Antisense Transcription of the Neurospora Frequency Gene Is Rhythmically Regulated by CSP-1 Repressor but Dispensable for Clock Function

Neurospora wild-type strain was acquired from Fungal Genetics Stock Center (FGSC) with stock number #2489. For race tube assay, strains with ras-1^bd mutation were used (Belden et al., 2007). Since the conventional frq deletion strain frq¹⁰ (Aronson et al., 1994) contains the enhancer site c-box and the qrf LRE, we created a full knockout of the frq locus in this study. The Δfrq bd, his-3 strain was created with the yeast in vivo recombination system as described previously (Colot et al., 2006) using the following knockout cassette primers:

The plasmids for the qrf knockout strains were created with the following primers:

The Δfrq bd, his-3 (histidine auxotroph) strain was used for transformation. All constructs were targeted to the his-3 locus via homologous recombination. Briefly, 5- to 7-day-old conidia were harvested and washed and pelleted at 2600 g at 4 °C for 10 min with 50 mL of 1 M cold sorbitol. The conidial pellet was mixed with 1- to 2-μg linearized plasmid DNA and incubated on ice for 10 min. Electroporation was applied at 1.5 kV/cm, 25 μF, 600 Ω. The cells were immediately resuspended in 1 mL of 1 M cold sorbitol and plated onto Vogel’s solid media (1× Vogel’s, 1% [w/v] agar, 1× FGS [20% (w/v) sorbose, 0.5% (w/v) glucose, and 0.5% (w/v) fructose]). After incubation at 30 °C for 3-5 days, the single colonies were picked. N. crassa cultures were grown in standard growth medium (2% glucose, 0.5% l-arginine, 1× Vogel’s medium, and 10-ng/mL biotin) at 25 °C with shaking at 115 rpm. In light induction experiments, the strains were grown into mats in Petri plates with 20 mL of standard medium. Mycelial disks (1 cm) were cut out and grown in standard growth medium for 1 day.

RNA was isolated with peqGOLD TriFAST (PeqLab). The reverse transcription was carried out with QuantiTect Reverse Transcription Kit (Qiagen) with indicated primers following manufacturer’s instructions. Relative transcript levels were quantified by quantitative real-time polymerase chain reaction (PCR) in 96-well plates with LightCycler 480 (Roche). The reaction was set by using qPCRBIO Probe Mix Hi-ROX (Nippongenetics) and TaqMan (5′: 6-FAM, 3′: TAMRA) or Universal Probe Library (UPL, Roche) probes. Primers and probes are listed below. Three replicates were used to calculate the mean threshold cycle (Ct) value. The relative enrichments were quantified relative to a housekeeping gene using the reverse transcription primers for cDNA synthesis:

And primers for qPCR are as follows:

The conidial suspension was inoculated from one end of the autoclaved glass tubes containing the media (1× Vogel’s, 0.1% glucose [w/v], 0.17% arginine [w/v], 50 ng/mL biotin, and 2% agar [w/v]) and was grown for 1 day prior to light to dark transfer. The conidial growth fronts were marked every 24 h.

Sorbose medium (1× FGS [0.05% fructose (w/v), 0.05% glucose (w/v), 1% sorbose (w/v)], 1× Vogel’s, 1% agarose [w/v], 10-ng/mL biotin, and 75-μM firefly luciferin) was used for the live-cell bioluminescence assay; 96-well plates were inoculated with 3 × 10⁴ conidia per well and incubated in dark at 25 °C for 2 days. The bioluminescence signal was recorded in constant darkness or in light-dark cycles at 25 °C for the indicated time windows with a multilabel plate reader. For dark recordings, the cells were synchronized by 1-h light pulse (LP) at 100-μmol photons m⁻² sec⁻¹ prior to the measurement in constant darkness; 0.3% glucose as carbon source was used when high glucose conditions are indicated. The light intensity titration (Figure 3e) was performed as described (Cesbron et al., 2015).

Previous studies (Kramer et al., 2003; Colot et al., 2005; Xue et al., 2014) and RNA-Seq data (Cemel et al., 2017) indicate that the frq locus directs expression of multiple species of overlapping sense and antisense transcripts (Figure 1a). Light-induced expression of frq and its antisense transcript, qrf, are driven from their corresponding LREs, LRE and qLRE, respectively (Froehlich et al., 2002; He et al., 2002; Smith et al., 2010; Xue et al., 2014). The majority of sense frq RNA species terminate between the qLRE and the annotated TSSs of the qrf promoter (Belden et al., 2011), and qrf transcripts terminate within the region of annotated TSSs of the frq promoter (Figure 1a).

To analyze the impact of antisense transcription on the circadian clock, we replaced the qrf promoter with 2 different transcription termination sequences, the 3′ UTR of the ccg-2 of N. crassa (Kramer et al., 2003) and the trpC terminator of Aspergillus nidulans (Mullaney et al., 1985), respectively. The sequences were inserted at the BssHII site to ensure deletion of all mapped qrf TSSs (Figure 1b, left). The modified frq genes and an unmodified frq gene with its natural qrf sequence (frq-qrf control), respectively, were inserted into the Neurospora genome at the his-3 locus of a Δfrq strain (see the “MATERIALS AND METHODS” section). RNA directed from these genes was quantified by quantitative reverse transcription polymerase chain reaction (qRT-PCR) (Figure 1b, right). In the frq-qrf control strain, antisense RNA was expressed at a high level in light and at a ~5-fold lower level in the dark. In contrast, in the frq-ccg-2 and frq-trpC strains, the antisense transcript levels were significantly reduced, indicating that deletion of the qrf promoter at the BssHII site compromised the expression of antisense RNA in light and in dark.

We then assessed the conidiation rhythms of the frq-qrf control strain and the mutant strains lacking antisense transcription. The frq-qrf control strain and frq-ccg-2 strain exhibited rhythmic conidiation in constant darkness (DD), but the phase of conidiation of the frq-ccg-2 strain was delayed by 2 h (Figure 1c). The observed phase delay is consistent with a previous report (Kramer et al., 2003), in which the qrf promoter had been replaced at the EcoRV site by the 3′ UTR of ccg-2. In contrast, conidiation in DD was apparently arrhythmic in the frq-trpC strain, indicating that the circadian clock was compromised. Thus, despite frq-ccg-2 and frq-trpC strains being both deficient in antisense transcription, conidiation of frq-ccg-2 was rhythmic while the overt rhythm of frq-trpC was lost.

We then generated frq-qrf, frq-trpC, and frq-trpC strains that expressed in the tubulin locus a destabilized luciferase gene under control of the frq promoter (frq-lucPEST) and measured luciferase activity as described previously (Gooch et al., 2008; Cesbron et al., 2013). Both, frq-ccg-2 and frq-trpC strains supported rhythmic expression of frq-lucPEST, albeit with a lower amplitude than the frq-qrf control strain (Figure 1d). However, compared with the race tube analysis (Figure 1c), the frq-lucPEST reporter gene did not display differences in circadian phase between frq -qrf and frq-ccg-2 strains.

Since the frq mRNA terminates downstream of the mapped qrf TSSs (Belden et al., 2011), replacement of the qrf promoter by ccg-2 or trpC altered the 3′ UTR of the frq sense transcript. We therefore asked how the choice of the DNA sequence, which was used to replace the qrf promoter, affected the stability of frq mRNA. To assess RNA turnover under physiological conditions, we allowed accumulation of high levels of the frq RNA by growing mycelial cultures in constant light. The light-induced transcription of frq was then shut down by a transfer of the cultures to the dark and the decrease of frq RNA was then measured over a time course of 120 min (Figure 2a). The frq RNA levels in the frq-qrf control strain decreased with a half-time (t_1/2) of 15 min, confirming that frq RNA is unstable (Diernfellner et al., 2005). The level of the frq-ccg-2 transcript decreased with t_1/2 ~ 25 min, while the frq-trpC transcript was substantially more stable (t_1/2 ~ 1 h). The data demonstrate, not surprisingly, that the 3′ UTRs affect the stability and hence the expression level of frq mRNA.

We then measured frq RNA and FRQ protein in light and after LD transfer of mycelial cultures (Figure 2b). In light (t = 0 h), frq transcript levels in the frq-ccg-2 and frq-trpC strains accumulated at a higher level than in the frq-qrf control strain, confirming previous reports that replacement of the antisense promoter results in accumulation of elevated levels of sense RNA (Xue et al., 2014). In constant darkness, the frq-qrf control strain expressed frq sense RNA in circadian fashion, with a peak about 12 h after transfer of the mycelial cultures into the dark (Figure 2b). FRQ protein displayed circadian abundance and phosphorylation rhythms (Figure 2c). In the frq-ccg-2 strain, frq RNA and FRQ protein also displayed circadian rhythms in DD (Figure 2b and 2c). In the frq-trpC strain, the levels of frq and FRQ in dark were elevated and did not display apparent rhythms on the time frame of the experiment (Figure 2b and 2c). Hence, the delayed conidiation rhythm of frq-ccg-2 and the arrhythmic conidiation of frq-trpC strains correlate with the expression profiles of frq and FRQ in these strains after LD transfer, but not with the phase and rhythm of the frq-lucPEST reporter.

Together, our results show that antisense (qrf) transcription at the frq locus is not required for rhythmicity of the core circadian clock. Rather, the circadian clock is sensitive to changes in the 3′ UTR that stabilize frq RNA, suggesting that rapid turnover of frq RNA appears to be critical for clock amplitude and therefore may influence overt rhythmicity.

Next, we characterized the transcriptional properties of the qrf promoter using luciferase reporter constructs. Three TSSs of the qrf promoter were previously mapped, 2 between the BssHII and EcoRV sites and 1 immediately downstream of the EcoRV site (Figure 3a). To include all TSSs, we fused the qrf promoter immediately after the stop codon of the FRQ ORF (1117 bp) to luc-PEST (Figure 3a). The DNA sequence transcribed by the qrf promoter contains several ATG codons, and 3 of these are present in the 5′-UTR of the chimeric qrf-lucPEST transcription unit. We changed these 3 upstream ATGs in the reporter gene to CTGs to ensure optimal translation of lucPEST ORF. The modified qrf-lucPEST reporter supported efficient bioluminescence expression and exhibited a low-amplitude circadian rhythm (Figure 3b). We also generated lucPEST reporter constructs with qrf promoters truncated at the EcoRV and BssHII site, respectively (Figure 3a). In comparison with the transcriptional activity of the full-length qrf-lucPEST reporter, the bioluminescence levels supported by qrf promoters truncated at EcoRV and BssHII, respectively, were substantially reduced (Figure 3b). The residual activity was arrhythmic and similarly low for the EcoRV and BssHII truncations. The data suggest that truncations at either restriction site abolish qrf rhythms and transcription levels to the same extent.

We then compared the transcription rhythms of the isolated qrf (antisense) reporter with the rhythm supported by a frq (sense) reporter gene (Figure 3c). The expression levels of the qrf-lucPEST reporter oscillated in antiphase to the circadian rhythm of the frq-lucPEST reporter.

Light-dependent expression of qrf is controlled by the qLRE (Smith et al., 2010; Xue et al., 2014). To assess whether the qLRE is required for the antiphasic transcription rhythm in the dark, we mutated the sequence in the qrf-lucPEST reporter and measured luciferase expression in the corresponding reporter strain, qrfΔqLRE. The qLRE mutation did not affect the evening-specific circadian expression rhythm of the qrfΔqLRE promoter (Figure 3d).

Finally, to compare the responsiveness and sensitivity of the frq and qrf promoters to light, the frq-lucPEST and qrf-lucPEST reporter strains were exposed to 1-min LPs of different intensities and bioluminescence was then recorded for 300 min (Figure 3e). The frq-lucPEST reporter reached peak expression levels about 50 min after the LP, while the qrf-lucPEST reporter reached maximal expression levels after about 75 min. Furthermore, expression levels of the frq-lucPEST reporter saturated at LP intensities above 2.7-µmol photon m⁻² sec⁻¹, while the qrf promoter saturated at much higher LP intensities (>21.3-µmol photon m⁻² sec⁻¹). The data indicate that the qrf promoter is less responsive and less sensitive to light cues.

In constant darkness, clock-controlled genes of Neurospora are rhythmically expressed in mainly 2 phases, subjective morning and evening, respectively (Bell-Pedersen et al., 1996; Sancar et al., 2015b). Morning-specific genes are expressed approximately in phase with the activity profile of the transcription activator WCC, while many evening-phased genes are regulated by the transcription repressor CSP-1 (Sancar et al., 2015b). Since the transcriptional activity of qrf is evening specific, that is, in antiphase to the rhythm of the WCC-driven frq promoter, we asked whether the qrf promoter is controlled by the morning-specific CSP-1 repressor. Analysis of CSP-1 ChIP-Seq (Sancar et al., 2011) and WC-2 ChIP-Seq (Sancar et al., 2015a) datasets revealed that CSP-1 binds overlapping with the LRE and qLRE of the frq and qrf promoters, respectively, and also close to the c-box of frq (Figure 4a). We have previously shown that frq expression is elevated and phase delayed in a Δcsp-1 strain (Sancar et al., 2011). To assess the impact of CSP-1 on the qrf promoter, we analyzed expression of qrf-lucPEST in a Δcsp-1 strain. Expression of qrf-lucPEST was elevated (p < 0.05), and its circadian rhythm was abolished. The data suggest that qrf is constitutively activated by at least one unknown TF. The activity of this TF is antagonized by CSP-1, which represses qrf transcription. Since CSP-1 is rhythmically expressed with a morning-specific phase, it generates an antiphasic evening-specific qrf rhythm (Figure 2d). Together, these results demonstrate that the isolated qrf promoter displays an evening-specific expression rhythm, which is dependent on the morning-specific activity of the repressor CSP-1.

CSP-1 is a short-lived repressor that is expressed in light-dependent fashion under transcriptional control of the WCC, and independently of WCC also in glucose-dependent fashion (Sancar et al., 2012). To characterize the impact of CSP-1 on qrf transcription, we examined qrf-lucPEST and qrfΔqLRE-lucPEST reporter strains. The strains were cultured in 96-well plates on agar medium with low and high glucose concentration. The cultures were exposed to repetitive 12 h:12 h LD cycles, and bioluminescence profiles were recorded for 84 h (Figure 4a and 4b). The qrf-luc-PEST strain displayed light-driven luciferase activity (bioluminescence) at low and high glucose concentration. On low glucose medium (Figure 3a, left), the bioluminescence levels decreased rapidly after the LD transition and approached dark levels within ~4 h. On high glucose medium (Figure 3b, right), the decrease in bioluminescence after the LD transition was biphasic: the initial rapid decrease of bioluminescence within the first 4 h in the dark was followed by a slower decrease during the 4- to 12-h time period after the LD transition. To our surprise, the bioluminescence supported by qrfΔqLRE-lucPEST was modulated by light in a glucose-dependent manner (Figure 4c). At low glucose concentration, qrfΔqLRE-lucPEST expression was essentially unaffected by light (Figure 4c, left). In contrast, at high glucose concentration in the medium, qrfΔqLRE-lucPEST expression displayed a specific response to light despite the absence of a functional qLRE (Figure 4c, right): after about 1 h in light, the bioluminescence supported by qrfΔqLRE-lucPEST increased steadily. After LD transition, the bioluminescence kept increasing for about 2 h and then declined steadily, resulting in a sawtooth-like bioluminescence rhythm that was delayed relative to the DL and LD transitions. Interestingly, the decrease in the dark of qrfΔqLRE-lucPEST expression superimposed with the slow decrease of qrf-lucPEST activity observed 4-8 h after LD transition (Figure 4c). The data suggest that light supports the expression of an unknown, glucose-dependent transcription activator of qrf, which accumulates steadily during the light phase and is degraded during the dark phase. Hence, the qrf promoter is regulated in a complex manner by at least two environmental cues, light and glucose. Light-dependent expression is directly controlled by the WCC via the qLRE, while glucose-dependent expression is controlled by an unknown activator and by the rhythmically expressed repressor CSP-1. Activity and presumably expression levels of both, activator and repressor, are dependent on light and glucose.

Taken together, our results demonstrate that antisense transcription is not required for the function of the circadian clock in constant darkness (Figures 1 and 2). Rather, the impact of qrf transcription on frq transcription in light via WCC (Kramer et al., 2003; Xue et al., 2014) (Figure 3), the direct impact of glucose via CSP1 on frq (Sancar et al., 2011) and qrf transcription (Figure 4a and 4b), and the indirect impact of glucose and light on qrf via an unknown transcription activator (Figure 4c) suggest that antisense transcription may help fine-tune and coordinate the light-dark phase of frq expression with the metabolic state of the fungus.