Effects of prd Circadian Clock Mutations on FRQ-Less Rhythms in Neurospora

Effects of prd Circadian Clock Mutations on FRQ-Less Rhythms in Neurospora Sanshu Li 1 and Patricia Lakin-Thomas 2 Department of Biology, York University, Toronto, Canada Abstract Rhythmic conidiation (spore formation) in Neurospora crassa provides a model system for investigating the molecular mechanisms of circadian rhythmicity. A feedback loop involving the frq, wc-1, and wc-2 gene products (FRQ/ WCC) is an important component of the mechanism; however, rhythmic conidiation can still be observed when these gene products are absent. The nature of the oscillator(s) that drives this FRQ-less rhythmicity (FLO) is an important question in Neurospora circadian biology. We have looked for interactions between FRQ/WCC and FLO by assaying the effects on FRQ-less rhythms of mutations known to affect the period in the presence of FRQ. We assayed 4 prd mutations (prd-1, prd-2, prd-3, and prd-4) under 2 conditions in frq null strains: long-period free-running rhythms in chol-1 strains grown without choline, and heat-entrainable rhythms in choline-sufficient conditions. We found effects of all 4 mutations on both types of FRQ-less rhythms. The greatest effects were seen with prd-1 and prd-2, which abolished free-running rhythms in the chol-1; frq 10 backgrounds and significantly affected entrained peak timing under heatentrainment conditions in frq 10 backgrounds. The prd-3 and prd-4 mutations had more subtle effects on period and stability of free-running rhythms in the chol-1; frq 10 backgrounds and had little effect on peak timing under heat-entrainment conditions in frq 10 backgrounds. These results, along with previously published evidence for effects of prd mutations on other FRQ-less rhythms, suggest that either there are common components shared between the FRQ/WCC oscillator and several FRQ-less oscillators or that there is a single oscillator driving all conidiation rhythms. We favor a model of the Neurospora circadian system in which a single FRQ-less oscillator drives conidiation and interacts with the FRQ/WCC feedback loop; the output or amplitude of the FRQ-less oscillator can be affected by many gene products and metabolic conditions that reveal FRQ-less rhythmicity. We propose that prd-1 and prd-2 are good candidates for components of the FRQ-less oscillator and that prd-3 and prd-4 act on the system mainly through effects on FRQ/WCC. Key words Neurospora, circadian, FLO, frq, prd, entrainment, temperature The conidiation (asexual spore formation) rhythm in the filamentous fungus Neurospora crassa has been studied for many years as a model system for understanding the molecular basis of circadian rhythmicity (Dunlap and Loros, 2006; Lakin-Thomas, 2006b; Liu and Bell-Pedersen, 2006). Analysis of a 1. Current address: Department of Molecular, Cellular and Developmental Biology, Yale University, New Haven, CT. 2. To whom all correspondence should be addressed: Patricia Lakin-Thomas, Department of Biology, York University, 4700 Keele Street, Toronto, ON M3J 1P3, Canada; e-mail: plakin@yorku.ca. JOURNAL OF BIOLOGICAL RHYTHMS, Vol. 25 No. 2, April 2010 71-80 DOI: 10.1177/0748730409360889 2010 SAGE Publications 71

72 JOURNAL OF BIOLOGICAL RHYTHMS / April 2010 small number of Neurospora clock genes, frq, wc-1, and wc-2, has led to a model of the circadian oscillator based on negative and positive feedback loops between the products of these genes (the FRQ/WCC oscillator) (Loros and Dunlap, 2001). However, rhythmic conidiation can be seen under many different conditions in strains carrying null mutations at the frq or wc loci (Aronson et al., 1994; Dragovic et al., 2002; Granshaw et al., 2003; Lakin-Thomas, 2006a; Lakin-Thomas and Brody, 2000; Lombardi et al., 2007; Loros and Feldman, 1986; Merrow et al., 1999; Schneider et al., 2009). This rhythmicity is said to be driven by one or more FRQ-less oscillator(s) (FLO) completely separate from the FRQ/WCC oscillator (de Paula et al., 2006; Iwasaki and Dunlap, 2000; Shi et al., 2007). An alternative view is that the FRQ/WCC loop is integrated into a circadian system in which conidiation rhythmicity is driven by an oscillator that can function without FRQ/WCC. The number and nature of FRQ-less oscillator(s) and their interaction with FRQ/WCC is currently the most salient question in Neurospora circadian research. This article reports a test of one prediction of the separate oscillators model. If the FRQ/WCC oscillator and the FLO(s) are completely separate, then mutations in genes affecting only one oscillator would not affect conidiation rhythmicity in the absence of that oscillator. We have tested this prediction using a series of mutations identified by Feldman and coworkers that affect the period of the circadian conidiation rhythm in frq + strains (Feldman, 1982; Morgan et al., 2001). The prd-1, prd-2, prd-3, and prd-4 mutations are in 4 independent loci and were found to either lengthen the period (prd-1, prd-2, and prd-3) or shorten the period (prd-4) in frq +. The prd-4 gene product has been identified as a protein kinase (the Neurospora ortholog of mammalian checkpoint kinase 2) that interacts with, and promotes phosphorylation of, FRQ protein (Pregueiro et al., 2006). The prd-3 gene product has been identified as a subunit of casein kinase 2; CK2 was found to directly phosphorylate the FRQ protein and to thereby affect temperature compensation of the period of the conidiation rhythm (Mehra et al., 2009). Identities of the prd-1 and prd-2 genes have not yet been published. A genetic analysis of interactions between prd mutations (Morgan and Feldman, 2001; Morgan et al., 2001) found epistatic and synergistic interactions among prd-1, prd-2, and prd-3 as well as interactions between long-period frq mutations and both prd-2 and prd-3. These results indicate that all 4 of these prd gene products participate in the same rhythmic system as FRQ/WCC. We have introduced each of these 4 prd mutations into strains carrying the null mutation frq 10 to assay effects of prds on FRQ-less rhythms. Here, we report analysis of rhythmicity in prd frq 10 strains using 2 conditions under which FRQ-less rhythmicity can be seen: 1) In the chol-1 (cholinerequiring) mutant, conidiation rhythms are seen in chol-1 strains carrying frq and wc null mutations under conditions of choline limitation. These rhythms can have periods in the circadian range on low levels of choline (10 µm) and as long as 40 to 60 hours when grown without any choline supplementation (Lakin-Thomas, 1998; Lakin-Thomas and Brody, 2000). 2) Cycles of high temperature pulses can entrain conidiation rhythms in frq and wc null mutants (Lakin-Thomas, 2006a). We have assayed free-running rhythmicity in chol-1 frq 10 prd strains without choline, and we have also assayed entrainment to heat pulses in chol-1 frq 10 prd strains on high choline when the chol-1 phenotype is not expressed. We have found that the prd mutations affect conidiation rhythmicity in both FRQ-sufficient and FRQ-deficient conditions. MATERIALS AND METHODS Strain Construction and Culture Methods All strains reported here carry the ras-1 bd (formerly named bd [Belden et al., 2007]) and csp-1 mutations in addition to the mutations listed in the figures and table. The ras-1 bd prd parental strains were obtained from the Fungal Genetics Stock Center (Kansas City, MO), and the chol-1 and frq 10 parental strains were previously described (Lakin-Thomas and Brody, 2000). The frq 10 strains carry a null mutation at the frq locus created by targeted gene disruption (Aronson et al., 1994). Multiple-mutant strains were constructed by standard crossing methods as previously described (Lakin-Thomas and Brody, 2000). The genotypes of all multiple-mutant strains reported here were verified by backcrossing to the ras-1 bd strain to confirm the expected segregation patterns (data not shown). Cultures were grown on maltose/arginine medium containing Vogel s salts, 0.5% maltose, 0.01% arginine, 2% agar, and either high choline (100 µm) or no choline supplementation, as described (Lakin-Thomas and Brody, 2000). On high choline, chol-1 strains are identical to chol + strains (Lakin-Thomas, 1998). Cultures were initially grown on agar plates in constant light (LL) at 30 C before transfer of small mycelial plugs to 30-cm race tubes, which were incubated for an additional 24 hours in LL at 30 C before transfer to constant darkness (DD) at

Li, Lakin-Thomas / FRQ-LESS RHYTHMS IN NEUROSPORA 73 22 C. The growth fronts of race tubes were marked once per day under red safelight, and the time of marking was recorded for calculation of growth rates and periods. Periods and growth rates in free-running conditions (DD, 22 C) were calculated by regression analysis as previously described (Lakin-Thomas, 1998). Entrainment to Temperature Cycles Cultures were grown on race tubes at 22 C and subjected to repeated T cycles of 2-hour pulses of either 32 C or 37 C as previously described (Lakin-Thomas, 2006a). T cycle refers to the time between repeated pulses. Temperature pulses were delivered as steps up from 22 C to either 32 C or 37 C, followed after 2 hours by a step down to 22 C. Density traces were collected and processed as previously described (Lakin-Thomas, 2006a). Density profiles of at least 5 replicate tubes were averaged. To calculate average days, 2 consecutive T cycles were averaged from at least 5 replicate tubes per experiment. The cycles chosen for averaging were the last 2 complete cycles for the fastest-growing strain in each experiment; thus, the cycles averaged for slower-growing strains were not the last cycles in those cultures. To calculate peak times, the conidiation density profiles for individual tubes were smoothed using a 41-point moving average (equal to 4.1 hours) as described (Lakin-Thomas, 2006a). Peak times were identified as the highest density value in the smoothed profile. Peak times were analyzed with 1-way ANOVA for effect of T cycle on peak time and with Student s t test for differences between peak times of prd strains and prd + controls. To calculate normalized days for each set of replicate density profiles (2 cycles from 5 or more replicate race tubes), the density profiles were normalized by setting the maximum pixel value equal to 1.0 and the minimum pixel value equal to 0.0. The standard error of the mean (SEM) at each time point was calculated for the average day, with N = number of individual cycles averaged, and confidence intervals were plotted as ±1 SEM. RESULTS Effects of prd Mutations on Free-Running Rhythms in chol-1 Multiple-mutant strains carrying a prd mutation as well as frq 10 were constructed by standard crossing methods, and all strains were backcrossed to a ras-1 bd strain. All multiple mutants produced the expected Figure 1. Conidiation phenotypes in free-running conditions. Cultures were grown on race tubes without added choline in constant conditions (DD, 22 C). Growth is from left to right. Two replicate cultures are shown for each genotype. Vertical black marks near the left end of the tubes are the first growth marks at the time of the light-to-dark transition. All other growth marks were removed before scanning for this figure. Horizontal black bars represent 24 hours of growth. All strains carry the ras-1 bd and csp-1 mutations in addition to the mutations indicated. w.t. = csp-1; ras-1 bd, wild type for chol-1, frq, and prd. ratios of progeny, verifying the genotypes (data not shown). The conidiation phenotypes of the multiplemutant strains were assayed in free-running conditions (DD, 22 C) on medium without choline. Choline-free medium reduces the growth rate of chol-1 strains and allows expression of long-period rhythmicity in both frq + and frq null strains (Lakin-Thomas, 1996, 1998; Lakin-Thomas and Brody, 2000; Ruoff and Slewa, 2002; Shi et al., 2007). In Figure 1, the wildtype strain (carrying the ras-1 bd and csp-1 mutations but wild type for chol-1, frq, and prd) displays the expected conidiation rhythmicity. The chol-1 strain is rhythmic with a long period, the frq 10 strain is arrhythmic under these conditions, and the double mutant chol-1; frq 10 is rhythmic, as previously reported

74 JOURNAL OF BIOLOGICAL RHYTHMS / April 2010 Table 1. Free-running periods and growth rates. Control strains High Choline No Choline Growth Rate, Growth Rate, Period, h mm/h Period, h mm/h w.t. 21.0 ± 0.08 (20) 1.22 ± 0.02 (20) 21.2 ± 0.06 (20) 1.29 ± 0.02 (20) chol-1 21.1 ± 0.08 (20) 1.20 ± 0.01 (20) 41.5 ± 1.0 (35) 0.68 ± 0.01 (35) frq 10 Arrhythmic 1.25 ± 0.02 (7) Arrhythmic 1.31 ± 0.02 (7) chol-1; frq 10 Arrhythmic 1.28 ± 0.01 (8) 55.2 a ± 2.2 (47) 0.69 ± 0.01 (47) prd strains prd-1; chol-1 25.0 a ± 0.11 (44) 0.74 ± 0.01 (44) 25 d 50 d 0.57 ± 0.01 (45) chol-1; prd-2 26.1 a ± 0.14 (30) 1.21 ± 0.01 (30) 41.4 e ± 2.06 (18) 0.75 ± 0.01 (30) prd-3; chol-1 23.6 a ± 0.13 (16) 1.14 ± 0.01 (16) 59.5 a ± 2.9 (33) 0.54 ± 0.01 (33) prd-4; chol-1 17.5 a ± 0.16 (10) 1.27 ± 0.01 (10) 40.9 c ± 0.9 (22) 0.66 ± 0.01 (22) prd-1; chol-1; frq 10 Arrhythmic 0.70 ± 0.02 (6) Arrhythmic 0.53 ± 0.02 (6) chol-1; prd-2; frq 10 Arrhythmic 1.12 ± 0.01 (6) Arrhythmic 0.82 ± 0.01 (6) prd-3; chol-1; frq 10 Arrhythmic 1.18 ± 0.01 (7) 61.9 e ± 6.3 (21) 0.60 ± 0.03 (23) prd-4; chol-1; frq 10 Arrhythmic 1.29 ± 0.01 (10) 41.0 b ± 1.6 (22) 0.65 ± 0.01 (22) period was not significantly different (p > 0.05) from chol-1; the prd-3; chol-1 period was significantly longer (p < 0.05) than chol-1; and both prd-1; chol-1 and chol-1; prd-2 displayed unstable and highly variable periodicity. When frq 10 was introduced into the triple mutants, the prd-1; chol-1; frq 10 and chol-1; prd-2; frq 10 strains became arrhythmic (as seen in Fig. 1), the period of the prd-3; chol-1; frq 10 strains became highly variable, and the period of prd-4; chol-1; frq 10 became significantly shorter (p < 0.05) than chol-1; frq 10. In summary, all 4 prd mutations significantly affected rhythmicity in the long-period chol-1 strains. The largest effects are seen with prd-1 and prd-2, while prd-3 has lesser effects, and prd-4 has the least effect. All strains carry the ras-1 bd and csp-1 mutations in addition to those in column 1. w.t. = csp-1; ras-1 bd, wild type for chol-1, frq, and prd. Cultures were grown in DD at 22 C on high choline (100 µm) or without choline supplementation. Values are reported as mean ± SEM (N), where N is the number of race tubes. Statistical comparisons were made using the 2-tailed Student s t test. a. Period significantly different from the chol-1 control (p < 0.05). b. Period significantly different from the chol-1 frq 10 control (p < 0.05). c. Period not significantly different from the chol-1 control. d. Highly variable; period shifts from about 25 hours to about 50 hours after several days of growth. e. Highly variable; some replicates are rhythmic, and some are not. (Lakin-Thomas and Brody, 2000). When either prd-1 or prd-2 is introduced into either chol-1 or chol-1; frq 10, rhythmicity is severely disrupted (Fig. 1). Some residual fluctuations can be seen in the chol-1 strains, but the chol-1; frq 10 strains become arrhythmic when prd-1 or prd-2 is introduced. Introduction of prd-3 or prd-4 has less effect on rhythmicity (Fig. 1). The periods and growth rates of these multiplemutant strains with and without choline are reported in Table 1. All cultures on high choline displayed the expected phenotypes. The chol-1 strains on high choline are indistinguishable from chol + strains (Table 1, Control Strains). Morgan and Feldman (2001) reported periods of 21.5, 25.8, 25.5, 25.1, and 18.0 hours for the prd +, prd-1, prd-2, prd-3, and prd-4 strains in a ras-1 bd background on minimal medium at 25 C, which are very similar to our results (21.0, 25.0, 26.1, 23.6, and 17.5 hours) for these mutations in the csp-1; ras-1 bd chol-1 background grown on high choline at 22 C (Table 1). All strains carrying frq 10 mutations were arrhythmic under these conditions with high choline. Without choline, the control strains behaved as expected, with long-period rhythms for the 2 chol-1 strains (frq + and frq 10 ) (Table 1). The prd double mutants with chol-1 displayed a range of period phenotypes without choline (Table 1). The prd-4; chol-1 Effects of prd Mutations on Entrained Rhythms in frq 10 To investigate the effects of prd mutations on FRQless rhythmicity under entrained conditions, the chol-1; frq 10 strains were grown on high choline to repair the chol-1 defect and were subjected to entraining heat pulses. It has been previously established that frq null strains can be entrained to cycles of heat pulses and behave as if a heat-entrainable oscillator is functioning in these FRQ-less strains (Lakin-Thomas, 2006a; Merrow et al., 1999; Roenneberg et al., 2005). We used cycles of 2-hour pulses of high temperature at 4 different T cycles: 16 hours, 20 hours, 24 hours, and 28 hours. A shift in the phase of the entrained rhythm relative to the entraining stimulus under different T cycles can reveal the presence of an entrainable oscillator (Johnson et al., 2003; Roenneberg et al., 2005). Two different pulse strengths were used: either 32 C (weak pulse) or 37 C (strong pulse). If there is a heat-entrainable oscillator in the cultures, then the shift in phase under different T cycles should be less for a strong pulse as compared to a weak pulse (Roenneberg et al., 2005).

Li, Lakin-Thomas / FRQ-LESS RHYTHMS IN NEUROSPORA 75 Figure 2. Entrainment to 32 C heat pulses, normalized days. Race tube cultures grown on high choline at 22 C were exposed to a 2-hour pulse at 32 C every T hours. Density traces for a set of 6 replicate tubes were averaged for 2 consecutive cycles from all replicate race tubes in each set and are plotted against hours after the beginning of the heat pulse. Regions of high density indicate thick areas of conidiation. Average density traces were normalized by setting the maximum pixel value to 1.0 and the minimum pixel value to 0. The average traces are plotted with ±1 SEM. Gray traces are the prd + control strain, which is repeated in all 4 panels in each column. The prd genotypes are indicated on the left. All strains carry the ras-1 bd, csp-1, chol-1, and frq 10 mutations in addition to the prd mutations indicated. T cycles are indicated above each column. In Figure 2, the density profiles for the average days for 32 C T cycles have been normalized so that all average profiles span a range of 0.0 to 1.0, and the averages have been plotted with confidence intervals of ±1 SEM. This procedure eliminates differences in amplitude between density profiles and allows comparison of peak timing. Each of the 4 prd strains is compared to the appropriate prd + control in each panel. There is an immediate response to the heat pulses that can be seen in most average days as a peak coincident with the 2-hour heat pulse. This immediate response may be a masking effect, or it may represent a response of oscillator components to the resetting stimulus. It will be referred to as the immediate response. This is usually followed by a second peak that is identified as the entrained peak for the purpose of calculating the peak times. The immediate response peak is not easily seen in the prd-1 traces and may be obscured by the early rise of the entrained peak. It can be seen in the top 2 rows that prd-1 and prd-2 affect the timing of the peaks relative to the prd + control strain at T = 16 and T = 20 and to a lesser extent at T = 28. The prd-3 and prd-4 mutations have less effect on peak timing (bottom 2 rows, Fig. 2). The phases of the entrained peaks for both 32 C and 37 C pulses were calculated in hours after the entraining pulse, as previously described (Lakin-Thomas, 2006a), and are presented in Figure 3. The peak times of the prd + ; frq 10 control strain are significantly affected by the T cycle for both 32 C and 37 C pulses (as shown by the results of 1-way ANOVA, testing for an effect of the T cycle on peak time within one strain, p < 0.01). This is the expected result if the conidiation peaks produced in frq 10 during entrainment are the output of a heat-entrainable oscillator (Merrow et al., 1999; Roenneberg et al., 2005) and are not simply a direct response to the

76 JOURNAL OF BIOLOGICAL RHYTHMS / April 2010 Phase (hours after heat pulse start) 14 12 10 8 6 4 2 0 14 12 10 8 6 4 2 0 A 32 B 32 prd+ prd-1 prd-2 16 20 24 28 prd+ prd-1 prd-2 pulse or the output of an hourglass process (Lakin-Thomas, 2006a). All 4 prd; frq 10 double mutant strains also showed significant effects (p < 0.01) of T cycle length on peak times for 32 C pulses (Fig. 3A and 3B), indicating that heat-entrainable oscillators are also operating in these strains. With a stronger stimulus (37 C pulses), the prd-1 and prd-2 double mutants did not show significant effects of T cycle length on peak time (Fig. 3C), unlike the prd + ; frq 10 control strain. This may indicate a stronger response to the 37 C pulse in these 2 prd frq 10 mutants compared to the prd + ; frq 10 strain. The prd-1; frq 10 and prd-2; frq 10 peak times (Fig. 3A and 3C) were significantly different (p < 0.05) from prd + ; frq 10 for all heat pulses except for prd-1; frq 10 at 32 C, T = 28 (Fig. 3A). The prd+ prd-3 prd-4 16 20 24 28 0 16 20 24 28 16 20 24 28 T-cycle (hours) C 37 D 37 8 14 12 10 8 6 4 2 0 14 12 10 6 4 2 T-cycle (hours) prd+ prd-3 prd-4 Figure 3. Peak times of cultures entrained with heat pulses. All strains carry the ras- 1 bd, csp-1, frq 10, and chol-1 mutations in addition to the indicated prd mutations. Cultures were grown on high choline (100 µm) at 22 C and received 2-hour high-temperature pulses (to 32 C [A, B] or 37 C [C, D]) at intervals of T = cycle length in hours. Peak times are calculated as hours after the start of the heat pulse. For each strain, peak times were calculated individually for 2 consecutive cycles from 5 or 6 race tubes per experiment, and in some cases, 2 or 3 experiments were pooled. Values are plotted as mean ± SEM, where N is the total number of cycles and ranged from 9 to 36. = peak time significantly different from the control strain (prd + ; frq 10 ) within that T cycle, at the 0.01 level of significance, 2-tailed t-test for 2 samples with equal variance. Note that in A, prd-1 did not produce clearly defined peaks for T = 20 hours and 24 hours. prd-3; frq 10 double mutant peak times were not significantly different from frq 10 ; prd + for any pulses (Fig. 3B and 3D). The prd-4; frq 10 pulse times were significantly different (p < 0.05) from prd + ; frq 10 only for 37 C pulses at T = 24 and 28 (Fig. 3D). In summary, the results in Figure 3 indicate that the prd-1 and prd-2 mutations significantly affect the entrainment behavior of frq 10. The prd-4 mutation has less of an effect, and prd-3 appears to have no effect on the entrained peak times. Effects of prd Mutations on Entrained Rhythms in frq + The results described above indicate effects of some prd mutations on heat-entrainment behavior in the absence of a functioning FRQ/WCC oscillator. To determine how prd mutations affect heat entrainment in the presence of FRQ, we carried out a set of entrainment experiments using 2-hour 37 C pulses at T = 24 with frq + strains. Figure 4 compares the normalized days of each prd strain with the prd + control. The frq 10 strains carry the chol-1 mutation but were grown on high choline to repair the chol-1 defect. The frq + strains are chol-1 + and were grown on medium without choline because they do not require choline for normal growth. In the frq + background, the prd-3 peak rise is somewhat delayed relative to prd +, and the prd-4 peak rise is somewhat advanced, which correlates with the lengthened period of prd-3 and the shortened period of prd-4 (Table 1, frq + strains, high choline). In the frq 10 background, there is little difference between the prd + control and either prd-3 or prd-4. Surprisingly, delays relative to the prd + control are not seen in the long-period prd-1 and prd-2 strains in either frq + or frq 10 backgrounds. The prd-2 mutation advances the peak time in frq 10 but has little effect on frq +, while prd-1 is advanced with a similar peak timing in both frq 10 and frq +. Comparing frq 10 and frq + in Figure 4, it can be seen that the absence of FRQ has some effect on the timing of most of the peaks. The exception is prd-1, in which the peak time is very similar in both frq + and frq 10 backgrounds.

Li, Lakin-Thomas / FRQ-LESS RHYTHMS IN NEUROSPORA 77 Figure 4. Entrainment to 37 C heat pulses, normalized days. Methods are as in Figure 2, except that 37 C heat pulses were used at T = 24. The average traces are plotted with ±1 SEM. Gray traces are the prd + control strain, which is repeated in all 4 panels in each column. The prd genotypes are indicated on the left. In the left column are the chol-1; frq 10 strains grown on high choline; in the right column are the chol-1 + ; frq + strains grown without choline. DISCUSSION Our results demonstrate that mutations (prds) isolated on the basis of an effect on the period of the conidiation rhythm in FRQ/WCC wild-type strains can also affect FRQ-less rhythms. We found that 4 prd mutations affect 2 different FRQ-less rhythms: the free-running long-period conidiation rhythm in frq 10 chol-1 on low choline, and the entrained conidiation rhythm in frq 10. Thus, these prd mutations affect FRQless rhythms both with and without choline starvation. The prd-1 and prd-2 mutations have the greatest effects, while prd-3 and prd-4 have lesser effects. We did not find simple epistatic relationships between prd mutations and frq 10 or chol-1; for example, we did not find that the phenotype of a chol-1 prd double mutant is identical to the chol-1 single mutant, in which case the chol-1 mutation (and the FRQ-less oscillator it reveals) would be epistatic to prd and the oscillator it influences. Such epistatic relationships can indicate a hierarchical pathway, with one gene product acting downstream of another. The failure to find epistasis could mean that these gene products interact in more complex ways and that the oscillators involved interact with each other. The prd mutations have been shown to affect other FRQ-less rhythms as well. Lombardi et al. (2007) found a similar gradient of effects of prds on conidiation rhythmicity in frq 10 revealed by farnesol and geraniol supplementation. With farnesol, the period was altered by prd-1 and prd-2, but FRQ-less rhythmicity was not abolished, while prd-3 and prd-4 had little effect on the period with farnesol (Lombardi et al., 2007). A fatty acid requiring strain, cel, expresses a long-period conidiation rhythm when supplemented with unsaturated fatty acids; this rhythm continues in cel frq 10 and is therefore a FRQless rhythm (Lakin-Thomas and Brody, 2000), and the period-lengthening effect of fatty acids in cel is disrupted by prd-1 (Lakin-Thomas and Brody, 1985). (The effects of other prds on cel were not determined.) The prd-1 mutation therefore affects FRQ-less rhythmicity in 4 different conditions. Several alternative explanations of the results can be considered and are presented in Supplementary Figure S1. In the first model, Supplementary Figure S1A, there are multiple noncircadian FLOs independent of a circadian FRQ/WC oscillator. In order to account for the results presented in this article, it is necessary to assume that several of those oscillators, as well as FRQ/WCC, must share prd gene products as common components while still behaving independently. Because prd-1 and prd-2 have greater effects than prd-3 and prd-4 in several FRQ-less rhythms, then the prd-1 and prd-2 gene products would play more important roles than prd-3 and prd-4 in all oscillators. Our results demonstrating effects of frq mutations on the periods of FRQ-less rhythms (cited below) make this model of independent FRQ/ WCC and FLOs unlikely. This model is also more

78 JOURNAL OF BIOLOGICAL RHYTHMS / April 2010 complicated than necessary, and by applying Occam s razor, we propose simpler models to account for the data. In the second model, Supplementary Figure S1B, there are 2 independent oscillators, FRQ/WCC and a single FLO, both capable of driving conidiation rhythms. To account for our results, it is necessary to propose that prd mutations affect both FRQ/WCC and FLO independently. This model is also difficult to reconcile with our data demonstrating effects of frq mutations on the periods of FRQ-less rhythms. The third model, Supplementary Figure S1C, proposes FRQ/WCC and FLO in series, with FLO providing the primary output to drive conidiation rhythms, and FRQ/WCC driving FLO. This accounts for effects of frq mutations on FRQless rhythms. In the absence of feedback from FLO to FRQ/WCC, the prd mutations are proposed to affect both oscillators to account for the effects of prd mutations on the period of both FRQ-sufficient and FRQ-less strains. We prefer a model of the Neurospora circadian system (Supplementary Fig. S1D) in which conidiation rythmicity is driven primarily by the output of an oscillator that may be identified with the FRQ-less oscillator (FLO). The prd-1 and prd-2 gene products appear to play an important role in this oscillator, based on our observations that mutations in these genes abolish rhythmicity in chol-1 frq 10 and alter entrainment behavior in frq 10. Because the prd-3 and prd-4 mutations have lesser effects on FRQ-less rhythms, their gene products may act on the system primarily through FRQ/WCC. They may also affect targets other than FRQ. This is suggested by the difference in phenotypes between chol-1; frq 10 and prd-3; chol-1; frq 10, and between chol-1; frq 10 and prd-4; chol-1; frq 10 (Table 1): If FRQ was the only target of prd-3 and prd-4, then there would be no difference between these strains and prd + strains when FRQ is missing in frq 10. In our model (Supplementary Fig. S1D), all treatments, conditions, and mutations that reveal rhythmicity in FRQ-less strains may act by increasing the amplitude of the FLO or by increasing the strength of the output pathway that drives rhythmic conidiation. FRQ-less rhythms are usually deficient in one or more of the properties that define circadian rhythms, such as temperature compensation and light entrainability, and therefore, the FRQ/WCC negative feedback loop is clearly critical to providing the full range of circadian properties to the system: In the absence of functional FRQ and WCC, the system as a whole is not a fully competent circadian clock. The data presented here indicate that these 4 prd gene products affect both FRQ-sufficient and FRQ-less rhythms, suggesting interactions between these rhythms. We have previously found that the long period of chol-1 is significantly shortened (p < 0.01) by a short-period frq 1 mutation (Lakin-Thomas, 1998) and is significantly lengthened (p < 0.05) by a wc-2 mutation (Lakin-Thomas and Brody, 2000). We reported a correlation between the periods of various alleles at the frq locus and the effects of these alleles on the long period of chol-1 (Lakin-Thomas, 1998), supporting the suggestion that the FRQ/WCC loop interacts with the oscillator driving FRQ-less rhythms. Table 1 provides additional evidence for interactions: 1) The period of chol-1 frq 10 is significantly different (p < 0.05) from chol-1. 2) Although chol-1 prd-1 and chol-1 prd-2 have highly variable periods on low choline, chol-1 frq 10 prd-1 and chol-1 frq 10 prd-2 are completely arrhythmic. In contrast to the evidence described above for interactions between FLO and FRQ/WCC, a paper by Shi et al. (2007) suggests that the rhythm of frq transcription is independent of the long-period conidiation rhythm in chol-1. Shi et al. (2007) used a frq-luciferase fusion gene to assay frq promoter activity in chol-1 cultures expressing a long-period (~60 hours) conidiation rhythm and found that the luciferase signal displayed a short-period (22 hours) rhythm. They found no effect of the frq 7 mutation on the period of chol-1, in agreement with our previous work (Lakin-Thomas, 1998), but did find the expected effect of frq 7 on the period of the molecular rhythm of frq promoter activity. They did not report any data for the frq 1 or wc-2 mutations, which we found to have significant effects on the chol-1 period (Lakin-Thomas, 1998; Lakin-Thomas and Brody, 2000). In contrast to our conclusions, these authors concluded that the FLO in chol-1 is in no way connected to the circadian system (Shi et al., 2007). Although these authors carried out periodogram analysis on the short-period luciferase data and failed to find long-period components in the frq expression rhythm, they did not carry out a similar analysis on the long-period conidiation rhythm to look for short-period components. It is possible to reconcile the 2 apparently contradictory conclusions by recognizing that the finding of shortperiod molecular rhythms in chol-1 does not preclude interactions such as relative coordination or frequency demultiplication between a short-period oscillator driving molecular rhythms and a longperiod process producing conidiation bands during choline depletion.

Li, Lakin-Thomas / FRQ-LESS RHYTHMS IN NEUROSPORA 79 In summary, these results provide support for a model of the Neurospora circadian system in which all FRQ-less conidiation rhythmicity is driven by a single FLO that closely interacts with FRQ/WCC. These results suggest mutual influences between FRQ/ WCC and FLO in an integrated circadian system that can be affected by mutations in many different genes and by many different growth conditions. Our results are also the first identification of potential candidates (prd-1 and prd-2) for components of the FLO. ACKNOWLEDGMENTS Many thanks to Liz Kafes, Kevin Ng, Fatima Panju, Neisma Peredo, and Nik Stephens for technical assistance, to Duane Lakin-Thomas for figure preparation, and to Stuart Brody for helpful comments on the article. This research is supported by NSERC Discovery Grant 250133-07. NOTE Supplementary material for this article (Supplementary Fig. S1) is available on the Journal of Biological Rhythms Web site at http://jbr.sagepub.com/supplemental. REFERENCES Aronson BD, Johnson KA, and Dunlap JC (1994) Circadian clock locus Frequency: protein encoded by a single open reading frame defines period length and temperature compensation. Proc Natl Acad Sci U S A 91:7683-7687. Belden WJ, Larrondo LF, Froehlich AC, Shi M, Chen CH, Loros JJ, and Dunlap JC (2007) The band mutation in Neurospora crassa is a dominant allele of ras-1 implicating RAS signaling in circadian output. Genes Dev 21:1494-1505. Christensen MK, Falkeid G, Loros JJ, Dunlap JC, Lillo C, and Ruoff P (2004) A nitrate induced frq-less oscillator in Neurospora crassa. J Biol Rhythms 19:280-286. de Paula RM, Lewis ZA, Greene AV, Seo KS, Morgan LW, Vitalini MW, Bennett L, Gomer RH, and Bell-Pedersen D (2006) Two circadian timing circuits in Neurospora crassa cells share components and regulate distinct rhythmic processes. J Biol Rhythms 21:159-168. Dragovic Z, Tan Y, Görl M, Roenneberg T, and Merrow M (2002) Light reception and circadian behavior in blind and clock-less mutants of Neurospora crassa. EMBO J 21:3643-3651. Dunlap JC and Loros JJ (2006) How fungi keep time: circadian system in Neurospora and other fungi. Curr Opin Microbiol 9:579-587. Feldman JF (1982) Genetic approaches to circadian clocks. Annu Rev Plant Physiol Plant Mol Biol 33:583-608. Granshaw T, Tsukamoto M, and Brody S (2003) Circadian rhythms in Neurospora crassa: farnesol or geraniol allow expression of rhythmicity in the otherwise arrhythmic strains frq 10, wc-1, and wc-2. J Biol Rhythms 18:287-296. Iwasaki H and Dunlap JC (2000) Microbial circadian oscillatory systems in Neurospora and Synechococcus: models for cellular clocks. Curr Opin Microbiol 3:189-196. Johnson CH, Elliot JA, and Foster RG (2003) Entrainment of circadian programs. Chronobiol Int 20:741-774. Lakin-Thomas PL (1996) Effects of choline depletion on the circadian rhythm in Neurospora crassa. Biol Rhythm Res 27:12-30. Lakin-Thomas PL (1998) Choline depletion, frq mutations, and temperature compensation of the circadian rhythm in Neurospora crassa. J Biol Rhythms 13:268-277. Lakin-Thomas PL (2006a) Circadian clock genes frequency and white collar-1 are not essential for entrainment to temperature cycles in Neurospora crassa. Proc Natl Acad Sci U S A 103:4469-4474. Lakin-Thomas PL (2006b) New models for circadian systems in microorganisms. FEMS Microbiol Lett 259:1-6. Lakin-Thomas PL and Brody S (1985) Circadian rhythms in Neurospora crassa: interactions between clock mutations. Genetics 109:49-66. Lakin-Thomas PL and Brody S (2000) Circadian rhythms in Neurospora crassa: lipid deficiencies restore robust rhythmicity to null frequency and white-collar mutants. Proc Natl Acad Sci U S A 97:256-261. Liu Y and Bell-Pedersen D (2006) Circadian rhythms in Neurospora crassa and other filamentous fungi. Eukaryotic Cell 5:1184-1193. Lombardi L, Schneider K, Tsukamoto M, and Brody S (2007) Circadian rhythms in Neurospora crassa: clock mutant effects in the absence of a frq-based oscillator. Genetics 175:1175-1183. Loros JJ and Dunlap JC (2001) Genetic and molecular analysis of circadian rhythms in Neurospora. Annu Rev Physiol 63:757-794. Loros JJ and Feldman JF (1986) Loss of temperature compensation of circadian period length in the frq-9 mutant of Neurospora crassa. J Biol Rhythms 1:187-198. Mehra A, Shi M, Baker CL, Colot HV, Loros JJ, and Dunlap JC (2009) A role for casein kinase 2 in the mechanism underlying circadian temperature compensation. Cell 137:749-760. Merrow M, Brunner M, and Roenneberg T (1999) Assignment of circadian function for the Neurospora clock gene frequency. Nature 399:584-586. Morgan LW and Feldman JF (2001) Epistatic and synergistic interactions between circadian clock mutations in Neurospora crassa. Genetics 159:537-543. Morgan LW, Feldman JF, and Bell-Pedersen D (2001) Genetic interactions between clock mutations in Neurospora crassa: can they help us to understand complexity? Philos Trans R Soc Lond B Biol Sci 356:1717-1724.

80 JOURNAL OF BIOLOGICAL RHYTHMS / April 2010 Pregueiro AM, Liu QY, Baker CL, Dunlap JC, and Loros JJ (2006) The Neurospora checkpoint kinase 2: a regulatory link between the circadian and cell cycles. Science 313:644-649. Roenneberg T, Dragovic Z, and Merrow M (2005) Demasking biological oscillators: properties and principles of entrainment exemplified by the Neurospora circadian clock. Proc Natl Acad Sci U S A 102:7742-7747. Ruoff P and Slewa I (2002) Circadian period lengths of lipid synthesis mutants (cel, chol-1) of Neurospora show defective temperature, but intact ph- compensation. Chronobiol Int 19:517-529. Schneider K, Perrino S, Oelhafen K, Li S, Zatsepin A, Lakin-Thomas PL, and Brody S (2009) Rhythmic conidiation in constant light in Vivid mutants of Neurospora crassa. Genetics 181:917-931. Shi M, Larrondo LF, Loros JJ, and Dunlap JC (2007) A developmental cycle masks output from the circadian oscillator under conditions of choline deficiency in Neurospora. Proc Natl Acad Sci U S A 104:20102-20107.