Genetic Architecture of Flowering-Time Variation inBrachypodium distachyon1[OPEN]

Daniel P. Woods, Ryland Bednarek, Frédéric Bouché, Sean P. Gordon, John P. Vogel, David F. Garvin,and Richard M. Amasino

Laboratory of Genetics, University of Wisconsin-Madison, Madison, Wisconsin 53706 (D.P.W., R.M.A.); UnitedStates Department of Energy Great Lakes Bioenergy Research Center, University of Wisconsin-Madison,Madison, Wisconsin 53706 (D.P.W., R.M.A.); Department of Biochemistry, University of Wisconsin-Madison,Madison, Wisconsin 53706 (D.P.W., R.B., F.B., R.M.A.); United States Department of Energy Joint GenomeInstitute, Walnut Creek, California 94598 (S.P.G., J.P.V.); and USDA-ARS Plant Science Research Unit,University of Minnesota, Department of Agronomy and Plant Genetics, St. Paul, Minnesota 55108 (D.F.G.)

ORCID IDs: 0000-0002-1498-5707 (D.P.W.); 0000-0002-8017-0071 (F.B.); 0000-0003-3431-5804 (S.P.G.); 0000-0003-1786-2689 (J.P.V.);0000-0003-3068-5402 (R.M.A.).

The transition to reproductive development is a crucial step in the plant life cycle, and the timing of this transition is an importantfactor in crop yields. Here, we report new insights into the genetic control of natural variation in flowering time in Brachypodiumdistachyon, a nondomesticated pooid grass closely related to cereals such as wheat (Triticum spp.) and barley (Hordeum vulgare L.). Arecombinant inbred line population derived from a cross between the rapid-flowering accession Bd21 and the delayed-floweringaccession Bd1-1 were grown in a variety of environmental conditions to enable exploration of the genetic architecture of floweringtime. A genotyping-by-sequencing approach was used to develop SNP markers for genetic map construction, and quantitative traitloci (QTLs) that control differences in flowering time were identified. Many of the flowering-time QTLs are detected across a rangeof photoperiod and vernalization conditions, suggesting that the genetic control of flowering within this population is robust.The two major QTLs identified in undomesticated B. distachyon colocalize with VERNALIZATION1/PHYTOCHROME C andVERNALIZATION2, loci identified as flowering regulators in the domesticated crops wheat and barley. This suggests thatvariation in flowering time is controlled in part by a set of genes broadly conserved within pooid grasses.

Proper timing of flowering is a major developmentaldecision in the life history of plants, and the geneticmanipulation of flowering time has played a crucialrole in the domestication and spread of cereal cropssuch as wheat (Triticum spp.), barley (Hordeum vulgareL.), rice (Oryza sativa), and maize (Zea mays; Greenupet al., 2009; Hung et al., 2012). Moreover, the modula-tion of flowering time has been important in the di-versification of temperate (pooid) grasses into higherlatitudes with colder winters (Woods et al., 2016;Fjellheim et al., 2014). An important environmental cuethat often affects flowering is day (d) length (photope-riod; Song et al., 2015). Many plants adapted to tem-perate regions flower in response to increasing daylengths (long-d plants), in contrast to many plants fromthe tropics that flower as day length decreases (short-dplants). In addition, some plants adapted to temperateclimates have taken on a biennial/winter annual lifehistory strategy in which plants become established inthe fall, then overwinter and flower rapidly in thespring as day lengths increase (Amasino 2010). Essen-tial to this strategy is the prevention of flowering beforewinter because cold temperatures could damage deli-cate floral structures, preventing reproduction. Thus,plants have evolved ways to repress flowering in thefall and alleviate this repression by sensing the pass-ing of winter to establish competence to flower. This

process, bywhich the block to flowering is alleviated byexposure to prolonged time in cold temperatures, isknown as vernalization (Chouard 1960).

Many varieties of wheat, barley, oats (Avena sativa),and rye (Secale cereale) require vernalization to flower.Winter annual cereal varieties require vernalization toflower, whereas varieties that can flower in the absenceof vernalization are referred to as “spring annuals”.Studying the allelic variation that exists between springand winter varieties has led to the identification ofgenes involved in the pooid grass vernalization regu-latory pathway. This molecular model of vernalizationresponsiveness involves a leaf-specific regulatory loop in-volvingVERNALIZATION1 (VRN1),VERNALIZATION2(VRN2), and VERNALIZATION3 (VRN3; Greenupet al., 2009; Distelfeld and Dubcovsky 2010). VRN3is homologous to Arabidopsis (Arabidopsis thaliana)FLOWERING LOCUS T (FT), a small mobile proteinknown as “florigen”, that moves from leaves to theshoot apical meristem to promote flowering (Corbesieret al., 2007; Zeevaart 2008). During the growth ofwinter-annual cereals in the fall, the CONSTANS-likegene VRN2 represses FT to prevent flowering, and theFRUITFULL-like MADS box transcription factor VRN1is transcribed at very low levels (Yan et al., 2004;Greenup et al., 2009). During winter, VRN1 is induced,causing the repression of VRN2 and the consequent

Plant Physiology�, January 2017, Vol. 173, pp. 269–279, www.plantphysiol.org � 2017 American Society of Plant Biologists. All Rights Reserved. 269 www.plantphysiol.orgon September 21, 2020 - Published by Downloaded from

Copyright © 2017 American Society of Plant Biologists. All rights reserved.

derepression of FT (Distelfeld and Dubcovsky 2010;Yan et al., 2003). In addition, FT requires long d to be-come activated by the pseudo-response regulator,PHOTOPERIOD1 (PPD1), through yet unknown mech-anisms; thus, flowering only occurs after winter duringthe lengthening days of spring and early summer(Turner et al., 2005; Shaw et al., 2013).

The above model is primarily based on the study ofepistatic relationships among VRN1, VRN2, PPD1, andFT in domesticated varieties of wheat and barley(Trevaskis et al., 2003; Yan et al., 2003, 2004, 2006;Dubcovsky et al., 2005; Karsai et al., 2005; Turner et al.,2005). Some varieties of spring barley and spring wheatthat do not require vernalization either carry deletionsof the VRN2 locus or point mutations within its con-served CCT domain (Yan et al., 2004; Dubcovsky et al.,2005; Karsai et al., 2005; von Zitzewitz et al., 2005).Therefore, the activity of VRN2 is necessary for a vernal-ization requirement thatwas recently proven in hexaploidwheat (Kippes et al., 2016). Other spring varieties havedominant alleles of VRN1 or FT that are constitutivelyactivated and epistatic to functional VRN2 alleles (Yanet al., 2003, 2004, 2006; Fu et al., 2005; Loukoianov et al.,2005; von Zitzewitz et al., 2005). In barley, allelic variationat the PPD1 locus results in two types of spring varietiesthat are either sensitive to photoperiod and early flow-ering (PPD-H1), or insensitive to photoperiod and laterflowering (ppd-h1; Turner et al., 2005).

Due to its small, fully sequenced diploid genome(IBI, 2010), its small physical stature, and its high re-combination rate (Brkljacic et al., 2011; Brutnell et al.,

2015), B. distachyon is an attractive model plant forstudying a number of different traits including flow-ering. Like wheat and barley, B. distachyon accessionsexhibit a considerable amount of natural variation inflowering responses (Ream et al., 2014; Tyler et al., 2014).However, unlike wheat and barley, which fall into twobroad categories for flowering requirements (winterand spring varieties), most of the B. distachyon acces-sions studied to date are likely to be some form of winterannual in their native environments because they allrequire vernalization to flower rapidly when grownin native photoperiods in growth chambers: under arti-ficial 20-h-long d, accessions such as Bd21 will flowerquite rapidly without vernalization (Vogel et al., 2006;Ream et al., 2014); however, when grown under 14 h d to15 h d, Bd21 requires a very short (2 to 3weeks) period ofvernalization to flower rapidly (Ream et al., 2014). Incontrast, many B. distachyon accessions such as Bd1-1 aredelayed in flowering even under artificially long d andrequire an extended period of cold (at least 6 weeks) tosaturate their vernalization requirement (Ream et al.,2014). This considerable natural variation in floweringtime can be used to explore the genetic architecture offlowering in an undomesticated pooid grass and con-tribute insights into the evolution of the vernalizationresponse within pooids. Furthermore, unlike wheatand barley, little is known about the molecular basis ofnatural variation in flowering-time responses in otherpooid grasses including B. distachyon.

In this study, we developed a recombinant inbredline (RIL) population from a cross between Bd21 (rapid-flowering accession) and Bd1-1 (delayed-flowering ac-cession) to explore the genetic architecture of floweringtime in B. distachyon. We observed significant variationin flowering behavior among the 142 RILs in responseto various environmental conditions. We used geno-type by sequencing (GBS) to create a genetic map for theRIL population, and then conducted a quantitative traitlocus (QTL) analysis to determine the genetic architec-ture of flowering regulation in this population. Weidentified six significant QTLs, several of which werepresent in multiple environments tested. Interestingly,VRN1 and VRN2 underlie two of the six QTLs. We alsoidentified QTL in which no flowering-time candidategenes are present, and thus represent novel loci regu-lating flowering time. The development and genotypingof this RIL populationmay prove useful in the dissectionof other traits in B. distachyon.

RESULTS

Development of a Recombinant Inbred Line Populationfrom a Cross between Two Diverse B. distachyonAccessions that Have Different Flowering Behaviors

The accessions Bd21 and Bd1-1 differ considerably inflowering time and requirement for vernalization. Wepreviously characterized Bd21 as “extremely rapidflowering” because it flowers in less than 40 d aftergermination in both 16-h d and 20-h day lengths

1 R.M.A.’s laboratory was funded by the National Science Foun-dation under grant no. IOS-1258126, and the Great Lakes BioenergyResearch Center (Department of Energy Biological and Environmen-tal Research Office of Science grant no. DE-FCO2-07ER64494); D.P.W.was funded in part by a National Institutes of Health-sponsored pre-doctoral training fellowship to the University of Wisconsin GeneticsTraining Program; F.B. thanks the Belgian American EducationalFoundation (BAEF) for their post-doctoral fellowship; J.P.V. andS.P.G. were funded by the U.S. Department of Energy Joint GenomeInstitute (a Department of Energy Office of Science User Facility),which is supported under contract no. DE-AC02-05CH11231, withadditional funding provided by Office of Biological and Environmen-tal Research, Office of Science, U.S. Department of Energy, underinteragency agreement no. DE-SC0006999; and D.F.G. was supportedby USDA-ARS CRIS project no. 5062-21000-030-00D.

* Address correspondence to amasino@biochem.wisc.edu.The author responsible for distribution of materials integral to the

findings presented in this article in accordance with the policy de-scribed in the Instructions for Authors (www.plantphysiol.org) is:Richard Amasino (richard.amasino@wisc.edu).

D.P.W., D.F.G., and R.M.A. conceived and designed researchplans; D.F.G. developed the RIL population and conducted prelimi-nary phenotyping on earlier generations during RIL development;D.P.W. and R.B. phenotyped population and prepared samples forsequencing; S.P.G. and J.P.V. developed the genetic map; F.B. per-formed QTL analysis and prepared all figures with input from D.P.W.,R.B., and R.M.A.; D.P.W., R.B., and R.M.A. wrote the article withcontributions and approval of all authors.

[OPEN] Articles can be viewed without a subscription.www.plantphysiol.org/cgi/doi/10.1104/pp.16.01178

270 Plant Physiol. Vol. 173, 2017

Woods et al.

www.plantphysiol.orgon September 21, 2020 - Published by Downloaded from Copyright © 2017 American Society of Plant Biologists. All rights reserved.

without vernalization (Ream et al., 2014; Fig. 1, A–D). Incontrast, Bd1-1 does not flower rapidly without vernal-ization (greater than 120 d to flower in 20-h and 16-hphotoperiod) and requires 6 weeks of cold to saturateits vernalization requirement (Ream et al., 2014; Fig. 1,A–D). A cross between these two phenotypically diverseaccessions was used to develop a RIL population (seeMaterial and Methods for details of RIL development).To develop a genetic map of the Bd21 X Bd1-1 RIL

population, we utilized reduced representation geno-type by sequencing (Elshire et al., 2011). After severalfiltering steps, 1693markers covering the entire genomewere selected (Fig. 2A). The analysis of the recombi-nation frequency betweenmarkers identified fivemajorlinkage groups corresponding to the five chromosomesof B. distachyon, and confirmed the absence of large-scale rearrangements between the genomes of Bd21and Bd1-1 (Fig. 2B). The final genetic map consists of1693 SNP markers and a cumulative size of 1456.4 cM(Fig. 2C, Supplemental Table S1), which is similar topreviously characterized RIL populations of Bd3-1 andBd21 (Cui et al., 2012; Des Marais et al., 2016), andconfirms the high recombination rate of B. distachyoncompared with other grass species. The observed het-erozygosity of selected markers (1.5%) matches ex-pected frequencies for an F7 population (1.7%).

Variation in Flowering Time in the Bd21 X Bd1-1 RIL Population

We characterized the F7 RILs using various vernali-zation and photoperiod treatments. Specifically, wegrew the RIL population in 16-h and 20-h photoperiodsafter 0, 2, 3, and 6 weeks of vernalization and scoreddays to heading and the number of leaves on the pri-mary culm at flowering (Supplemental Figs. S1 andS2; 6-week vernalization data not shown because noflowering-time variation in the population was ob-served). Previous studies in several species have founda strong positive correlation between days to headingand the number of leaves at flowering (leaf numberprovides a developmental assessment; Salomé et al.,2011; Ream et al., 2014), indicating that these two traitsare highly correlated in natural accessions. Under allconditions we observed a similar linear relationshipbetween days to heading and leaf number, indicatingthat later flowering plants were indeed developmen-tally delayed (Supplemental Fig. S3). We observedflowering variation within the RIL population under allconditions except after 6weeks of vernalization, inwhichall plants flowered rapidly as expected (Fig. 3; 6-weekvernalization data not shown). The greatest range inflowering timeswas observed in 20-h nonvernalized and

Figure 1. Vernalization time course in B. distachyon accessions Bd21 and Bd1-1. A and C, Photographs of representative plantstaken after 60 d of growth. Imbibed seeds of Bd21 and Bd1-1 were cold treated at 5°C in soil in an 8-h photoperiod for the in-dicated length of time (weeks), followed by outgrowth in a growth chamber set to 16-h light/8-h dark (A) or 20-h light/4-h dark (C).B and D, Flowering time measured as the number of days to spike emergence from the end of cold treatment. Arrows indicatetreatments where plants did not flower within 120 d of the experiment.

Plant Physiol. Vol. 173, 2017 271

Flowering-Time QTL in B. distachyon

www.plantphysiol.orgon September 21, 2020 - Published by Downloaded from Copyright © 2017 American Society of Plant Biologists. All rights reserved.

16-h 2-week vernalized conditions, in which some linesflowered as early as 15 d while others failed to flower bythe endof the 120 d experiment (Fig. 3; Supplemental Fig.S1). Additionally, considerable flowering-time variationwas observed in the RIL population under 20-h 2-weekvernalization conditions, in which flowering occurred asearly as 17 d in some lines and as late as 109 d in otherlines (Fig. 3). The majority of nonvernalized lines grownunder 16 h of light flowered between 60 and 120 d,whereas after 3weeks of vernalization themajority of thepopulations flowered between 20 d and 30 d with thelatest lines flowering around d 50 (Fig. 3).

QTL Mapping of Flowering Time in B. distachyon

To identify regions of the B. distachyon genome con-tributing to the observed flowering variation in the RILpopulation, we performed a QTL analysis on the pop-ulation for those photoperiod and vernalization treat-ments resulting in phenotypic variation for flowering.We detected two major flowering QTLs that correlatewith flowering-time differences in multiple environ-mental conditions (Fig. 4; Supplemental Table S2).Flowering-time related traits such as days to headingand leaf number were highly correlated and confi-dence intervals of QTL peaks for these traits usuallyoverlapped within the various environmental condi-tions (Fig. 4). The two major QTLs were found onchromosome 1 (QTL1) and chromosome 3 (QTL2) andwere robustly observed across several different envi-ronments. QTL1was detected under 20 h nonvernalized,20-h 2-week vernalized, and 16-h 3-week vernalizedconditions. Also, QTL1 was close to reaching thecomputed significance threshold under 16-h 2-weekvernalized conditions. Depending upon the condition,QTL1 (peak marker Bd1_6006000) explains between1.8% and 20.5% of the phenotypic variance observedin this mapping population (Supplemental Table S2).Although several genes are within the QTL interval

(46 cM to 61cM in 20 h nonvernalized conditions), thetightly linked flowering-time genes VRN1 and PHYCare likely candidates because previous reverse andforward genetic studies have shown that both genesplay important roles in flowering-time regulationin B. distachyon as well as in wheat (Woods et al.,2014b, 2016; Chen and Dubcovsky 2012; Chen et al.,2014). The major QTL on chromosome 3 (QTL2) wasdetected only under 16-h nonvernalized and 20-h non-vernalized conditions with peak LOD scores under16-h (LOD 5.42) and 20-h (LOD 5.23; Fig. 4). QTL2(peak marker Bd3_8505000) explains between 6.5%and 20.1% of the phenotypic variance observed withinthe mapping population and, like QTL1, its mappinginterval (60cM-78cM in 16-h nonvernalized) spansseveral genes including the floral repressor VRN2(Woods et al., 2016), which is a likely candidate genefor the flowering-time differences (Fig. 4).

When data from the 20-h nonvernalized condi-tion were analyzed using a two-QTL model, we ob-served strong additive effects of QTL1 and QTL2(Supplemental Fig. S5). Bd21 alleles at QTL1 and QTL2were associated with rapid flowering, whereas RILswith the Bd1-1 alleles at both QTL1 and QTL2 weretypically the most delayed flowering lines within thepopulation (Fig. 5). Despite this general trend fordelayed flowering with the Bd1-1 alleles for QTL1 andQTL2, there were several RILs, which were still rapid-flowering (Fig. 5). Furthermore, if a particular RIL wasmixed for either the Bd21 or Bd1-1 genotype at QTL1and QTL2, this resulted on average in an intermediateflowering time of approximately 50 d, but with highvariability (Fig. 5). Thus, although the Bd21 genotype atQTL1 and QTL2 is associated with rapid flowering andthe Bd1-1 genotype at QTL1 and QTL2 is associatedwith delayed flowering, there are exceptions, suggest-ing the presence of additional loci that are likely tobe contributing to flowering-time variation betweenB. distachyon accessions.

Figure 2. Graphical representation of the estimated linkage map for the Bd21 X Bd1-1 RIL mapping population. A, Physicalposition of selected markers along B. distachyon chromosomes. B, Pairwise recombination fraction (upper-left triangle) and LODscores for all pairs of markers ordered according to their position on the genome shown in (A). Yellow indicates linked markers;dark blue indicate unlinked markers. Axes show marker numbers. C, Genetic map of selected markers. Distances are shown incentimorgans. Mb, megabase; cM, centimorgans.

272 Plant Physiol. Vol. 173, 2017

Woods et al.

www.plantphysiol.orgon September 21, 2020 - Published by Downloaded from Copyright © 2017 American Society of Plant Biologists. All rights reserved.

In addition to the two major QTLs described above,we also detected fourminorQTLs present in at least oneenvironmental condition (Fig. 4; Supplemental TableS2). QTL3 was detected under 20-h nonvernalized and20-h 2-week vernalized conditions. QTL3 explains 1.5%to 14.3% of the phenotypic variance observed in thismapping population. Interestingly QTL3 colocalizeswith FD, approximately 10 cM from the end of chro-mosome 3 (Fig. 4). FD is a basic Leu zipper domaintranscription factor that, in Arabidopsis, interacts withFT to turn on floral homeotic genes (Abe et al., 2005).Three other minor QTLs (QTL4, 5, and 6) are also pre-sent on chromosome 3 and contribute from 5.8% to12.8% of the phenotypic variance in vernalized popu-lations, and no previously identified flowering-time lociare within these QTL intervals (Fig. 4; SupplementalFig. S6; Supplemental Table S2).

Sequence Variants in VRN1, PHYC, VRN2, and FD

We explored the sequence variation around candi-date flowering-time genes VRN1, PHYC, VRN2, andFD, which colocalize to the most significant QTL peaks(QTL1, 2, and 3). We identified several sequence vari-ants in different alleles of these genes; however, we didnot find any obvious variants that might disrupt genefunction such as large deletions within the coding re-gion or nonsynonymous changes leading to a prema-ture stop codon, which have been found between theVRN1 and VRN2 genes of spring and winter annualvarieties of wheat and barley (Supplemental Fig. S7).We did, however, find several indels within both thepromoter and the first intron of VRN1 (SupplementalFig. S7). In wheat and barley, the first intron has beenshown to play an important regulatory role, and indels

within this intron have been associated with thespring annual habit (Yan et al., 2003; Fu et al., 2005;von Zitzewitz et al., 2005; Yan et al., 2004). Indeedprevious reports have shown that VRN1 mRNA levelsare higher in Bd21 than Bd1-1 under 20-h nonvernalizedconditions and after 3 weeks of vernalization, whichmay reflect the effect of the sequence variants detected(Ream et al., 2014).

DISCUSSION

In this studywe developed a RIL population betweena rapid (Bd21) and a delayed flowering (Bd1-1) acces-sion of B. distachyon (Ream et al., 2014; Vogel et al., 2006)and used it to evaluate the genetic architecture offlowering time under a range of environmental condi-tions. For this study, we developed a high-resolutiongenetic map containing 1693 SNP markers using geno-typing by sequencing. Flowering times of RILs rangedfrom as early as 20 d to greater than 120 d in some en-vironments. We found two major QTLs (QTL1 andQTL2) and four minor-effect QTLs (QTL3 to QTL6) thataccount for most of the observed flowering-time differ-ences. The two major QTLs coincide with the location ofthe genes VRN1 and PHYC on chromosome 1 (QTL1)and VRN2 on chromosome 3 (QTL2). These genes havepreviously been shown to play important roles inflowering-time regulation in B. distachyon, and contrib-ute to variation in flowering-time responses in wheatand barley (Woods et al., 2014b, 2016; Woods andAmasino 2015; Distelfeld et al., 2009). Thus, it appearsthat allelic variation in VRN1 and VRN2 likely contrib-utes tofloweringdifferences in an undomesticated pooidgrass. However, further fine mapping and functionalanalyses will be required to unequivocally establish the

Figure 3. Flowering-time distribution within the RIL population under five environmental treatments: 16-h long d nonvernalized(16-h LD NV), 20-h long d nonvernalized (20-h LD NV), 16- and 20-h long d after 2-week vernalization (16- and 20-h LD 2WVern), and 16-h long d after 3 weeks of vernalization (16-h LD 3W Vern). Days to heading (x axis) indicate the number of days tospike emergence once plants germinated. The number of lines within the RIL population that flowered within ranges of 10 d isindicated by the y axis. White arrows indicate the average days to heading for Bd21 plants (n = 12) and black arrows indicate theaverage days to for Bd1-1 plants (n = 12).

Plant Physiol. Vol. 173, 2017 273

Flowering-Time QTL in B. distachyon

www.plantphysiol.orgon September 21, 2020 - Published by Downloaded from Copyright © 2017 American Society of Plant Biologists. All rights reserved.

genes responsible for the QTLs identified. Additionalminor-effectQTLswe identified indicates that additionalloci contribute to the flowering-time difference betweenBd21 and Bd1-1.

The Genetic Architecture of Flowering Time inB. distachyon

Characterizing this RIL population enabled us toevaluate if flowering was controlled by many geneswith small effects, such as inmaize (Buckler et al., 2009),or by a few genes with large effects, such as in Arabi-dopsis (Salomé et al., 2011), and to determine if the locicontrolling flowering are the same under different daylengths and vernalization treatments. Under a givenenvironment, only two to three significant QTLs wereidentified, indicating that flowering-time variation

between Bd21 and Bd1-1 is controlled by only a fewgenes, similar to what has been shown in wheat, barley,and Arabidopsis (Distelfeld et al., 2009; Salomé et al.,2011). QTL1 was the only locus that was uniformlyidentified under both vernalized and nonvernalizedconditions and under both 16-h d and 20-h d. In con-trast, QTL2 was only identified under nonvernalizedconditions in both 16-h and 20 h conditions. The Bd21genotype at QTL1 and QTL2 is associated with rapidflowering RILs and the Bd1-1 allele is associated withdelayed flowering RILs. Although the Bd1-1 genotypeat QTL1 and QTL2 was strongly associated withdelayed flowering, there were several RILs containingthese QTLs that still flowered rapidly, suggesting ad-ditional genes contribute to promoting flowering, andthese RILs provide an entry point for molecularlyidentifying these additional loci.

Figure 4. Location of flowering-time QTLs under five different environmental conditions. QTLs based on days to heading areindicated by a solid line whereas the QTLs based on leaf count on the parent culm are indicated by a dotted line. The redhorizontal line represents the threshold of significance (see Materials and Methods). The phenotyping of the mapping populationwas repeated, which resulted in similar segregation of flowering-time phenotypes and QTL peaks (data not shown). The nameattributed to the different QTLs (QTL1 to QTL6) and the candidate flowering-time genes underlying each QTL (bold red line)are shown at the bottom of the diagram. Orange vertical lines indicate candidate flowering-time genes that are not correlatedwith QTLs.

274 Plant Physiol. Vol. 173, 2017

Woods et al.

www.plantphysiol.orgon September 21, 2020 - Published by Downloaded from Copyright © 2017 American Society of Plant Biologists. All rights reserved.

An additional study in this focus issue on floweringand reproduction also characterized flowering time in aRIL population from a cross between Bd21 and anotherdelayed flowering accession ABR6 (Bettgenhaeuseret al., 2016). They also found only a fewQTLswith largeeffect. One of their major QTL overlapswith QTL2 fromthis study, which colocalizes with VRN2, and anotherQTL overlaps with QTL4 from this study, which con-tains no candidate flowering-time genes. Interestingly,Bettgenhaeuser et al. (2016) did not identify a QTL thatoverlaps with QTL1, but they identified a major QTLthat colocalizes under FT. This indicates that VRN2 andthe unknown loci underlying QTL4 are robust acrossdifferent environmental conditions and may contributetoflowering-time variation broadlywithin Brachypodium.However, it also highlights that variation in differentgenes can also influence flowering-time variation indifferent populations that have unique genetic histo-ries and adapted to different environments.Genome-wide association studies (GWAS) are an-

other approach to decipher the genetic architecture oftraits (Weigel and Nordborg, 2015). Recently, twoflowering-time GWAS in B. distachyon identified nineand five associated peaks, none of which overlap withthe QTLs identified in this study (Tyler et al., 2016;Wilson et al., 2015). Thus, the QTL identified in our RILpopulation may represent rare alleles that do not sur-face in GWAS, or the GWAS that was conducted con-tains too few accessions or compounding populationstructures (issues that are common when conductingGWAS in in-breeding species; Weigel and Nordborg2015). For example, initial GWAS flowering studiesin Arabidopsis had difficulty identifying FLOWERING

LOCUS C and FRIGIDA, two genes responsible formuch of the flowering-time variation among accessionsof Arabidopsis. FRIGIDA and FLOWERING LOCUS Cwere later identified after additional well admixedArabidopsis populations were included (Saloméet al., 2011). The increase in the number of sequencedB. distachyon accessions will help enhance the resolutionof GWAS, thus improving the identification of flower-ing candidate genes.

Candidate Flowering-Time Genes Underlying the QTL

As described above, two previously identified ver-nalization genes, VRN1 and VRN2, colocalized toQTL1 and QTL2, respectively. However, no candidateflowering-time genes underlie QTL4 to QTL6, and thusthese QTLs are likely to be novel loci contributing toflowering-time variation in B. distachyon. There was nosequence variation within the coding region of VRN1between Bd21 and Bd1-1, suggesting that VRN1 isfunctional in both accessions. However, we did findseveral indels and SNPs within the first intron of VRN1(Supplemental Fig. S6). Studies of allelic variation ofVRN1 in wheat and barley have shown that indelswithin the first intron, influence VRN1 expression (Fuet al., 2005; Yan et al., 2006). Indeed, there are differ-ences in BdVRN1 expression patterns between Bd21and Bd1-1 that correlate with differences in floweringbehavior. For example, the length of cold required tosaturate the vernalization response is 2 to 3 weeks inBd21 but in Bd1-1 is 6 to 7 weeks (Ream et al., 2014).Correspondingly, after a 2- to 3-week cold exposure,BdVRN1 levels are higher in Bd21 than Bd1-1. In ad-dition, in the absence of vernalization, Bd21 flowersrapidly in 20-h day lengths but Bd1-1 is extremelydelayed. This is correlated with the higher levels ofBdVRN1 mRNA in nonvernalized Bd21 versus Bd1-1.These expression differences coincide with the QTL1peak identified after 3 weeks of cold and 20-h non-vernalized conditions, suggesting that the more rapidflowering of Bd21 after shorter vernalization treatmentsfollowed by 16-h day lengths and in 20-h day lengthswithout vernalization may be due to the elevated ex-pression of VRN1. VRN1 and FT form a positive feed-back loop to promote flowering in B. distachyon, wheat,and barley (Ream et al., 2014; Woods et al., 2016; Yanet al., 2006; Sasani et al., 2009; Shimada et al., 2009;Distelfeld and Dubcovsky 2010); however, we did notidentify a QTL peak around FT, so the elevated FT ex-pression may be due to the indirect effect of variation inVRN1 in this population.

PHYC is another candidate gene underlying QTL1,which might also influence flowering in this mappingpopulation. PHYC is an essential light receptor forphotoperiodic flowering in pooids (Woods et al., 2014b;Chen et al., 2014) and allelic variation of PHYC in barleyand pearl millet (Pennisetum glaucum) has been impli-cated in influencing flowering in these species (Nishidaet al., 2013; Pankin et al., 2014; Saïdou et al., 2014). Allelic

Figure 5. Phenotype by genotype influence on flowering for non-vernalized plants grown in 20-h-long d. Phenotype by genotype plot forthe two major loci (QTL1, VRN1/PHYC candidate and QTL2, VRN2candidate) influencing flowering time in the Bd21 X Bd1-1 RIL popu-lation grown in 20 h nonvernalized conditions in which both QTLs aresimultaneously present. Days to heading results indicate that presenceof the Bd21 genotype at both QTL1 and QTL2 results in rapid floweringwhereas presence of the Bd1-1 genotype at both loci results in delayedflowering. Difference in letters above box plots (a, b, a,b, c) indicatestatistical significance based on the mean days to heading values be-tween the various genotypes at QTL1 and QTL2 computed with anANOVA Tukey’s HSD test (P # 0.01).

Plant Physiol. Vol. 173, 2017 275

Flowering-Time QTL in B. distachyon

www.plantphysiol.orgon September 21, 2020 - Published by Downloaded from Copyright © 2017 American Society of Plant Biologists. All rights reserved.

variation at PHYC included a single nonsynonymousvariant as well as several SNPs within the promoter re-gion between Bd21 and Bd1-1, which might influencePHYC function. Mutations in PHYC result in a nonflow-ering phenotype in B. distachyon even after prolongedexposure to cold (Woods et al., 2014b); thus, the variantsare unlikely to cause a loss of PHYC function given thatboth Bd21 and Bd1-1 respond to vernalization.

As discussed above, VRN2, the candidate gene un-derlying QTL2, is a floral repressor that is conserved ingrasses (Woods et al., 2016). It was initially hypothe-sized that Bd21 has a nonfunctional VRN2 allele dueto the rapid flowering of Bd21 when grown under16-h d or 20-h d (Higgins et al., 2010; Schwartz et al.,2010). However, Bd21 and other “spring” or “rapid-flowering” accessions, such as Bd21-3 and Bd3-1, failto flower without vernalization when grown undershortermore “native” inductive photoperiods (e.g. 14-h d)without vernalization, revealing a facultative vernali-zation requirement; thus, all B. distachyon accessionsstudied to date may behave as winter annuals in nativeenvironments (Ream et al., 2014; Colton-Gagnon et al.,2014). Consistent with this winter-annual behavior,Bd21 has a functionalVRN2 ortholog (Ream et al., 2012;Woods et al., 2016). In support of BdVRN2 being arepressor of flowering in rapid flowering accessions,in Bd21-3 BdVRN2 RNAi knock-down lines aremore rapid-flowering and overexpression of BdVRN2delays flowering (Woods et al., 2016). Despite being afloral repressor, BdVRN2 mRNA levels in both Bd21and Bd1-1 do not decrease during vernalization as inwheat and barley, but in fact increase during the cold(Ream et al., 2014). Furthermore, BdVRN2 is not re-pressed by BdVRN1 as it is in wheat and barley (Woodset al., 2016). Thus, the role of VRN2 as a repressor offlowering that is necessary for a vernalization require-ment is conserved in pooids, but the cold-mediated re-pression of VRN2 by VRN1 most likely occurred afterB. distachyon diverged from core pooids comprised ofTritaceae and Poeae tribes (Woods et al., 2016).

These QTL results as well as the recent functionalstudies (described above) demonstrating BdVRN2 isindeed a repressor of flowering in B. distachyon sug-gest that BdVRN2 provides a basal repressive signalpreventing flowering in the absence of vernalization.After vernalization, this repression is overcome by thestrong flowering inductive signal provided by BdVRN1and the BdPHYC-mediated photoperiod pathway. Thismodel is consistent with our finding that the BdVRN2peak is only significant in the absence of vernalizationwhereas the BdVRN1/PHYC peak is significant underhighly inductive conditions (i.e. 20-h photoperiodwithout vernalization and 16-h photoperiod with priorshort vernalization). Hence, we hypothesize that anincrease in the signals from the photoperiod and ver-nalization pathways overcomes the BdVRN2-mediatedrepression of flowering.

As noted above, our QTL results indicate that allelicvariation in a region containing BdVRN2 influences thedifference in flowering time between Bd21 and Bd1-1.

Because BdVRN2 mRNA levels do not vary betweenrapid and delayed flowering accessions and the ex-pression before, during, and after cold is the same be-tween Bd21 and Bd1-1 (Ream et al., 2014), it is temptingto speculate that perhaps BdVRN2 is less biochemicallyactive in Bd21 versus Bd1-1 and thus, the repressionby BdVRN2 is easier to overcome by BdVRN1/PHYCleading to more rapid flowering in Bd21 relative toBd1-1 without vernalization. If this hypothesis is cor-rect, it is unlikely that variation within the promoterregion or intron of BdVRN2 contributes to flowering dif-ferences between Bd21 and Bd1-1. We did, however, findamino acid variation within the conserved CCT domainbetween Bd21 and Bd1-1. A point mutation within thisdomain in diploid wheat results in a spring-annual habit(Yan et al., 2004), and thus structural variants identified inthe CCT domain might also be causative for flowering-time differences between Bd21 and Bd1-1.

In conclusion, we developed a RIL population be-tween two genotypically and phenotypically diverseaccessions (Vogel et al., 2009; Gordon et al., 2014) andused this population to explore the genetic architectureof flowering time. We identified six QTLs that controlflowering within this population. Three of these QTLsare not associatedwith previously described flowering-time genes, and thus offer an opportunity to expand ourmolecular understanding of the control of flowering-time regulation in grasses. Two other QTLs colocalizewith well-described flowering-time genes, demonstrat-ing evolutionary conservation for some molecular as-pects offlowering-time regulation between domesticatedand undomesticated pooid grasses. The development ofthe RIL population together with its high-density SNPmap should also greatly help in deciphering the geneticcontrol of other plant traits that vary between Bd21 andBd1-1 such as lemma hair formation, cell wall composi-tion, height, and dormancy, to name but a few (Vogelet al., 2009; Cass et al., 2016; Ream et al., 2014; Barreroet al., 2012).

MATERIALS AND METHODS

Development of the Bd21 X Bd1-1 Recombinant InbredLine Population

A Brachypodium distachyon RIL population was generated from a cross be-tween inbred lines Bd21 (female) and Bd1-1 (male). A single F1 plant was self-pollinated and the resulting F2 seedswere propagated by single-seed descent tothe F6 generation. Individual F6 plants were then selfed to produce 142 F6:7 RILsfor use in genetic analysis and gene mapping (Fig. 3A). Several F7:8 plants perline were planted to bulk F8 seed.

Growth Conditions and Flowering-Time Phenotyping

Seeds imbibed distilled water overnight at 5°C before they were sowed.Individual plants were grown in MetroMix 360 (Sungrow) in square 6.5 cmplastic pots and fertilized every other week after one month of growth withPeters Excel 15-5-15 Cal-Mag and Peters 10-30-20 Blossom Booster (RJ Peters).Growth chamber temperatures averaged 22°C during the light period and 18°Cduring the dark period. Plants were grown under four T5 fluorescent bulbs(5000 K; Phillips), and light intensities averaged approximately 150 mmol m22 s–1

at plant level.

276 Plant Physiol. Vol. 173, 2017

Woods et al.

www.plantphysiol.orgon September 21, 2020 - Published by Downloaded from Copyright © 2017 American Society of Plant Biologists. All rights reserved.

Eight individuals of each of the 142 RILs and both parental lines were growninboth20and16hafter0, 2, 3, and6weeksofvernalizationas imbibedseed in soilat 5°C under 8-h day lengths. To minimize light intensity differences, individ-uals within a given RIL and different RILs were randomly assigned differentlocations throughout the growth chamber, and plants were rotated two timesper week throughout the growth chamber. (Note: with an imbibed seed, thephotoperiod does not influence the vernalization response (Ream et al., 2014)).None of the lines had germinated at the end of the vernalization treatment.Nonvernalized RILs imbibed for 2 d at 4°C, and were sown at the end of thevernalization treatment for nonvernalized and vernalized plants to be grown inparallel. Phenotyping of the RIL population in all conditions was repeated withsimilar results (data not shown).

Flowering time (days to heading) wasmeasured as the number of days fromemergence of the coleoptile to d 1 that emergence of the spike was detected(Zadoks scale = 50; Zadoks et al., 1974). The number of leaves derived from themain (parent) culm was recorded at the time of heading to provide a devel-opmental assessment. Most often the first culm to flower was derived from themain (parent) culm. Lines that did not flower by the end of the 120 d experimentwere scored as 120 d and 20 leaves, as all plants had greater than or equal to20 leaves. See Supplemental Table S4 for raw phenotypic data.

Development of a Genetic Map for the Bd21XBd1-1 RIL Population

DNA Extraction and Sequencing

Leaves from 12 F8 plants per RIL line were harvested and bulked to re-constitute the F7genotype. Specifically, a 2 cmportion of each leafwasharvestedand placed into 1.2 mL polycarbonate tubes (designed for multiple sampleprocessing) in liquid nitrogen. Samples were stored at 280°C before beingpulverized to a fine powder by adding a single metal bead and 2X CTAB-PVPextraction buffer (0.1 M Tris-HCl pH 8.0, 1.4 M NaCl, 0.02 M EDTA, polyvi-nylpyrrolidone 0.1%, CTAB 2%) to each tube followed by shaking in a beadmill for 3 min. Samples were then placed at 65°C for 1 h before conducting achloroform/isoamyl alcohol (24:1) extraction followed by DNA precipitationusing NaCl and 95% ethanol. DNA concentration was verified using theQuant-iT PicoGreen dsDNA Kit (Life Technologies).

Libraries were prepared as in Elshire et al. (2011) at the WI-MadisonBiotechnology Center with minimal modification. In short, 50 ng of DNAwas digested using the 5-bp cutter ApeKI (New England Biolabs) afterwhich bar-coded adapters amenable to Illumina sequencing were added byligation with T4 ligase (New England Biolabs). The 96 adapter-ligatedsamples were pooled, amplified to provide sufficient DNA (2 mM) for se-quencing, and adapter dimers were removed by SPRI bead purification.Quality and quantity of the finished libraries was assessed using the AgilentBioanalyzer High Sensitivity Chip (Agilent Technologies) and QubitdsDNA HS Assay Kit (Life Technologies), respectively. Cluster generationwas performed using HiSeq SR Cluster Kit v3 cBot kits (Illumina). Flowcellswere sequenced using single read, 100 bp sequencing and a HiSeq SBS Kit v3(50 Cycle; Illumina) on a HiSeq2000 sequencer. Images were analyzed usingthe standard Illumina Pipeline software (v1.8.2).

SNP Development and Genotyping

Deep Illumina resequencing data of the parental inbred lines Bd21and Bd1-1 was used to identify SNP variants (Gordon et al., 2014). Parentalreads were mapped to the Bd21 version 2 B. distachyon genome assembly(Goodstein et al., 2012) using BWA (v0.6.2, Li and Durbin 2010) and thegenotype and position of SNP markers were determined using SAMtools(v0.1.19; Li and Durbin 2010). Barcoded RIL data were demultiplexed usingbarcode sequences and ApeKI cut site using GBSX (v1.0.1, Herten et al., 2015)and partitioned for each line. Illumina data for each RIL individual wasqueried for 813,363 parental markers and a genotype was assigned whenassayed. To improve the accuracy of genotyping and summarize the geno-typing of markers with close physical position, consensus genotypes for 7 kbwindows tiling the genome were determined by calculating genotype ratiosin each window and assigning the consensus genotype according to majorityrule, requiring that the consensus genotype be supported by twice thenumber of individual genotyped sites as the next most probable consensusgenotype, if there was one. This analysis yielded 7469 consensus genotypestiling the B. distachyon genome assembly, some with high levels of missingdata across the population.

Data Filtering

Genotypic data were compiled for 2664 SNP markers spanning the entireB. distachyon genome. Because RIL populations should be nearly homozygous,heterozygous calls made by markers were scored as missing data. We tested2664 markers and removed markers with more than 15% of missing data(Supplemental Figs. S4, A and B), for a total of 1693 markers used in the anal-ysis. The parental genotypes were both equally represented across all 142 in-dependent lines (Supplemental Figs. S4, C and D); however, 18 lines wereremoved, 17 due to greater than 30% of missing genotypic data and one due togreater than 90% identity to another RIL. The final QTL analysis was completedusing GBS data from 124 independent lines of the Bd21 X Bd1-1 RIL population.

Genetic Map

A high-density genetic map was built using the ML objective function andHaldane distance function of MSTmap (Wu et al., 2008) using the 1693 filteredmarkers described above. Inferred marker order from the genetic data agreedwith the physical map of the B. distachyon genome assembly, supporting theaccuracy of the population genotyping. See Supplemental Table S3 for geno-typic dataset.

QTL Analysis

QTL analysis was performed in R using the R\qtl package (Broman et al.,2003). First, QTLmappingwas computed using simple interval mapping, usingthe Haley and Knott regression method (Haley and Knott, 1992). The empiricalLOD threshold was computed using 10,000 permutations (a = 0.05), whichresulted in a LOD score around 3.2. Graphs were created using the ggplot2package (Wickham, 2009). Peak interaction was computed using a two-dimensional genome scan (scantwo() function; Haley and Knott regression),and the significance of QTL interaction was computed using 1000 permutations(a = 0.05).The contribution of individual peaks to the observed phenotypicvariance was obtained using the fitqtl() function (Haley and Knott regression)and a single-peak model.

Identification of Variants within Candidate Flowering-Time Genes

Variants were identified using a Variant Call Format (VCF) file comparingBd1-1 to the Bd21 reference genome (Gordon et al., 2014). We used VCFtoolsand Vcflib command line tools to retrieve the region of interest, select ho-mozygous variants, and remove low-quality calls (i.e. calls with a GQ valueless than 90) from the VCF file. All gene models (PHYC: Bradi1g08400.v2.1;VRN1: Bradi1g08340.v2.1; VRN2: Bradi3g10010.v2.1 ; FD: Bradi3g00300),which include 2 Kb upstream and 200 bp downstream of the transcribed re-gion, were obtained from the Brachypodium genome Version 2 (Bd2.1_V283;http://phytozome.jgi.doe.gov; Goodstein et al., 2012).

The supplemental data sets, including the raw phenotypic and raw geno-typic data as well as the R script used for the QTL analysis can be found here:https://zenodo.org/record/61660.

Accession Numbers

Raw Illumina sequencing files have been uploaded to the NCBI sequenceread archive SUB1982345.

Supplemental Data

The following supplemental materials are available.

Supplemental Figure S1. Days to heading of individual recombinant in-bred lines.

Supplemental Figure S2. Number of leaves at flowering for individualrecombinant inbred lines.

Supplemental Figure S3. Correlation between the number of days and theleaves at flowering.

Supplemental Figure S4. Filtering steps for the selection of markers andrecombinant inbred lines included in the analysis.

Supplemental Figure S5. Interaction between QTL peaks using a two-QTLmodel.

Plant Physiol. Vol. 173, 2017 277

Flowering-Time QTL in B. distachyon

www.plantphysiol.orgon September 21, 2020 - Published by Downloaded from Copyright © 2017 American Society of Plant Biologists. All rights reserved.

Supplemental Figure S6. Location of the flowering-time candidate genescompared to the QTL experiment.

Supplemental Figure S7. Sequence variant comparison in candidate genesunderlying QTL1, QTL2, and QTL3.

Supplemental Table S1. SNP-based genetic map.

Supplemental Table S2. QTL variance explained.

Supplemental Table S3. Genotypic data.

Supplemental Table S4. Phenotypic data.

Supplemental Material. VRN2 alignment file.

ACKNOWLEDGMENTS

The authors thank the University of Wisconsin Biotechnology Center DNASequencing Facility for providing GBS facilities and services. Mention of tradenames or commercial products in this publication is solely for the purpose ofproviding specific information and does not imply recommendation or en-dorsement by the U.S. Department of Agriculture.

Received July 28, 2016; accepted October 10, 2016; published October 14, 2016.

LITERATURE CITED

Abe M, Kobayashi Y, Yamamoto S, Daimon Y, Yamaguchi A, Ikeda Y,Ichinoki H, Notaguchi M, Goto K, Araki T (2005) FD, a bZIP proteinmediating signals from the floral pathway integrator FT at the shootapex. Science 309: 1052–1056

Amasino R (2010) Seasonal and developmental timing of flowering. Plant J61: 1001–1013

Barrero JM, Jacobsen JV, Talbot MJ, White RG, Swain SM, Garvin DF,Gubler F (2012) Grain dormancy and light quality effects on germi-nation in the model grass Brachypodium distachyon. New Phytol 193:376–386

Bettgenhaeuser J, Corke FM, Opanowicz M, Green P, Hernández-PinzónI, Doonan JH, Moscou MJ (2016) Natural variation in Brachypodiumlinks vernalization and flowering time loci as major flowering deter-minants. Plant Physiol 2016 Sep. 20. pii: pp.00813.2016

Brkljacic J, Grotewold E, Scholl R, Mockler T, Garvin DF, Vain P, BrutnellT, Sibout R, Bevan M, Budak H, Caicedo AL, Gao C, et al (2011)Brachypodium as a model for the grasses: today and the future. PlantPhysiol 157: 3–13

Broman KW, Wu H, Sen S, Churchill GA (2003) R/qtl: QTL mapping inexperimental crosses. Bioinformatics 19: 889–890

Brutnell TP, Bennetzen JL, Vogel JP (2015) Brachypodium distachyon andSetaria viridis: model genetic systems for the grasses. Annu Rev PlantBiol 66: 465–485

Buckler ES, Holland JB, Bradbury PJ, Acharya CB, Brown PJ, Browne C,Ersoz E, Flint-Garcia S, Garcia A, Glaubitz JC, Goodman MM, HarjesC, et al (2009) The genetic architecture of maize flowering time. Science325: 714–718

Cass CL, Lavell AA, Santoro N, Foster CE, Karlen SD, Smith RA, Ralph J,Garvin DF, Sedbrook JC (2016) Cell wall composition and biomass re-calcitrance differences within a genotypically diverse set of Brachypodiumdistachyon inbred lines. Front Plant Sci 7: 708

Chen A, Dubcovsky J (2012) Wheat TILLING mutants show that the ver-nalization gene VRN1 down-regulates the flowering repressor VRN2 inleaves but is not essential for flowering. PLoS Genet 8: e1003134

Chen A, Li C, Hu W, Lau MY, Lin H, Rockwell NC, Martin SS, JernstedtJA, Lagarias JC, Dubcovsky J (2014) Phytochrome C plays a major role inthe acceleration of wheat flowering under long-day photoperiod. ProcNatl Acad Sci USA 111: 10037–10044 10.1073/pnas.1409795111

Chouard P (1960) Vernalization and its relations to dormancy. Annu RevPlant Physiol 11: 191–238

Colton-Gagnon K, Ali-Benali MA, Mayer BF, Dionne R, Bertrand A, DoCarmo S, Charron J-B (2014) Comparative analysis of the cold accli-mation and freezing tolerance capacities of seven diploid Brachypodiumdistachyon accessions. Ann Bot (Lond) 113: 681–693

Corbesier L, Vincent C, Jang S, Fornara F, Fan Q, Searle I, Giakountis A,Farrona S, Gissot L, Turnbull C, Coupland G (2007) FT protein

movement contributes to long-distance signaling in floral induction ofArabidopsis. Science 316: 1030–1033

Cui Y, Lee MY, Huo N, Bragg J, Yan L, Yuan C, Li C, Holditch SJ, Xie J,Luo M-C, Li D, Yu J, et al (2012) Fine mapping of the Bsr1 barley stripemosaic virus resistance gene in the model grass Brachypodium distachyon.PLoS One 7: e38333

Des Marais DL, Razzaque S, Hernandez KM, Garvin DF, Juenger TE(2016) Quantitative trait loci associated with natural diversity in water-use efficiency and response to soil drying in Brachypodium distachyon.Plant Sci 251: 2–11

Distelfeld A, Dubcovsky J (2010) Characterization of the maintainedvegetative phase deletions from diploid wheat and their effect on VRN2and FT transcript levels. Mol Genet Genomics 283: 223–232

Distelfeld A, Li C, Dubcovsky J (2009) Regulation of flowering in tem-perate cereals. Curr Opin Plant Biol 12: 178–184

Dubcovsky J, Chen C, Yan L (2005) Molecular characterization of the allelicvariation at the VRN-H2 vernalization locus in barley. Mol Breed 15:395–407

Elshire RJ, Glaubitz JC, Sun Q, Poland JA, Kawamoto K, Buckler ES,Mitchell SE (2011) A robust, simple genotyping-by-sequencing (GBS)approach for high diversity species. PLoS One 6: e19379

Fjellheim S, Boden S, Trevaskis B (2014) The role of seasonal floweringresponses in adaptation of grasses to temperate climates. Front Plant Sci5: 431

Fu D, Szucs P, Yan L, Helguera M, Skinner J, Hayes P, Dubcovsky J (2005)Large deletions in the first intron of the VRN-1 vernalization gene areassociated with spring growth habit in barley and polyploid wheat. MolGenet Genomics 273: 54–65

Goodstein DM, Shu S, Howson R, Neupane R, Hayes RD, Fazo J, MitrosT, Dirks W, Hellsten U, Putnam N, Rokhsar DS (2012) Phytozome: acomparative platform for green plant genomics. Nucleic Acids Res 40:D1178–D1186

Gordon SP, Priest H, Des Marais DL, Schackwitz W, Figueroa M, MartinJ, Bragg JN, Tyler L, Lee CR, Bryant D, Wang W, Messing J, et al (2014)Genome diversity in Brachypodium distachyon: deep sequencing of highlydiverse inbred lines. Plant J 79: 361–374

Greenup A, Peacock WJ, Dennis ES, Trevaskis B (2009) The molecularbiology of seasonal flowering-responses in Arabidopsis and the cereals.Ann Bot (Lond) 103: 1165–1172

Haley CS, Knott SA (1992) A simple regression method for mappingquantitative trait loci in line crosses using flanking markers. Heredity(Edinb) 69: 315–324

Herten K, Hestand MS, Vermeesch JR, van Houdt JK (2015) GBSX: atoolkit for experimental design and demultiplexing genotyping by se-quencing experiments. BMC Bioinformatics 16: 73

Higgins JA, Bailey PC, Laurie DA (2010) Comparative genomics of flow-ering time pathways using Brachypodium distachyon as a model for thetemperate grasses. PLoS One 5: e10065

Hung H-Y, Shannon LM, Tian F, Bradbury PJ, Chen C, Flint-Garcia SA,McMullen MD, Ware D, Buckler ES, Doebley JF, Holland JB (2012)ZmCCT and the genetic basis of day-length adaptation underlying thepostdomestication spread of maize. Proc Natl Acad Sci USA 109: E1913–E1921

International Brachypodium Initiative (2010) Genome sequencing andanalysis of the model grass Brachypodium distachyon. Nature 463: 763–768

Karsai I, Szűcs P, Mészáros K, Filichkina T, Hayes PM, Skinner JS,Láng L, Bedő Z (2005) The Vrn-H2 locus is a major determinantof flowering time in a facultative x winter growth habit barley(Hordeum vulgare L.) mapping population. Theor Appl Genet 110:1458–1466

Kippes N, Chen A, Zhang X, Lukaszewski AJ, Dubcovsky J (2016) De-velopment and characterization of a spring hexaploid wheat line withno functional VRN2 genes. Theor Appl Genet 129: 1417–1428

Li H, Durbin R (2010) Fast and accurate long-read alignment withBurrows-Wheeler transform. Bioinformatics 26: 589–595

Loukoianov A, Yan L, Blechl A, Sanchez A, Dubcovsky J (2005) Regula-tion of VRN-1 vernalization genes in normal and transgenic polyploidwheat. Plant Physiol 138: 2364–2373

Nishida H, Ishihara D, Ishii M, Kaneko T, Kawahigashi H, Akashi Y, SaishoD, Tanaka K, Handa H, Takeda K, Kato K (2013) PHYTOCHROME C isa key factor controlling long-day flowering in barley. Plant Physiol 163:804–814

278 Plant Physiol. Vol. 173, 2017

Woods et al.

www.plantphysiol.orgon September 21, 2020 - Published by Downloaded from Copyright © 2017 American Society of Plant Biologists. All rights reserved.

Pankin A, Campoli C, Dong X, Kilian B, Sharma R, Himmelbach A, SainiR, Davis SJ, Stein N, Schneeberger K, von Korff M (2014) Mapping-by-sequencing identifies HvPHYTOCHROME C as a candidate gene for theearly maturity 5 locus modulating the circadian clock and photoperiodicflowering in barley. Genetics 198: 383–396

Ream TS, Woods DP, Amasino RM (2012) The molecular basis of ver-nalization in different plant groups. Cold Spring Harb Symp Quant Biol77: 105–115

Ream TS, Woods DP, Schwartz CJ, Sanabria CP, Mahoy JA, Walters EM,Kaeppler HF, Amasino RM (2014) Interaction of photoperiod andvernalization determines flowering time of Brachypodium distachyon.Plant Physiol 164: 694–709

Saïdou A-A, Clotault J, Couderc M, Mariac C, Devos KM, Thuillet A-C,Amoukou IA, Vigouroux Y (2014) Association mapping, patterns oflinkage disequilibrium and selection in the vicinity of the PHYTO-CHROME C gene in pearl millet. Theor Appl Genet 127: 19–32

Salomé PA, Bomblies K, Laitinen RAE, Yant L, Mott R, Weigel D (2011)Genetic architecture of flowering-time variation in Arabidopsis thaliana.Genetics 188: 421–433

Sasani S, Hemming MN, Oliver SN, Greenup A, Tavakkol-Afshari R,Mahfoozi S, Poustini K, Sharifi HR, Dennis ES, Peacock WJ, et al(2009) The influence of vernalization and daylength on expression offlowering-time genes in the shoot apex and leaves of barley (Hordeumvulgare). J Exp Bot 60: 2169–2178

Schwartz CJ, Doyle MR, Manzaneda AJ, Rey PJ, Mitchell-Olds T, AmasinoRM (2010) Natural variation of flowering time and vernalization respon-siveness in Brachypodium distachyon. Bioenerg Res 3: 38–46

Shaw LM, Turner AS, Herry L, Griffiths S, Laurie DA (2013) Mutant allelesof Photoperiod-1 in wheat (Triticum aestivum L.) that confer a late flow-ering phenotype in long days. PLoS One 8: e79459

Shimada S, Ogawa T, Kitagawa S, Suzuki T, Ikari C, Shitsukawa N, Abe T,Kawahigashi H, Kikuchi R, Handa H, Murai K (2009) A genetic network offlowering-time genes in wheat leaves, in which an APETALA1/FRUITFULL-like gene, VRN1, is upstream of FLOWERING LOCUS T. Plant J 58: 668–681

Song YH, Shim JS, Kinmonth-Schultz HA, Imaizumi T (2015) Photope-riodic flowering: time measurement mechanisms in leaves. Annu RevPlant Biol 66: 441–464

Trevaskis B, Bagnall DJ, Ellis MH, Peacock WJ, Dennis ES (2003) MADSbox genes control vernalization-induced flowering in cereals. Proc NatlAcad Sci USA 100: 13099–13104

Turner A, Beales J, Faure S, Dunford RP, Laurie DA (2005) The pseudo-response regulator Ppd-H1 provides adaptation to photoperiod in barley.Science 310: 1031–1034

Tyler L, Fangel JU, Fagerström AD, Steinwand MA, Raab TK, WillatsWG, Vogel JP (2014) Selection and phenotypic characterization of a corecollection of Brachypodium distachyon inbred lines. BMC Plant Biol 14: 25

Tyler L, Lee SJ, Young ND, DeIulio GA, Benavente E, Reagon M, SysophaJ, Baldini RM, Troìa A, Hazen SP, et al (2016) Population structure inthe model grass Brachypodium distachyon is highly correlated withflowering differences across broad geographic areas. Plant Genome J 9:1–20

Vogel JP, Garvin DF, Leong OM, Hayden DM (2006) Agrobacterium-mediated transformation and inbred line development in the modelgrass Brachypodium distachyon. Plant Cell Tissue Organ Cult 84: 199–211

Vogel JP, Tuna M, Budak H, Huo N, Gu YQ, Steinwand MA (2009) De-velopment of SSR markers and analysis of diversity in Turkish popu-lations of Brachypodium distachyon. BMC Plant Biol 9: 88

von Zitzewitz J, Szűcs P, Dubcovsky J, Yan L, Francia E, Pecchioni N,Casas A, Chen THH, Hayes PM, Skinner JS (2005) Molecular andstructural characterization of barley vernalization genes. Plant Mol Biol59: 449–467

Weigel D, Nordborg M (2015) Population genomics for understandingadaptation in wild plant species. Annu Rev Genet 49: 315–338

Wickham H (2009) ggplot2: Elegant Graphics for Data Analysis. Springer,New York

Wilson P, Streich J, Borevitz J (2015) Genomic diversity and climate ad-aptation in Brachypodium. In PJ Vogel, ed, Genetics and Genomics ofBrachypodium. Springer International Publishing, Basel, Switzerland, pp107–127

Woods DP, Amasino RM (2015) Dissecting the control of flowering time ingrasses using Brachypodium distachyon. In PJ Vogel, ed, Genetics andGenomics of Brachypodium. Springer International Publishing, Basel,Switzerland, pp 259–273

Woods DP, McKeown MA, Dong Y, Preston JC, Amasino RM (2016)Evolution of VRN2/Ghd7-like genes in vernalization-mediated repres-sion of grass flowering. Plant Physiol 170: 2124–2135

Woods DP, Ream TS, Amasino RM (2014a) Memory of the vernalized statein plants including the model grass Brachypodium distachyon. Front PlantSci 5: 99

Woods DP, Ream TS, Minevich G, Hobert O, Amasino RM (2014b)PHYTOCHROME C is an essential light receptor for photoperiodicflowering in the temperate grass, Brachypodium distachyon. Genetics 198:397–408

Wu Y, Bhat PR, Close TJ, Lonardi S (2008) Efficient and accurate con-struction of genetic linkage maps from the minimum spanning tree of agraph. PLoS Genet 4: e1000212

Yan L, Fu D, Li C, Blechl A, Tranquilli G, Bonafede M, Sanchez A, ValarikM, Yasuda S, Dubcovsky J (2006) The wheat and barley vernalizationgene VRN3 is an orthologue of FT. Proc Natl Acad Sci USA 103: 19581–19586

Yan L, Loukoianov A, Blechl A, Tranquilli G, Ramakrishna W, SanMiguelP, Bennetzen JL, Echenique V, Dubcovsky J (2004) The wheat VRN2gene is a flowering repressor down-regulated by vernalization. Science303: 1640–1644

Yan L, Loukoianov A, Tranquilli G, Helguera M, Fahima T, Dubcovsky J(2003) Positional cloning of the wheat vernalization gene VRN1. ProcNatl Acad Sci USA 100: 6263–6268

Zadoks JC, Chang TT, Konzak CF (1974) Decimal code for growth stagesof cereals. Weed Res 14: 415–421

Zeevaart JA (2008) Leaf-produced floral signals. Curr Opin Plant Biol 11:541–547

Plant Physiol. Vol. 173, 2017 279

Flowering-Time QTL in B. distachyon

www.plantphysiol.orgon September 21, 2020 - Published by Downloaded from Copyright © 2017 American Society of Plant Biologists. All rights reserved.