Title: Association analysis reveals genes underlying natural variation in circadian rhythms in Arabidopsis

Authors: Hannah Rees^1,2, Ryan Joynson¹, James K.M. Brown³, Anthony Hall¹

Corresponding author: Anthony Hall anthony.hall@earlham.ac.uk

Contact information: ¹Earlham Institute, Norwich Research Park, Norwich, NR4 7UG, UK; ²Institute of Integrative Biology, University of Liverpool, Crown Street, Liverpool, L69 7ZB, UK; ³John Innes Centre, Norwich Research Park, Norwich, NR4 7UH, UK

Funding: This project was supported by the BBSRC via the Earlham institute CSP (BB/P016774/1, AH, HR) and BBSRC Design Future Wheat (BB/P016855/1, AH, RJ).

Abstract: Circadian clocks have evolved to resonate with external day and night cycles. However, these entrainment signals are not consistent everywhere and vary with latitude, climate and seasonality. This leads to divergent selection for clocks which are locally adapted. To investigate the genetic basis for this circadian variation, we used a Delayed Fluorescence (DF) imaging assay to screen 191 naturally occurring Swedish Arabidopsis accessions for their circadian phenotypes. We demonstrate that period length co-varies with both geography and population sub-structure. Several candidate loci linked to period, phase and Relative Amplitude Error (RAE) were revealed by genome-wide association mapping and candidate genes were investigated using TDNA mutants. We show that natural variation in a single non-synonymous substitution within COR28 is associated with a long-period and late-flowering phenotype similar to that seen in TDNA knock-out mutants. COR28 is a known coordinator of flowering time, freezing tolerance and the circadian clock; all of which may form selective pressure gradients across Sweden. We demonstrate the effect of the COR28 -58S SNP in increasing period length through a co-segregation analysis. Finally, we show that period phenotypic tails remain diverged under lower temperatures and follow a distinctive ‘arrow-shaped’ trend indicative of selection for a cold-biased temperature compensation response.

Keywords: Circadian clock, temperature compensation, Arabidopsis , natural variation, Sweden, 1001 genomes project, GWA-mapping,COR28 , ELF3

Acknowledgements: We thank Paul W. Goedhart (Wageningen University) for advice about circular phase statistics using the RCIRCULAR procedure and Grant Calder (John Innes Centre) for his assistance developing the 96-well ROI selection using FIJI. We thank Susan Duncan for help and advice with the flowering time investigation.

We also thank the Dean Research group (JIC) for providing the 191 Swedish accessions and the Liu Research group (Shanghai Institutes for Biological Sciences) for providing the cor28-2 , cor27-1and cor27-1/28-2 mutants. We also thank Ewan Holmes and Georgia Scutter for help with seed harvesting.

Main text:

Introduction

Plants are highly adapted to survive and exploit the daily fluctuations in light and temperature experienced as the earth spins on its axis. The circadian clock plays an intrinsic role in this; integrating temporal cues from the environment to inform photosynthetic, metabolic and developmental processes (Covington et al., 2008; Harmer et al., 2000). A robustly oscillating circadian clock which is highly synchronised to external day-length contributes to the overall fitness of the plant, giving it an edge over competitors, predators and pathogens (Dodd et al., 2005; Green et al., 2002; Ingle et al., 2015; Michael et al., 2003). Circadian rhythms can be quantified by their period (the time taken to complete one cycle), their phase (the time of day in which the rhythm peaks), their amplitude (the change in intensity from a baseline) and their Relative Amplitude Error (RAE) which is the amplitude error divided by the overall amplitude of the rhythm and can be equated to rhythm robustness. In our study, circadian phenotyping was achieved using delayed fluorescence (DF) imaging. Delayed fluorescence levels are circadian regulated and reflect the changes in the photosynthetic state of photosystem II (Gould et al., 2009). DF imaging has been validated as a reliable and flexible tool to measure circadian rhythms in a range of plant models but has not previously been used for phenotyping of a large population of individuals on the scale required for genome-wide association mapping.

Arabidopsis thaliana is a model plant system which has been extensively studied in circadian biology. The Arabidopsis clock is comprised of a series of interlocking negative transcriptional feedback loops connected by key activators that control the oscillation of clock gene expression(Hernando et al., 2017). Key genes within this network include CIRCADIAN CLOCK ASSOCIATED 1 (CCA1) andLATE ELONGATED HYPOCOTYL (LHY) which transcriptionally repressTIMING OF CAB EXPRESSION (TOC1) (Alabadi et al., 2001). Downstream genes interact with the clock to communicate circadian rhythmicity to physiological outputs. Examples include photoperiodic regulation of flowering time and the temporal gating of cold acclimation responses (Kinmonth-Schultz et al., 2013). Flowering under long days is instigated through accumulation of the floral promoter CONSTANS (CO) controlled by the circadian clock component GIGANTEA (Suárez-López et al., 2001). In winter-annuals, flowering is also dependent on the vernalization response of FLOWERING LOCUS C (FLC) andFRIGIDA (FRI) (Bastow et al., 2004; Johanson et al., 2000). Cold tolerance is diurnally activated through alternative splicing of CCA1 which co-regulates the expression of COLD REGULATED (COR) genes alongside light and temperature stress pathways (James et al., 2012; Kinmonth-Schultz et al., 2013).

The endogenous core circadian network is entrained by external stimuli; most notably light and temperature. Day-length, light composition and light intensity have all been shown to affect circadian rhythms (Aschoff, 1979; Más et al., 2003; Pittendrigh & Minis, 1964; Yanovsky & Kay, 2002). Temperature also has a well-documented effect on entrainment of circadian rhythms (Edwards et al., 2005; Gould et al., 2006; Salome & McClung, 2005) with a degree of period shortening expected under higher temperatures. Circadian rhythms are said to be ‘temperature compensated’; they resist large changes in period-length in response to temperature (Pittendrigh & Harvey, 1954). The extent of temperature compensation has been shown to vary betweenArabidopsis accessions (Gould et al., 2006; Kusakina et al., 2014). Rhythm robustness is also strongly affected by temperature, although the temperature which produces the most rhythmic oscillations appears to be dependent on the species and circadian assay used (Kusakina et al., 2014; Rees et al., 2019). Plants with clocks which resonate with environmental conditions are typically fittest, however it has been suggested that in climates with large seasonality there may be a compromise for clocks which are adaptable to rapidly changing day-lengths (Michael et al., 2003).

Natural populations of several important model organisms that exhibit extensive diversity in their circadian behavior have been documented. Pupal eclosion rhythms of Drosophila are latitude dependent; with shorter rhythms, earlier phases and less robust rhythms observed in Northern latitudes (Allemand & David, 1974; Lankinen, 1986). InArabidopsis , Michel et al. conducted a global study of leaf movement rhythms in 150 accessions and found that day-length of origin country correlated positively with period, but not phase or amplitude. They also identified several loci in the TOC1/PRR family which determined natural variation in period, phase and amplitude independently (Michael et al., 2003). Other investigations have shown allelic diversity of several clock related genes including FLC(Swarup et al., 1999) GI (de Montaigu et al., 2015) andEARLY FLOWERING 3 (ELF3) (Anwer et al., 2014) which contribute to natural circadian phenotypes without fully disrupting the clock mechanism. The positive relationship with period and latitude has also been observed in natural populations of Mimulus guttatus(Greenham et al., 2017) and in cultivated varieties of soybean and tomato (Greenham et al., 2017; Müller et al., 2016), potentially due to unintentional selection for circadian rhythms which function best at particular latitudes.

In this study, we focused on a collection of Arabidopsisaccessions from across a large latitudinal range in Sweden; a country with variations in climatic, anthropogenic and day-length factors all possibly influencing clock adaption. Northern latitudes around 63°N have permanent snow cover during winter months and the growing season is much cooler and shorter than in the South. At the solstice, there is almost 3 hours difference in day-length between the North and South. This divergence of climate has led to selection for ecotypes with adapted growth and flowering strategies (Bloomer & Dean, 2017; Shindo et al., 2005). Analysis of the global population structure of Arabidopsisaccessions has previously identified Swedish accessions as being genetically distinct from the wider population, with further differentiation within the country between North and South (Horton et al., 2012; Huber et al., 2014; Long et al., 2013; Nordborg et al., 2005). Accessions from South Sweden have high genetic diversity within a relatively small area, perhaps suggesting a historic emigration from central Europe following glacial retreat (Alonso-Blanco et al., 2016). Northern Swedish accessions have lower genetic diversity but larger genome sizes (Long et al., 2013) and also carry a surprising reservoir of drought tolerance genes (Exposito-Alonso et al., 2018). Completion of the 1001 genomes project in 2006 has facilitated a recent expansion inArabidopsis genome wide association (GWA) studies made possible through the provision of high-quality re-sequenced genotype data. These accessions are publicly available, geo-referenced and genetically inbred making it easy for researchers to perform experiments over several generations under a variety of conditions (Weigel & Mott, 2009). Many of the accessions in this study have also been characterized for other phenotypic traits such as seed dormancy (Kerdaffrec et al., 2016), flowering time (Sasaki et al., 2015), and freezing tolerance (Horton et al., 2016).

The objectives for this research were firstly; to investigate whether circadian diversity exists in Swedish Arabidopsis and to understand to what extent this variation can be explained by either latitude or the underlying population structure. We next wanted to identify and validate genetic variation in candidate genes which might explain the observed circadian variation. To our knowledge, this work provides the first example of a GWA study using circadian phenotypes in plants. Our final objective was to measure circadian rhythms under 10°C and 16°C to test whether divergence of circadian traits persists across a range of temperatures.

Methods