The frequency Gene Is Required for Temperature-Dependent Regulation of Many Clock-Controlled Genes in Neurospora crassa

Corresponding author: Jay C. Dunlap, HB7400, Dartmouth Medical School, Hanover, NH 03755., jay.c.dunlap{at}dartmouth.edu (E-mail)

Communicating editor: M. S. SACHS

	ABSTRACT

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

The circadian clock of Neurospora broadly regulates gene expression and is synchronized with the environment through molecular responses to changes in ambient light and temperature. It is generally understood that light entrainment of the clock depends on a functional circadian oscillator comprising the products of the wc-1 and wc-2 genes as well as those of the frq gene (the FRQ/WCC oscillator). However, various models have been advanced to explain temperature regulation. In nature, light and temperature cues reinforce one another such that transitions from dark to light and/or cold to warm set the clock to subjective morning. In some models, the FRQ/WCC circadian oscillator is seen as essential for temperature-entrained clock-controlled output; alternatively, this oscillator is seen exclusively as part of the light pathway mediating entrainment of a cryptic "driving oscillator" that mediates all temperature-entrained rhythmicity, in addition to providing the impetus for circadian oscillations in general. To identify novel clock-controlled genes and to examine these models, we have analyzed gene expression on a broad scale using cDNA microarrays. Between 2.7 and 5.9% of genes were rhythmically expressed with peak expression in the subjective morning. A total of 1.4–1.8% of genes responded consistently to temperature entrainment; all are clock controlled and all required the frq gene for this clock-regulated expression even under temperature-entrainment conditions. These data are consistent with a role for frq in the control of temperature-regulated gene expression in N. crassa and suggest that the circadian feedback loop may also serve as a sensor for small changes in ambient temperature.

THE ascomycete Neurospora crassa has a long standing as a model organism for the investigation of circadian rhythms (DUNLAP 1999 ). The circadian system of N. crassa is composed of several interdependent molecular feedback loops (ARONSON et al. 1994A ; CROSTHWAITE et al. 1997 ; LEE et al. 2000 ; CHENG et al. 2001 ; DENAULT et al. 2001 ; GOERL et al. 2001 ; HEINTZEN et al. 2001 ) that regulate the time-of-day-specific expression of a number of output genes, thereby generating distinct phenotypes, e.g., the clock-dependent rhythm of macroconidiation. The clock integrates input signals, including light and temperature changes, to keep the organism in tune with the environment. Both light and temperature have similar entrainment effects on the circadian clock in that 12 hr/12 hr cycles of light/dark or warm/cold result in a 24-hr rhythm of conidiation with the peak of conidiation occurring just prior to the transition from dark to light or from cold to warm, the time point defined as subjective dawn. Known players involved in circadian and light regulation of these processes include the products of the frq, wc-1, wc-2, and vvd genes (LOROS and DUNLAP 2001 ). However, there is evidence that other genes also contribute to the circadian system in Neurospora and might comprise independent, if not circadian, feedback loops (e.g., LOROS and FELDMAN 1986 ; MERROW et al. 1999 ; LAKIN-THOMAS and BRODY 2000 ; RAMSDALE and LAKIN-THOMAS 2000 ).

Part of clock-controlled gene expression in Neurospora is exercised at the level of transcription (LOROS and DUNLAP 1991 ). Over the last decade, clock-controlled genes (ccgs), which exhibit cycling mRNA levels, have been identified by subtractive hybridization, differential screening of cDNA libraries, and expressed sequence tag (EST) sequencing (LOROS et al. 1989 ; BELL-PEDERSEN et al. 1996B ; ZHU et al. 2001 ). Another method for the identification of genes that are differentially regulated is microarray technology. In recent years, microarrays have been widely used to analyze gene expression in a growing number of organisms, including plants, mammals, bacteria, and yeast (SCHENA et al. 1995 ; DERISI et al. 1997 ; WILSON et al. 1999 ; HARMER et al. 2000 ; PEROU et al. 2000 ). For filamentous fungi, cDNA microarrays have been established for Trichoderma reesei (CHAMBERGO et al. 2002 ) and for N. crassa (LEWIS et al. 2002 ).

Here we report the use of microarrays for the identification of novel clock-controlled genes in N. crassa and for a comparison of temperature control of gene expression in the wild-type and a frq null mutant strain. This analysis allowed us to evaluate existing models for temperature-influenced clock-regulated gene expression. One recent model (MERROW et al. 1999 ) states that FRQ/WCC are required only for light input, with temperature input and output not requiring FRQ. To be consistent with this model, temperature regulation of clock-controlled genes should be largely independent of frq and therefore similar in the knockout strain frq¹⁰ and in the wild type. However, we found that clock-controlled genes that are also temperature-regulated lose both temperature and clock regulation in the frq¹⁰ strain. This demonstrates that at least part of temperature-regulated gene expression in Neurospora depends on frq and on a fully functional circadian clock.

	MATERIALS AND METHODS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Strains and growth conditions:
The following N. crassa strains were used: frq⁺ (wild type) strain 87-3 (bd; a), long period mutant 585-7 (bd; frq⁷ a), and frq knockout strain 86-1 (bd; frq¹⁰ A; ARONSON et al. 1994B ). Neurospora media (Vogel's) were as described (DAVIS and DESERRES 1970 ). Culture conditions for rhythmic RNA analysis and light-induction experiments were similar to published methods (HEINTZEN et al. 2001 ) with the following modifications: For analysis under free-running conditions in constant darkness (DD), strains were grown in 100-ml Erlenmeyer flasks in 50 ml of liquid culture medium with 1x Vogel's salts, 2% glucose, 0.5% arginine, and 50 µg/liter biotin at 25°. For time courses, the following times in constant darkness were used for array experiments: DD12, DD16, DD20, DD24, and DD28 for strain 87-3 and DD17, DD22, DD27, DD32, and DD37 for strain 585-7. Samples were kept in constant light prior to transfer to constant darkness, and transfer to DD was staggered to ensure that samples were of similar developmental age at the time of harvest (LOROS and DUNLAP 1991 ). Neurospora does not respond to light with wavelengths >550 nm (red light), so for temperature-entrainment experiments mycelial mats were grown in constant darkness or red light (DD) for 48 hr at 22°. Mycelial disks (1 cm in diameter) were cut from mats in DD and transferred to liquid medium. These samples were then transferred to entrainment conditions (cycles of 12 hr at 22°/12 hr at 27°) and harvested after at least 44 hr under entrainment (for a representation of the temperature entrainment, see Figure S1 in supplementary material at http://www.genetics.org/supplemental/). For array experiments, the following five time points were used (time in hours given after the last change of temperature from high to low or low to high, respectively; for exact times, see individual experiments): 10–11.5 hr at 27°, 2–6 hr at 22°, 10–11.5 hr at 22°, 2 hr at 27°, and 4–6 hr at 27°.

Preparation and analysis of RNA and proteins:
RNA was prepared as described previously (YARDEN et al. 1992 ; HEINTZEN et al. 2001 ) and poly(A) RNA was extracted from total RNA with a poly(A) tract kit according to the manufacturer's protocol (Promega, Madison, WI). Integrity of all RNAs were verified by agarose gel and Northern blot analysis using genes with known expression patterns as probes prior to poly(A) RNA extraction. Northern blots were prepared and hybridized according to standard techniques (MANIATIS et al. 1982 ) using DNA probes or riboprobes synthesized with DECAprime II kit or Strip-EZ RNA T3/T7 kit (Ambion, Austin, TX). Preparation of protein extracts, SDS-PAGE, and Western blot analysis were performed as described before (DENAULT et al. 2001 ).

Generation of a unigene library and microarray preparation:
The generation of two time-of-day-specific N. crassa cDNA libraries and the sequencing and analysis of 13,000 clones from these libraries have been described (ZHU et al. 2001 ). A unigene library was established by choosing a single clone representing each gene in the libraries. The unigene library consists of 1100 clones in 12 microtiter plates (plates UA–UL). As it was found that there were errors in the original annotation of the morning and evening libraries (a common problem in EST libraries, KNIGHT 2001 ), most of the clones of the unigene library have been resequenced at the Advanced Center for Genome Technology (University of Oklahoma, Norman, OK) as described previously (ZHU et al. 2001 ). A copy of the unigene library was deposited with the Fungal Genetics Stock Center (University of Kansas Medical Center, Kansas City). Sequence information can be found in the online supplementary material at http://www.genetics.org/supplemental/. Amplification of the cDNA inserts from the library was done according to HEDGE et al. 2000 using the universal forward primer (5' GACGTTGTAAAACGACGGCC) and the universal reverse primer (5' CACAGGAAACAGCTATGACC). PCR results were verified by agarose gel electrophoresis and the PCR products were purified using a QIAquick multiwell PCR purification kit according to the manufacturer's protocol (QIAGEN, Hilden, Germany). PCR products were then dissolved in 3x SSC buffer and spotted onto CLONTECH (Palo Alto, CA) DNA-ready type II slides or dissolved in 50% DMSO and spotted onto GAPSII slides (Corning, Acton, MA) with a GMS 417 arrayer (Affymetrix, Santa Clara, CA). Each PCR product was spotted twice on each slide. Postprinting slide procedures were performed according to the manufacturer's recommendations.

Microarray target preparation, hybridization, and scanning:
For time course experiments in DD, microarray targets were made using a CLONTECH Atlas glass fluorescent labeling kit and FluoroLink Cy3 or Cy5 monofunctional dyes (Amersham Pharmacia, Piscataway, NJ) according to the CLONTECH protocol. A total of 3–5 µg of DNaseI-treated poly(A) RNA was used as starting material for each target, and a mixture of 2 µg oligo(dT) and 2 µg random hexamer oligonucleotides (both from GIBCO Invitrogen, Carlsbad, CA) was used as a primer for reverse transcription. For time course experiments in DD, each slide was hybridized with two targets, one "experimental target," which was made from the RNA of interest and varied for each slide, and one "reference RNA," which consisted of a combination of RNAs from different growth conditions and was the same for each slide used in a given experiment. (RNAs used as references were extracted from mycelia grown as described above and harvested at different DD times, with some of them being given a 5-min light pulse 15 min to 1 hr prior to harvest.) "Experimental" and "reference" RNAs were labeled with Cy3 and Cy5 dyes, respectively, and the dyes were switched in one of the biologically independent replicas of each experiment. Hybridization was done for 16 hr at 50° in CLONTECH GlassHyb hybridization solution using a CLONTECH hybridization chamber, and slides were washed in wash solution according to the manufacturer's recommendations (CLONTECH).

For temperature-entrainment experiments, microarray targets were made from DNase-treated total RNA or poly(A) RNA (7 or 1 µg, respectively). Reverse transcription was done in the presence of aminoallyl-dUTP (Sigma, St. Louis), 2 µg oligo(dT), 5 µg random hexamer oligonucleotides, and 600 units Superscript II reverse transcriptase (GIBCO, Gaithersburg, MD) in a 30-µl reaction at 42°. Cy3 or Cy5 dyes were coupled in a second step in 0.5 M sodium bicarbonate buffer (pH 9.0). Targets were cleaned with QIAquick PCR purification kit (QIAGEN) and vacuum dried. For hybridization, slides were prehybridized in 5x SSC, 0.1% SDS, 1% BSA for 45 min at 42°. Targets were resuspended in 14-µl hybridization solution (50% formamide, 5x SSC, 0.1% SDS) and hybridized in the presence of 0.1 µg/µl ssDNA and 0.2 µg/µl tRNA (both Sigma) for 16 hr at 42°. For each temperature experiment, five or six samples corresponding to five or six different time points for frq wild type and frq¹⁰, respectively, were labeled with Cy3 and Cy5, respectively, and hybridized to five different slides so that the corresponding time points of the two different strains were hybridized to the same slide. In repeat experiments, dyes were exchanged between the strains. Slides were washed twice (1 min and 5 min) in 2x SSC, 0.1% SDS, 10 min in 0.1x SSC, 0.1% SDS, and 1 min in 0.1x SSC. They were briefly rinsed first in distilled water and then in 95% ethanol and dried by a brief centrifugation step. All washes were done at room temperature except the first one (42°). Slides were scanned using a GMS 418 Scanner (Affymetrix) and stored as tiff files.

Primary microarray data analysis:
Analysis of tiff files from arrays was done with ScanAlyze (written by Michael Eisen, Stanford University, http://rana.lbl.gov/EisenSoftware.htm). Grids were predefined and manually adjusted to ensure optimal spot recognition. Spots with dust or locally high background were discarded. The resulting data files were further analyzed in Excel (Microsoft, Redmond, WA) or by using Cluster and Treeview (EISEN et al. 1998 ). Thresholds for CH1GTB1 and CH2GTB1 values calculated by ScanAlyze were set to 0.55 or 0.65 to eliminate spots that had signals not significantly above background levels. For the final data analysis, data points were averaged from two replicates for each PCR product on each slide. To correct for differences between slides or for uneven loss of samples during target preparation, one of the following normalization methods was employed: For time courses in DD, the "experimental RNA" value (for each spot on the slide) was divided by the "reference RNA" value for each spot. Alternatively, for normalization within time courses, the average fluorescence value for the whole slide was determined for each slide within an experimental series and a normalization factor was determined. The resulting expression patterns after normalization for previously characterized control genes were similar to the expected behavior, thereby verifying that our microarray hybridizations could reproducibly detect changes in gene expression patterns.

For the time course experiments, the corrected value for each cDNA clone was divided by the value of DD24 [24 hr in darkness, equal to circadian time (CT) 15 for strain 87-3] or the value of DD32 (equal to CT15 for strain 585-7), thereby setting the CT15 value for each clone at 1. The CT15 value is typically the trough of the rhythm. The resulting values give the amplitude of change for each cDNA clone within a time course relative to CT15. CT is a formalism whereby the endogenous circadian biological day (22 hr in the wild type, 29 hr in frq⁷) is divided into 24 equal parts so that equivalent phases in the cycle can be compared in strains having different period length. By convention, CT0 corresponds to subjective dawn on a 12 hr/12 hr light/dark cycle, so CT15 corresponds roughly to 3 hr after subjective dusk. Normalization of the temperature-entrainment experiments was done similarly to the experiments under free-running conditions (DD) in that the expression value for each gene at one of the five or six time points (usually 10 hr or 11.5 hr at 27°) was set to 1. Expression values at the other four time points are then given in relation to this time point. Alternatively, expression levels were compared between the wild type and the frq¹⁰ mutant strain by dividing the wild-type value for each cDNA clone by the frq¹⁰ value. Using this method of normalization allowed us to focus on changes within the expression of a given gene by eliminating information about absolute expression levels. Normalized expression values for all microarray experiments can be found in the supplemental Table S1 at http://www.genetics.org/supplemental/.

Microarray data analysis to identify clock-controlled and temperature-regulated genes:
After normalization as described above, microarray data from free-running and temperature-entrainment time courses were subjected to the following analysis to identify cycling genes: The resulting data files were further analyzed in Excel (Microsoft) and using Cluster and Treeview (EISEN et al. 1998 ) to find clones that fulfilled the following criteria: (1) Expression patterns are consistent with clock control (one peak and one trough 12 hr apart in wild type connected by increasing or decreasing values, respectively, within the observed time span) in at least three out of the four experiments; (2) amplitudes are at least 2-fold in one and at least 1.8-fold and 1.5-fold in the other two experiments; and (3) peaks and troughs of expression for the individual experiments are in phase (no more than 4 hr apart). Data files from free-running experiments were additionally analyzed using the CORRCOS algorithm (M. Straume, University of Virginia Center for Biomathematical Technology, Charlottesville, VA) as initially described in HARMER et al. 2000 and also as used by DUFFIELD et al. 2002 to identify clock-controlled genes. CORRCOS empirically tests for statistically significant (P < 0.05) cross-correlation between the time series arising from hybridization to each microarray probe and cosine waves of specific period and phase. A significant correlation with a cosine wave of an optimal period between 18 and 28 hr is reported by the algorithm. The analysis is independent of signal strength and amplitude of change. The CORRCOS algorithm was used to identify rhythmic expression patterns in the four free-running time course experiments. Genes that were identified in at least two out of the four experiments were carefully reexamined to determine whether they were consistent with the criteria described under 1–3. For the free-running experiments, both the CORRCOS analysis, which is amplitude independent, and the analysis using Excel, Cluster, and Treeview, identified several genes that generally had cycling expression patterns but did not meet the amplitude criteria or were rhythmic in only two out of four experiments. These genes, which might comprise a set of weak clock-controlled genes, can be found in supplemental Table S2 at http://www.genetics.org/supplemental/.

Supplementary data and tables:
Primary data and analyses too extensive to be printed in the journal have been deposited in a publicly accessible form at http://www.genetics.org/supplemental/.

	RESULTS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

We have assembled an N. crassa unigene library from two existing cDNA libraries (ZHU et al. 2001 ). The unigene library represents 1000 different genes and was used to generate cDNA microarrays. With these arrays, we examined the changes that occur in the abundance of transcripts during the course of the circadian day and during temperature-entrainment conditions. A table with hybridization results can be found in the supplementary material (supplemental Table S1 at http://www.genetics.org/supplemental/). In general, we found that microarrays accurately reported qualitative changes in gene expression known to occur from previous work, but quantitatively revealed smaller responses than prior Northern blot or real time-PCR approaches did, as had also been found by other studies in our own and other laboratories (DUFFIELD et al. 2002 ).

Identification of clock-controlled genes:
For time courses in DD, five different time points spanning a circadian cycle were investigated in each experiment. Three independent experiments were carried out with tissues from a clock wild-type strain (strain 87-3, frq⁺, period length 21.6 hr) and one experiment was performed using a frq⁷ mutant (strain 585-7). This mutant has a period length (29 hr) longer than that of the wild type and is used to verify that gene expression patterns that appear to be rhythmic are truly under control of the circadian clock rather than of developmental regulation (LOROS et al. 1989 ; BELL-PEDERSEN et al. 1996B ). For the experiments with the wild-type and the frq⁷ strains, DD times were chosen to correspond to similar CTs, thereby allowing a direct comparison between the wild-type and the frq⁷ samples. The time courses used for array hybridization experiments were chosen to cover almost an entire circadian day (CT1–CT19) to allow observation of at least one peak and trough of transcript abundance for each clock-regulated gene. Genes with expression patterns consistent with clock control were identified using two different data analysis methods: In one approach, genes were analyzed according to their expression patterns and amplitudes using Excel, Cluster, and Treeview (see MATERIALS AND METHODS for details); in the second approach, we used the CORRCOS algorithm to identify cycling genes. With both methods, several previously known clock-controlled genes, e.g., ccg-1, eas (ccg-2), ccg-13, and lyz (LOROS et al. 1989 ; ZHU et al. 2001 ) were among the genes identified. In general, CORRCOS identified fewer rhythmic genes than the other approach, probably reflecting the relatively short time courses that were used for the four experiments.

Using a combination of these two approaches, 27 genes were identified as robustly clock controlled (Fig 1, Table 1). [ Table 1 associates unigene/EST clone names with corresponding unique locus identifier (NCU) numbers found in the annotated Neurospora genomic sequence at http://www-genome.wi.mit.edu/annotation/fungi/neurospora/.] Six of these were found to be strongly rhythmically expressed in all four experiments; among these are the known robustly clock-controlled genes ccg-1, eas (ccg-2), ccg-13, and ccg-14 (LOROS et al. 1989 ; ZHU et al. 2001 ). The other two strongly rhythmic genes (unigene clones UBh05 and ULd10, neither of which has significant homology to genes previously associated with a phenotype) had not previously been identified. Some other genes that have been identified previously as clock controlled did not meet the threshold criteria, but we found in a previous study that ccgs vary greatly with respect to the level of clock control (ZHU et al. 2001 ), and our use of stringent thresholds might lead to exclusion of more weakly clock-controlled genes. Also, tissues used in this investigation were grown in medium with 2% glucose, a condition that increases the amplitudes of several known clock-controlled genes but decreases amplitudes of others (BELL-PEDERSEN et al. 1996B ). All the putative clock-controlled genes that were identified show peaks in expression late in the subjective night or early in the subjective morning (CT19–CT6). These data establish a broad baseline characterization by identifying 2.7% of the genes as being regulated by circadian rhythms at the level of RNA abundance. An additional 32 genes that showed rhythmic expression patterns but did not meet the threshold criteria with respect to amplitude, or showed cycling expression in only two of four experiments, might comprise a set of weakly clock-controlled genes (see the supplemental Table S2 at http://www.genetics.org/supplemental/).

View larger version (60K): In this window In a new window Download PPT slide   Figure 1. Clock-controlled and temperature-regulated gene expression. Cluster and Northern blot analysis of clock-controlled and temperature-dependent gene expression of a subset of genes from the microarrays. (A) Cluster analysis and visualization were done with Cluster and Treeview (EISEN et al. 1998 ). The graph shows three out of four array experiments carried out under free-running conditions, or temperature-entrainment conditions, and the ratios for frq+/frq10 at the end of the warm temperature period (first time point in each temperature-entrainment experiment) for all four experiments. Growth conditions are indicated above each experiment. The cluster is divided into three sections, the top section containing genes that are both clock controlled and temperature regulated in the wild type, the middle section containing genes that are clock controlled but not robustly temperature regulated, and the bottom section contains a selection of genes that are not regulated under either condition. Unigene clone numbers are given on the right, and their identities, when known, are given in Table 1 along with their unique genome annotation locus identifiers. (B) Representation of one of the genes from each subset from A; labeling of the x-axis indicates hours in the respective temperature for temperature-entrainment experiments or CT times for free-running experiments. (C) Northern blot verification of two newly identified clock-controlled genes. Total RNA (30 µg/lane) was extracted from the frq7 strain after growth in darkness for the indicated times in DD. The corresponding CT is given below the DD values. Below each blot, values (normalized to ethidium-bromide-stained rRNAs on the gels) are plotted. The lowest value was set to 1. Riboprobes used are indicated in the graph. For each gene, two independent RNA sets were hybridized; only one is shown in this figure.

To verify our array results, we tested six genes by Northern blot analysis (Fig 1C; due to space constraints, only two time courses are shown). For all genes, at least two Northern blots with RNAs from independent time courses were investigated; one time course originated from the wild type and one from a frq⁷ mutant (strain 585-7). The Northern analyses confirmed our array data and demonstrated that amplitudes of 1.5- to 2-fold on arrays, which could reproducibly be followed when found in repeated independent array experiments, correspond to 3- to 10-fold amplitudes when assayed by alternative means.

Temperature entrainment of many clock-controlled genes is dependent on frq:
Temperature is one of the main signals that keep the circadian clock in synchrony with the environment (LIU et al. 1998 ). In this study, we investigated the effects of a 12 hr at 22°/12 hr at 27° temperature regime on gene expression. This is a temperature change capable of yielding robust circadian entrainment but well below that eliciting a heat-shock response. Five to six time points covering changes from high to low temperature and vice versa within a subjective day were investigated in each of four independent array experiments. The strains we used for these experiments were a frq⁺ strain and a frq knockout mutant (frq¹⁰ allele) that lacks normal circadian rhythmicity in constant darkness (ARONSON et al. 1994A , ARONSON et al. 1994B ). Normalization of the temperature-entrainment experiments was similar to the experiments under free-running conditions (DD) in that the expression value for each gene at one of the five or six time points was set to 1. The time point chosen as a reference is the end of the warm period (Fig 1, 10–11.5 hr at 27°) within each time course. This time point corresponds to subjective evening and marks the trough expression levels for most of the clock-controlled genes. Expression values at the other four or five time points are then given in relation to this time point. Additionally, we compared expression levels of the wild type vs. frq¹⁰ at the end of the warm period when FRQ protein in the wild type is at its peak (Fig 2 and Fig 3). For this analysis, wild-type transcript levels for each gene were divided by frq¹⁰ transcript levels, thereby setting frq¹⁰ levels to 1.

View larger version (59K): In this window In a new window Download PPT slide   Figure 2. Under temperature entrainment, clock-controlled genes peak when FRQ protein is at its trough. Northern analysis of total RNA (30 µg/lane) from the wild type (frq+) and the frq10 mutant were probed with riboprobes for ccg-1 as an example of a clock- and temperature-controlled gene (Fig 1A, top) or the gene for ribosomal protein L6 as an example of a non-clock-, non-temperature-controlled gene (Fig 1A, bottom). Western blots of protein extracts (80 µg/lane) from the wild-type strain were probed with anti-FRQ-antibody. Hours of incubation at the indicated temperature are given above each lane. Below, normalized values are plotted for ccg-1 in the wild type and in frq10 and for FRQ protein in the wild type. Normalization was to ethidium-bromide-stained rRNAs on the gel or to Coomassie-stained proteins on the membranes. The lowest value was set to 1.

View larger version (51K): In this window In a new window Download PPT slide   Figure 3. A comparison of transcript levels between frq+ and frq10 under temperature-entrainment conditions shows opposite responses to loss of frq among frq-dependent genes. The chart at the left gives the average ratio and standard deviation (sd) of ccg transcript levels from four independent experiments comparing frq+/frq10 transcript levels at the end of the warm period. Only genes that are under temperature control and also under clock control under free-running conditions are shown here. The first column gives the unigene clone designation; the second column gives significant BLAST homologies. For more information about putative homologs, see Table 1. Horizontal lines separate groups according to frq+/frq10 ratios. On the right are Northern blot analyses of ccg-1, eas (ccg-2), and the gene for ribosomal protein L6, examined under two different temperature conditions (end of the warm period, 27°; end of the cold period, 22°). Northern analysis was performed with 30 µg of total or 0.1 µg of poly(A) RNA.

The temperature-entrainment experiments led to the identification of 14 genes that respond to temperature changes in the wild type and have expression patterns consistent with circadian regulation under entrained conditions in at least three out of four experiments with the same thresholds that were used for the experiments under free-running conditions (Fig 1). We confirmed the expression of five of the genes by Northern analysis and found the same patterns as with the array analysis. Three of the genes tested were among the temperature-regulated genes and two genes were among the nonregulated genes (Fig 3; only blots for ccg-1 and the gene for ribosomal protein L6 are shown).

All 14 temperature-regulated genes show peak expression at the end of the 12 hr at 22° period (cold period), which would represent late night in the subjective circadian day and coincides with the trough of FRQ protein under temperature-entrainment conditions (Fig 2). The group of genes consists solely of genes already identified as clock controlled, and among them are all the genes seen to be most robustly clock controlled, that is, those identified as rhythmic in all array experiments carried out under free-running conditions. Surprisingly, then, no genes that were not already identified as clock regulated were identified as temperature regulated. From the 32 genes that were identified as weakly rhythmic under free-running conditions, only four were temperature controlled (supplemental Table S2 at http://www.genetics.org/supplemental/). Another group of 22 genes peaked 2–6 hr after the onset of the 27° (warm) period, but these genes showed regulation in only two out of four experiments, although not all of them were regulated in the same experiments. None of these genes are clock controlled under free-running conditions, but as they did not show consistent temperature-dependent expression patterns, we did not investigate them further.

Overall, this means that of the 27 clock-controlled genes comprising 2.7% of the genes on the arrays, 52% (14 of 27) respond dependably to temperature regulation, whereas none of the non-clock-controlled genes do so consistently even though they comprise 98% of the genes examined. Interestingly, the 14 clock-controlled genes that are also temperature regulated completely lost temperature-regulated expression in frq¹⁰ and displayed uniform or randomly changing expression levels over the time course (Fig 1). The same was true for the four weakly clock-controlled genes that were temperature regulated in the wild type (supplemental Table S2 at http://www.genetics.org/supplemental/). Thus, among the genes on our arrays, the clock-controlled genes that are also temperature controlled invariably lost their temperature responses in frq¹⁰, a strain in which the clock gene frq is deleted.

To see whether there was any diversity in the regulation of the temperature-responsive genes, we further analyzed these genes by comparing their transcript levels in wild type to those seen in frq¹⁰ at the end of the warm period (Fig 3). At this time point, FRQ protein in the wild type reaches its peak (Fig 2) and transcript levels for all the rhythmic genes investigated reach their troughs (Fig 1 and Fig 2). By choosing this time point, we thereby compare basal levels of gene expression and also compare at a stage where large amounts of FRQ protein are present in the wild type. Genes were sorted into groups depending on whether they were downregulated, upregulated, or similar compared to frq¹⁰ (Fig 3). The results were also verified by Northern blot analysis (Fig 3). Ten of the genes were downregulated from 1.5- to 6.5-fold in the wild type compared to frq¹⁰. One, eas (ccg-2), was significantly upregulated, and three did not show any difference. Implications of these findings are discussed below.

	DISCUSSION

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Microarray analysis of N. crassa reveals novel clock-controlled genes:
Much progress is being made toward the unraveling of the molecular mechanisms of the Neurospora circadian clock (LOROS and DUNLAP 2001 ). Nevertheless, control of output gene expression by the clock is probably the area of clock research that is least understood. Clock-regulated genes have been identified by several different screening approaches (LOROS et al. 1989 ; BELL-PEDERSEN et al. 1996B ; ZHU et al. 2001 ), but the extent to which the circadian clock controls gene expression within the whole genome remains unclear. The development of microarray technologies allows investigation of gene expression patterns within large parts of genomes or within complete genomes (DUGGAN et al. 1999 ; BLOHM and GUISEPPI-ELIE 2001 ). For Neurospora, microarrays have been used to investigate light-regulated and wc-1-dependent gene expression, thereby allowing analysis of light signal transduction through the circadian clock (LEWIS et al. 2002 ).

In this study, we used microarrays to analyze changes in gene expression patterns in N. crassa over the course of the circadian day. Using microarrays providing signals for 1000 different genes, we were able to identify 27 genes (2.7%) that had expression profiles consistent with clock control. An additional 32 genes (3.2%) seem to be weakly clock controlled, consistent with previous findings that clock control in Neurospora can vary depending on the gene and culture conditions used for the investigation (ZHU et al. 2001 ). Our results are in a range similar to findings in Arabidopsis, where Harmer and co-workers identified 6% of genes as cycling (HARMER et al. 2000 ); in Drosophila, where 1–5% of genes are rhythmic (CLARIDGE-CHANG et al. 2001 ; MCDONALD and ROSBASH 2001 ); and in mammalian cells (GRUNDSCHOBER et al. 2001 ; AKHTAR et al. 2002 ; DUFFIELD et al. 2002 ), in which 2–9% of genes are seen to be clock regulated. Northern blot analysis of several genes showed excellent agreement with our array results under the experimental conditions we investigated with respect to expression patterns, although the amplitude on the microarrays was generally somewhat lower than that observed on Northern blots.

Surprisingly, the genes found to be under control of the circadian clock do not seem to preferentially belong to specific pathways but instead cover a range of cellular functions (Table 1). All of the clock-controlled genes peak in the subjective late night or morning, as is the case with previously identified ccgs (LOROS et al. 1989 ; BELL-PEDERSEN et al. 1996B ; ZHU et al. 2001 ). We did not find any evening-specific genes; a possible reason for this might be that the subset of genes on our arrays does not necessarily reflect all subsets of genes found in the Neurospora genome (see below). Recent findings in Drosophila have suggested the existence of clusters of functionally related genes in the genome that are also coordinately controlled by the circadian clock (MCDONALD and ROSBASH 2001 ). To find out whether this is also true for Neurospora, we determined the genomic locations of the clock-controlled genes on our arrays by comparing their sequences to the latest release of the whole Neurospora genome sequence (http://www.genome.wi.mit.edu/annotation/fungi/neurospora/). For the genes for which linkage group and position could be determined (supplemental Table S2 at http://www.genetics.org/supplemental/), no apparent clustering of clock-controlled genes was observed.

It should be obvious on the basis of first principles, but is worth stating explicitly, that we do not believe these data are the last and final word concerning circadian-regulated gene expression in Neurospora. Our conclusions are based on a careful analysis of 1000 genes identified by deep EST sequencing using two libraries made from nondifferentiating vegetative tissue slowly starving in the dark. For this reason, genes involved in asexual development (spore formation, aerial hyphae extension, septation) and sexual development will be underrepresented, and light-induced genes will be severely underrepresented; these are three classes of genes known from previous work (LOROS et al. 1989 ; BELL-PEDERSEN et al. 1996B ; LOROS and DUNLAP 2001 ) to be highly correlated with circadian-regulated genes. It is likely that a similar study executed with an alternative set of genes would identify a different proportion of genes under clock control.

Temperature entrainment of many clock-controlled genes is dependent on the frequency gene:
Having defined a set of clock-controlled genes on our microarrays, we investigated their expression patterns under temperature-entrainment conditions, temperature being among the most important signals that reset or entrain the circadian clock (LOROS and DUNLAP 2001 ). In these experiments, we compared a frq⁺ strain and a frq knockout mutant (frq¹⁰ allele). Another frq loss-of-function mutant, frq⁹, has been shown to retain some conidial banding every 19 hr when driven under 9.5 hr 22°/9.5 hr 27° temperature-entrainment conditions (MERROW et al. 1999 ). On the basis of this and related findings, Merrow and co-workers have proposed a model in which the FRQ/WCC oscillator is involved solely in processing light signals, whereas yet-unidentified factors comprise the circadian "rhythm generator" and mediate temperature responses and circadian output from the system (MERROW et al. 1999 , MERROW et al. 2001A ). To be consistent with this model, temperature regulation of clock-controlled genes, to a large degree, should be independent of frq and therefore similar in knockout strain frq¹⁰ and in the wild type. However, we found a significant loss of temperature-regulated gene expression in frq¹⁰ among the genes on our arrays: All 14 genes that are consistently temperature controlled in the wild type do not show any significant temperature regulation in frq¹⁰. They are under clock control under free-running conditions in the wild type, indicating that this entire set of clock-controlled genes depends on the presence of frq for temperature-regulated expression. Exclusive of circadian regulatory issues, an interesting and unexpected corollary of this result is that the FRQ/WCC-based circadian regulatory system appears likely to mediate much of the organism's transcriptional responses to small temperature changes, acting as a thermal sensor in the organism. This derivative function of a circadian oscillator as a sensory module has not been previously imagined.

There is now considerable evidence that other oscillatory loops in addition to the frq-dependent oscillator contribute to the manifestation of overt circadian rhythmicity in Neurospora (e.g., LOROS and FELDMAN 1986 ; LOROS et al. 1989 ; DUNLAP 1998 ; MERROW et al. 1999 ; LAKIN-THOMAS and BRODY 2000 ; RAMSDALE and LAKIN-THOMAS 2000 ). A model that tries to integrate these results is shown in Fig 4. The findings presented here, together with previous data (e.g., LOROS and FELDMAN 1986 ; ARONSON et al. 1994B ; MERROW et al. 1999 , MERROW et al. 2001B ; LAKIN-THOMAS and BRODY 2000 ), may indicate that the frq-dependent oscillator as well as other noncircadian feedback loops may each be involved in processing a given signal but regulate a distinct portion of output genes. The loss of part of the temperature regulation may explain why the overt rhythm on a race tube, the circadian developmental phenotype, in loss-of-function mutants of frq is not as robust as seen in wild type under temperature-entrainment conditions (MERROW et al. 1999 ). Furthermore, previous studies (BELL-PEDERSEN et al. 2001 ) and this analysis indicate that frq-dependent regulation itself occurs through several pathways: ccg expression levels in the wild type at the peak time of FRQ protein can be up- or downregulated compared to frq¹⁰, depending on the gene (Fig 3). Also, it is quite likely that output genes that are controlled through a certain pathway under one condition may be controlled by a different pathway under another condition; combinatorial control of gene expression by activators is well known (e.g., BELL-PEDERSEN et al. 1996A ), and this may also occur with circadian expression under varying conditions. A free run at a constant temperature could clearly constitute a growth condition different from that associated with a repeating 5° temperature cycle. This difference might be an explanation for the finding that some of the clock-controlled genes on our arrays do not show any significant regulation under temperature-entrainment conditions in either the wild type or the frq¹⁰ mutant. This seems to be especially the case for genes that are not robustly clock controlled under free-running conditions; that is, weakly rhythmic genes were often not temperature regulated, whereas genes that were strongly clock controlled in all four array experiments were also temperature regulated. This observation is further supported by the fact that among the 32 genes identified as weakly rhythmic, only 4 are temperature regulated in the wild type (and were not regulated in frq¹⁰), while the other 28 do not show significant temperature-dependent expression in either the wild type or the frq¹⁰ mutant.

View larger version (23K): In this window In a new window Download PPT slide   Figure 4. Model summarizing temperature-, light-, and clock-dependent gene regulation in N. crassa. Clock-regulated light signal transduction is mediated by the FRQ/WCC oscillator (CROSTHWAITE et al. 1995 ). WC-1 is the circadian blue light photoreceptor (FROEHLICH et al. 2002 ; HE et al. 2002 ), and light-regulated phenotypes depend on the products of the white-collar genes, WC-1 and WC-2 (BALLARIO et al. 1996 ; CROSTHWAITE et al. 1997 ; LINDEN and MACINO 1997 ; TALORA et al. 1999 ), VVD (HEINTZEN et al. 2001 ; SCHWERDTFEGER and LINDEN 2003 ), or possibly other uncharacterized photoreceptors (DRAGOVIC et al. 2002 ), but not directly on FRQ. Temperature changes affect the FRQ/WCC oscillator (LIU et al. 1998 ), and in this study, we have shown that within a sample of 1000 genes, all temperature-entrained, clock-controlled gene expression depends on FRQ for relay to the output genes; an alternative model had suggested that another oscillator (FLO, FRQ-less oscillator) mediated circadian temperature regulation (MERROW et al. 1999 , MERROW et al. 2001A ). Solid lines indicate regulatory relationships for which molecular components have been identified; dotted lines indicate parts that can be inferred from physiological experiments but for which molecular components have yet to be found, where alternative pathways cannot be distinguished, or where regulation can be imagined. Signals besides light and temperature, e.g., nutritional changes, are not shown here, but have been demonstrated to influence overt rhythmicity under certain conditions (e.g., LAKIN-THOMAS and BRODY 2000 ).

Given the appearance of even weak noncircadian developmental rhythms in a frq loss-of-function mutant, however, it remains plausible that some genes may be under temperature control through means other than the FRQ/WCC oscillator even though they were not identified in this screen of a thousand genes. This hypothesis is especially appealing in view of the fact that only a small portion of genes seems to be regulated at the level of transcription under temperature-entrainment conditions. This might be due to gene selection on our arrays, and additionally, parts of temperature-dependent regulation of the circadian system are known to occur at a post-transcriptional level (LIU et al. 1998 ) and would therefore not be apparent when using cDNA microarray analysis. Given the complexity of circadian regulation, a considerable overlap may exist in input integration as well as in output control among different oscillatory loops at various levels of gene expression.

	FOOTNOTES

¹ Present address: Lehrstuhl für Allgemeine und Molekulare Botanik, Ruhr-Universität Bochum, 44780 Bochum, Germany.

	ACKNOWLEDGMENTS

The authors thank Carol Ringelberg for excellent help with the microarray experiments; Doris Kupfer, Bruce A. Roe, and Hua Zhu for sequencing of the unigene library clones; Deanna Denault for help with the Western blots; Martin Straume for the generous provision of the CORRCOS algorithm; and Hildur Colot for helpful discussion of the manuscript. This research was supported by grants MH44651 to J.C.D. and J.J.L., National Science Foundation grant no. MCB-0084509 to J.J.L. and no. R37GM34985 to J.C.D. M.N. received an Emmy Noether fellowship from the Deutsche Forschungsgemeinschaft (Bonn, Germany) and G.E.D. received a Prize Travelling Research Fellowship (058332/B/99/Z) from the Wellcome Trust.

Manuscript received November 11, 2002; Accepted for publication February 21, 2003.

	LITERATURE CITED

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

AKHTAR, R. A., A. B. REDDY, E. S. MAYWOOD, J. D. CLAYTON, and V. M. KING et al., 2002  Circadian cycling of the mouse liver transcriptome, as revealed by cDNA microarray, is driven by the suprachiasmatic nucleus. Curr. Biol. 12:540-550.[Medline]

ARONSON, B., K. JOHNSON, J. J. LOROS, and J. C. DUNLAP, 1994a  Negative feedback defining a circadian clock: autoregulation in the clock gene frequency.. Science 263:1578-1584.[Abstract/Free Full Text]

ARONSON, B. D., K. A. JOHNSON, and J. C. DUNLAP, 1994b  Circadian clock locus frequency: a single ORF defines period length and temperature compensation. Proc. Natl. Acad. Sci. USA 91:7683-7687.[Abstract/Free Full Text]

BALLARIO, P., P. VITTORIOSO, A. MAGRELLI, C. TALORA, and A. CABIBBO et al., 1996  White collar-1, a central regulator of blue-light responses in Neurospora crassa, is a zinc-finger protein. EMBO J. 15:1650-1657.[Medline]

BELL-PEDERSEN, D., J. C. DUNLAP, and J. J. LOROS, 1996a  Distinct cis-acting elements mediate clock, light, and developmental regulation of the Neurospora crassa eas (ccg-2) gene. Mol. Cell. Biol. 16:513-521.[Abstract]

BELL-PEDERSEN, D., M. SHINOHARA, J. LOROS, and J. C. DUNLAP, 1996b  Circadian clock-controlled genes isolated from Neurospora crassa are late night to early morning specific. Proc. Natl. Acad. Sci. USA 93:13096-13101.[Abstract/Free Full Text]

BELL-PEDERSEN, D., Z. A. LEWIS, J. J. LOROS, and J. C. DUNLAP, 2001  The Neurospora circadian clock regulates a transcription factor that controls rhythmic expression of the output eas(ccg-2) gene. Mol. Microbiol. 41:897-909.[Medline]

BLOHM, D. H. and A. GUISEPPI-ELIE, 2001  New developments in microarray technology. Curr. Opin. Biotechnol. 12:41-47.[Medline]

CHAMBERGO, F. S., E. D. BONACCORSI, A. J. S. FERREIRA, A. S. P. RAMOS, and J. RIBAMAR FERREIRA, JR. et al., 2002  Elucidation of the metabolic fate of glucose in the filamentous fungus Trichoderma reesei using expressed sequence tag (EST) analysis and cDNA microarrays. J. Biol. Chem. 277:13983-13988.[Abstract/Free Full Text]

CHENG, P., Y. YANG, and Y. LIU, 2001  Interlocked feedback loops contribute to the robustness of the Neurospora circadian clock. Proc. Natl. Acad. Sci. USA 98:7408-7413.[Abstract/Free Full Text]

CLARIDGE-CHANG, A., H. WIJNEN, F. NAEF, C. BOOTHROYD, and N. RAJEWSKY et al., 2001  Circadian regulation of gene expression systems in the Drosophila head. Neuron 32:657-671.[Medline]

CROSTHWAITE, S. C., J. J. LOROS, and J. C. DUNLAP, 1995  Light-induced resetting of a circadian clock is mediated by a rapid increase in frequency transcript. Cell 81:1003-1012.[Medline]

CROSTHWAITE, S. C., J. C. DUNLAP, and J. J. LOROS, 1997  Neurospora wc-1 and wc-2: transcription, photoresponses, and the origins of circadian rhythmicity. Science 276:763-769.[Abstract/Free Full Text]

DAVIS, R. L. and D. DESERRES, 1970  Genetic and microbial research techniques for Neurospora crassa.. Methods Enzymol. 27A:79-143.

DENAULT, D. L., J. J. LOROS, and J. C. DUNLAP, 2001  WC-2 mediates WC-1-FRQ interaction within the PAS protein-linked circadian feedback loop of Neurospora. EMBO J. 20:109-117.[Medline]

DERISI, J. L., V. R. IYER, and P. O. BROWN, 1997  Exploring the metabolic and genetic control of gene expression on a genomic scale. Science 278:680-686.[Abstract/Free Full Text]

DRAGOVIC, Z., Y. TAN, M. GORL, T. ROENNEBERG, and M. MERROW, 2002  Light reception and circadian behavior in "blind" and "clock-less" mutants of Neurospora. EMBO J. 21:3643-3651.[Medline]

DUFFIELD, G. E., J. D. BEST, B. H. MEURERS, A. BITTNER, and J. J. LOROS et al., 2002  Circadian programs of transcriptional activation, signaling, and protein turnover revealed by microarray analysis of mammalian cells. Curr. Biol. 12:551-557.[Medline]

DUGGAN, D., M. BITTNER, Y. CHEN, P. MELTZER, and J. TRENT, 1999  Expression profiling using cDNA microarrays. Nat. Genet. 21:10-14.[Medline]

DUNLAP, J. C., 1998  An end in the beginning. Science 280:1548-1549.[Free Full Text]

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

EISEN, M. B., P. T. SPELLMAN, P. O. BROWN, and D. BOTSTEIN, 1998  Cluster analysis and display of genome-wide expression patterns. Proc. Natl. Acad. Sci. USA 95:14863-14868.[Abstract/Free Full Text]

FROEHLICH, A. F., Y. LIU, J. J. LOROS, and J. C. DUNLAP, 2002  WHITE COLLAR-1, a circadian blue light photoreceptor, binding to the frequency promoter. Science 297:815-819.[Abstract/Free Full Text]

GOERL, M., M. MERROW, B. HUTTNER, J. JOHNSON, and T. ROENNEBERG et al., 2001  A PEST-like element in FREQUENCY determines the length of the circadian period in Neurospora crassa.. EMBO J. 20:7074-7084.[Medline]

GRUNDSCHOBER, C., F. DELAUNAY, A. PÜHLHOFER, G. TRIQUENEAUX, and V. LAUDET et al., 2001  Circadian regulation of diverse gene products revealed by mRNA expression profiling of synchronized fibroblasts. J. Biol. Chem. 276:46751-46758.[Abstract/Free Full Text]

HARMER, S. L., J. B. HOGENESCH, M. STRAUME, H.-S. CHANG, and B. HAN et al., 2000  Orchestrated transcription of key pathways in Arabidopsis by the circadian clock. Science 290:2110-2113.[Abstract/Free Full Text]

HE, Q., P. CHENG, Y. YANG, L. WANG, and K. GARDNER et al., 2002  WHITE COLLAR-1, a DNA binding transcription factor and a light sensor. Science 297:840-842.[Abstract/Free Full Text]

HEDGE, P., R. QI, K. ABERNATHY, C. GAY, and S. DHARAP et al., 2000  A concise guide to cDNA microarray analysis. Biotechniques 29:548-562.[Medline]

HEINTZEN, C., J. J. LOROS, and J. C. DUNLAP, 2001  The PAS protein VIVID defines a clock-associated feedback loop that represses light input, modulates gating, and regulates clock resetting. Cell 104:453-464.[Medline]

KNIGHT, J., 2001  When the chips are down. Nature 410:860-861.[Medline]

LAKIN-THOMAS, P. L. and S. BRODY, 2000  Circadian rhythms in Neurospora crassa: lipid deficiencies restore robust rhythmicity to null frequency and white-collar mutants. Proc. Natl. Acad. Sci. USA 97:256-261.[Abstract/Free Full Text]

LEE, K., J. J. LOROS, and J. C. DUNLAP, 2000  Interconnected feedback loops in the Neurospora circadian system. Science 289:107-110.[Abstract/Free Full Text]

LEWIS, Z. A., A. CORREA, C. SCHWERDTFEGER, K. L. LINK, and X. XIE et al., 2002  Overexpression of white collar-1 (wc-1) activates circadian clock-associated genes, but is not sufficient to induce most light-regulated gene expression in Neurospora crassa.. Mol. Microbiol. 45:917-931.[Medline]

LINDEN, H. and G. MACINO, 1997  White collar-2, a partner in blue-light signal transduction, controlling expression of light-regulated genes in Neurospora crassa.. EMBO J. 16:98-109.[Medline]

LIU, Y., M. MERROW, J. J. LOROS, and J. C. DUNLAP, 1998  How temperature changes reset a circadian oscillator. Science 281:825-829.[Abstract/Free Full Text]

LOROS, J. and J. C. DUNLAP, 1991  Neurospora crassa clock-controlled genes are regulated at the level of transcription. Mol. Cell. Biol. 11:558-563.[Abstract/Free Full Text]

LOROS, J. J. and J. C. DUNLAP, 2001  Genetic and molecular analysis of circadian rhythms in Neurospora. Annu. Rev. Physiol. 63:757-794.[Medline]

LOROS, J. J. and J. F. FELDMAN, 1986  Loss of temperature compensation of circadian period length in the frq-9 mutant of Neurospora crassa.. J. Biol. Rhythms 1:187-198.[Abstract/Free Full Text]

LOROS, J. J., S. A. DENOME, and J. C. DUNLAP, 1989  Molecular cloning of genes under the control of the circadian clock in Neurospora.. Science 243:385-388.[Abstract/Free Full Text]

MANIATIS, T., E. F. FRITSCH and J. SAMBROOK, 1982 Molecular Cloning: A Laboratory Manual. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY.

MCDONALD, M. J. and M. ROSBASH, 2001  Microarray analysis and organization of circadian gene expression in Drosophila. Cell 107:567-578.[Medline]

MERROW, M., M. BRUNNER, and T. ROENNEBERG, 1999  Assignment of circadian function for the Neurospora clock gene frequency.. Nature 399:584-586.[Medline]

MERROW, M., T. ROENNEBERG, G. MACINO, and L. FRANCHI, 2001a  A fungus among us: the Neurospora crassa circadian system. Semin. Cell Dev. Biol. 12:279-285.[Medline]

MERROW, M., L. FRANCHI, Z. DRAGOVIC, M. GÖRL, and J. JOHNSON et al., 2001b  Circadian regulation of the light input pathway in Neurospora crassa.. EMBO J. 20:307-315.[Medline]

PEROU, C. M., T. SORLIE, M. B. EISEN, M. VAN DE RIJN, and S. S. JEFFREY et al., 2000  Molecular portraits of human breast tumors. Nature 406:747-752.[Medline]

RAMSDALE, M. and P. L. LAKIN-THOMAS, 2000  sn-1,2-Diacylglycerol levels in the fungus Neurospora crassa display circadian rhythmicity. J. Biol. Chem. 275:27541-27550.[Abstract/Free Full Text]

SCHENA, M., D. SHALON, R. W. DAVIS, and P. O. BROWN, 1995  Quantitative monitoring of gene expression patterns with a complementary DNA microarray. Science 270:467-470.[Abstract/Free Full Text]

SCHWERDTFEGER, C. and H. LINDEN, 2003  VIVID is a flavoprotein and serves as a fungal blue light photoreceptor for photoadaptation. EMBO J. in press.

TALORA, C., L. FRANCHI, H. LINDEN, P. BALLARIO, and G. MACINO, 1999  Role of a white collar-1-white collar-2 complex in blue-light signal transduction. EMBO J. 18:4961-4968.[Medline]

WILSON, M., J. DERISI, H. H. KRISTENSEN, P. IMBODEN, and S. RANE et al., 1999  Exploring drug-induced alterations in gene expression in Mycobacterium tuberculosis by microarray hybridization. Proc. Natl. Acad. Sci. USA 96:12833-12838.[Abstract/Free Full Text]

YARDEN, O., M. PLAMANN, D. EBBOLE, and C. YANOFSKY, 1992  cot-1, a gene required for hyphal elongation in Neurospora crassa, encodes a protein kinase. EMBO J. 11:2159-2166.[Medline]

ZHU, H., M. NOWROUSIAN, D. KUPFER, H. V. COLOT, and G. BERROCAL-TITO et al., 2001  Analysis of expressed sequence tags from two starvation, time of day-specific libraries of Neurospora crassa reveals novel clock-controlled genes. Genetics 157:1057-1065.[Abstract/Free Full Text]

This article has been cited by other articles:

M. A. Woelfle, Y. Xu, X. Qin, and C. H. Johnson Circadian rhythms of superhelical status of DNA in cyanobacteria PNAS, November 20, 2007; 104(47): 18819 - 18824.