Abstract
Viable circadian clocks help organisms to synchronize their development with daily and seasonal changes, thereby providing both evolutionary fitness and advantage from an agricultural perspective. A high-resolution mapping approach combined with mutant analysis revealed a cereal ortholog of Arabidopsis thaliana LUX ARRHYTHMO/PHYTOCLOCK 1 (LUX/PCL1) as a promising candidate for the earliness per se 3 (Eps-3Am) locus in einkorn wheat (Triticum monococcum L.). Using delayed fluorescence measurements it was shown that Eps-3Am containing einkorn wheat accession KT3-5 had a distorted circadian clock. The hypothesis was subsequently confirmed by performing a time course study on central and output circadian clock genes, which showed arrhythmic transcript patterns in KT3-5 under constant ambient conditions, i.e., constant light and temperature. It was also demonstrated that variation in spikelet number between wild-type and mutants is sensitive to temperature, becoming negligible at 25°. These observations lead us to propose that the distorted clock is causative for both early flowering and variation in spike size and spikelet number, and that having a dysfunctional LUX could have neutral, or even positive, effects in warmer climates. To test the latter hypothesis we ascertained sequence variation of LUX in a range of wheat germplasm. We observed a higher variation in the LUX sequence among accessions coming from the warmer climate and a unique in-frame mutation in early-flowering Chinese T. turgidum cultivar ‘Tsing Hua no. 559.’ Our results emphasize the importance of the circadian clock in temperate cereals as a promising target for adaptation to new environments.
THE circadian clock is an intrinsic regulator of biological processes oscillating within an ∼24-hr period (Pittendrigh 1993). It is considered to be the main mechanism by which plants recognize the optimal photoperiod for seasonal flowering (Imaizumi 2009). Transcriptional regulation of the circadian clock has been well described in Arabidopsis (Pokhilko et al. 2012) with the latest model emphasizing the importance of the Evening Complex (EC) composed of EARLY FLOWERING 3, EARLY FLOWERING 4, and LUX ARRHYTHMO/PHYTOCLOCK 1 (ELF3, ELF4, and LUX/PCL1) proteins (Onai and Ishiura 2005; Nusinow et al. 2011; Pokhilko et al. 2012). The EC directly represses the function of PSEUDO RESPONSE REGULATOR 9 (PRR9) (Helfer et al. 2011) and acts antagonistically to the elements expressed in the morning, including LATE ELONGATED HYPOCOTYL (LHY) and CIRCADIAN CLOCK ASSOCIATED 1 (CCA1) (Pokhilko et al. 2012). LHY, CCA1, and PRR9 form the so-called morning loop, which becomes arrhythmic when the EC is impaired (Hazen et al. 2005; Dixon et al. 2011; Nusinow et al. 2011). The EC also down-regulates transcription of evening genes such as GIGANTEA (GI) and TOC1 (Hazen et al. 2005; Dixon et al. 2011; Pokhilko et al. 2012).
Genetic studies have shown that the recently cloned maturity-a, (mat-a; syn. early maturity 8, eam8) locus is an ortholog of AtELF3 in barley (Hordeum vulgare L.). eam8 appears to be epistatic to eam10 (syn. easp), which in turn seems to be a possible ortholog of LUX/PCL1 (Gallagher et al. 1991; Zakhrabekova et al. 2012; Campoli et al. 2013). This would be consistent with Eam10 and Eam8 in barley forming a complex similar to AtLUX and AtELF3 (and AtELF4) in Arabidopsis (Nusinow et al. 2011). Interestingly, the genetic location of eam10 in barley and earliness per se 3Am (Eps-3Am) in einkorn wheat (Triticum monococcum L.) appears to be syntenic (Gallagher et al. 1991; Börner et al. 2002; Gawroński and Schnurbusch 2012). Both mutants also display similar phenotypic features and flower early under both long and short day conditions, thus resembling mutants in the Arabidopsis EC (Hicks et al. 2001; Shindo and Sasakuma 2001; Doyle et al. 2002; Hazen et al. 2005; Zakhrabekova et al. 2012). Moreover, recent analysis of eam8 (HvELF3) and eam10 (HvLUX1) mutants showed misexpression of circadian clock gene-related transcripts (Faure et al. 2012; Campoli et al. 2013), which was in line with previous findings in Arabidopsis (Hicks et al. 2001). So far, cereal PCL1/LUX was proposed as a candidate conferring early flowering in eam10 and KT3-5 mutants (Mizuno et al. 2012; Campoli et al. 2013). Authors have concluded that the mechanism of the PCL1/LUX protein action is similar in both einkorn wheat and barley; it represses expression of Vrn3/FT through the negative regulation of Ppd-1/Ppd-H1, thus delaying flowering. When PCL1/LUX is knocked-out, it causes the early heading phenotype, especially when the mutant harbors the wild-type (wt) allele of Ppd-H1; i.e., photoperiod sensitivity (Mizuno et al. 2012; Campoli et al. 2013).
Recently, delayed fluorescence (DF) has been proposed as a method for quick but high-resolution analysis of the circadian clock in virtually any species (Gould et al. 2009). The DF phenomenon, discovered by Strehler and Arnold (1951) is a luminescence produced by the photosynthetic apparatus after excitation with ambient light as a result of charge recombination in the photosystem II (PSII) and subsequent emission of a photon (Rutherford et al. 1984). DF allows measuring the intrinsic oscillation in chlorophyll fluorescence (PSII) that remains under circadian control (Gould et al. 2009). The value of DF measurements in cereal mutant research has not, to our knowledge, been reported in the literature. However, it appears to have considerable potential for high-throughput analysis in circadian clock research.
Here we present more detailed genetic and physiological analyses of the Eps-3Am mutant in diploid wheat KT3-5. Fine mapping and comparative genetic analyses with barley and bread wheat provide more confirmation that the TmLUX/PCL1 is the most sensible candidate conferring the early flowering phenotype. Moreover, we present physiological evidence that KT3-5 possesses a distorted circadian clock and has a decreased phenotypic plasticity, in agreement with the view that variation in the clock is an effective adaptive mechanism (Dodd et al. 2005). We also mark the importance of the circadian clock for adaptation to temperature, in addition to photoperiod perception as discussed elsewhere (Faure et al. 2012; Zakhrabekova et al. 2012).
Materials and Methods
High-resolution mapping
Seeds of the F2 population consisting of 658 individuals, as well as parental lines recombinant inbred line (RIL) RILWA25 and RILWA71, were germinated on Petri dishes. To synchronize germination, soaked seeds were kept at 4° for 2 days. Seedlings of selected recombinant individuals plus 38 random plants were transplanted and placed in a cool room (15°, photoperiod of 11 hr light (L)/13 hr dark (D)) for 2 weeks, followed by cold treatment (4–8°) for 3 weeks and a photoperiod of 10 hr L/14 hr D. For acclimation, plants were moved back to cool temperature (15°, photoperiod of 11 hr L/13 hr D) for 2 weeks and were transplanted to 1.1-liter pots. After transplanting, the LD treatment (16 hr/8 hr) started at 18°/15° until harvest. Each pot was manually randomized every 3–4 days to avoid positional bias. Heading time was scored when the visible part of awns on the main culm was about 1 cm long. A total of 658 individuals were screened with flanking markers INDEL_271_272 and INDEL_201_202 giving size polymorphism (Supporting Information, Table S1). Plants with a recombination event between flanking markers were further genotyped with markers listed in Table S1 and Table S2. Marker development was performed according to the procedure described elsewhere (Gawroński and Schnurbusch 2012).
Analysis in einkorn wheat mutants
Growing conditions (greenhouse), phenotypic analyses, DNA extraction and PCR genotyping were conducted as previously reported (Gawroński and Schnurbusch 2012). Mutants and wild-types analyzed are listed in Table S3.
Delayed fluorescence measurement
Leaf samples were taken from 4-week-old plants, cut into 2-cm pieces and floated on sterile distilled water containing 18 mg/liter fungicide Dithane poured into a 25-mm compartmental Petri dish. At dusk (10:00 pm, ZT = 12, where ZT means “Zeitgeber time”) dishes were put into a Sanyo MIR-553 cooled incubator (Sanyo Gallenkamp) and the DF was imaged using an ORCA-II-BT 1024 16-bit camera (Hamamatsu Photonics) cooled to −80°. However, the “zero” time point was set at dawn next day (10:00 am, ZT = 0). Leaf pieces were kept in darkness intermitted every hour with a pulse of red/blue light (80 µmol/m2⋅s) lasting for 1 min. Immediately after the pulsing, the picture was captured allowing for assaying the circadian clock output with 1-hr resolution. The whole process was automated by using Wasabi software (Hamamatsu Photonics) for controlling the camera and light source (light-emitting diodes). The temperature during the experiments was kept constant at 22° or 17°. The pictures were analyzed in Metamorph 6.0 (Universal Imaging) and the numerical light intensities were extracted. Obtained data were normalized and detrended using Excel (Microsoft). Regression equation of the polynomial trend line in the order of six was used to subtract Y values (delayed fluorescence) thus removing the trend. The data showed a pattern of up to six peaks of DF. The first peak of DF needed to be discarded as it was out of phase. Data collected between hours 36 and 132 (normalized but not detrended) were used to calculate periods corresponding to the four middle peaks of the delayed fluorescence. The periods at 95% confidence interval and relative amplitude errors (RAEs) were calculated in BRASS (http://millar.bio.ed.ac.uk/PEBrown/BRASS/BrassPage.htm) by running fast Fourier transformed nonlinear least-square analysis (Plautz et al. 1997).
Time-course RT-qPCR
Seeds of DH_BG284E11-PP1, BG353/1E15 (File S1), T. monococcum KT3-1 and KT3-5, T. turgidum ‘Tsing Hua no. 559’ and ‘Fo Shou Mai,’ and azygous segregants of BG284E11-38 were soaked in short day (SD) water and kept in the cold (4°–8°) to synchronize germination. After 2 days, young seedlings were moved to room temperature for 2 days and planted in soil (150-ml wells). Plants were grown in the greenhouse at 20°/17° day/night under a photoperiod of 16 hr for 4 weeks. In the actual phase of the experiment, plants were moved to the incubator Heraeus Vötsch, type HPS 1500/S. For the following 2 days conditions were changed to 12 hr/12 hr at 18°/22° to synchronize the circadian clocks of the plants. In the last stage of the experiment, constant light and temperature (22°) were set for 3 days. Sampling was performed on 31-day-old plants starting at 6:00 am of the second day under constant light and temperature. Leaves were harvested every 3 hr (LL conditions) from at least three plants per time point per genotype (biological replicates) and immediately frozen in liquid nitrogen. RNA extraction was performed by using PureLinkRNA mini kit (Invitrogen) in combination with TRIzol reagent (Invitrogen) as described by the manufacturer. QuantiTect Reverse Transcription kit (Qiagen) was applied to synthesize cDNA using 1 µg of total purified RNA. qPCR reactions were performed on the ABI 7900HT Fast Real-Time PCR system (Applied Biosystems) employing QuantiTect SYBR Green PCR kit (Qiagen) at the following thermal profile: 94° for 15 min, five cycles of 94° for 30 s, 65° for 30 s decreasing by 1° per cycle, and 72° for 30 s, and then 40 cycles of 94° for 30 s, 60° for 30 s, 72° for 30 s, and 72° for 1 min. Primers used to amplify clock genes are listed in the Table S1. Gene nomenclature and sequences used for primer development were as published elsewhere (Campoli et al. 2012). TtLUX-A and TtLUX-B, A and B genome specific primers, respectively, were developed initially by using the LUX sequences from following databases: http://www.wheatgenome.org/ and http://www.cshl.edu/genome/wheat. The genome specificity was confirmed by amplification of the TtLUX from flow-sorted chromosome arms of cv. ‘Chinese Spring.’
Detailed methods for physical mapping, DNA gel blot analysis, growing conditions of KT3-1 and KT3-5, targeting induced local lesions in genomes (TILLING), development and analysis of knock-down lines, and LUX resequencing are included in File S1 (Materials and Methods).
Results
Phenotypic characterization of KT3-5
KT3-5 is an Eps-3Am early flowering einkorn wheat mutant derived from a donor line called KT3-1. We phenotypically analyzed KT3-5 and KT3-1 in different environments and showed that KT3-5 flowered early in our experimental conditions, irrespective of photoperiod (Table 1 and File S2), consistent with observations made in a previous study (Shindo and Sasakuma 2001). We found that differences between wild-type and mutant plants were exaggerated when the plants were exposed to SD treatment (8-hr photoperiod, Table 1). Immature spike investigations under long-day (LD, 16 hr/8 hr day/night) and cool (16°/14°) growing conditions revealed that KT3-5 transitioned to the generative phase (double-ridge stage, P = 0.028) 10 days earlier than KT3-1 and to the terminal spikelet stage 7 days in advance (terminal spikelet stage, P = 0.0085).
It has previously been postulated that eps loci can be negatively associated with spike characters including spike length and spikelet number depending on temperature (Lewis et al. 2008; Gawroński and Schnurbusch 2012). At LD (16°/14°) we observed that KT3-1 produced on average four spikelets more than KT3-5 (P = 8.3 × 10−8) but this did not significantly affect the difference for the kernel number per spike (KNP, P = 0.81) (Figure 1A, Table 1). At higher temperatures during early development, fewer spikelets developed on both KT3-1 and KT3-5 inflorescences. When plants were grown at LD and 26°/24°, both KT3-1 and KT3-5 produced a similar number of tillers, similar number of spikelets per spike, and spike sizes (Table 1, File S2, and Figure 1A). However, the KT3-5 mutant carrying Eps-3Am headed 20 days earlier and produced significantly more grains per spike (KNP). Thus, Eps-3Am does not appear to negatively influence yield-related spike characters when plants are exposed to high temperatures during early spike development and there was a significant interaction between genotype and temperature during spike development (Table 1 and Figure 1A).
Phenotypic analysis of the Eps-3Am locus. (A) Spikelet numbers per spike of wild-type KT3-1 (Triticum monococcum L.,) and mutant KT3-5 were plotted to show interactions with temperature. Presence of the interaction was confirmed with two-way factorial ANOVA at P < 0.0001. At 15°, KT3-1 developed significantly more spikelets than KT3-5 (P < 0.0000001, paired Student’s t-test). Seven plants per genotype and temperature treatment were analyzed. Error bars indicate SEM. (B) Frequency distribution of heading time among 38 F2 recombinants and 38 randomly selected individuals grouped according to the presence/absence of the TmLUX gene. (C) Days to heading of F3 families with closest recombination events were similarly determined by the genotypic status at the Eps-3Am locus; (a + b) plants with mutant allele (solid) headed earlier than those carrying the wild-type allele (shaded). Error bars represent SEM.
Genetic analysis of Eps-3Am
Initial phenotyping and high-density mapping of Eps-3Am was performed in a recombinant inbred line (RIL) population derived from a cross between KT3-5 (spring type) and wild-type line KT1-1 of T. boeoticum Boiss. (winter type) (Gawroński and Schnurbusch 2012). Two selected spring-type lines from this population, RILWA25 and RILWA71, were then crossed to create an F2 population for high-resolution mapping. The population consisted of 658 individuals and was screened with two markers flanking Eps-3Am (File S3). Thirty-eight recombinants were identified in this initial mapping and these were subsequently used to delimit Eps-3Am to a sub-cM level. As for eam10 in barley (Campoli et al. 2013), a putative ortholog of AtLUX ARRHYTHMO (Hazen et al. 2005; Onai and Ishiura 2005) (i.e., TmLUX) cosegregated with the flowering phenotype. Phenotypic distribution of Eps-3Am (8 early:30 late) was not significantly different from 1:3, thus being in agreement with a single-gene-model inheritance (P = 0.57, χ2 = 0.316).
We grew all recombinant plants plus 38 randomly chosen F2’s and showed that the distribution of flowering time agreed completely with the genotypic status at the TmLUX gene (Figure 1B). PCR-based screens of barley BACs using molecular markers flanking Eps-3Am identified barley contig_95 on the recently released physical map (Mayer et al. 2012) (Table S2, File S4, and Figure 2A). Additional marker analyses then enabled us to narrow the region containing Eps-3Am to a syntenic region in barley that contained only two putative genes, HvPUMILIO (GenBank accession no. KF769445) and HvLUX (Figure 2A). F2 individuals having the closest recombination events (seven proximal, one distal), were advanced to the F3 generation (File S3). This allowed us to confirm the association between heading time (Eps-3Am) and presence/absence of TmLUX (Figure 1C). We were able to discard HvPUMILIO as Eps-3Am by TILLING. Screening for mutations in HvPUMILIO identified a single mutant that carried a premature stop codon (Figure S1, File S5). Importantly this mutant did not exhibit an early flowering phenotype (Figure S1). Finally, to address the possibility that other nonsyntenic genes might have been present at Eps-3Am in einkorn wheat and may be responsible for the phenotype, we built a physical map from bread wheat chromosome 3A (T. aestivum L. cv. ‘Chinese Spring,’ (Figure 2A, File S6, and File S7) and identified two contigs that were syntenic to Eps-3Am. In terms of sequence conservation, we observed that the Eps-3Am locus in einkorn wheat had higher resemblance to the syntenic barley sequence than to that of bread wheat (Figure 2A and Table S2).
Comparative analysis of the Eps-3Am locus. (A) Genetically mapped Eps-3Am locus (3Am) was integrated with physical ctg_95 from barley chromosome 3H and two contigs, ctg_1331 and ctg_1512 from bread wheat chromosome 3A. Wheat contigs did not overlap (gap). Only BAC addresses selected for the minimum tiling path (MTP) are shown. Putative genes (color-coded) annotated from the barley and wheat sequence could be mapped in F2 population of einkorn wheat, revealing higher synteny with barley. New recombinations found in the F2/F3 population (rectangles filled with number of recombinations) delimited the Eps-3Am locus to only two genes: TmLUX ARRHYTHMO (TmLUX, red) and TmPUMILIO (orange). Both genes had been deleted from the genomes of the early heading T. monococcum KT mutants. Genes mapped proximally to the locus (KIN, HMA, and LEG) did not have putative orthologs in the collinear part of rice chr1 (Os01) and the syntenic relationship was thus broken at this site. RFP, a putative ring finger protein and ZT, a putative zinc transporter flanked the Eps-3Am locus from the distal site. A marker developed based on the putative transcription factor (BTB/POZ) could not be linked with the locus in einkorn wheat. (B) Amino acid sequence of the MYB domain from LUX was conserved between wheat and barley. Resequencing of LUX in selected germplasm revealed deletion and substitution mutants at a highly conserved nucleic acid binding motif ‘SHxQK’(Hwang et al. 2002) ‘Tsing Hua no. 559’ in T. turgidum L. (A genome copy of LUX), and ‘Super Precoz 2H’ (eam10) in barley (Campoli et al. 2013), respectively.
During our genetic characterization of Eps-3Am, we noted that two adjacent marker loci, TmLUX (PAV_261_262) and TmPUMILIO (PAV_295_296), were scored as null alleles in KT3-5 (Figure 2A), suggesting a putative deletion of both genes in agreement with previous findings (Mizuno et al. 2012). DNA gel blot analysis confirmed this result (Figure S2). We then identified 15 additional X-ray early mutants of einkorn wheat with null alleles at both loci (Table S3). Haplotype analysis revealed the presence of four groups, with null allele at marker flanking Eps-3Am in KT3-10, suggesting a minimum of two independent deletion events being present in the deletion panel (Table S4). All 15 mutants flowered significantly earlier than KT3-1, with the differences ranging from 25 up to 60 days (Table S4).
Circadian clock distortion in KT3-5
DF measurements have been shown to mirror CAB2:LUC reporter analysis in circadian clock mutants of Arabidopsis. Importantly, DF has no requirement for transgenesis, which was an important advantage while working with einkorn wheat (Gould et al. 2009). We hypothesized that if TmLUX was functionally orthologous to AtLUX (Hazen et al. 2005; Onai and Ishiura 2005), then a complete knock-out of the TmLUX should cause circadian clock distortion. To test this hypothesis, we measured DF in KT3-5 and its wild-type KT3-1 (File S8). At 22° DF oscillation generally ceased in the KT3-5 after 2 days (Figure 3, A and B). Samples that remained rhythmic displayed significant amplitude lengthening to 28 hr compared to 23 hr in the wild-type (Figure 3, B and C, Table S5). At 17° the differences were less pronounced (Figure 3D). Nevertheless we conclude that DF is appropriate for measuring circadian clock distortion in einkorn wheat, easily revealing the perturbed clock phenotype in the KT3-5 einkorn wheat mutant.
Delayed fluorescence (DF) measurements were performed to compare wild-type and mutant circadian clock outputs to the chloroplast. Oscillation of the DF detected in the wild-type KT3-1 (A) (T. monococcum L.) lasted longer than that of the mutant KT3-5 (B). Leaf samples were kept under 22° in the darkness interrupted every hour with a light pulse lasting for 1 min prior to capturing the picture. Light intensities were extracted and normalized and trends for the DF curves were removed. Error bars indicate standard deviations for 16 replicates. (C and D) Relative amplitude errors (RAEs) of the DF oscillation plotted against the periods generated from the BRASS software. Periods were calculated for the DF from 36 to 132 hr. Each data point represents the time course measurement at a single region on the leaf from wild-type KT3-1 or mutant KT3-5 plants. Experiments were performed in two different temperature regimes: 22° (C) and 17° (D). In both cases, wild-type samples were more tightly clustered around the expected value of the period equaling 24 hr. (C) At 22°, mutant KT3-5 showed greater period lengthening than KT3-1, despite the higher RAE values found in both lines. The temperature equaling 22° was thus an appropriate condition to recognize a mutant clock phenotype.
It had already been shown that KT3-5 had altered levels of circadian clock-related transcripts (Mizuno et al. 2012); however, the arrhythmia could not be detected because of the alternating sampling conditions in the previous study (Mizuno et al. 2012). To enable a proper comparison between KT3-5 and LUX mutants in Arabidopsis (Hazen et al. 2005; Onai and Ishiura 2005), the einkorn lines must be kept under constant ambient conditions, i.e., constant light and temperature. This is a basic prerequisite to analyze the circadian clock, as, according to its definition, it has to be maintained under constant environments (de Montaigu et al. 2010). In our study, einkorn wheat mutant KT3-5 showed a severely dampened amplitude and/or arrhythmia in transcript patterns of both morning (TmCCA1, TmPRR95, TmPRR73 (Figure 4, A–C, File S9) and evening (TmGI, TmPpd1, TmPRR59, TmTOC1, TmLUX, and TmElf3) elements (Figure 4, D–I, File S9). As KT3-5 is a null mutation, it did not express TmLUX at all (Figure S2 and Figure 4H). However, consistent with an early flowering phenotype, KT3-5 displayed elevated levels of TmFT—on average 10 times higher in KT3-5 than wt KT3-1 (Figure 4I).
Relative transcript levels of circadian clock genes in einkorn wheat. (A) TmCCA1, (B) TmPRR95, (C) TmPRR73, (D) TmGI, (E) TmPpd1 (F) TmPRR59, (G) TmTOC1, (H) TmLUX, (I) TmElf3, and (J) TmFT were measured from a 1-day time-course RT-qPCR study on wild-type KT3-1 (T. monococcum L., blue diamonds) and early heading mutant KT3-5 (T. monococcum L., red squares). Sampling was performed on 31-day-old plants starting at 6:00 am (ZT = 0) of the second day under constant light and temperature. Leaves were harvested every 3 hr (LL conditions) from at least three plants per time point per genotype (biological replicates). Error bars indicate SEM.
To gain a complementary proof of function of TmLUX, we had already generated transgenic HvLUX-RNAi (RNA interference) plants in barley cv. ‘Golden Promise.’ We therefore took the opportunity to analyze clock gene expression in two of these transformants, DH_BG284E11-PP1 and BG353/1E15, which showed clear down-regulation of HvLUX transcripts compared to their azygous segregant BG284E11-38 (Figure S3D, File S9). In contrast to the KT3-5 mutant, the transcript abundances of HvCCA1, HvPRR95, HvGI, and HvLUX were not arrhythmic (Figure S3). Transcript levels of HvCCA1 were slightly up-regulated in DH_BG284E11-PP1 and BG353/1E15, while those of HvPRR95 and HvGI were quite strongly up-regulated (Figure S3, A–D). Furthermore, neither transformants exhibited an early flowering phenotype (Table S6, File S10).
Resequencing of LUX ortholog in 96 wheat accessions
To study sequence variation of LUX in wheat, we selected 96 wild and cultivated accessions covering the majority of the observed phenotypic variation in flowering time and resequenced the putative LUX ortholog (Figure S4A, File S11). While focusing on A and B genome copies of the gene (GenBank accession nos. KF769443 and KF769444), which were mostly represented (Table S7), we found a higher number of haplotypes in wild wheat accessions—18 vs. 8 and 5 vs. 4 for A and B copy, respectively (Table S8, File S11, File S12, and File S13). Because we found that the KT3-5 mutant was thermosensitive (Table 1), we also grouped the accessions according to the climatic conditions at the site where they had been collected (Figure 1A, Table 1). This grouping revealed more sequence variation being present in warmer climates within the A copy of LUX—16 vs. 12 (warm/cool) haplotypes and one haplotype more in the B copy, 7 vs. 6 (Table S9). This trend might suggest increased wheat LUX haplotype diversity in warmer climates. However, most haplotypes of the cultivated accessions came from the cooler climate (7 vs. 4), whereas most haplotypes of the wild accessions were collected from the warmer climate (14 vs. 9) (Table S10). Thus, higher sequence variation among the wild wheat accessions grown in warmer conditions, at least in part, might explain this observation (Table S8).
Another finding was the discovery of a TtLUX-A (A-genome homoeolog of TtLUX) allele containing a seven-amino acid in-frame deletion within the MYB domain (Figure 2B, File S12). Importantly, a tetraploid wheat cultivar harboring this mutation, ‘Tsing Hua no. 559,’ headed among the early lines (May 31), while the range for the whole collection was from May 23 to June 30 (Figure S4A, File S11). The mutated TtLUX-A was considered to be nonfunctional and the circadian clock of the cultivar was examined in a time-course RT-qPCR (Figure S4, B–E, File S9). Transcript levels of both TtLUX-A and TtLUX-B (B-genome homoeolog of TtLUX) were up-regulated in ‘Tsing Hua no. 559’ compared to ‘Fou Shou Mai’ (Figure S4B). As previous studies had shown that functional AtLUX suppresses its own expression (Hazen et al. 2005; Onai and Ishiura 2005; Helfer et al. 2011), up-regulated transcript levels of TtLUX in ‘Tsing Hua no. 559’ may be a direct consequence of a nonfunctional TtLUX-A. Also, because the viable TtLUX-B most likely rescued the effect of mutated TtLUX-A, no clock distortion was observed (Figure S4, B–E). This idea is consistent with patterns of oscillation of other clock genes, TtGI, TtPRR95, and TtCCA1, which were very similar between both genotypes (Figure S4, C–E). However, we cannot completely rule out the possibility that a dysfunctional TtLUX-A has a partial phenotypic effect on flowering, which would be consistent with a role for eps genes in fine tuning flowering in polyploid wheats (Snape et al. 2001).
Discussion
Mutants in LUX ARRHYTHMO have been associated with early flowering in Arabidopsis where genetic and functional studies allowed the gene to be placed within the evening loop of the circadian clock (Hazen et al. 2005; Onai and Ishiura 2005). LUX participated in forming an EC together with two other proteins, ELF3 and ELF4 (Nusinow et al. 2011). Recently, WPCL1, a gene homologous to Arabidopsis PCL1/LUX was proposed as a candidate conferring early heading in einkorn wheat mutant KT3-5 (Mizuno et al. 2012). The prediction was based on high-density QTL mapping and synteny with barley genome (Mizuno et al. 2012). Preliminary results showed that a possible deletion of WPCL1 could be responsible for the observed phenotype in KT3-5 (Mizuno et al. 2012). Within the present study, we found that WPCL1 is the most sensible candidate for Eps-3Am containing KT3-5. By using fine mapping, we also delimited the locus to only two genes when compared to the syntenic region in the barley genome. This included putative TmPCL1/TmLUX and TmPUMILIO (Figure 2A). We performed TILLING of both genes; however, only for TmPUMILIO did we obtain one mutation potentially affecting gene function (Figure S1). Although as expected, we did not observe an early flowering phenotype in those plants, the obtained result needs to be further verified by introgressing the mutation into the nonmutant background to rule out the possible masking effect of other genome aberrations. Such an approach has already been successfully applied in barley mutant research (Druka et al. 2011).
Previously discovered differential regulation of clock-related transcripts allowed us to hypothesize that early flowering in KT3-5 was due to a disrupted circadian clock (Mizuno et al. 2012). However, in the previous study, plants were only analyzed under short day conditions, which is insufficient to fully explain the circadian clock distortion in KT3-5 (Mizuno et al. 2012). In the present study, we kept the plants in a constant environment, i.e., constant light and temperature, and this revealed arrhythmia in transcript patterns of clock-related genes (Figure 4). We thus demonstrated that the KT3-5 mutant is very similar to eam10 as both are probably caused by knock-out of the orthologous gene PCL1/LUX (Campoli et al. 2013).
HvLUX-RNAi plants generated in our study helped us to confirm the position of HvLUX in the circadian clock network. In agreement with expectations, we observed higher transcript levels of evening elements HvPRR95 and HvGI in these plants (Figure S3, B and C). This suggests the repressive action of HvLUX by direct binding to the HvPRR9 promoter as already well investigated between AtLUX and AtPRR9 in Arabidopsis (Helfer et al. 2011). Besides, the effect of HvLUX knock-down on HvLHY transcript level was much smaller, probably due to the missing direct interaction between these elements (Figure S3A) (Pokhilko et al. 2012). However, slight up-regulation of HvLHY transcript is against prediction from the most recent clock model, in which PRR9 and PRR7 proteins down-regulate LHY (Pokhilko et al. 2012). On the other hand, the reciprocal positive effect of LHY on PRR9 may be responsible for the observed low transcript levels of TmPRR95 in KT3-5 (Figure 4B), (Pokhilko et al. 2012). We hypothesize that under constant conditions and complete absence of TmLUX action, the level of TmPRR95 protein initially rises and leads to severely dampened LHY transcript level through the negative feedback. However, low TmLHY protein levels cannot fulfill the required activating effect on the TmPRR95 and high transcript levels of TmPRR95 cannot be maintained, resulting in becoming low and arrhythmic (Figure 4B).
It has been suggested that the early heading of KT3-5 was driven by elevated Ppd1 expression and subsequent accumulation of WFT transcript (Mizuno et al. 2012). Evidence found by Campoli et al. (2013) also supports that HvLUX1 (barley ortholog of PCL1/LUX) acts on HvFT through Ppd-H1, as it shows a clear interaction with this gene, e.g., eam10 mutant lines harboring the wild-type Ppd-H1 allele flowered earlier than those with the recessive ppd-H1 mutant allele (Campoli et al. 2013). In our study, we also observed elevated transcript level of TmPpd-1 in KT3-5 during the subjective night/dawn period (Figure 4E) and in turn higher TmFT/TmVrn3 transcript levels (Figure 4J). In addition, our transgenic HvLUX-RNAi plants were generated in ‘Golden Promise,’ which possesses the ppd-H1 mutant allele, i.e., photoperiod insensitivity (Faure et al. 2012), and this may be another decisive factor for the missing early flowering phenotype in our transgenic plants (Table S6). Hence, we conclude that HvLUX knock-down is not sufficient to abolish the circadian rhythm under constant environment and to induce an early flowering phenotype in photoperiod insensitive barley cv. ‘Golden Promise.’
Disruptive mutations in cereal LUX seem to be favorable for spring cultivars that are dedicated for late sowing. In Japan, mutant KT3-5 performed better than KT3-1 as it could escape the humid and hot summer season and reach full maturity due to earlier flowering (Utsugi et al. 2006). In the case of winter cultivars, a functional LUX seems to be required to provide full yield potential, a requirement consistent with low temperature treatment during vegetative growth, which significantly increased yield of wild-type plants. We speculate that the low photoperiodic requirement of cereal LUX mutants could also be advantageous for a better adaptation to shorter growing seasons, shorter photoperiods, and warmer climates. This makes it very similar to HvELF3 mutations (Faure et al. 2012; Zakhrabekova et al. 2012).
It is thought that the eps trait is connected with adverse effects on spike size and decreased plant yield (Lewis et al. 2008). We observed a similar relationship in the KT3-5 mutant, but only in the cooler temperature regime (15°). We found that spike development also depends on temperature, with the most temperature-sensitive period being the early phase of development. We observed much greater phenotypic variation in spikelet number per spike in KT3-1 (wt) than in KT3-5 when grown in different environmental regimes, indicating that KT3-5 has a decreased phenotypic plasticity.
Consistent with this finding that at warmer temperatures KT3-5 mutant performed better than the donor wild-type KT3-1, we hypothesized finding higher sequence variation in LUX in warmer climates. We in fact observed this association in wheat within the sequence of the A-genome copy of LUX. Besides, we found one putative nonfunctional mutation in the Chinese cultivar ‘Tsing Hua no. 559.’ This cultivar was initially classified as being from a cooler climate, although the average temperature value was just below the chosen threshold (17.5°). Apart from this allele, we did not find any yet known functional variation in wheat homoeologs of LUX, which generally is very similar to the situation among barley accessions (Campoli et al. 2013). Future studies need to be carried out to further test the relationship of an adaptive advantage between temperature and circadian clock in temperate cereals.
In this study on LUX ARRHYTHMO in einkorn wheat, we have provided further justification for circadian clock research in modern crops and a clear link to environmental adaptation and yield optimization. We have also shown that the most efficient route to follow can be screening for clock mutants by using the DF measurements and subsequently confirming putative mutations with the time course RT-qPCR. Further investigations are required, including those at the protein level, to bring more direct evidence for the existence and molecular function of an EC in cereal crops.
Acknowledgments
We appreciate help from Hana Simkova, Institute of Experimental Botany (Olomouc, Czech Republic), for sharing the 3AL, 3B, and 3DL DNA samples. We thank Robbie Waugh, James Hutton Institute (Dundee, United Kingdom), for critically reading previous versions of the manuscript. We acknowledge the help of Susanne König, Ravi Koppolu, Marzena Kurowska, Dimitar Douchkov, Patrick Schweizer, Andrew J. Millar, Anna A. Filatenko, and excellent technical assistance from Kathrin Gramel-Eikenroth, Enk Geyer, Corinna Trautewig, Jelena Perovic, Anne Kusserow, Ulrike Beier, Carola Bollmann, Sabine Sommerfeld, Andrea Müller, Ingrid Otto, Ute Krajewsky, Heike Harms, Christiane Kehler, Marita Nix, Birgit Dubsky, and Kerstin Wolf. This work was supported by grants from the US Department of Agriculture National Institute of Food and Agriculture grant 2008-35300-04588 to BSG; German Research Foundation (Deutsche Forschungsgemeinschaft, DFG) Priority Programme SPP1530 to BK; BARLEX grant no. 0314000 to NS; and DFG grant no. SCHN 768/3-1 and the German Federal Ministry of Education and Research GABI-FUTURE Start Program grant no. 0315071 to TS.
Footnotes
Communicating editor: J. Borevitz
- Received November 10, 2013.
- Accepted January 9, 2014.
- Copyright © 2014 by the Genetics Society of America