The Arabidopsis sickle Mutant Exhibits Altered Circadian Clock Responses to Cool Temperatures and Temperature-Dependent Alternative Splicing

The Arabidopsis SICKLE gene plays a role in pre-mRNA metabolism that influences alternative splicing of core circadian clock transcripts, temperature entrainment, and temperature compensation.

Abstract

The circadian clock allows plants to anticipate and respond to daily changes in ambient temperature. Mechanisms establishing the timing of circadian rhythms in Arabidopsis thaliana through temperature entrainment remain unclear. Also incompletely understood is the temperature compensation mechanism that maintains consistent period length within a range of ambient temperatures. A genetic screen for Arabidopsis mutants affecting temperature regulation of the PSEUDO-RESPONSE REGULATOR7 promoter yielded a novel allele of the SICKLE (SIC) gene. This mutant, sic-3, and the existing sic-1 mutant both exhibit low-amplitude or arrhythmic expression of core circadian clock genes under cool ambient temperature cycles, but not under light-dark entrainment. sic mutants also lengthen free running period in a manner consistent with impaired temperature compensation. sic mutant alleles accumulate LATE ELONGATED HYPOCOTYL (LHY) and CIRCADIAN CLOCK ASSOCIATED1 (CCA1) splice variants, among other alternatively spliced transcripts, which is exacerbated by cool temperatures. The cca1-1 lhy-20 double mutant is epistatic to sic-3, indicating the LHY and CCA1 splice variants are needed for sic-3 circadian clock phenotypes. It is not expected that SIC is directly involved in the circadian clock mechanism; instead, SIC likely contributes to pre-mRNA metabolism, and the splice variants that accumulate in sic mutants likely affect the circadian clock response to cool ambient temperature.

INTRODUCTION

The rhythms produced by the endogenous circadian clock allow biological systems to anticipate and respond to daily and seasonal environmental cycles. The circadian clock regulates many fundamental processes in plants, including metabolism, growth, development, and defense responses (reviewed in Bendix et al., 2015; Greenham and McClung, 2015). An appropriately functioning circadian system that entrains to the environment confers an adaptive advantage to Arabidopsis thaliana (Dodd et al., 2005).

The Arabidopsis molecular circadian clock comprises interlocked regulatory feedback loops that incorporate transcriptional, posttranscriptional, and posttranslational control mechanisms (reviewed in Nagel and Kay, 2012; Fogelmark and Troein, 2014; Hsu and Harmer, 2014; McClung, 2014). Beginning at dawn of a single circadian cycle, clock-driven and light-stimulated expression of CIRCADIAN CLOCK ASSOCIATED1 (CCA1) and LATE ELONGATED HYPOCOTYL (LHY) rises (Schaffer et al., 1998; Wang and Tobin, 1998); these Myb-like transcription factor genes belong to the larger REVEILLE (RVE) gene family (Chaudhury et al., 1999). CCA1 and LHY repress expression of TIMING OF CAB EXPRESSION1 (TOC1) and the evening complex components LUX ARRHYTHMO (LUX), EARLY FLOWERING3 (ELF3), and ELF4 (Nusinow et al., 2011). Dawn also promotes RVE8 expression (Rawat et al., 2011), and afternoon accumulation of RVE8 activates expression of PSEUDO-RESPONSE REGULATOR9 (PRR9), PRR5, TOC1, GIGANTEA (GI), LUX, and ELF4 (Farinas and Mas, 2011; Rawat et al., 2011; Hsu et al., 2013). Together, the action of RVE8 and sequential expression of PRR9, PRR7, and PRR5 throughout the day suppress the morning transcriptional program (Matsushika et al., 2000; Farré et al., 2005; Rawat et al., 2011; Nakamichi et al., 2012). Evening accumulation of TOC1 maintains repression of CCA1, LHY, and earlier expressed PRRs and feeds back to repress its own expression (Gendron et al., 2012; Huang et al., 2012). The evening complex also represses expression of PRR9, PRR7, and GI at this time (Dixon et al., 2011; Helfer et al., 2011; Chow et al., 2012; Herrero et al., 2012), as well as its own expression through inhibition of LUX (Helfer et al., 2011). As dawn approaches, reduced TOC1 and PRR5 activity, together with repression of PRR9 and PRR7 expression, allows CCA1 and LHY expression to rise and begin the next circadian clock regulatory cycle.

Circadian clocks share three fundamental properties: entrainment by environmental cues, self-sustaining rhythms, and temperature compensation (Johnson et al., 2004). Entrainment is the capacity of the circadian clock to adjust the timing, or phase, of rhythms to synchronize daily internal processes with external light and temperature cues (Salome and McClung, 2005a). The rhythms generated by the circadian clock persist with an ∼24-h period in free-running conditions (i.e., constant temperature and continuous light or dark). Free-running rhythms are temperature compensated to maintain a uniform period over a range of ambient temperatures (Pittendrigh, 1954), in contrast to the general rule that the rate of enzymatic processes changes with temperature (Garcia-Viloca et al., 2004). In Arabidopsis, mutants in the core circadian clock genes PRR7, PRR9, CCA1, and LHY exhibit altered temperature compensation behavior (Edwards et al., 2005; Gould et al., 2006; Salomé et al., 2010). Repression of LHY and CCA1 by PRR7 and PRR9 is important for maintenance of the circadian clock period under varying temperatures (Gould et al., 2006; Salomé et al., 2010). The circadian clock in a prr7 prr9 double mutant overcompensates at warm temperatures so that the period is long at 30°C and close to that of the wild type at 12°C (Salomé et al., 2010). Overcompensation in prr7 prr9 depends on LHY and CCA1, since the addition of either a cca1 or lhy mutant to prr7 prr9 suppresses the long period at 30°C (Salomé et al., 2010). Temperature compensation also requires CCA1 at cool temperatures and LHY at high temperatures (Edwards et al., 2005; Gould et al., 2006). Despite these findings, an integrated understanding of the molecular mechanisms providing temperature compensation and temperature entrainment to the Arabidopsis circadian clock is still unclear.

Recent work shows that alternative splicing of transcripts from circadian clock genes is an important regulatory mechanism for circadian rhythms in Arabidopsis (reviewed in Staiger and Green, 2011; Staiger and Brown, 2013; Cui et al., 2014; Filichkin et al., 2015). Spliceosomes are large ribonucleoprotein complexes that produce mature mRNAs from pre-mRNAs through removal of introns based on 5′- and 3′-splice site sequences at exon-intron junctions (Wang and Burge, 2008; Matera and Wang, 2014). Alternative splicing occurs when spliceosomes use different 5′- and 3′-splice site combinations to yield two or more structurally different transcripts, or splice variants, from the same gene. Arabidopsis mutants in four different spliceosome-associated genes exhibit altered circadian clock period and alternative splicing of transcripts from core circadian clock genes. The SNW/Ski interacting (skip), spliceosomal timekeeper locus1 (stipl1), and protein arginine methyl transferase5 (prmt5) mutants have a lengthened clock period (Hong et al., 2010; Sanchez et al., 2010; Jones et al., 2012; Wang et al., 2012), while the gemin2 mutant has a shortened clock period (Schlaen et al., 2015). The skip and gemin2 mutants also have impaired circadian clock temperature compensation (Wang et al., 2012; Schlaen et al., 2015). Furthermore, all four mutants exhibit altered alternative splicing of core circadian clock transcripts, which affects the abundance of splice variants from those transcripts (Hong et al., 2010; Sanchez et al., 2010; Jones et al., 2012; Wang et al., 2012; Schlaen et al., 2015). Thus, the Arabidopsis circadian clock requires full spliceosome activity and alternative splicing of circadian clock transcripts to establish period and temperature compensation.

The temperature environment also influences alternative splicing of Arabidopsis circadian clock transcripts (Filichkin et al., 2010). The abundance of splice variants from many circadian clock genes, including CCA1, LHY, PRR7, and PRR9, changes in response to external temperature cues (James et al., 2012). It remains unclear how the accumulation of circadian clock splice variants mechanistically affects circadian clock function and its responses to temperature cues.

Here, we describe characterization of a new mutant allele of the SICKLE (SIC) gene, sic-3, and the existing sic-1 allele with respect to their effects on the Arabidopsis circadian clock. sic mutants exhibit diminished cool temperature responses of the Arabidopsis circadian clock in terms of temperature entrainment, maintenance of rhythms, and temperature compensation. Alternative splicing of LHY, CCA1, PRR7, and ELF3 transcripts is elevated at all temperatures in sic-3 and sic-1, but cool temperatures markedly stimulate splice variant production. SIC is a nuclear protein of unknown function, previously implicated in the control of alternative splicing, microRNA (miRNA) biogenesis, and stress responses (Zhan et al., 2012). These observations establish that SIC is needed to regulate splice variant production from core circadian clock transcripts, and without its activity, the circadian clock is impaired under cool temperature conditions.

RESULTS

warp2 Is a Mutant Allele of SIC That Alters Circadian Clock Temperature Responses

To identify genes necessary for correct temperature perception by the Arabidopsis circadian clock, EMS-mutagenized seedlings were screened for alterations in temperature-induced activation of the PRR7 promoter (Supplemental Figure 1A). In wild-type seedlings, a 28°C temperature pulse of 3 h generates a strong and reproducible induction the endogenous PRR7 gene (Thines and Harmon, 2010). The temperature responsiveness of the PRR7 promoter is faithfully reported by the ProPRR7:LUC transgenic reporter construct (Supplemental Figure 1A). The warm acute response of PRR7 2 (warp2) mutant was selected from the screen based on a weak response to the 28°C temperature pulse (Supplemental Figure 1B). Subsequent experiments revealed that warp2 lengthens circadian clock period, indicating that this mutant affects circadian clock function (see below).

The location of the mutation in warp2 was mapped based on its long circadian period phenotype. Next-generation mapping (Austin et al., 2011) employing whole-genome resequencing of pooled genomic DNA from mutant individuals indicated the mutation was within a 400-kb region of chromosome 4 (Supplemental Figure 2). The warp2 long period phenotype segregated 1:3 in the F2 mapping population, indicating warp2 is a fully recessive allele in a single nuclear gene. Sequencing of candidate genes in the mapping interval identified the warp2 mutation as a G → A transition at position +552 (relative to the transcriptional start site) in the AT4G24500 gene (Supplemental Figure 3A). This mutation eliminates the GU pair of the 5′-splice site for intron 2 in AT4G24500 transcripts (Supplemental Figure 3A), which causes mis-splicing of exon 2 to produce two different transcripts (Supplemental Figures 3B and 3C). One transcript potentially encodes a protein with an internal deletion of 76 amino acids near the N terminus, and the second potentially encodes a truncated protein made up of the N-terminal 61 amino acids (Supplemental Figure 3D). In agreement with the mapping results, quantitative PCR (qPCR) showed AT4G24500 transcript levels are substantially reduced in warp2 (Supplemental Figure 3E). The AT4G24500.1 transcript is the predominant transcript in the wild type (Supplemental Figures 3A and 3B). AT4G24500 expression increases somewhat in the evening under photocycles, indicating potentially weak diurnal regulation (Supplemental Figure 3E).

Two types of genetic tests confirmed that the mutation found in AT4G24500 is responsible for the warp2 phenotype. First, F1 seedlings from crosses between warp2 and sic-1, a previously identified mutation in AT4G24500, maintain the long period phenotype characteristic of warp2 (Figure 1A; Supplemental Data Set 1). Furthermore, sic-1 has a long circadian period (Figure 1A, Table 1). sic-1 is an EMS-induced G → A transition at position +1190 of AT4G24500 that changes Trp143 to a premature stop codon (Supplemental Figures 3A to 3D), and it behaves as a recessive allele (Zhan et al., 2012). warp2 has no obvious growth- or development-related phenotypes (Supplemental Figures 3F and 3G), while sic-1 plants have reduced rosette diameter, as well as serrated and downward curled leaves (Supplemental Figure 3F) (Zhan et al., 2012). Second, transgenic introduction of a 2296-bp genomic region of AT4G24500 from the wild type into the warp2 mutant background complemented the warp2 period phenotype (Figure 1B; Supplemental Data Set 1). These results conclusively demonstrate that the mutation found in AT4G24500 is the cause of the warp2 long period phenotype and indicate that warp2 is a mutant allele of SIC and therefore will be referred to as sic-3.

Figure 1.

Mutation of AT4G24500/SIC Is Responsible for the warp2 Phenotype.

(A) Noncomplementation of long period ProPRR7:LUC rhythms in an F1 cross between warp2/sic-3 and sic-1 (warp2/sic-3 x sic-1 F1; dashed black line) compared with the wild type (WT; solid black line), warp2/sic-3 (solid green line), and sic-1 (dotted blue line). Seedlings entrained under LD|22°C and released into LL|22°C. Data are representative of three independent experiments.

(B) Complementation of long period rhythms in warp2/sic-3 with a stable transgene carrying the AT4G24500/SIC genomic region from the wild type (gATG24500:warp2/sic-3; dashed black line), compared with the wild type (solid black line) and warp2/sic-3 (solid green line). Seedlings were first entrained in photocycles (LD|22°C) and then released into free run at LL|22°C. ProPRR7:LUC activity was monitored for 5 d.

Table 1. Estimated Circadian Clock Parameters under Different Entrainment and Free-Running Conditions.

Condition	Genotype	Mean Period, h ± sd	Mean RAE ± sd	Mean Phase, CT ± sd	Percent Rhythmic	Total n
LL|22–18°C	Wild type	24.1 ± 0.5	0.40 ± 0.10	11.6 ± 2.4	98%	277
	sic-3	25.7 ± 1.9	0.56 ± 0.20	11.2 ± 7.7	60%	220
	sic-1	25.5 ± 1.5	0.60 ± 0.15	9.6 ± 6.0	52%	141
LL|22–18°C	Wild type	23.8 ± 0.5	0.17 ± 0.06	12.2 ± 1.2	100%	277
→ LL|22°C	sic-3	25.2 ± 1.0	0.20 ± 0.08	13.4 ± 1.7	99%	220
	sic-1	25.8 ± 1.4	0.34 ± 0.16	14.6 ± 2.8	97%	141
LD|22°C	Wild type	24.5 ± 0.7	0.40 ± 0.09	10.9 ± 1.8	95%	314
	sic-3	24.4 ± 0.4	0.31 ± 0.07	10.7 ± 1.1	99%	305
	sic-1	25.2 ± 0.5	0.39 ± 0.09	10.3 ± 1.6	98%	257
LD|22°C	Wild type	24.2 ± 0.6	0.24 ± 0.08	10.8 ± 1.6	100%	314
→ LL|22°C	sic-3	25.6 ± 0.8	0.23 ± 0.07	11.6 ± 1.8	100%	305
	sic-1	26.3 ± 1.6	0.30 ± 0.10	14.3 ± 3.8	100%	257

Rhythmic ProPRR7:LUC activity assayed for 4 to 5 d in the condition indicated in bold and the indicated genotypes. Mean period, mean RAE, and mean phase were calculated from all rhythmic individuals. Rhythmic seedlings are those with RAE < 0.6. CT, circadian time in hours.

sic Mutation Lengthens the Circadian Clock Period

Compared with the wild type, sic-1 and sic-3 exhibit longer periods of ProRR7:LUC rhythms after release into free-running conditions (LL|22°C; constant light and constant 22°C) from either thermocycle (LL|22–18°C; constant light and 12-h long cycles of 22°C and 18°C) or photocycle (LD|22°C; constant 22°C and 12-h long cycles of light and dark) entrainment (Figure 2; Supplemental Figures 4A and 4B). Similarly, peak expression of eight core circadian clock genes (PRR7, CCA1, LHY, PRR9, TOC1, GI, ELF3, and LUX) and three output genes (CAB [CHLOROPHYLL A/B BINDING PROTEIN], CAT3 [CATALASE3], and CCR2 [COLD-CIRCADIAN RHYTHM-RNA BINDING2]) is delayed in sic-3 seedlings during the 24 to 48 h (Zeitgeiber Time [ZT] 24 to ZT48) after release into LL|22°C (Figure 3; Supplemental Figures 5A and 5B) following either thermocycle or photocycle entrainment conditions. Quantitative estimation of period length for ProPRR7:LUC rhythms with FFT-NLLS (fast Fourier transform nonlinear least square) curve fitting analysis (Plautz et al., 1997) confirmed that the period for sic-3 and sic-1 is an average of 1 and 2 h longer, respectively, than the wild type (Table 1, Figures 2B and 2D). Thus, sic mutants have altered circadian clock periods.

Figure 2.

sic Lengthens the Circadian Clock Period in Free-Running Conditions.

(A) and (C) Individual traces of ProPRR7:LUC activity in wild-type (black lines) and sic-3 (green lines) seedlings under LL|22°C beginning at ZT0 after entrainment with LL|22–18°C (A) or LD|22°C (C) (n = 8 for all). Data are representative of three independent experiments.

(B) and (D) RAE as a function of period of ProPRR7:LUC activity for wild-type (black circles), sic-3 (green squares), and sic-1 (blue triangles) seedlings under LL|22°C after entrainment under LL|22–18°C (B) and LD|22°C (D) (n total indicated in Table 1). Dotted horizontal line indicates RAE = 0.6 threshold, above which traces of ProPRR7:LUC activity are considered arrhythmic. Data are from three independent experiments.

Figure 3.

sic Lengthens Expression Waveforms of Core Circadian Clock Genes in Free-Running Conditions.

Expression levels of the indicated transcripts in the wild type (black circles, solid line) and sic-3 (green squares, dotted green line) grown under LL|22°C after entrainment under LL|22–18°C ([A] to [C]) or LD|22°C ([D] to [F]). Transcript levels were determined with qPCR. In each biological replicate, relative expression for each transcript was calculated by normalization to the highest expression value for the wild type. Each point is the mean of three independent experiments, and error bars are sd.

sic Mutation Alters Circadian Clock Temperature Responses

ProPRR7:LUC rhythms were assessed in sic-3, sic-1, and the wild type during LL|22 to 18°C and LD|22°C to assess circadian rhythms during entrainment. Individual sic-3 and sic-1 seedlings lacked clear rhythms for ProPRR7:LUC expression during 5 d under LL|22–18°C (Figures 4A and 4B; Supplemental Figures 4A and 4C). Approximately 40 to 50% of sic-1 and sic-3 individuals had arrhythmic ProPRR7:LUC expression (Figure 4B; Supplemental Figure 4C and Supplemental Data Set 1), defined as ProPRR7:LUC expression waveforms with a relative amplitude error (RAE) value >0.6 from FFT-NLLS analysis (McWatters et al., 2000). In addition, periods for rhythmic individuals differed by as much as 12 h between sic individuals (Figure 4B), which is reflected by large standard deviations for the mean period of sic-1 and sic-3 (Table 1). Moreover, under LL|22–18°C, the time at which peak expression of ProPRR7:LUC occurs, or phase, was highly variable across the sic populations (Supplemental Figure 4C; Table 1). By comparison, populations of wild-type seedlings under LL|22–18°C exhibited few arrhythmic individuals, and rhythms were in close agreement in terms of both period length and phase (Figures 4A and 4B, Table 1; Supplemental Figure 4C). In contrast to thermocycle conditions, sic-1 and sic-3 seedlings under LD|22°C had robust ProPRR7:LUC rhythms similar to those of the wild type in terms of both period and phase, with >95% of the individuals exhibiting rhythmic behavior (Figures 4C and 4D, Table 1; Supplemental Figures 4B and 4E).

Figure 4.

sic Weakens Clock Function in Cool Temperature Thermocycles.

(A) and (C) Individual traces of ProPRR7:LUC activity in wild-type (black lines) and sic-3 (green lines) seedlings during LL|22–18°C (A) or LD|22°C (C) over 4 d (n = 8 for all). Blue shading represents periods of 18°C, and gray shading represents periods of dark.

(B) and (D) RAE as a function of period of ProPRR7:LUC activity for wild-type (black circles), sic-3 (green squares), and sic-1 (blue triangles) seedlings during LL|22–18°C (B) and LD|22°C (D) (n total indicated in Table 1). Dotted horizontal line indicates RAE = 0.6 threshold, above which traces of ProPRR7:LUC activity are considered arrhythmic. Data are from three independent experiments.

To determine whether cycling temperature, and not overall cool temperature, was the cause of arrhythmia in sic, rhythms for wild-type and sic-1 seedlings were tested under LD|18°C. The wild type and sic-1 showed comparable period length and no arrhythmic individuals (Supplemental Figures 6A and 6B and Supplemental Data Set 1). Rhythms of sic-3 and sic-1 seedlings entrained under warm LL|28–22°C thermocycles were comparable to those of the wild type (Supplemental Figures 6C and 6D and Supplemental Data Set 1). These observations show that sic interferes with the establishment of circadian rhythms only when (1) temperatures cycle and (2) thermocycles are cool.

To confirm that ProPRR7:LUC rhythmicity in sic-3 individuals during thermocycles reflected general clock dysfunction, the expression level and waveform for the same eight core circadian clock genes and three output genes were evaluated over a 24-h time course in sic-3 and the wild type exposed to LL|22–18°C and compared with behavior under LD|22°C (Figure 5; Supplemental Figures 5C and 5D). Under LL|22–18°C, PRR7 transcript expression in sic-3 was characterized by a low and broad peak of expression compared with the wild type, together with high variability at each time point (Figure 5A). The sic-3 mutation similarly changed the expression profiles of LHY, TOC1 (Figures 5B and 5C) CCA1, PRR9, GI, ELF3, LUX, CAB, CAT3, and CCR2 (Supplemental Figure 5C). Low-amplitude rhythms and variable transcript expression in pooled sic-3 seedlings indicated the same variable behavior observed for ProPRR7:LUC activity under LL|22–18°C (Figures 4A and 5A). Under LD|22°C, however, the expression of the same genes was comparable between sic-3 and the wild type (Figures 5D to 5F; Supplemental Figure 5D).

Figure 5.

sic Disrupts the Expression Waveforms of Core Circadian Clock Genes in Cool Temperature Thermocycles.

Expression levels of the indicated transcripts in wild type (black circles, solid line) and sic-3 (green squares, dotted green line) grown under LL|22–18°C ([A] to [C]) or LD|22°C ([D] to [F]). Blue regions in (A) to (C) indicate periods of 18°C, and gray regions in (D) to (F) indicate periods of darkness. Transcript levels were determined with qPCR. In each biological replicate, relative expression for each transcript was calculated by normalization to the highest expression value for the wild type. Each point is mean of three independent experiments, and error bars are sd.

sic Mutation Causes Sensitivity to Cool Temperature Cycles

To better understand the nature of the temperature signaling defect in sic, the next experiments evaluated whether sic was impaired in perceiving specific ambient temperatures (i.e., cool versus warm temperatures). Regardless of prior thermocycle composition, sic-3 and sic-1 mutants always had longer and more variable free-running periods at 22°C than the wild type (Figures 6A to 6C; Supplemental Data Set 1), as indicated by the larger sd for mutant populations. Exposure of sic seedlings to cool thermocycles during entrainment (LL|22–12°C, LL|22–16°C, and LL|22–18°C) accentuated the variation in period across the population, as indicated by sd (Figures 2B, 6A, and 6B; Supplemental Data Set 1). In addition, sic individuals were more likely than the wild type to exhibit arrhythmic ProPRR7:LUC expression in free-running LL|22°C conditions after cool temperature cycle entrainment: 13% of sic-3 and sic-1 seedlings were arrhythmic after LL|22–12°C entrainment, while <1% of mutant seedlings were arrhythmic following LL|28–22°C entrainment (Supplemental Data Set 1). Thus, exposure of sic to cool thermocycles during entrainment had lasting negative effects on the period and amplitude of free-running circadian rhythms. These results indicate that the Arabidopsis circadian clock requires SIC activity to appropriately respond to cool temperature cues.

Figure 6.

sic Impairs Temperature Compensation and Reduces Circadian Rhythm Quality, Particularly in Cool Thermocycles.

(A) to (C) RAE as a function of period at 22°C for individual wild-type, sic-3, and sic-1 seedlings after entrainment with LL|22–12°C (A) (n for wild type = 90, sic-3 = 90, and sic-1 = 53), LL|22–16°C (B) (n for wild type = 67, sic-3 = 63, and sic-1 = 47), or LL|28–22°C (C) (n for wild type = 102, sic-3 = 98, and sic-1 = 80). Points above dotted horizontal line at RAE = 0.6 are considered arrhythmic. Data are from three independent experiments.

(D) Mean period length of ProPRR7:LUC activity in free-running conditions of LL|X°C after LD|22°C entrainment in wild-type, sic-3, and sic-1 seedlings, where X stands for either 16, 22, or 28°C. Each point is the mean of all rhythmic individuals (RAE < 0.6) from three independent experiments (n at 16°C: wild type =108, sic-3 = 94, and sic-1 = 28; at 22°C: wild type = 314, sic-3 = 305, and sic-1 = 257; at 28°C: wild type = 83, sic-3 = 80, and sic-1 = 35). Error bars are sd and asterisk indicates significance P value < 0.0001 from ANOVA with Tukey’s multiple comparison test between all genotypes at the same temperature.

sic Interferes with Circadian Clock Temperature Compensation

A potential explanation for the imprecise rhythms in sic during cool thermocycles and the long period under free-running conditions was the presence of an underlying problem with temperature compensation. To test temperature compensation in wild-type, sic-1, and sic-3 seedlings, free-running period was determined at constant 28°C (LL|28°C), 22°C (LL|22°C), and 16°C (LL|16°C), after either LL|22–18°C or LD|22°C entrainment. Following either type of entrainment, the free-running period of the wild type exhibited temperature compensation, while those of sic-3 and sic-1 did not (Figure 6D; Supplemental Figure 7). The period of sic-3 was directly correlated with ambient temperature, so that the mean period under LL|16°C was 2 to 3 h longer than that under LL| 22°C, and the mean period under LL| 22°C was 1 h longer than that under LL|28°C. In fact, the period of sic-3 under LL|28°C approached that of the wild type, but remained significantly different from the wild type (Figure 6D; Supplemental Figures 7A, 7F, and 7G and Supplemental Data Set 1). The average period of a sic-1 population under LL|16°C was 3 to 5 h longer than that under LL|22°C (Figure 6D; Supplemental Figures 7A to 7C and Supplemental Data Set 1). However, sic-1 responded differently to LL|28°C: Its average period at this condition was similar to that under LL|22°C (Figure 6D; Supplemental Figures 7A and 7D to 7G and Supplemental Data Set 1). Together, these findings indicate that both sic alleles interfere with the circadian clock temperature compensation mechanism.

In addition, sic individuals entrained with cool thermocycles were more likely to be arrhythmic under cool temperature free-running conditions than under warm conditions. When entrained with LL|22–18°C thermocycles, 33% of sic-3 and 15% of sic-1 seedlings were arrhythmic under LL|16°C (Supplemental Figure 7B and Supplemental Data Set 1); on the other hand, fewer than 4% of individuals from either mutant genotype were arrhythmic when released into LL|28°C conditions (Supplemental Figure 7F and Supplemental Data Set 1). Each sic allele continued to show broad dispersion in period length across the population regardless of the temperature used for free-running conditions, which is clear from the larger sd values for mutant populations compared with the wild type (Supplemental Figure 7 and Supplemental Data Set 1). Thus, a combination of cool temperature thermocycles and cool free-running conditions promotes arrhythmicity in sic, and warm free-running conditions suppress this arrhythmicity.

SIC Encodes a Conserved Proline/Serine-Rich Protein Found in Nuclear Foci

SIC is a 319-amino acid protein of unknown molecular and biochemical function (Zhan et al., 2012). The protein is enriched in proline and serine residues, which constitute 23% of all amino acids. A stable translational fusion of SIC and YFP (Pro35S:SIC-YFP) overexpressed in wild-type plants localized exclusively to the nucleus within punctate foci (Figure 7A). These foci are distributed throughout the nucleoplasm and are distinct from the nucleolus.

Figure 7.

SIC Is a Conserved Proline-Rich Protein That Accumulates in Nuclear Foci.

(A) Subcellular localization of a SIC-YFP fusion in nuclei of root cells detected by fluorescence microscopy. Images are bright-field (BF) and false colored YFP fluorescence (YFP). Bars indicate scale as indicated.

(B) MUSCLE alignment of Arabidopsis SIC to SIC-like proteins from P. trichocarpa, O. sativa japonica, Z. mays, M. acuminata, and Amborella trichopoda. Conserved amino acid regions are highlighted in green, corresponding to Arabidopsis SIC amino acids 1 to 20 (region a), 58 to 70 (region b), 239 to 260 (region c), 273 to 284 (region d), and 300 to 319 (region e). Short vertical bars indicate the level of percent amino acid similarity corresponding to 100% (black), 80 to 100% (dark gray), and 60 to 80% (light gray).

Sequence and phylogenetic analysis of SIC identified it as an angiosperm-specific protein with no homologs in gymnosperms, ferns, bryophytes, or algae (Supplemental Figure 8). Alignment of 70 SIC protein homologs (Supplemental File 1) identified five highly conserved amino acid regions (regions a to e) located at the N- and C-terminal portions of the protein (Figure 7B), while the central part of the protein is variable. SIC has several predicted functional motifs, including a nuclear localization signal (NLS) at the N terminus, consistent with the observed nuclear accumulation of SIC, and an MPLKIP (M-phase-specific PLK1-interacting protein) motif (SerMetXGluAspXXLeuXPro) near the C terminus (Zhang et al., 2007) (Figure 7B).

The 70 known SIC homologs share 30% overall amino acid similarity, and the conserved amino acid regions together share 61.5% similarity. Individually, region a has 60% similarity, region b has 72% similarity, region c has 61% similarity, region d has 49% similarity, and region e has 72% similarity (Supplemental File 1). A phylogenetic tree of SIC protein homologs shows it is a singleton in most species except for in maize (Zea mays), California poplar (Populus trichocarpa), desert poplar (Populus euphratica), banana (Musa acuminata), eucalyptus (Eucalyptus grandis), flax (Linum usitatissimum), Jatropha (Jatropha curcas), cocoa (Theobroma cacao), soybean (Glycine max), and Brassica rapa (Supplemental Figure 8).

sic Mutation Changes Alternative Splicing of Circadian Clock Transcripts, Particularly under Cool Temperatures

The temperature dependence of alternative splicing for several circadian clock transcripts was evaluated in wild-type, sic-3, and sic-1 seedlings. Specific splice variants from CCA1, LHY, PRR9, PRR7, GI, ELF3, TOC1, PRR5, and PRR3 transcripts were tested at LD|28°C, LD|22°C, and LD|16°C (Figures 8A to 8E; Supplemental Figure 9). Splice variants were detected using splice variant-specific primers and RT-PCR in cDNA generated from a pool of time points taken over a 24-h period. Labeling of PCR products with the FAM fluorophore allowed relative quantification of splice variant accumulation (Figures 8F to 8J). Bulk levels of each transcript were also assessed to confirm that changes in splice variant levels were explained by alternative splicing instead of changes in overall transcript expression (Supplemental Figure 10).

Figure 8.

sic Enhances Intron Retention in Transcripts of LHY, CCA1, ELF3, and PRR7.

(A) to (D) Discovery of the splice variants LHY I1R (A), CCA1 I4R (B), PRR7 I4R (C), and ELF3 I2R (D) in wild-type, sic-3, and sic-1 seedlings grown under LD|28°C, LD|22°C, or LD|16°C by RT-PCR. Red arrows indicate the specific RT-PCR product.

(E) UBC21 served as the transcript level reference gene. Bands detected by ethidium bromide staining and images inverted for clarity. Images are representative of three independent experiments.

(F) to (J) Quantification of FAM-labeled RT-PCR products from the wild type, sic-3, and sic-1 grown under LD|28°C, LD|22°C, or LD|16°C. Data are the mean of three independent experiments. Error bars are sd and asterisk indicates significance P value < 0.05 from two-tailed t test between the wild type and the indicated genotype at the same temperature.

A splice variant of LHY previously shown to arise in cool (12°C) to cold (4°C) temperatures is derived from retention of intron 1 (LHY I1R) (James et al., 2012). In wild-type seedlings, LHY I1R accumulated to higher levels as ambient temperature was reduced from LD|28°C to LD|16°C (Figures 8A and 8F). Increased amounts of LHY I1R were visually apparent in sic-3 and sic-1 at all temperatures. LHY I1R reaches its highest levels in mutants grown under LD|22°C and LD|16°C (Figures 8A and 8F). Bulk LHY transcript levels in the sic mutants were somewhat lower than in the wild type, indicating the observed changes represent authentic elevation of LHY I1R accumulation (Supplemental Figure 10A).

Retention of CCA1 intron 4 (CCA1 I4R) was proportional to ambient temperature in the wild type, so that the most CCA1 I4R transcript appeared at LD|28°C and the least occurred at LD|16°C (Figures 8B and 8G), in agreement with previous observations (James et al., 2012; Seo et al., 2012). sic-3 and sic-1 exhibited clear accumulation of CCA1 I4R at all temperatures, and levels were significantly higher in sic-1 (Figures 8B and 8G). It is notable that CCA1 I4R levels in sic-1 were significantly higher than in the wild type at LD|16°C, although this was not the condition where CCA1 I4R reaches peak levels for either genotype. The increase in CCA1 I4R levels was not due to higher bulk CCA1 transcript levels (Supplemental Figure 10D).

Both sic alleles permitted the accumulation of a PRR7 splice variant from retention of intron 4 (PRR7 I4R) and an ELF3 splice variant from retention of intron 2 (ELF3 I2R). PRR7 I4R levels in the wild type did not vary in a temperature-dependent manner (Figures 8C and 8H), but the level of this splice variant was elevated in sic-3 at LD|22°C and LD|16°C (Figures 8C and 8H), and it appeared at all temperatures in sic-1. ELF3 I2R in the wild type had a cool temperature-dependent accumulation pattern in which levels were highest under LD|16°C (Figures 8D and 8I). sic-3 and sic-1 accumulated ELF3 I2R under LD|22°C, and sic-1 generated significantly higher levels of this splice variant at LD|16°C (Figures 8D and 8I). Like the bulk transcripts from the other genes tested, increased expression of PRR7 and ELF3 does not explain the splice variant increase in the sic mutants (Supplemental Figures 10B and 10C). Strikingly, accumulation of other described splice variants for CCA1, LHY, PRR7, and ELF3, as well as known splice variants for PRR9, GI, TOC1, PRR5 and PRR3, appeared unchanged in either sic allele (Supplemental Figure 9). However, it is possible that the pooling strategy used here masked the presence of low abundance splice variants. Nevertheless, these results indicate that sic mutants have increased abundance of splice variants for LHY, CCA1, ELF3, and PRR7. In addition, sic alleles have a broader range of temperature conditions under which these splice variants occur, particularly at cool temperatures.

The Period and Thermocycle Entrainment Phenotypes Caused by sic-3 Require LHY and CCA1

To ask whether LHY and CCA1 transcripts are required for clock impairment in the sic mutant background, we generated lhy-20, cca1-1, lhy-20 sic-3, cca1-1 sic-3, and cca1-1 lhy-20 sic-3 mutants carrying the ProPRR7:LUC reporter. As expected, the lhy-20, cca1-1, and cca1-1 lhy-20 mutants had significantly shortened free running period compared with the wild type under LL|22°C after either LL|22–18°C thermocycle or LD|22°C photocycle entrainment (Figure 9A; Supplemental Figure 11A and Supplemental Data Set 1). Addition of sic-3 to either lhy-20 or cca1-1 lengthened period to a degree consistent with an additive phenotype (Figure 9A; Supplemental Figure 11A and Supplemental Data Set 1). In contrast, the period of the cca1-1 lhy-20 sic-3 was significantly shorter than the period of either sic-3, lhy-20 sic-3, or cca1-1 sic-3 (Figure 9A; Supplemental Figure 11A and Supplemental Data Set 1). Thus, the period lengthening effects of sic-3 were dependent on the presence of CCA1 and LHY transcript. Other hallmark sic-associated clock phenotypes also required CCA1 and LHY. For example, cca1-1 lhy-20 and cca1-1 lhy-20 sic-3 populations under LL|22–18°C thermocycles exhibited comparable mean period lengths, limited period variability, and few arrhythmic individuals (Figure 9B; Supplemental Figure 11C and Supplemental Data Set 1), unlike the broad distribution of periods observed for a sic-3 population (Figures 4A, 4B, and 9B). Therefore, the absence CCA1 and LHY suppressed the thermocycle entrainment phenotype of sic-3.

Figure 9.

Long Period and Weakened Rhythms under Thermocycles in sic Depend on CCA1 and LHY.

(A) Mean free-running period of ProPRR7:LUC activity for the wild type, sic-3, lhy-20, lhy-20 sic-3, cca1-1, cca1-1 sic-3, cca1-1 lhy-20, and cca1-1 lhy-20 sic-3 under LL|22°C following entrainment under LL|22–18°C. Error bars are sd. All genotypes are significantly different from sic-3 based on P value < 0.01 from ANOVA followed by Tukey’s multiple comparison test.

(B) RAE as a function of period of ProPRR7:LUC activity for sic-3 (green squares), cca1-1 lhy-20 (filled purple diamonds), and cca1-1 lhy-20 sic-3 (open purple diamonds) under LL|22–18°C entrainment (n for sic-3 = 126, cca1-1 lhy-20 = 57, and cca1-1 lhy-20 sic-3 = 64). Dotted horizontal line indicates RAE = 0.6 threshold, above which traces of ProPRR7:LUC activity are considered arrhythmic. Data are from three independent experiments.

DISCUSSION

sic mutants exhibit a lengthened circadian clock period, most notably at cool ambient temperatures. Furthermore, sic mutants have low amplitude rhythms or arrhythmic expression of core circadian clock genes in cool thermocycle conditions, but not in cool photocycle conditions. Altered circadian clock function of sic is accompanied by an increased abundance of specific splice variants of the LHY, CCA1, ELF3, and PRR7 transcripts. CCA1 and LHY are required for the circadian clock phenotypes of sic-3, indicating that these CCA1 and LHY splice variants potentially interfere with circadian clock function. Thus, SIC is important for the control of pre-mRNA metabolism, and the splice variants that accumulate in sic mutants may affect regulation of temperature compensation and temperature entrainment.

sic Exhibits Weakened Circadian Clock Function at Cool Temperatures

SIC is needed for the circadian clock to integrate cool temperature cycles to achieve entrainment and to produce temperature-compensated rhythms. sic mutant seedlings have impaired temperature compensation (Figure 6D; Supplemental Figure 7). In addition, exposure of sic to cool temperatures cycles during entrainment produces low amplitude rhythms (Figures 2A and 2B) and causes lasting negative effects on clock period and rhythm amplitude (Figures 6A and 6B). In contrast, sic has high amplitude rhythms with similar periods both during thermocycles of LL|28–22°C (Supplemental Figure 6C) and afterward in free-running conditions of LL|22°C (Figure 6C). On the other hand, SIC is not needed for entrainment in photocycle conditions, even under photocycles at 18°C (Supplemental Figures 6A and 6B). These observations agree with the notion that clock function is impaired under cool temperature cycles and that splice variant production for core clock transcripts is elevated under cool temperatures in sic (Figure 8). However, sic is not a general low temperature-sensitive mutant, since it does not exhibit significantly altered thermoresponsive flowering (Supplemental Figure 3G).

SIC Suppresses Alternative Splicing Events Induced by Cool Temperatures

The combination of circadian clock and alternative splicing phenotypes present in sic demonstrate that the Arabidopsis circadian clock requires SIC to control alternative splicing of LHY, CCA1, ELF3, and PRR7 transcripts brought on by cool temperature conditions (Figure 8). The accumulation of CCA1 I4R and LHY I1R splice variants in sic and the requirement for CCA1 and LHY transcripts for the period lengthening effects and cool thermocycle arrhythmia of sic (Figure 9) indicate these splice variants likely contribute to clock impairment. It is possible that accumulation of CCA1 I4R and LHY I1R in sic slows the pace of the clock, even in the presence of the full-length transcripts (Supplemental Figures 10A and 10D). Such an effect is unexpected given the short period caused by cca1-1, lhy-20, and cca1-1 lhy-20 mutations (Figure 9A) (Green and Tobin, 1999; Mizoguchi et al., 2002; Salomé et al., 2010).

The mechanistic consequences of aberrant splice variant accumulation in sic remain unclear. Splice variants frequently have alternate upstream open reading frames with premature termination codons (Lareau et al., 2007; Filichkin and Mockler, 2012). Premature termination codons cause stalling of ribosomes that induces transcript degradation by nonsense-mediated decay (NMD) (Lewis et al., 2003; McGlincy and Smith, 2008; Kalyna et al., 2012). On the other hand, splice variants may encode truncated, partially inactive protein isoforms that interfere with full-length protein activity (Seo et al., 2012; Syed et al., 2012; Filichkin et al., 2015). Premature termination codons can be predicted for ELF3 I2R and PRR7 I4R, but there is no evidence that these splice variants are subject to NMD or result in the production of truncated proteins. The CCA1 I4R splice variant is not degraded by NMD (James et al., 2012); instead, it is translated to produce a truncated CCA1 protein that interferes with the activity of full-length protein (Seo et al., 2012; Filichkin et al., 2015). LHY I1R is degraded by NMD under some conditions (James et al., 2012). Unlike the other splice variants prevalent in sic, LHY I1R does not introduce a premature termination codon or change the coding potential of the transcript because the retained intron interrupts the normal 5′ untranslated region. It is possible that this alternate 5′ untranslated region interferes with ribosome loading or has reduced translation initiation that ultimately triggers NMD degradation.

Previously, the sic-1 allele was shown to have reduced tolerance to chilling (i.e., 4°C) (Zhan et al., 2012), which may be related to, but distinct from, the cool ambient temperature phenotypes observed here. In addition, sic-1 exhibits reduced accumulation of certain miRNAs and therefore was implicated in miRNA biogenesis (Zhan et al., 2012). Whole-genome tiling array analysis of sic-1 in the same study found intron retention splice variants for many transcripts but not those observed here for LHY, CCA1, PRR7, and ELF3 (Zhan et al., 2012). Together, these observations indicate that the miRNA accumulation defect reported for sic-1 is more likely a consequence of altered transcript splicing than a direct effect on miRNA biogenesis. Furthermore, it is doubtful that the circadian clock phenotypes in sic arise from changes in miRNA biogenesis, since the circadian clock transcripts with disrupted expression in sic are not targeted by miRNAs.

SIC Protein Is a Conserved Nuclear Protein of Unknown Function

SIC has five highly conserved amino acid regions (regions a to e) that are shared with homologous proline/serine-rich proteins in angiosperms (Figure 7B; Supplemental Figure 8 and Supplemental File 1). SIC accumulates in punctate nuclear foci distributed throughout the nucleus, but it appears to be excluded from the nucleolus (Figure 7A). SIC foci were shown to colocalize with foci formed by HYL1 (Zhan et al., 2012), an RNA binding protein important for the generation of miRNAs (Song et al., 2007). Colocalization with HYL1 foci, together with the shape and location of the SIC-containing foci, indicates these are RNA processing-associated nuclear bodies, possibly “dicing bodies” involved in miRNA biogenesis or sites of spliceosome assembly/activity (Shaw and Brown, 2004).

Support for the idea that SIC participates in RNA processing is provided by a recent coimmunoprecipitation-mass spectrometry experiment that found SIC associated with the two Arabidopsis PRMT4/CARM1 protein arginine methyl transferases (Karampelias et al., 2016). PRMT4a and PRMT4b are thought to be important for the regulation of RNA splicing, similar to PRMT5 (Bedford and Richard, 2005), possibly through direct methylation of spliceosomal proteins or by influencing the coupling between transcription and RNA processing (Cheng et al., 2007; Kuhn et al., 2011). Other proteins in the SIC-PRMT4 protein complex include an RNA lariat debranching enzyme (AT4G31770) (Wang et al., 2004), a possible intron binding protein (AT2G38770), a protein with an Isy1-like splicing motif (AT3G18790), consistent with a role in optimization of splicing, and two proteins with RNA recognition motifs (AT1G13690 and AT5G28740) (Karampelias et al., 2016). Thus, SIC interacts with multiple proteins that are highly likely to participate in and/or regulate RNA processing.

Interestingly, the prmt4a;4b mutant does not exhibit a circadian clock phenotype (Hernando et al., 2015), in contrast to the long period phenotype of prmt5 mutants (Hong et al., 2010; Sanchez et al., 2010). Thus, the circadian clock phenotypes caused by sic are not simply due to a loss of PRMT4 activity. Clearly, a more complete understanding of the biochemical and molecular activities of SIC will be critical to rationalize these paradoxical observations.

Unlike predictable regulators of the spliceosome or spliceosomal components that affect the circadian clock when absent, such as PRMT5, SKIP, STIPL1, and GEMIN2 (Hong et al., 2010; Sanchez et al., 2010; Jones et al., 2012; Wang et al., 2012; Schlaen et al., 2015), the role of SIC is difficult to predict based on its amino acid sequence. Instead of acting specifically within the core clock mechanism, SIC appears to be necessary at cool temperatures to maintain control of splice variant production for specific clock-associated transcripts (Figure 10). In this model, appropriate levels of certain splice variants are critical for temperature entrainment and temperature compensation. Ambient temperature influences spliceosomal activity in such a way as to change levels of these splice variants, as previously proposed (Schlaen et al., 2015). Two formal possibilities for SIC function are (1) SIC is directly involved in spliceosome assembly or modulation of spliceosomal activity, or (2) SIC is necessary for NMD-promoted removal of clock-associated splice variants that are an undesirable consequence of cool temperatures (Figure 10). We favor the first role for SIC over the second. This role is consistent with the reported association of SIC with PRMT4a, PRMT4b, and other splicing-related proteins (Karampelias et al., 2016). In addition, nuclear-localized SIC is more likely to contribute to processes within the nucleus, and NMD is a translation-dependent process that occurs outside the nucleus (He and Jacobson, 2015). Discovery of the biochemical and molecular activities of SIC is certain to reveal important details about the functional link between alternative splicing and temperature responses within the Arabidopsis circadian clock and holds the potential to provide new insights into the molecular processes involved in pre-mRNA processing.

Figure 10.

Model for How SIC Affects Circadian Clock Temperature Responses.

See text for details. Blue ovals represent proteins. Blue ellipsoids indicate spliceosome component proteins. Thin arrows represent indirect protein interaction with the spliceosome. Green bars represent exons, and yellow bars represent introns for mRNA. Thick arrows represent sequential order of events. Dotted lines represent potential functions of SIC as a regulator of splice variant accumulation or a promoter of NMD.

METHODS

Plant Material and Growth Conditions

All experiments used the Arabidopsis thaliana Columbia-0 accession, except for the warp2 mapping population created by an outcross to the Landsberg erecta-0 accession. All lines carried a ProPRR7:LUC reporter construct that consisted of the PRR7 promoter driving expression of firefly luciferase (Salomé and McClung, 2005b). sic-1 seeds were obtained from The Arabidopsis Information Resource (TAIR), and this mutant is described by Zhan et al. (2012). Crossing of sic-1 to sic-3 ProPRR7:LUC generated plants homozygous for sic-1 and the ProPRR7:LUC construct.

For all experiments, seeds were surface sterilized by 10 min treatment with a solution of 30% bleach and 0.01% SDS, followed by extensive washing with sterile water, and were then sown on MS plates composed of 1× Murashige and Skoog basal salt medium (pH 5.7 to 5.8) with 0.8% Type I micropropagation agar (Caisson Labs). After stratification in the dark at 4°C for 3 d, the plates were transferred to constant light and 22°C to promote germination. After 3 d, germinating seedlings were transferred to entrainment for 5 d in the indicated light and temperature conditions; all entrainment treatments were divided into alternating 12-h cycles of either light:dark or different temperatures. Free-running conditions were always constant light and the indicated temperature. Surface sterilization of seeds and all transfers (i.e., germination, entrainment, and free run) occurred at ZT0. For RNA expression analysis, seedlings were grown in closely spaced groups on MS plates. Conditions for flowering time experiments were 12 h light and 12 h dark, along with the indicated constant temperature. Total number of rosette leaves was counted when the inflorescence was 1 cm tall. Cool white fluorescent bulbs supplied light at 50 μmol·m⁻²·s⁻¹ in growth chambers from Percival Scientific.

Bioluminescence Assays

Each MS plate was sprayed with 1 mL of 5 mM firefly luciferin (Biosynth; Gold Biotechnology) prepared in 0.01% (v/v) Triton X-100 (Sigma Aldrich) applied 24 h before imaging with a luciferase imaging system built by BioImaging Solutions, employing an ORCAII camera (Hamamatsu Photonics) housed in a Percival incubator (Percival Scientific) for temperature control. Each experiment had three to nine plates together in the camera. Halogen bulbs provided 50 μmol·m⁻²·s⁻¹ of white light. Seedlings were imaged every 2.5 h, and bioluminescence of individual seedlings was collected using MetaMorph software (Molecular Devices). Bioluminescence data from images were extracted with MetaMorph software and compiled in Microsoft Excel with the Biological Rhythms Analysis Software System 3.0 (BRASS), an Excel workbook for the analysis of rhythmic data series (Locke et al., 2005; Southern and Millar, 2005). Within BRASS, FFT-NLLS was used to estimate circadian period, RAE, and phase values from the rhythmic bioluminescence data (Plautz et al., 1997). Phase values were corrected to a 24-h timescale by dividing the FFT-NLLS calculated phase value of each seedling by its estimated period and multiplying this value by 24. Corrected phase values were plotted with a modified version of an R script employing the “polar.plot” function written by Michael Covington (Harmer and Kay, 2005).

warp2/sic-3 Mutant Discovery

Col-0 ProPRR7:LUC seeds were mutagenized with EMS. The M2 seeds were germinated for 3 d and entrained in LL|22:18°C for 4 d. The M2 seedlings were then transferred to constant darkness and constant 22°C in the camera and administered a 28°C pulse 28 h later. The 5328 M2 seedlings were screened for a fold change (FC) in bioluminescence lesser or greater than twice the sd of the wild type mean (FC = signal at 31 h/signal at 28 h) (Supplemental Figure 1A). Putative M2 seedlings were rescreened twice, and the phenotype was confirmed in the M3 generation. One mutant with attenuated FC was designated warp2/sic-3 (Supplemental Figure 1B).

Mapping of warp2/sic-3

warp2/sic-3 was mapped with next-generation EMS mapping (Austin et al., 2011), using the default settings (Supplemental Figure 2). The mapping population consisted of the F2 generation from warp2/sic-3 ProPRR7:LUC in the Col-0 accession backcrossed to wild-type plants of the Ler-0 accession. A pool of genomic DNA from 80 F2 individuals with the warp2/sic-3 phenotype was single end sequenced on a single lane of Illumina HiSeq 2000 with 50-bp read length by the QB3 Vincent J. Coates Genomics Sequencing Laboratory at University of California, Berkeley (http://qb3.berkeley.edu/gsl). Genomic DNA was randomly sheared genomic by Covaris S2 and used to prepare a sequencing library with the Apollo 324 NGS Library Prep System (WaferGen Biosystems). The average library size was 482 bp. Reads were mapped to the Arabidopsis TAIR10 genome sequence with bwa (Burrows-Wheeler Alignment Tool) (Li and Durbin, 2009) using the “index” command (-a bwtsw option), followed by the “aln” command (with default options), and alignments generated in the SAM (aequence alignment/map) format with the “samse” command (with default options). Alignments were converted to the BAM format with SAMtools (Li et al., 2009; Li, 2011). This BAM file is available at the NCBI Sequence Read Archive with BioProject accession number PRJNA314711. The mpileup command (-E -ugf options) in SAMtools followed by BFCtools was used to create a VCF (variant call format) file from the BAM file for input into the Next Generation EMS Mapping pipeline (Austin et al., 2011). Candidate genes in the sic-3 mapping interval were Sanger sequenced at the UC Berkeley DNA sequencing facility (mcb.berkeley.edu/barker/dnaseq/home) to confirm predicted EMS-induced mutations.

Construction of Transgenic Plants

Phusion High-Fidelity DNA Polymerase (New England Biolabs) was used to amplify the desired PCR product with the appropriate primers (Supplemental Data Set 2). PCR products were cloned into the pENTR/D-TOPO vector (Thermo Fisher Scientific) and transformed into One Shot TOP10 Chemically Competent Escherichia coli (Thermo Fisher Scientific). Binary vectors for plant transformation were generated by site-specific recombination with Gateway LR Clonase II enzyme mix (Thermo Fisher Scientific). Binary constructs were transformed into Agrobacterium tumefaciens strain GV3101 by electroporation, and Arabidopsis plants were transformed by the floral dip method (Clough and Bent, 1998). T1 transformants were selected on 1× MS plates (pH 5.7 to 5.9) supplemented with either 50 μg/mL kanamycin (Sigma-Aldrich) or 16 μg/mL phosphinothricin (Gold Biotechnology). Transgenic lines were genotyped by PCR with appropriate primers (Supplemental Data Set 2).

For the complementation tests, a 2296-bp amplicon, which included the full SIC (AT4G24500) gene and 134 bp of sequence upstream encompassing the predicted promoter region, was PCR amplified from wild-type genomic DNA and cloned into the pMDC100 binary vector (Curtis and Grossniklaus, 2003). T1 transformants in the sic-3 background were identified as kanamycin-resistant individuals.

To generate Pro35S:SIC-YFP lines, a 957-bp amplicon of the SIC coding sequence (lacking the endogenous stop codon) was PCR amplified from wild type cDNA and cloned into the pEarleyGate101 binary vector (Earley et al., 2006). T1 transformants in the wild-type background were identified as phosphinothricin-resistant individuals. Expression of the SIC-YFP fusion protein in the T3 generation was confirmed by protein gel blot with anti-GFP Living Colors A.v. Monoclonal Antibody (JL-8; Clontech) at 1:20,000 dilution and goat anti-mouse IgG HRP antibody (Santa Cruz Biotechnology) at 1:10,000 dilution, as described previously (Harmon et al., 2008). Crossing of the Pro35S:SIC-YFP construct into the sic-3 background rescued the long period phenotype. To determine SIC-YFP subcellular localization, Pro35S:SIC-YFP lines were grown under LD|22°C, and YFP signal was visualized in 13-d-old seedlings. Fluorescence and bright-field microscopy were performed on roots with a Leica DM4000B microscope (Leica Microsystems). Cloning and sequencing of the SIC transcript in wild-type and sic alleles employed Taq DNA Polymerase to amplify the desired PCR product from cDNA (see Supplemental Data Set 2 for primers). PCR products were cloned into the pCR2.1-TOPO vector (Thermo Fisher Scientific) and sequenced as described.

Gene Expression Analysis with qPCR

Seedlings were grown as described. For analysis in entrainment conditions, seedling tissue was collected on day 5 of entrainment (i.e., 8 d old) at ZT0, ZT3, ZT6, ZT9, ZT12, ZT15, ZT18, and ZT21 in the indicated entrainment conditions. For analysis in free-running conditions, seedlings were transferred from entrainment to LL|22°C free run on day 4 (i.e., 7 d old), and tissue was collected on day 2 of free run (i.e., 8 d old) at ZT24, ZT27, ZT30, ZT33, ZT36, ZT39, and ZT42. After collection, the tissue was immediately placed in liquid N₂. Total RNA was extracted with Plant RNA Reagent according to the manufacturer’s recommendations (Thermo Fisher Scientific). Genomic DNA was removed from total RNA with the Ambion TURBO DNA-free Kit (Thermo Fisher Scientific). First-strand cDNA was synthesized from 10 μg of total RNA with oligo(dT) primers and the Maxima H Minus first-strand cDNA synthesis kit (Thermo Fisher Scientific). qPCR was performed for three independent experiments with a CFX real-time system (Bio-Rad Laboratories) as described previously (Harmon et al., 2008). Transcript levels of target genes were calculated from the average C_q of two technical replicates using the equation , where C_q for each amplification curve was calculated by the “regression” mode in Bio-Rad CFX Manager 3.0 software (Bio-Rad Laboratories). The normalization control was the geometric mean of C_q values for IPP2 (AT3G02780) and PP2A (AT2G42500), calculated as √C_q(AT3G02780)*C_q(AT2G42500). Primers sequences are shown in Supplemental Data Set 2.

Phylogenetic Analysis

Arabidopsis SIC homologs were identified using BLASTP to identify homologous amino acid sequences (Altschul et al., 1990) in the UniProt Database (UniProt Consortium, 2015), Phytozome 11 (Goodstein et al., 2012), and EnsemblPlants (Kersey et al., 2016). Homologs were compiled using Geneious V. 9.0.5 software with the default settings (Kearse et al., 2012). This analysis generated a multiple sequence alignment with the MUSCLE tool based on a BLOSUM 62 matrix (Edgar, 2004), which was manually edited for obvious mismatched amino acid alignment errors in Mesquite V.3.04. Mesquite was also used to trim the alignment to each separate domain and reuploaded to Geneious to determine percent similarity. A maximum likelihood tree was built using PhyML 3.0 (Guindon et al., 2010) with the aBayes Fast likelihood-based method with 1000 bootstrap replicates. Data were visualized in FigTree v1.4.2. Predicted protein domains were identified using the PredictProtein tool (Yachdav et al., 2014).

Identification and Quantification of Transcript Splice Variants

Wild-type, sic-3, and sic-1 seedlings were grown as described. Entrainment conditions were either LD|16°C, LD|22°C, or LD|28°C, and experiments were always done concurrently. Tissue from 8-d-old seedlings was collected at ZT0, ZT6, ZT12, and ZT18, followed by RNA extraction and cDNA synthesis as described. cDNA pools for each entrainment condition and genotype were made by combining equal amounts of cDNA from each time point. Splice variants were detected by RT-PCR with primers designed with Primer3 software (Rozen and Skaletsky, 2000) or previously published (James et al., 2012). The desired RT-PCR products were amplified from cDNA pools with Taq polymerase and 25 to 35 cycles of amplification depending on the primer set. PCR products were separated in 1.2% agarose gels (Sigma-Aldrich) in 1× TAE buffer and bands visualized with a Typhoon FLA 7000 biomolecular imager (GE Life Sciences). Qualitative detection was performed by staining with 0.5 μg/mL of ethidium bromide. Quantitative detection employed FAM fluorophore labeling of PCR products as previously described (Schuelke, 2000; Lu et al., 2011), with the indicated modifications. Forward primers had an additional 5′ leader corresponding to the M13(-21) universal sequence (5′-TGTAAAACGACGGCCAGT-3′). Immediately after completion of a standard RT-PCR, a mixture of universal FAM-labeled forward primer (5′-[FAM]-TGTAAAACGACGGCCAGT-3′) and reverse primer was spiked into the reaction and amplification allowed to proceed for an additional eight cycles. FAM-labeled bands were quantified with ImageQuant TL 8.1 software (GE Life Sciences). Splice variant abundance for each sample was calculated by subtracting background from the band intensity for the FAM-labeled PCR product. Background was the intensity of an empty section of the same lane identical in area to the PCR band.

Accession Numbers

Gene models from this article can be found in the Arabidopsis Genome Initiative or GenBank/EMBL data libraries under the following accession numbers: SIC, AT4G24500; LHY, AT1G01060; CCA1, AT2G46830; RVE8, AT3G09600; TOC1, AT5G61380; PRR9, AT2G46790; PRR7, AT5G02810; PRR5, AT5G24470; PRR3, AT5G60100; GI, AT1G22770; LUX, AT3G46640; ELF3, AT2G25930; ELF4, AT2G40080; CAB2, AT1G29920; CAT3, AT1G20620; CCR2, AT1G06820; SKIP, AT1G77180; STIPL1, AT1G17070; PRMT5, AT4G31120; PP2A, AT5G60100; IPP2, AT3G02780; UBC21, AT5G25760; AT4G31770; AT2G38770; AT3G18790; AT1G13690; and AT5G28740. Sequence data from this article can be found in the NCBI Sequence Read Archive under BioProject accession number PRJNA314711.

Supplemental Data

• Supplemental Figure 1. Screen for warp (warm acute response of PRR7) mutants.

• Supplemental Figure 2. Mapping warp2/sic-3 with bulk segregant analysis and next-generation EMS mapping.

• Supplemental Figure 3. sic mutants affect AT4G24500, which encodes a conserved proline/serine-rich protein.

• Supplemental Figure 4. sic mutants perturb the accuracy and precision of circadian clock period and phase.

• Supplemental Figure 5. sic disrupts expression waveforms of core circadian clock genes and clock-regulated output genes.

• Supplemental Figure 6. sic does not affect rhythms in warm thermocycles and in cool photocycles.

• Supplemental Figure 7. Exposure of sic to cold temperature cycles impairs temperature compensation and rhythm quality.

• Supplemental Figure 8. Phylogenetic tree of SIC homologs in angiosperms.

• Supplemental Figure 9. sic does not change all splice variants possible for transcripts from CCA1, LHY, GI, ELF3, TOC1, PRR9, PRR7, PRR5, and PRR3.

• Supplemental Figure 10. Expression of LHY, ELF3, PRR7, and CCA1 transcripts in sic and the wild type under conditions for detection of splice variants.

• Supplemental Figure 11. Long period and weakened rhythms under thermocycles in sic depend on CCA1 and LHY.

• Supplemental Data Set1. Circadian clock parameters determined from ProPRR7:LUC rhythms in the indicated conditions.

• Supplemental Data Set2. Primers used in this study.

• Supplemental References. References for supplemental data.

• Supplemental File 1. Alignment of SIC homologs.

Glossary

miRNA

microRNA

LL

constant light

LD

long-day

FFT-NLLS

fast Fourier transform nonlinear least square

RAE

relative amplitude error

NMD

nonsense-mediated decay

MS

Murashige and Skoog

FC

fold change