Wheat EARLY FLOWERING 3 affects heading date without disrupting circadian oscillations

Abstract

Plant breeders have indirectly selected for variation at circadian-associated loci in many of the world's major crops, when breeding to increase yield and improve crop performance. Using an eight-parent Multiparent Advanced Generation Inter-Cross (MAGIC) population, we investigated how variation in circadian clock-associated genes contributes to the regulation of heading date in UK and European winter wheat (Triticum aestivum) varieties. We identified homoeologues of EARLY FLOWERING 3 (ELF3) as candidates for the Earliness per se (Eps) D1 and B1 loci under field conditions. We then confirmed a single-nucleotide polymorphism within the coding region of TaELF3-B1 as a candidate polymorphism underlying the Eps-B1 locus. We found that a reported deletion at the Eps-D1 locus encompassing TaELF3-D1 is, instead, an allele that lies within an introgression region containing an inversion relative to the Chinese Spring D genome. Using Triticum turgidum cv. Kronos carrying loss-of-function alleles of TtELF3, we showed that ELF3 regulates heading, with loss of a single ELF3 homoeologue sufficient to alter heading date. These studies demonstrated that ELF3 forms part of the circadian oscillator; however, the loss of all homoeologues was required to affect circadian rhythms. Similarly, loss of functional LUX ARRHYTHMO (LUX) in T. aestivum, an orthologue of a protein partner of Arabidopsis (Arabidopsis thaliana) ELF3, severely disrupted circadian rhythms. ELF3 and LUX transcripts are not co-expressed at dusk, suggesting that the structure of the wheat circadian oscillator might differ from that of Arabidopsis. Our demonstration that alterations to ELF3 homoeologues can affect heading date separately from effects on the circadian oscillator suggests a role for ELF3 in cereal photoperiodic responses that could be selected for without pleiotropic deleterious alterations to circadian rhythms.

Figure 1

Variation in time to reach growth stage 55 across a United Kingdom wheat MAGIC population. Time in days for individual MAGIC recombinant inbred lines (RILs) to reach growth stage 55 (GS55) for the (A) 2012/2013 growing season and (B) 2013/2014 growing season. The number of days after sowing to reach GS55 for the eight MAGIC parents is indicated.

Figure 2

EARLY FLOWERING 3 is a candidate gene for Earliness per se-B1 (Eps-B1). (A) Syntenic relationships of the Eps-B1 QTL to the Eps-D1 quantitative trait locus (QTL) and the T. monococcum Eps-A^m1 QTL. ELF3 highlighted in green. (B) TaELF3-B1 gene model with exons represented by filled rectangles, untranslated regions (UTRs) as white rectangles. The two promoter cytosine repeat polymorphisms with a location relative to the transcription start site (TSS) are indicated with the cytosine (C) repeat number for Claire, Robigus, Cadenza, and Paragon. The Ser674Gly single-nucleotide polymorphism (SNP) is situated within exon 4 of ELF3-B1.

Figure 3

A subtelomeric chromosomal introgression and inversion containing EARLY FLOWERING 3-D1 is likely to be the causal polymorphism underlying the Earliness per se-D1 quantitative trait loci. (A) Syntenic relationships between the distal end of Chinese Spring (CS) 1D chromosome (IWGSCv1.1) and Jagger (Jag) 1D (PGSBv2.0), TaELF3-D1, labelled "ELF3" and highlighted in green, whereas as each unlabelled black arrow corresponds to a gene in the forward (pointing right) or reverse (pointing left) strand. (B) the Cadenza/Jagger ELF3-D1 contains a deletion within intron 2 which is not present in CS, F, and R, referring to relative primer binding sites for PCR amplification shown in (C). (D) Agarose gel separation of polymerase chain reaction (PCR) products from (C). A PCR product of 776 base pairs (bp) is indicative of the presence of a subtelomeric chromosomal introgression and inversion.

Figure 4

Circadian rhythms of delayed chlorophyll fluorescence in the MAGIC parents. (A) Mean delayed fluorescence (DF) luminescence in counts per second (cps) normalized to −1 to 1 using Biodare2 with error bars representing standard deviation (n = 27–33). Circadian amplitude (B), period (C), and relative amplitude error (RAE) (D) calculated using FFT-NLLS (Biodare2). Upper and lower hinges represent the first and third quartiles (25th and 75th percentiles), the middle hinge represents the median value, whiskers represent the third quartile + 1.5*interquartile range (IQR) and the first quartile – 1.5*IQR, red triangle represents the mean value, and black dots represent individual replicates. LL is constant light.

Figure 5

Mutation of single ELF3 homoeologues affects heading date but not circadian rhythms of chlorophyll a fluorescence and Ttelf3-null is arrhythmic in continuous light. (A) GS55 for Ttelf3-null, Ttelf3-Anull, Ttelf3-Bnull, and TtELF3-WT defined as the days after sowing (DAS) at which half of the first ear has emerged past the ligule. Upper and lower hinges represent the first and third quartiles (25th and 75th percentiles), the middle hinge represents the median value, whiskers represent the third quartile + 1.5*interquartile range (IQR) and the first quartile – 1.5*IQR, triangle represents the mean value, and black dots represent individual replicates (n = 5). (B) Mean F_m' of lines shown in (A) in LL with SEM bars (n = 24). (C) relative amplitude error (RAE) where points below the dashed line are considered rhythmic (D) circadian period length (hours) and (E) amplitude for genotypes inplot (B). Box jitter plots as described above, note plotted points in (C-E) correspond to those where data were successfully fitted by the FFT-NLLS model in Biodare2. Significant differences (P < 0.05) calculated in R using the Kruskal–Wallis test followed by post-hoc Dunn's test.

Figure 6

Abundance of wheat circadian clock transcripts in the TtELF3-WT and Ttelf3-null lines in light-dark cycles and constant light. (A–G) Mean abundance of circadian oscillator transcripts (n = 3–5) in TtELF3-WT (circles) and Ttelf3-null (triangles) in a 24 h light and dark (LD) cycle in long day conditions (16 h light at 250 µmol m⁻² s⁻¹, 20°C: 8 h dark 16°C, represented by a horizontal grey bar) followed by constant light (LL) and temperature (20°C) from time 24 to 96 h. Transcript abundance (ΔΔCq) is relative to RP15 and RPT5A, (A) TtLHY, (B) Ppd-1, (C) TtPRR73, (D) TtTOC1, (E) TtGI, (F) TtLUX, and (G) TtELF3. The white bar represents light, the black bar represents darkness, and the grey bar represents subjective night.

Figure 7

Functional LUX is required for the maintenance of robust circadian oscillations under constant light in hexaploid wheat. (A) Mean F_m' for Chogokuwase, Geurumil, and Minaminokomugi in constant light (LL) (n = 16). (B) Mean period and (C) relative amplitude error (RAE) for (A) calculated using FFT-NLLS (Biodare2). Upper and lower hinges represent the first and third quartiles (25th and 75th percentiles), the middle hinge represents the median value, whiskers represent the third quartile + 1.5*interquartile range (IQR) and the first quartile – 1.5*IQR, triangle represents the mean value and dots represent individual replicates. (D) Mean Chogokuwase, Geurumil, and Minaminokomugi leaf temperature relative to the background in LL (n = 4). (E) mean period and (F) RAE for plot (D). Error bars are SEM, for (D) every 20th point plotted for clarity.

Figure 8

ELF3 functions in the circadian oscillator and regulates the heading of wheat. Genes associated with the circadian oscillator are positioned and shaded relative to the time of day at which their transcript abundance is highest in the diel cycle. The progression of the circadian oscillator is indicated by the circle. Lines indicate potential regulatory links with outputs and other pathways. Pointed arrow head represent activation and blunt arrowhead represents repression.