2]. Criteria have thus been established for candidate clock components and oscillators [12]. Briefly, (i) mutations in the component should affect canonical clock properties and null mutations should abolish normal rhythmicity; (ii) the amount (activity) of the component must oscillate in a self-sustained manner with an appropriate periodicity; (iii) induced changes in the amount (activity) of the component must (by feedback) act to change the amount (activity) of the component; (iv) the phase of the component's oscillation must be reset by shifts in the light-dark growth regimen, and conversely, the overt rhythm must be reset by changes in the amount (activity) of the component; and (v) prevention of the component's oscillation should result in loss of the overt rhythm. In particular, there should be no degree of constitutive expression that will support rhythmicity. Thus, to establish the identity of a clock component, it is necessary to demonstrate that an oscillation in the component is necessary for the overt rhythm, that the activity of the component feeds back to affect its own activity, and that step changes in the components activity set the phase of the clock.
Two candidate clock genes fulfill several of the criteria listed above, frq and the Drosophila gene period [13]. Although feedback regulation of these genes was suspected, the first reported cycling of a candidate clock component involved the PER protein [14], which was then extended to the per transcript [15]. These data establish the potential of feedback, but this was not explicitly shown and the other points have not been established. Particularly, in no case has cycling of a putative clock component been established as necessary for circadian function.
We now show the existence of a molecular feedback loop in which the product of frq satisfies all of the criteria expected for a central clock component. The frq gene encodes a product that negatively feeds back to regulate its own transcript, resulting in a daily oscillation of frq transcript amount with a peak in the subjective morning. Because frq and its product function within individual cells or syncytia rather than intercellularly, these results establish frq as encoding a central component, a state variable, in a cellular circadian oscillator.
Rhythmic expression of frq mRNA.^
Northern (RNA) analysis was done to examine frq mRNA at 4-hour intervals over 48 hours under constant conditions Figure 1A. In a clock wild-type genetic background [16], the transcript showed a rhythmic accumulation that peaked at 12 to 16 and 32 to 36 hours in the dark. This corresponds to circadian time (CT) 0 to 5, that is, subjective morning [17]. Conversely, the trough in mRNA abundance was at 0 to 4 hours and again at 24 to 28 hours in the dark (subjective early evening). The period length of the mRNA rhythm was approximately 22 hours, the same period as the N. crassa conidiation rhythm [2]. If frq cycling is regulated by the cellular oscillator, then the period length of the frq RNA rhythm should reflect the period length of the overt conidiation rhythm. This was shown to be the case when RNA isolated from a frq7 strain (29-hour period length) was used in an identical RNA hybridization experiment Figure 1B. Under the culture conditions used, the amount of frq transcript at the peak was about 5 to 10 times the amount at the trough in both frq sup ``plus'' and frq7, although expression in frq7 was reproducibly greater than in frq sup ``plus''. The amplitude and phase of the RNA rhythm of frq was similar to that of two other N. crassa clock-controlled genes, ccg-1 and ccg-2 [18]. Thus, frq is a morning-specific clock-controlled gene that acts to control the period length of the clock, and thus the timing of its own expression.
If frq RNA cycling is necessary for overt rhythmicity, then mutations in frq that result in loss of normal rhythmicity should result in loss of frq RNA cycling. This was confirmed with RNA harvested from a strain bearing frq9, a recessive loss-of-function allele that displays reduced conidiation and no stable circadian rhythmicity [19]. The pattern of frq mRNA expression in this strain was distinct from the frq sup ``plus'' and frq7 cases; frq RNA did not cycle in the frq9 background, although the amounts did fluctuate Figure 1B in a manner reminiscent of the unstable conidial rhythms observed in such frq null backgrounds [9,19]. In frq9, frq mRNA was two to three times more plentiful than the peak amounts in a frq sup ``plus'' strain Figure 1B. The observation that frq mRNA in a loss-of-function mutant did not show a stable rhythm and was elevated as compared to the peak of wild-type expression was consistent with a model in which the cycling of the frq transcript is the result of negative feedback regulation, in a manner similar to that suggested for the per transcript [14,15]. Here then, the protein FRQ appears to act directly or indirectly to depress the abundance of its own mRNA, thus creating a feedback loop with a period length of approximately 1 day.
Because frq was a central component in this feedback loop, and mutations in frq lead to alterations in the circadian clock [2], frq is a candidate for being or encoding a state variable in the circadian oscillator. However, these and similar data [15,20] are equally compatible with genes such as frq providing a necessary support function to a separate oscillator, analogous to an enzyme supplying a critical substrate for a clock protein or a channel required to maintain a necessary intracellular ionic environment. If frq is a central component of the clock rather than an auxiliary gene, then normal rhythmic expression should be required for overt rhythmicity. Three predictions follow from this: (i) just as loss-of-function mutations eliminate normal circadian rhythmicity, constitutive elevated expression from a heterologous inducible frq gene should eliminate rhythmicity by silencing expression from the endogenous, feedback-regulated copy, thereby short-circuiting the autoregulatory loop; (ii) constitutive frq expression, in any amount, should not rescue overt rhythmicity in a frq-null strain; (iii) the phase of the rhythmic expression of frq must define the phase of the overt rhythm. Thus, in a frq sup ``plus'' strain bearing a second inducible copy of a FRQ-encoding gene, a step from constitutive elevated inducible frq expression to the normal feedback-regulated amount of frq mRNA should reset the clock to a phase corresponding to the low point in the normal oscillation. Each of these three predictions was tested.
Requirement for oscillating quantities of frq mRNA.^
The prediction that frq RNA cycling, rather than simply its expression, is necessary for proper operation of the circadian clock was tested by fusion of the heterologous, inducible N. crassa promoter, qa-2 [21], to the frq open reading frame Figure 2A [22]. A construct, designated qa-2pFRQ, was made in which expression of the FRQ open reading frame was dependent on the presence of an inducer, quinic acid. The qa-2pFRQ construct was transformed [23] into a frq sup ``plus'' strain and examined for a conditional quinic acid-dependent phenotype by the race tube assay Figure 2B. Nine transformed strains were extensively analyzed [24] to test if constitutive expression of FRQ-encoding RNA eliminated overt rhythmicity. All nine were internally consistent: frq sup ``plus'' strains transformed with the qa-2pFRQ construct showed wild-type rhythmicity in the absence of the inducer; however, in the presence of 10 sup ``minus''5 M quinic acid Figure 2B or greater (up to 10 sup ``minus''2 M was tested), no rhythmic conidial banding was observed in any of nine independent qa-2pFRQ transformants. Instead there was constant conidiation down the length of the race tube. This confirms the first prediction, that constitutive elevated expression of FRQ-encoding RNA should result in arrhythmicity.
The threshold concentration of inducer that was required to observe the constant conidiation phenotype was determined. Complete arrhythmicity of all frq sup ``plus'' (qa-2pFRQ) isolates [24] Figure 2B was observed at quinate concentrations (10 sup ``minus''5 M) that induced the heterologous qa-2pFRQ transcript to concentrations lower than the trough amounts of endogenous frq. The phenotypic response to increasing amounts of inducer is shown for one representative strain, frq sup ``plus'' (qa-2pFRQec#10) Figure 2B. In some cases, at inducer concentrations below 10 sup ``minus''5 M, several days of growth on the race tube were required before arrhythmic constant conidiation was evident. At inducer concentrations below the threshold value required to observe complete arrhythmicity, the period length and phase of conidial banding was unaffected by the inducer.
To test the second prediction, frq9 null mutants transformed with the qa-2pFRQ construct were examined for inducer-dependent rescue of normal rhythmicity [23,24]. Four his-3-targeted transformants and four ectopic insertion strains were examined for conditional rhythmicity by serially diluting the inducer over a range of concentrations (10 sup ``minus''2 M to 10 sup ``minus''7 M) that spanned the dynamic range of the promoter [21,25]; rhythmicity was not restored under these conditions. The four his-3-targeted strains [24] were then subjected to a more refined analysis. The inducer concentration was gradually raised in twofold steps, from a basal concentration of 3 ``times'' 10 sup ``minus''7 M quinic acid to 10 sup ``minus''5 M; this covered the entire range of inducer in which the conditional phenotype of frq sup ``plus'' (qa-2pFRQhis-3) strains was observed. There were no concentrations of inducer at which stable overt rhythmicity was seen Figure 2B. These data suggest that the lack of complementation of the frq9 phenotype by the qa-2pFRQ construct results from a lack of rhythmic expression of the mRNA, consistent with the second prediction that rhythmic expression is essential for overt rhythmicity.
Feedback repression by qa-2pFRQ.^
The effect of expression of the inducible frq gene on endogenous frq mRNA was examined in both a frq sup ``plus'' and a frq9 background Figure 3. At 1.5 ``times'' 10 sup ``minus''2 M quinic acid and 2 percent glucose the inducible transcript was present and expression of the endogenous gene was depressed by approximately 70 percent; this concentration of inducer had no effect on the frq mRNA in the frq sup ``plus'' control Figure 3A. A sharper but similar repression of frq mRNA by induction of qa-2pFRQ was observed in the frq9 background Figure 3B although higher concentrations of inducer were needed. The endogenous transcript was not detected at the highest inducing conditions (1.5 ``times'' 10 sup ``minus''2 M quinic acid, no glucose) in any of the three frq9 (qa-2pFRQhis-3) isolates examined. The higher concentrations of inducer necessary to observe the repression of the endogenous frq mRNA in the frq9 background may be related to the higher endogenous frq expression in the frq9 strain Figure 1B. Also, unlike the frq sup ``plus'' case, there was no contribution to overall functional FRQ concentrations by the endogenous frq9 gene. Induction of frq from a heterologous promoter thus acted to repress expression of the endogenous copy, a finding consistent with the frq9 rhythmic Northern blot result Figure 1B and in support of the interpretation that frq is regulating its own expression. The frq mRNA is part of a feedback loop, apparently with the protein FRQ acting in a negative fashion to control the abundance of its own mRNA.
Setting the phase of the rhythm with step changes of inducer.^
The third prediction of the feedback model was that, in a frq sup ``plus'' strain bearing an inducible copy of qa-2pFRQ, a step from constitutive induced frq expression to the normal, feedback-regulated amounts, should reset the phase of the clock. The new phase established at the time of inducer withdrawal should correspond to the low point in the normal frq mRNA oscillation (CT 12, subjective evening). Additionally, it was important to eliminate the possibility that the elevated expression of frq was simply inducing conidiation, thereby masking the normal expression of the clock. To test the third prediction and to rule out the masking possibility, liquid culture experiments [26] were done. This enabled the addition and subsequent withdrawal of the inducer from the culture under conditions in which the clock could be monitored Figure 4. The experimental genotype and condition was frq sup ``plus'' (qa-2pFRQ) grown in medium containing quinic acid. The controls were frq sup ``plus'' grown in the presence or absence of quinic acid and frq sup ``plus'' (qa-2pFRQ) grown in the absence of quinic acid. Groups of all four cultures were transferred from light to dark at successive 5-hour intervals, and then all of the cultures were washed in fresh medium to remove the inducer and transferred to race tubes to measure the phase of the rhythm [26,27]. The phase of control cultures [frq sup ``plus'' strain with and without inducer as well as the frq sup ``plus'' (qa-2pFRQ) strain with no inducer] was set by the light to dark transition, as shown by the 5-hour staggered phasing of the banding of the successive light-dark cultures Figure 4A. The final phase of the inducible frq strain, however, was not set by light. Instead, all of these cultures, regardless of the time of light to dark transfer, recovered from the quinic acid treatment in synchrony at an identical phase, the inducer release-driven phase Figure 4B. This shows that the withdrawal of inducer from the media, not the light to dark transition, dictated the phase of the rhythm in the experimental strain; this would not be the case if elevated frq was simply acting to mask the normal rhythm. Because the center of a conidial band occurs at approximately CT 0 [28], the phase of the rhythm at the time of quinic acid release could be calculated in independent experiments to fall between CT 9 and CT 11.5 (subjective evening), close to the observed minimum in the normal frq oscillation, and at the phase predicted by a feedback model. Thus, induction of the qa-2pFRQ gene fusion strongly affected the phase of the rhythm; the phase was fully set by the withdrawal of the inducer, rather than the light-dark transfer. The phase was set to the early evening hours, similar to the phase established by a light to dark transfer. These data confirm the third prediction.
Increasing amounts of frq mRNA in the frq sup ``plus'', frq7, and frq9 allelic series is consistent with the proposal that long period length mutants are hypomorphic [9]. Thus, one function of frq is to regulate its own expression. Loss of this function results in runaway expression of frq mRNA as shown in the frq9 case. From this perspective, the frq7 mutation would result in a protein with partial loss of this autoregulatory function. More mRNA must accumulate before the requisite amount of protein activity is produced for autoregulation, resulting in a longer cycle (29 hours instead of 21.5 hours).
The phase of peak frq mRNA occurs at CT 0 to 5, which corresponds to subjective early morning. This coincides with peak mRNA amounts of the morning-specific clock-controlled genes targeted for regulation by the clock: ccg-1, ccg-2 [18,30], ccg-4 [31], and ccg-7 [32]. It is not clear whether FRQ is acting directly or indirectly (through the clock) to regulate these clock-controlled genes. The similarity of peak times of mRNA accumulation is consistent with rapid translation of frq mRNA and the subsequent direct action of FRQ upon expression of the clock-controlled genes. Because the products of several morning-specific ccg's have been associated with the conidiation process [31,33], a developmental pattern regulated by the clock, the observation that constitutive elevated expression of frq tended to promote more conidiation Figure 2 is consistent with a direct role of FRQ in activating these genes. Alternatively, the near coincidence of the peaks of frq and ccg expression could reflect the simultaneous action of other factors that promote the expression of both frq and the clock-controlled genes.
The time of day at which steady-state amounts of frq mRNA reach a peak coincides with the phase at which the N. crassa clock shows maximum sensitivity to inhibitors of transcription and translation [27,34,35]. This suggests that it may be the RNA and protein products of frq that are the direct targets of these clock-resetting agents and that the generalized resetting effects of chemicals such as these [36] may be visualized in terms of their effects in disrupting a feedback loop. Two effects seen at low frq induction in the frq sup ``plus'' background, the gradual loss of rhythmicity observed during continuous exposure to low concentrations of inducer and the lack of period effects, remain unexplained. It may be that FRQ is quite unstable and requires more than a day to reach an effective concentration. Although effects on the period might have been expected based on per gene dosage studies in Drosophila [37], N. crassa heterokaryons carrying a nuclear ratio of 10 percent frq sup ``plus'' and 90 percent frq9 have normal period lengths [19]. Generally, dosage effects on period length are not required by all multicomponent mathematical models describing temperature-compensated oscillators.
On the basis of our data on the regulation of frq, we propose a model of the feedback oscillator generating the N. crassa circadian rhythm Figure 5 in which frq transcript and its product are envisioned as components, state variables, of the clock. The clock is functional in single germinated conidiospores on both solid and liquid media [38] and is thus a product of intracellular regulation. The frq gene and product have multiple roles; although autoregulation of frq transcript is seen as central to the clock mechanism and our data confirm the importance of transcript cycling for the rhythm, they do not yet prove that this cycling is derived solely through regulation of the frq promoter as suggested in Figure 5. In no case is the actual number of regulatory steps within the clock or from the clock to a controlled process known; specifically the existence of other state variables within the clock is anticipated [39]. Although all transcriptionally regulated clock-controlled genes will possess circadian clock responsive elements (CCRE's) Figure 5 [2,5], these DNA elements may be different for different genes activated at different times. Several of the clock-regulated genes are also regulated by light, both acutely and indirectly through light regulation of the clock [4,31,33,40].
Despite these distinctions, however, several important common features describing circadian rhythms and clock genes remain. Both of the genetically pertinent clock genes, frq and per, are self-regulating in terms of their transcripts, and in the case of frq we have shown that this self-regulation naturally follows from the position of FRQ within the loop comprising the clock. In frq and per, these rhythms in transcript cycling are associated with, and at least for frq, essential for, the overt rhythms. Other regulation, post-transcriptional and translational [44], will surely be required for the assembly of a compensated, entrainable clock, as these transcripts are turned into trans-active proteins that are, in time, turned over. These transcript and protein cycles provide a ready explanation for the universal effectiveness of transcriptional and translational inhibitors in clock resetting. This universality, in turn, suggests that the clock-gene transcript and protein cycling seen and implied here may be a universal feature of circadian oscillators.
REFERENCES AND NOTES^
1. L. E. Edmunds Jr., Cellular and Molecular Bases of Biological Clocks (Springer-Verlag, New York, 1988). [Context Link]
2. J. C. Dunlap, Annu. Rev. Physiol. 55, 683 (1993); ExternalResolverBasic Bibliographic Links Trends Genet. 6, 159 (1990); P. Lakin-Thomas, G. Cote, S. Brody, CRC Crit. Rev. Microbiol. 17, 365 (1990). [Context Link]
3. M. C. Moore-Ede, F. M. Sulzman, C. A. Fuller, The Clocks That Time Us (Harvard Univ. Press, Cambridge, MA, 1982). [Context Link]
4. J. W. Hastings, Annu. Rev. Microbiol. 13, 297 (1959); S. Michel, M. E. Geusz, J. J. Zaritsky, G. D. Block, Science 259, 239 (1993). [Context Link]
5. J. Takahashi, Curr. Opin. Genet. Dev. 3, 301 (1993). ExternalResolverBasic Bibliographic Links [Context Link]
6. J. F. Feldman and M. Hoyle, Genetics 75, 605 (1973). ExternalResolverBasic Bibliographic Links [Context Link]
7. C. R. McClung, B. A. Fox, J. C. Dunlap, Nature 339, 558 (1989). [Context Link]
8. M. Merrow and J. C. Dunlap, EMBO J., in press. [Context Link]
9. B. D. Aronson, K. A. Johnson, J. C. Dunlap, Proc. Natl. Acad. Sci. U.S.A., in press (1994). [Context Link]
10. At least two nonoverlapping divergent transcripts map to the frq locus. A smaller transcript (1.5 kb) has no extended open reading frame and no clock-relevant functions for this transcript have been uncovered (K. A. Johnson and J. C. Dunlap, unpublished observations). The central role of the larger transcripts (``approximate''4.5 kb) in the circadian clock has been established [9]. The transcriptional product of frq referred to in our work is the ``approximate''4. 5-kb clock-relevant transcript. [Context Link]
11. T. Pavlidis and W. Kauzman, Arch. Biochem. Biophys. 132, 338 (1969); ExternalResolverBasic Bibliographic Links B. C. Goodwin, Temporal Organization in Cells (Academic Press, London, 1963). [Context Link]
12. M. Zatz, Discoveries Neurosci. 8, 67 (1992); D. J. Chadwick and K. Ackrill, Eds., Circadian Clocks and Their Adjustment (CIBA Foundation Symposium 183, Wiley, Chichester, UK, in press). [Context Link]
13. R. J. Konopka and S. Benzer, Proc. Natl. Acad. Sci. U.S.A. 68, 2112 (1971). [Context Link]
14. K. K. Siwicki, C. Eastman, G. Petersen, M. Rosbash, J. C. Hall, Neuron 1, 141 (1988); ExternalResolverBasic Bibliographic Links D. M. Zerr, J. C. Hall, M. Rosbash, K. K. Siwicki, J. Neurosci. 10, 2749 (1990). [Context Link]
15. P. E. Hardin, J. C. Hall, M. Rosbash, Nature 343, 536 (1990). ExternalResolverBasic Bibliographic Links [Context Link]
16. The standard clock wild-type genetic background used in all experiments includes the band (bd) mutation that permits clear observation of the conidial banding phenotype. This mutation has no effect on the clock itself. [Context Link]
17. The concept of circadian time (abbreviated CT) has been formulated to allow circadian pacemakers with different endogenous periodicities to be compared. Biological time is thus normalized by dividing the circadian cycle into 24 equal parts, each a circadian hour. By convention, CT 0 is subjective dawn and CT 12 corresponds to the end of the light period in a 12:12 light:dark cycle or to the time of transfer from constant light to constant darkness. CT 6 is thus always midcircadian day and CT 18 is midnight. [Context Link]
18. J. J. Loros, S. A. Denome, J. C. Dunlap, Science 243, 385 (1989). [Context Link]
19. J. J. Loros and J. F. Feldman, J. Biol. Rhythms 1, 187 (1986); J. J. Loros, A. Richman, J. F. Feldman, Genetics 114, 1095 (1986). [Context Link]
20. I. Edery, J. E. Rutila, M. Rosbash, Science 263, 237 (1994). [Context Link]
21. N. H. Giles et al., Microbiol. Rev. 49, 338 (1985). ExternalResolverBasic Bibliographic Links [Context Link]
22. An exact fusion between the qa-2 promoter and the frq open reading frame was created by polymerase chain reaction (PCR) [R. M. Horton, H. D. Hunt, S. N. Ho, J. K. Pullen, L. R. Pease, Gene 77, 61 (1989)] such that the nucleotide sequence that was downstream of the initiating methionine residue was from frq and the sequences upstream belonged to qa-2. Oligonucleotides were used to amplify the qa-2 promoter and 5``prime'' untranslated region: BDA14, CATTGTGTTTGGTACCTCTG (qa-2 ATG, coordinates ``plus''3 to ``minus''16); and BDA15, GCTCTAGAGTCCTTGACAAGCAGT, qa-2 upstream, coordinates ``minus''1131 to ``minus''1112, and a small region of the frq open reading frame; BDA16, GTACCAAACACAATGGCGGATAGTGGGGATAA, frq ATG with qa-2 overlap, coordinates ``plus''1 to ``plus''20; and BDA17, CTCCTCTGGGATGTCGATTC, frq COOH-terminal, coordinates 641 to 660. The bold sequence corresponds to the initiating methionine of either qa-2 (BDA14) or frq (BDA16). All coordinates are listed using the A of the initiating ATG as ``plus''1. The primer at the frq ATG was designed to include 12 bp of the qa-2 promoter region, providing overlap for the second round of PCR. The spliced product from the second round of PCR was then subcloned into pSKII sup ``minus'' (Stratagene) and further subcloned into a vector carrying the entire frq open reading frame. This clone, designated pBA40, has the entire frq open reading frame under qa-2 promoter control. The pBA40 insert was further subcloned into vector pDE3 [23], compatible with targeted transformations at the his-3 locus, and designated pBA50. All regions of pBA40 and pBA50 that had been subjected to PCR were sequenced. A single base pair alteration (C to T at coordinate 2249) in the qa-2 promoter was detected; this coincided with the most distal and weakest of the four qa-1F binding sites. This alteration is predicted to have no effect on quinic acid induction because in the wild-type case this position is nonconsensus. The inducer, quinic acid, is a well characterized utilizable carbon source for wild-type N. crassa. In the presence of glucose, quinic acid can be considered nutritionally gratuitous and at the concentrations used here its only effects should be submaximal induction of the elements of the quinic acid metabolic pathway as well as the qa-2pFRQ. The qa-2 promoter can be induced up to 500 times by growth in the presence of quinic acid [21]; however, because the promoter is catabolite-repressed, induction of only 25 to 50 times basal amounts are achieved when both glucose and quinic acid are present [21]. All experiments were done in the presence of glucose unless otherwise noted. [Context Link]
23. In N. crassa, transforming DNA can be introduced at random ectopic sites in the genome [S. J. Vollmer and C. Yanofsky, Proc. Natl. Acad. Sci. U.S.A. 83, 4869 (1986) ExternalResolverBasic Bibliographic Links] or targeted to the his-3 locus [D. Ebbole, Fungal Genet. Newsl. 37, 15 (1990)] in order to eliminate chromosomal positioning effects. Transformants of both types were used here. The formal strain designations for transformants in this study are: bd; (qa-2pFRQ)ec pBA40 for random ectopic insertions of the qa-2pFRQ construct into a frq sup ``plus'' strain, or bd; his-3(his-3 sup ``plus'' qa-2pFRQ)hom pBA50 for his-3-targeted insertions. For convenience, a shorthand notation is used. The ectopic strains are designated frq sup ``plus'' (qa-2pFRQec) and his-3-targeted strains are designated frq sup ``plus'' (qa-2pFRQhis-3). Similarly the formal frq9 designations are: bd;frq9; (qa-2pFRQ)ec pBA40 and bd;frq9; his-3 (his-3 sup ``plus'' qa-2pFRQ)hom pBA50 but the shorthand notation is frq9 (qa-2pFRQec) and frq (9) (qa-2pFRQhis-3) respectively. Where appropriate, individual transformant strains are identified by an isolate number that follows the name. [Context Link]
24. Nine frq sup ``plus'' (qa-2pFRQ) isolates were extensively studied; five of these were ectopic insertions and four were his-3-targeted. The pBA40 construct was introduced into N. crassa by cotransformation with a BmlR plasmid pSV50.These ectopic BmlR primary pBA40 transformants were analyzed on race tubes in the absence of inducer (QA) as well as the presence of 10 sup ``minus''2 M QA. In the case of frq sup ``plus'' recipients, the conditional loss of rhythmicity was apparent and occurred in 11 of 80 BmlR transformants. For targeted insertions, the pBA50 construct was introduced into the his-3 N. crassa recipient and transformants were selected for histidine prototrophy. Of 11 frq sup ``plus'' recipients that were transformed to His sup ``plus'', six had the conditional arrhythmic phenotype, and all six were shown by Southern (DNA) analysis to carry a single copy of pBA50, appropriately targeted by insertion at his-3. One His sup ``plus'' transformant examined that did not display a conditional phenotype was shown by Southern analysis not to carry transforming DNA, and the other five transformants that did not display the conditional phenotype were not further characterized. The frq9 (qa-2pFRQ (ec)) candidates were screened by PCR for the presence of the qa-2pFRQ gene. His sup ``plus'' frq9 transformants were directly tested by Southern analysis and 6 of 10 showed appropriately targeted transforming DNA; four correctly targeted isolates were chosen for extensive phenotypic analysis. Specifically, all four of the his-3-targeted frq9 strains used in attempts to rescue rhythmicity through constitutive expression of a FRQ-encoding RNA were shown to have a properly targeted qa-2pFRQ construct, and of the three examined by Northern analysis, all produced high amounts of inducible qa-2pFRQ mRNA. [Context Link]
25. J. Chaleff, J. Gen. Microbiol. 81, 357 (1974). ExternalResolverBasic Bibliographic Links [Context Link]
26. H. Nakashima, Plant Cell Physiol. 22, 231 (1981). ExternalResolverBasic Bibliographic Links [Context Link]
27. J. C. Dunlap and J. F. Feldman, Proc. Natl. Acad. Sci. U.S.A. 85, 1096 (1988). ExternalResolverBasic Bibliographic Links [Context Link]
28. S. Dharmananda and J. F. Feldman, Plant Physiol. 63, 1049 (1979). ExternalResolverBasic Bibliographic Links [Context Link]
29. J. J. R. Tyson, in The Molecular Basis of Circadian Rhythms, J. W. Hastings and H.-G. Schweiger, Eds. (Abakon Verlagsgesellschaft, Berlin, 1976) pp. 85-108. [Context Link]
30. J. Loros and J. C. Dunlap, Mol. Cell. Biol. 11, 558 (1991). ExternalResolverBasic Bibliographic Links [Context Link]
31. D. Bell-Pedersen, J. C. Dunlap, J. J. Loros, Genes Dev. 6, 2382 (1992). ExternalResolverBasic Bibliographic Links [Context Link]
32. _____, unpublished observations. [Context Link]
33. F.-R. Lauter and C. Yanofsky, Proc. Natl. Acad. Sci. U.S.A. 90, 8249 (1993); ExternalResolverBasic Bibliographic Links F. Lauter, V. Russo, C. Yanofsky, Genes Dev. 6, 2373 (1992). [Context Link]
34. J. C. Dunlap, S. Crosthwaite, J. J. Loros, J. F. Feldman, in preparation. [Context Link]
35. H. Nakashima, J. Perlman, J. F. Feldman, Am. J. Physiol. 241, R31 (1981). ExternalResolverBasic Bibliographic Links [Context Link]
36. J. W. Jacklet, Science 198, 69 (1977); ExternalResolverBasic Bibliographic Links S. J. Yeung and A. Eskin, J. Biol. Rhythms 3, 225 (1988); S. B. S. Khalsa and G. D. Block, Proc. Natl. Acad. Sci. U.S.A. 89, 10862 (1992); J. C. Dunlap, W. R. Taylor, J. W. Hastings, J. Comp. Physiol. 138, 1 (1980). [Context Link]
37. X. Liu et al., Neuron 6, 753 (1991); R. F. Smith and R. J. Konopka, Mol. Gen. Genet. 185, 30 (1982); M. K. Baylies, T. A. Bargiello, F. R. Jackson, M. W. Young, Nature 326, 390 (1987). [Context Link]
38. K. Lindgren, J. C. Dunlap, J. J. Loros, unpublished observations. [Context Link]
39. Although in these experiments the clock is reset by high amounts of FRQ-encoding transcript to the point predicted if frq and FRQ were the only state variables, these data do not support a contention that feedback between frq and FRQ exclusively comprises the clock. Rather we expect that there are additional state variables yet to be discovered, but this raises the issue of whether the inducer release-driven phase was predictable a priori, because the final phase of the clock after release would also depend on the values of the other state variables at the time the oscillation was reinitiated. Despite this caveat, the fact that the phase was predictable is noted, and we suggest the following interpretation for our data. The extremely strong clock resetting stimulus represented by the high concentration of FRQ may artificially drive not only frq and FRQ, but the other state variables as well, to unique phase points they might assume when frq is fixed at its low point in the normal cycle. The analogy can be drawn to resetting by light. A short pulse of light minutes in length will reset (advance or delay) the clock toward the daytime part of the cycle, no matter where in the cycle the clock is when light is perceived. However, the clock is driven not to any one point in the day, but rather to phases lying broadly within the day--a span of some hours--the final phase depending on where in the cycle the clock was before the pulse. A long pulse of light (over 12 hours), though, will appear to drive the clock to a unique phase point, so that upon transfer to darkness, the clock will begin from subjective dusk (a span of about 1 hour), regardless of where the clock was when the long pulse of light began. In the present case, the ``pulse'' of FRQ was more than 12 hours long so it may be that FRQ is driving all of the state variables rather than just frq, in a manner analogous to the action of light. [Context Link]
40. A. Millar and S. Kay, Plant Cell 3, 541 (1991); G. Arpaia, J. J. Loros, J. C. Dunlap, G. Morelli, G. Macino, Plant Physiol. 102, 1299 (1993). [Context Link]
41. P. E. Hardin, J. C. Hall, M. Rosbash, Proc. Nat. Acad. Sci. U.S.A. 89, 11711 (1992). ExternalResolverBasic Bibliographic Links [Context Link]
42. J. Ewer, M. Rosbash, J. C. Hall, Nature 333, 82 (1988). ExternalResolverBasic Bibliographic Links [Context Link]
43. B. Frisch, P. E. Hardin, M. J. Hamblen-Coyle, M. Rosbash, J. Hall, Neuron, in press (1994). [Context Link]
44. I. Edery, L. Zweibel, M. E. Dembinska, M. Rosbash, Proc. Nat. Acad. Sci. U.S.A., in press. [Context Link]
45. All cultures were grown at 25 ``degrees'' C. The RNA was transferred to a nylon-based membrane (Nitropure, Micron Separation, Inc.) and hybridized with a32 P-labeled frq-specific RNA. The RNA probe was generated by in vitro transcription of a frq template corresponding to a 1.8-kb fragment from the frq open reading frame [coordinates 1389 to 3203, in which 1 corresponds to the A of the initiating ATG of the FRQ protein [9]]. Hybridization was done at 60 ``degrees'' C under standard conditions and blots were washed at high stringency (68 ``degrees'' C, 0. 1 ``times'' SSPE, 0. 1 percent SDS). Northern analysis with ribosomal RNA and sod-1 [P. Chary, R. A. Hallewell, D. O. Natvig, J. Biol. Chem. 265, 18961 (1990)] for normalization of RNA was done in the same manner, except that a DNA probe was used and hybridization was done at 42 ``degrees'' C and washing at 50 ``degrees'' C. For all experiments, multiple autoradiographic exposures were collected so that for any lane or band, several films could be identified in which the grain density was within the linear range of autoradiographic density. For comparisons between gels, autoradiographic exposures of identical duration were collected for blots hybridized with the same probe, and in different experiments, the normalization data with the sod-1 transcript and ribosomal RNA were always collected from the same gel used to quantify frq. The DNA fragment was labeled with random hexamers [A. P. Feinberg and B. Vogelstein, Anal. Biochem. 137, 266 (1983)]. [Context Link]
46. B. D. Aronson, K. M. Lindgren, J. C. Dunlap, J. J. Loros, Mol. Gen. Genet., in press. [Context Link]
47. Race tubes were layered with glucose-arginine media [19] supplemented with the indicated amounts of quinic acid. Media was titrated to pH 6.0 after addition of quinic acid. Strains were inoculated at the left end (as shown) of the tubes with conidia from 4- to 7-day-old cultures. The inoculum was grown in constant light for about 24 hours and the tubes were then moved to constant dark conditions (25 ``degrees'' C). In Figure 2Bthe first hash mark represents the time at which the tubes were transferred from light to dark, and subsequent hash marks were placed at 24-hour intervals there-after. White regions of the race tube represent areas of dense conidiation. Race tube analysis from one representative frq9 (qa-2pFRQ) transformant at two concentrations of inducer are shown in Figure 2B, and results from other concentrations and other strains showed an identical lack of overt rhythmicity. Although recovery of the rhythm was not observed at any inducer concentration in any frq9 (qa-2pFRQhis-3) strain, the elevated expression of frq in these strains was not without effect on the pleiotropic frq9 phenotype. There appeared to be steadier and more robust conidiation in these strains (but not in nontransformed controls) when quinic acid was present. The frq sup ``plus'' control strain showed no evidence of arrhythmicity even at inducer concentrations of 1. 5 ``times'' 10 sup ``minus''2 M. [Context Link]
48. Mycelial cultures from appropriate strains were grown in 2.0 percent glucose (except where noted), 1 ``times'' Vogel's salts, plus various concentrations of inducer. After 15 hours of induction (where appropriate) and 12 hours in the dark (subjective morning, CT 1), the cultures were harvested, RNA was isolated and Northern (RNA) analysis was done as in Figure 1.Because of catabolite repression or inducer exclusion, when cultures were grown in 2. 0 percent glucose, induction of the qa-2pFRQ transcript was limited [21,22]. One of two different riboprobes was used for hybridizations. The riboprobe that detects endogenous frq mRNA, but not qa-2pFRQ mRNA, was derived from an Eco RI-Bgl II fragment (coordinates ``minus''18 to ``minus''1371) that lies entirely within the 5``prime'' untranslated region and that was therefore absent from the qa-2pFRQ construct. The riboprobe that detected both endogenous frq transcript and the qa-2pFRQ transcript was the same probe used in Figure 1RF 43*. [Context Link]
49. Conidia from the experimental frq sup ``plus'' (qa-2pFRQ) and control frq sup ``plus'' or frq sup ``plus'' (qa-2pFRQ) strains were inoculated into 96-well microtiter plates at a concentration of 5 ``times'' 104 conidia per milliliter in 0.1 percent glucose, 0.17 percent arginine, either with 10 sup ``minus''4 M quinic acid or without quinic acid, and then grown for 36 hours in constant light. Mycelial pads (seven for each light to dark transfer) were taken from the microtiter wells and placed in 75 ml of fresh medium in a 15 by 150 petri dish, and the first transfer of samples was made from light to dark (t ``equals'' 0). For each strain and experimental condition, replicate samples were then transferred at successive 5-hour intervals (a total of five light to dark transfers), thus spanning approximately one circadian cycle. Three hours after the last light to dark transfer, the pads were blotted of excess media and were washed in 75 ml of 0.1 percent glucose, 0.17 percent arginine, 1 ``times'' Vogel's salts (pH 6.0) [R. Davis and D. deSerres, Methods Enzymol. 27A, 79 (1970)]. After blotting excess wash media, the pads were inoculated onto race tubes containing the same media used to wash the pads plus 1. 5 percent agar. The sidereal time of the center of conidial bands was determined with an automated program [7] and was then normalized to circadian days to adjust for small differences in period length (``tau'') [in this experiment, ``tau'' of frq sup ``plus'' (qa-2pFRQ) ``equals'' 23. 2 and ``tau'' of frq sup ``plus'' strain ``equals'' 23. 0]. [Context Link]
50. Supported by NIH grant GM 34985 and the National Institute of Mental Health (J.C.D.), the Norris Cotton Cancer Center Core Grant, and NIH fellowship GM14465 (B.D.A.). We thank D. Natvig for sod-1 clones and DNA and G. Block and J. Hall for critical discussion of the data.
